Stand-alone Panelsand

Terminal ControllersUp- to-date information on functional and friendlystand-alone panels and the need to improve air-

terminal con trollers so that both can beeffectively integrated in to single sys terns

W hen the book on the 20thcentury is closed a scantdecade from now, a strong

case will be made that the mostprofound technological develop-ment of the century has been thedigital computer. Although one sel-dom sees a substantial impact ofthe technology in the controls ofbuildings today, the situation israpidly changing. Development ofdigital controls for building sys-tems lagged behind other indus-tries, but the last several years haveseen enormous growth in the powerand economy of direct digital con-trol (DDC) systems.

Today, DDC systems are fast be-

By THOMAS HARTMAN, PE,The Hartman Co.,Seattle, Wash.

coming the leading innovative forcein building design. So many DDCimprovements have been releasedrecently that some have suggestedmanufacturers should pause to letthe rest of the industry digest allthat is new before embarking onfur ther deve lopments . Such apause in development at this timeis most inappropriate because inreality the building controls indus-try, like Robert Frost’s traveler,still has many promises to keep andmiles to go.

In this article, we will review thereasons DDC controls, as they aregenerally available today, have be-come so successful. We will alsolook at why some of the promises ofextending DDC to terminal controlhave yet to be realized. Finally, wewill review suggestions that have

been offered to ensure that thepromise of extending DDC controlto terminal HVAC units and light-ing systems is realized.

DDC systems today

Users, and some suppliers, arestill working with DDC systemsthat were designed a decade ormore ago. Many of these peoplehave difficulty understanding thechanges that have taken place inDDC systems over the last fewyears. From a hardware perspec-tive, systems are much simpler.They are easier to install, start up,calibrate, operate, and trouble-shoot. Today, operators with basicskill levels and training can main-tain modern DDC system applica-tions in-house. Or users can signextended warranty contracts that,

Heating/Piping/Air Conditioning , November 1990 41

DDC controls

when procured competitively, costas little as $4 to $8 per DDC systempoint per year (it used to be that$50 to $100 per DDC point per yearwas considered by many to be areasonable range for DDC mainte-nance costs). The costs of systemsthemselves have fallen dramatical-ly. A decade ago, designers oftenused $1000 per point to estimatethe installed cost of DDC systems.In some recent competitively pro-cured retrofi t applications, in-stalled DDC system costs havebeen quoted below $200 per point.

Meanwhile, the performance ofDDC systems continues to im-prove. Programming features makeit easier than ever for an operatorto create, document, or adjust con-trol programs, view the value andstatus of system points, and addnew points or calibrate existingones.

In the last few years, the real suc-cess story in the HVAC control in-dustry is the development and re-f inement o f DDC s tand-a lonepanels. They have become the plat-form for almost all recent DDC im-provements. By understanding theperformance attributes that havemade stand-alone DDC panels suc-cessful, we can better appreciatewhere the industry is now and whatthe focus should be for further im-provements.

Panel hardwareStand-alone panels make up the

nucleus of a modern DDC system.How the stand-alone panel per-forms is the single most importantfactor determining the overall per-formance of a DDC system. A basicstand-alone panel typically has thecapacity to connect directly toabout 50 total input and outputpoints, and most panels also havethe capacity for expander boards orslave I/O multiplex boards that canincrease the point capacity almostindefinitely. The limiting factor inexpansion is usually the amount ofpanel memory available to supportapplications programs, trend logs,displays, etc., for the points con-nected. When procured compet-

42

itively, stand-alone panels typi-cally cost the end user about $2500to $5000 (uninstalled) for a 50-point configuration.

There are several approaches toconfiguring stand-alone panels inthe industry today. One approachadopted by manufacturers employsa fully configured stand-alonepanel on a single printed circuitboard. Fig. 1 is an example of a sin-gle board panel. The idea is that aDDC system composed of fewerparts is easier and more economicalfor the manufacturer to make, thedistributor to stock, the contractorto install, and the user to maintain.Indeed, our firm’s experience isthat single board panel config-urations often provide the lowestfirst cost.

To make certain the panel servesa variety of applications, most sin-gle board panels permit DDC con-nection points to be configuredwith great flexibility. Each pointcan typically be analog or digital,but output points and input pointsare generally not interchangeable.

The downside to the single boardapproach is that any componentfailure requires complete board re-placement. Although plug-on ter-minations and fast reload pro-cedures make panel replacementnearly as fast as replacing a singlecomponent, it can be expensive. So

1 Example of a DDC stand-alone panel on asingle board.

2 Example of a modular DDC stand-alonepanel.

long as failures are few and manu-facturers or third-party firms pro-vide economical board repair, thesingle board approach will be suc-cessful.

A second approach is to build apanel from modules. Separatemodules are inserted into a motherboard to configure each input oroutput, and only those modulesthat are required for the actual I/Opoints in each panel are installed.Power supplies, communicationcontrollers, and other functions arealso modularized by some manu-facturers. An example of a modularpanel is shown in Fig. 2. Some ad-vantages exist for the modularpanel because it can be configuredto meet the exact I/O combinationrequired. However, because thesum of the individual modulestends to be more costly than thesingle board panel, the real advan-tage of the modular approach is theenhanced ability to isolate and re-pair failures quickly and inexpen-sively. How much of an advantagethis can become depends on thecosts of troubleshooting and re-placing a single module comparedto replacing the whole panel.

Heating/Piping/Air Conditioning l November 1990

A third approach to stand-alonepanel hardware is to make a com-promise between the panel on aboard and the modular panel.Manufacturers adopting this ap-proach wish to use advantagesfrom both the other approaches.The semi -modula r pane l s , a sshown in Fig. 3, have both advan-tages and disadvantages comparedto the others when various featuresare considered, but the differencesamong all the approaches are cer-tainly not significant. In fact, ourfirm has never found any compel-ling reason to specify one of thesetypes of panels over another.

The most effective means of de-termining the right panel for agiven application is to require along-term extended warranty pro-posal along with the DDC systemprice. The system cost added to thepresent value of the extended war-ranty provides the best picture ofthe true system cost over time.Whether or not the user intends tosign an extended warranty agree-ment, a warranty proposal from thevendor in a competitive environ-ment is probably the best estimateof the true cost of maintaining thesystem. And in the final analysis,cost is the primary reason for se-lecting one hardware configurationover another.

Panel functionStand-alone panels typically

contain all the capabilities neededto provide DDC control, includingcoord ina t ing communica t ionsamong panels, executing controlprograms, PID loop control, stor-ing trend log values and status, etc.Operator interface devices can allbe disconnected without affectingautomatic control in true stand-alone panels because all real-timefunctions are contained at thepanel level.

Important progress has beenmade in the last few years withstand-alone panel technologies.T h o u g h m a n u f a c t u r e r s h a v eadopted a variety of philosophiesregarding operating features, anumber of features have been uni-

versally recognized and incorpo-rated in stand-alone panel designs.Recent improvements have beenadvanced in:l Reliability- A few years ago,

stand-alone panels had a reputa-tion for being sensitive to power-l ine p rob lems . E leva to r s andemergency generators were but twoof a myriad of reasons (a betterword might be excuses) given forunexplained DDC system crashesand other failures. In recent years,the reliability of stand-alone panelshas improved remarkably. Today,it is not unusual to find large sys-tems operating several years with-out a failure serious enough to re-quire even reloading a single panel.l Panel memory- M e m o r y

poses the same problems to DDCsystem manufacturers as it does tocomputer manufacturers in other

to utilize valuable software featuresfully.l Programming languages-

Not so long ago, it was widely heldby vendors that canned programswith simple input parameters wereall a building operator needed (ormore to the point, all the buildingoperator could handle) to imple-ment control strategies in build-ings. Users’ demands for morefunctional systems have at lastbeen heard. Though there areseveral different viewpoints aboutthe form applications programsshould take to be most functional,the power and versatility of controllanguages in most products havebeen significantly upgraded in thelast few years.l Panel-to-panel communica-

tions- Although users have longsupported the distributed control

3 Example of a semi-modular DDC stand-alone panel.

industries-they have a tough timetelling how much is enough. InDDC systems, point databases, ap-plications programs, and trend logsall compete for available memory.Early stand-alone panels were no-torious for their limited memory.Users were frustrated when neededfeatures were severely limited bythe amount of memory available.Newer panels supply much im-proved memory or memory up-grade capabilities that permit users

concept of stand-alone panels, theyalso demand that a DDC systemperform as a single integrated en-tity and not as a series of small sep-arate systems. When a program iswritten in one panel that requirespoint information from another,the operator shouldn’t be requiredto set up the network required forthe information to be shared be-tween the panels. A fast, fully au-tomatic communications networkis an important ingredient for suc-

Heating/Piping/Air Conditioning. November 1990 43

DDC controls

cessful stand-alone panel config-urations. Recent product releasesand updates confirm that the in-dustry understands the need toprovide such networks.l Graphic displays and trend

logs- Graphic displays of real-timepoint data and historically trendeddata vastly improve an operator’sability to utilize the ever-growingsupply of information availablefrom a DDC system. Only a fewyears ago, many systems had verylimited graphics features or graph-ics that were so slow they were im-practical. Now, manufacturers ap-pear to have focused on graphicfeatures that really work and areeasy to set up as well.

Needed improvementsWith all the improvements of the

last few years and the continuingtrend to lower DDC system prices,one might wonder what more isneeded. But for many who have re-cently started up full DDC systems(a full DDC system is one that con-

trols the entire system, including have been designed with the idea ofthe terminal units that directly controlling small actuators and de-supply occupied areas), a whole vices, some come with only digitalnew series of problems is threat- outputs, in which case two physicalening to renew the old complaints outputs are employed to controlthat systems are complicated, have each analog output device. Termi-slow response, and lack flexibility. nal controllers are almost alwaysTo see what the problem is and single board products. An examplewhat can be done about it, let’s look of a terminal controller board isat the relatively new DDC products shown in Fig. 4, Figs. 5 and 6 showthat have made full DDC systems typical terminal controllers withpossible. enclosures.

Terminal controllersThe relatively recent develop-

ment of terminal controllers offersnew opportunities to HVAC de-signers and system operators. Ter-minal controllers typically offerlimited point capacity (8 inputsand 8 outputs or less) and usuallyhave no expansion capability. Un-installed, a terminal controllerpanel sells for $150 or more, de-pending on manufacturer and con-figuration.

Because terminal controllers

Typically, terminal controllersare configured on a separate trunkthat is supported by one of thestand-alone panels. The communi-cation of this trunk is generallyslower than the stand-alone paneltrunk and doesn’t employ peer-to-peer communications. Dependingon the manufacturer, 30 to severalhundred terminal controllers canbe, in theory at least, attached to asingle stand-alone panel. Fig. 7shows how terminal controllers areusually configured in DDC systemstoday.

T e r m i n a l c o n t r o l l e r s h a v eseveral problems that can causeconsiderable headaches for DDCsystem operators who have grownaccustomed to the power, flex-ibility, and simplicity of stand-alone panels now available. Theproblems terminal controllers cancause are:

4 Example of a DDC terminal controller board. controller with enclosure.

l Lack of consistency-Fig. 7shows that in a full DDC system,input and output points may beconnected to either a stand-alonepanel or a terminal controller.However, many DDC manufactur-ers see the terminal controllers asbeing application specific and puttight constraints on the type ofpoints that can be connected to ter-minal controllers and how they areaccessed. For the configuration inFig. 7, assume temperature sensorsare attached to several stand-alonepanels and to the terminal control-lers. Because of inherent differ-ences between the stand-alonepanel and the terminal controller,the operator may have to employtwo very different techniques toconnect the sensors, define theirpoint database, calibrate them,

44 Heating/Piping/Air Conditioning l November 1990

manually override them, program fices is supplied by a single variablethem, or display their values, de- air volume (VAV) box. Traditionalpending on how they are connected controls would place a single spaceto the system. thermostat in one office to operate

Some full DDC configurations the box damper and reheat coilmore closely resemble two distinct with the hope that all offices wouldDDC systems: one consisting of have consistent heating and cooling

Printer Host computer Modem 1 DDC svstem

points lnput/output points Input/output points p o

I component level4 Operator

interface

4 Stand-alonepanel

4 Terminal-controller

4 Device

7 Typical DDC system architecture.

stand-alone panels and anotherconsisting of terminal controllers.This makes the operator’s task ofremembering the different oper-ating rules and procedures and howor where they apply most difficult.Manufacturers who have devel-oped techniques that make theirstand-alone panels simple andstraightforward to operate appearto have underestimated the needfor consistency of operation tomeet these goals in DDC systemsconsisting of stand-alone panelsand terminal controllers together.l Lack of function- The prob-

lem of consistency noted abovewould not be so serious if one onlywished to replace the simple con-trol functions of present terminalcontrol devices. But replacementwith no added function is not acompelling reason to upgrade toDDC terminal controls. The reasonmost designers and users desire tomove up to full DDC systems is toprovide more functional control atthe terminal level to solve problemsinherent with mechanical systems.

Fig. 8 illustrates a common con-trol problem. A group of small of-

loads and thus would require thesame HVAC response to remaincomfortable. Building engineersknow all too well that this type ofcontrol is unsatisfactory. Sooner orlater an occupant in one of therooms without the thermostat willcompla in about comfor t . Thebuilding engineer has no satis-factory means to solve such prob-lems. Can the thermostat be movedwithout offending that occupant?Will a rebalance of the outlets tosend more or less flow to the of-fending space be anything morethan a short term fix? These aresituations designers have not hadthe tools to resolve, and they resultin operating problems that are al-most never entirely solved.

With a functional full DDC con-trol system, the opportunities toprovide a comfortable, high qualityenvironment and at the same timeimprove the energy efficiency of thesystem are much improved. DDCsystem temperature sensors arevery low cost items, making it fea-sible to install a temperature sensorin every office. To control lightingand the HVAC terminal unit, each

Heating/Piping/Air Conditioning . November 1990

office might also employ a push-button or occupancy sensor to no-tify the DDC system that the spaceis occupied.

If a pushbutton is the occupancydevice, occupancy may be assumedto be continuous for the day whenactivated during normal businesshours or for an agreed fixed time ifactivated outside business hours.An occupancy sensor offers furthercontrol by providing a signal onlywhen someone is in the room. Ashort delay in switching to unoccu-pied condition after the room is va-cated permits stable operation.

The occupancy signal controlslighting (for each office) and theVAV terminal box that serves allthree offices. If only one office is oc-cupied, then the box operates tosatisfy that office. When multipleoffices are occupied, the box oper-ates to satisfy the average of the oc-cupied offices unless an office ex-ceeds high or low space temper-ature limits; in that instance, thebox acts to bring the office tempera-ture within limits, then returns toaveraging control.

The box need have no set mini-mum air flow. If all rooms are unoc-cupied, the box may shut off en-tirely so long as all offices arewithin suitable unoccupied tem-perature limits. When one or morespace is occupied, the minimum airflow is set by the DDC system toprovide the required outside air tothe zone-based on the percentageof outside air being supplied to thebox at the time.

Such an operating sequence, to-gether with additional featuressuch as warm-up and night purgecycles, can lead to comfort and airquality levels for the occupantsthat are vastly superior to thoseprovided by traditional terminalcontrols. The marginal DDC sys-tem cost to implement better ter-minal control is increasingly com-p e n s a t e d f o r b y e n e r g y c o s tsavings. Further energy savings arepossible by employing terminalregulated air volume (TRAV) con-trol strategies for the central fan(see HPAC, July 1989). The prob-

45

DDC controls

lem facing designers is that veryfew of the terminal control prod-ucts on the market today have suf-ficient functions to accomplish ef-fectively the sequence describedabove. Of those that can, most re-quire extraordinary databases orprogram manipulations to do so,making such sequences difficult toimplement and difficult for an op-

not practically be accomplished atall in many terminal controllers.

ROM-based backup programs interminal controllers are a goodidea. ROM programs allow termi-nal units to be started and testedbefore the DDC system has beeninstalled. ROM programs ensurethat compressors or fans in certaintypes of terminal units will not be

4

@ Occupancy sensor

0 Temperature sensorA

VAV box

8 HVAC terminal unit serving multiple offices. Such situations represent potential problemsfor traditional control svstems but can be handled effectively and efficiently with full DDCcontrol.

erator to support,l Lack of flexibility-When

manufacturers began designingterminal controllers, it appearedthe primary goal was to replace ex-isting pneumatic controllers with aDDC product whose only advan-tage was remote monitoring. Thisunambitious goal has resulted inthe industry’s being deluged withterminal controllers whose applica-tion software is preprogrammedwith simple control sequences inread-only memory (ROM). Manyof the products available today notonly lack the functional capacity tosolve the design problem outlinedabove but also lack the flexibilityfor the operator to make even sim-ple changes or adjustments in theirsequence of operation once theyhave been installed. Any changethe operator wishes to make that isnot a preprogrammed option can-

damaged by short cycling if theunit is operated without a program.However, ROM-based programsshould be limited to providingbackup for startup and defaultconditions or executing very simplecontrol strategies. A flexible appli-cation program language that fol-lows the same basic rules as theprogramming language in stand-alone panels is needed to provideeffective terminal controller pro-grams.

A flexible programming languageshould be available to write pro-grams in terminal controllers in thesame form as those in stand-alonepane l s . However , ROM-basedbackup programs should be pre-served for startup and fail-safe con-trol.

Some manufacturers have builtin the ability to “unbundle” termi-nal controller points and operatethem as if they were points connec-ted to the stand-alone panel towhich the terminal controller isconnected. There are two problemswith the unbundling approach thatmake it unworkable for typical ter-minal control applications. In mostsystems, unbundling is a cum-

Finally, it would be worthwhilefor manufacturers to consider add-ing an area of memory in stand-alone panels to store the databasesof all connected terminal control-lers. This would simplify DDC sys-tem backup and reload procedures,Loading a stand-alone panel wouldautomatically load all the terminalcontrollers attached to i t . Thestand-alone panel could monitorthe state of all attached terminalcontrollers, reload any that losetheir data, and alert the operator toany that are having difficultymaintaining their databases.

Manufacturers sometimes buildin limitations to terminal devicesbecause the margin of profit is less

46 Heating/Piping/Air Conditioning l November 1990

bersome process because eachpoint that is to be unbundled mustbe redefined in the stand-alonepanel. Still, in some instances theunbundled points cannot be oper-ated the same as panel points. Also,unbundled t e rmina l con t ro l l e rpoints add to the point count of thestand-alone panel, which can se-verely restrict the number of termi-nal controllers a single stand-alonepanel can serve.

Controller developmentThere are some lessons learned

during development efforts ofstand-alone panels that may helpsolve many problems the industryis now experiencing with terminalcontrollers.

First, manufacturers should beencouraged to further develop theirterminal control products so that,to the greatest extent possible, allpoints in the DDC system, whetherconnected to stand-alone panels orto terminal controllers, can be de-fined, calibrated, programmed, andoverridden, with the same parame-ters and by the same simple pro-cesses.

on these products and they do notwant them to compete with theirmore expensive stand-alone panels.This is a short-sighted view be-cause the potential market for fullDDC systems will expand enor-mously as the two products are bet-ter integrated into efficiently con-figured DDC systems.

Opportunities aheadAs more functional and versatile

terminal controllers are developed,the opportunities for improving thecomfort and economy of HVACsystems will open to entirely newareas. The recent ASHRAE stan-dard for ventilation has been crit-icized because it does little to en-sure that outside air is actuallydelivered to each of the building’soccupants. Using methods similarto the terminal control exampleearlier, full DDC systems can en-sure on a zone-by-zone basis that

adequate outside air is supplied toevery occupant and at the sametime improve the comfort and en-ergy efficiency of each building.

Meanwhile, the development ofDDC and associated technologies iscontinuing to reduce the cost of ap-plying terminal control products.Electric “quasi-modulating” tech-niques provide good modulationcharacteristics at low cost for ter-minal reheat and other terminalmodulation requirements. Thecosts of standard temperature, hu-midity, air flow, and occupancysensing devices have all decreasedsubstantially in the last few years.

The prospects for expandingbuilding DDC systems to terminalcontrol is indeed very promising.But to be certain the promises re-garding opportunities for full DDCare kept, users must encourageDDC system manufacturers to im-prove the function and flexibility of

their terminal control products andto develop operational featuresmore consistent with the stand-alone panels to which they are con-nected. Once such powerful prod-ucts are available, designers can usethe added power to improve build-ing comfort, air quality, and costefficiency up to the levels ownersand occupants now demand.

Next month we will explore op-tions in programming languages forDDC controls today. We will dis-cuss the advantages and disadvan-tages of each basic approach to pro-gramming DDC panels and suggestwhere the future may lead systemmanufacturers in this critical DDCsystem feature. a

The author appreciates DDC manufactur-ers’ cooperation in providing photographsto help the descriptions in this article. Thephotographs have been chosen to be repre-sentative of the industry and are not en-dorsements of any kind.

analysis of your requirements, call us toll-free.

we’ll tell you what you could be saving with our two complete lines Aof heaters. Lines that are not only versatile but require littlemaintenance.

So give us a call. Let us prove we’re not just a lot of hot sic

lnlra-red heat warms people and oblects directly-not wasted spaces

Detroit Radiant Products Company21400 Hoover Road, Warren MI 48089

In Michigan, call collect: (313) 7%09%

Gas. America’s best energy va1ue.A0 1990 American Gas Association

Circle 339 on Reader Service CardHeating/Piping/Air Conditioning. November 1990 47

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .DDC CONTROLS TODAY

PROGRAMMING LANGUAGESFOR DDC SYSTEMS

An examination of the current and future state ofprogramming languages and the various optionsavailable for programming DDC panels

By THOMAS HARTMAN, PE,The Hartman Co.,Seattle, Wash.

QAStar Trek movie findsCapt. Kirk, through nofault of his own, mys-te r ious lv t r anspor ted

back to the primitive present. Tocorrect the situation and return tohis proper place in time, he solicitsthe use of a computer. When showna powerful computer, one of Kirk’slieutenants immediately tries toinitiate a dialog by commanding“Computer!” After a moment of un-comfortable silence, the operatorpoints out the mouse, which Kirk’slieutenant raises to his mouth andagain tries to command the com-puter verbally.

I was reminded of that scene re-cently at a presentation of a newDDC (direct digital control) prod-uct. The vendor was nervously pre-senting what he thought were ad-vanced programming features to agroup of experienced DDC users.The questions and criticisms flewfast and furiously. Finally the exas-perated rep asked, “Well, how doyou want to program DDC sys-tems?” One of the participants re-sponded immediately, “We want totell the system what we want it todo, and we want it to understand

and do it.”I never had a chance to ask that

fellow if he got this idea from themovie, but I suspect it wouldn’tmake any difference. Virtually ev-eryone who uses computers be-lieves they could be vastly morefriendly to use than they are. DDCoperators are no exception. Untilsystems operate as the Star Trekcrew expected, the pressure for im-provements will continue to be verystrong indeed.

In the last few years, the DDC in-dustry has learned a great dealabout performance and the ele-ments of successful programminglanguages. A number of ideas havebeen tried and many have beenvery successful. In this article wewill examine the current state ofprogramming languages for DDCsystems, what the various optionsare for programming DDC today,and the benefits and drawbacks ofeach. Finally, we will look to the fu-ture and suggest a path for con-tinued improvement of DDC sys-tem programming languages.

The beginningIn the early days of computer-

based building control, most pro-gramming languages offered veryfew features and even less flex-ibility. As a result, the notion devel-oped that the system operator

should be more a specialist in com-puters than in HVAC systems.Many systems were supplied withprograms written at the factory.These programs were provided in alow-level assembly type languagethat allowed the operator only afew of what we call hooks into thesystem. For example, typical pro-grams permitted the operator todefine occupancy schedules andadjust certain set points, but thesequence of control was fixed andcould only be changed by re-compiling the program, which usu-ally had to be done off site. Prob-lems that could not be adjustedaway with the built-in hooks re-quired elaborate schemes to cor-rect. Operators who were veryknowledgeable in the operation ofthe computer could sometimes ad-just certain database parameters,fooling the system into performingmore satisfactorily.

However, users became veryfrustrated by the inflexibility ofthese systems. Building controlproblems that seemed quite simpleand straightforward often requiredelaborate measures and a computerspecialist to solve. A flurry of activ-ity took place by manufacturers,users, and the building design com-munity to improve the success ofcomputer-based building controlsystems. Some of the initial solu-

Programming languages

tions proposed (such as tri-servicesspecification) failed because theytried to treat the symptoms andnot the cause. But when the dustsettled early in the 8Os, two new ap-proaches to building control pro-gramming were being offered thatallowed operators who were notskilled in computers to supportbuilding automation systems moreeffectively than ever before.

Line programmingOne of the approaches to im-

proved programming capabilitiesoffered by some manufacturers wasthe ability to write sequences ofoperation in standard line-programformats. Line programming hadbeen employed for many years inthe general computing industry.The formats of these DDC lan-guages look very much like thehigh-level general computing lan-guages (such as BASIC) exceptthat certain additional functionsare added to permit the language toissue start and stop commands,control outputs to PID (propor-tional/integral/derivative) algo-rithms, tie into occupancy sched-ules, etc. Some languages are com-piled and some interpretive; but alloffer similar flexible control capa-bilities, and they can be easily de-veloped and altered on site by thesystem operator. A sample line pro-gram in the form our firm typicallyemploys to calculate the supply airset point for simple fan systems isshown in Fig. 1.

Some line-programming lan-guages contain all the features ofpowerful high-level languages andinclude formatting features thatmake programs written in theselanguages quite readable. Somenewer releases also include addi-tional features, such as full-screeneditors and on-line error checking,permitting operators to view, edit,change, and debug virtually anycontrol sequence quickly and eas-ily.

It is important to note, however,that there are wide variationsamong l ine -programming lan-guages. Some are crude, inflexible,

and very difficult to use, thoughtheir suppliers still claim they pro-vide high-level line programming.As with other features of DDC sys-tems, equals in languages do notexist. Users and consultants shouldmake themselves knowledgeableabout the features of any program-ming language to be supplied witha system before it is purchased.

The primary advantage to lineprogramming rests in its power andflexibility. Line-program functionshave already proved themselves insolving diverse general computingproblems. With the addition of afew special functions for buildingcontrol, system designers andbuilding operators find they havethe tools they need to develop justabout any control sequence(s) for

particular HVAC system controlrequirements.

Another advantage of some lineprograms is that they are self-docu-menting. When a designer or oper-ator determines that a programchange is necessary, a printout ofthe program provides an accurateand fairly readable description ofthe new control sequence. And be-cause line programs are in the formof general computing languages,many operators have already hadsome experience with this type oflanguage at school or home.

The primary disadvantage citedfor line programming is that someoperators have trouble under-standing and writing line programswithout specific training. Indeed,some of the line languages have

ro!rr/lllrNlorl ,‘OK’ 7.1

“CALCULATE MAXIMUM, MINIMUM AND AVERAGE SPACE TEMPS"

STMAX = MAX(ST1 ,ST2,ST3,ST4,ST5)STMIN = MIN(ST1 ,ST2,ST3,ST4,ST5)STAVE = AVE(ST1,ST2,ST3,ST4,ST5)

“CALCULATE SUPPLY AIR SETPOINT BASED ON AVERAGE SPACE TEMP”

SASP = 65 - 3*(STAVE-STOBJ)

“ADJUST SUPPLY AIR SETPOINT FOR PROJECTED HIGH OUTDOOR TEMP”

SASP = SASP - (PHT-50)/5

“ADJUST SUPPLY AIR SETPOINT FOR COLD DAY MODE OPERATION”

IF CDM = ON THEN SASP = SASP + 2

“ADJUST SUPPLY AIR SETPOINT FOR HIGH OR LOW SPACE TEMPS

IF STMAX > 74 THEN SASP = SASP - (STMAX-74)*3IF STMIN < 70 THEN SASP = SASP + (70-STMIN)*3

LEGEND:

SASPSTMAXSTMIN

STAVE

STOBJST1 -ST5PHT

SUPPLY AIR TEMPERATURE SETPOINTMAXIMUM ZONE SPACE TEMPERATUREMINIMUM ZONE SPACE TEMPERATURE

AVERAGE ZONE SPACE TEMPERATURE

SPACE TEMPERATURE OBJECTIVE (CALCULATED ELSEWHERE IN PROGRAM)SPACE TEMPERATURES IN AREAS SUPPLIED BY AIR SYSTEM

DAY’S PROJECTED HIGH OUTSIDE AIR TEMP (CALCULATED ELSEWHERE IN THEPROGRAM)

CDM COLD DAY MODE (LOGICALLY DETERMINED ELSEWHERE IN THE PROGRAM)

NOTE: “.....” ARE COMMENTS THAT ARE FOR THE BENEFIT OF THE OPERATOR, AND ARE IG-NORED BY THE PROGRAM

1 Line program for calculating supply air temperature set point.

Programming languages

long lists of rules regarding theiruse, and operators are often frus-trated when they cannot easilymake occasional program changes.However, line-based languageswith fewer rules, which permit theuse of comments and special for-matting, are usually supported suc-cessfully by building operators.

Another disadvantage of lineprograms is that software develop-ment for typical projects can be-come time consuming by requiringentire programs to be rewritten formultiple systems even though theyall may operate very much thesame. Fortunately, certain copyingfeatures and editing aids mitigatethis disadvantage in the more ad-vanced line program languages.

Function-block programmingA second approach taken to im-

prove success with applicationssoftware involves refining the pre-programmed approach to give itsome additional flexibility in aform that is structured specificallyfor typical HVAC applications.The manufacturers that adoptedthis approach decided to breakdown the factory programmed ap-plications into small programblocks that can be linked togetherand have parameters assigned bythe designer or operator. By assem-bling these preprogrammed blocksin various combinations and pro-viding some flexibility in assigningpoints and parameters, manufac-turers believe they can satisfy mosttypical DDC applications whilemaintaining a simple and easy-to-use program format.

Fig. 2 is a sample function-blockprogram. This function block pro-vides space temperature reset ofthe supply air temperature setpoint of a simple fan system. Somein the industry call this “fill in theblanks” programming because onlya limited number of fields in eachprogram block need to be entered.

The primary advantage for func-tion-block programming is its sim-plicity in standard HVAC applica-t ions . Indeed , i f the cont ro lsequence happens to call for the ex-

act functions provided by the func- point of a simple fan system astion blocks included with the sys- space conditions change. However,tern, the programming effort is very the line program in Fig. 1 also in-easy. The disadvantage of the func- cludes outdoor weather factors andtion-block approach is that when- could easily be changed to accom-ever sequences are required that do modate different relationships ornot match ava i lab le func t ion additional factors-all in this one

TEMPERATURE LOOP RESET FUNCTION BLOCK

NAME OF FUNCTION BLOCKCONTROL LOOP TO BE RESETDEFAULT TEMPERATURE

RESET POINTS AND LIMITS

POINT LOW LIMIT HIGH LIMIT

RATE OF RESET LOW TEMP: <-ii----) HIGH TEMP: c-?---->

LEGEND:

SASP SUPPLY AIR TEMPERATURE SETPOINT (NAME OF THIS FUNCTION BLOCK)AH1 SATLP ANALOG CONTROL LOOP TO BE RESET (A SEPARATE FUNCTION BLOCK)ST1 -ST5 SPACE TEMPERATURES IN AREAS SUPPUED BY AIR SYSTEMc-) AREAS OF FUNCTION BLOCK THAT CAN BE CHANGED BY THE OPERATOR

2 Function-block program for supply air reset.

blocks, the programmer must em-ploy custom blocks or employblocks for purposes other thanthose for which they were devel-oped. This usually results in morecomplicated programs and reducedHVAC system performance.

As a result of the relative advan-tages and disadvantages of the twoprogramming approaches, line pro-gramming-based systems are usu-ally far more effective in high-per-formance building applicationsthat require more in-depth controlstrategies. Systems with function-block programming are generallylimited to applications employingsimple or traditional pneumaticstrategies.

To see the differences in the twoapproaches, consider the programsin Figs. 1 and 2. The line programin Fig. 1 and the reset-functionblock in Fig. 2 are both intended toreset the supply air temperature set

program. By contrast, the functionblock cannot implement a numberof the factors employed in Fig. 1.Function-block programs usuallydo permit linking blocks togetherfor additional factors in calcu-lations. However, linking causesthe calculation to be broken into anumber of small relationships thatdo not appear together on a singlescreen and are therefore difficultfor the operator to follow.

DDC language trendsBecause our firm focuses its ef-

forts on high-performance build-ing-control applications, we havefavored DDC systems employingline programming. Line programsoffer greater power and expandedfunctions that are necessary in ourhigh-performance DDC projects.With our experience, we have longunders tood the p rob lems andshortcomings of line-programming

~(E41’l~~i/~‘ll’lN~;/~fli~ONI~I1’IONINti n DECEMRER l!#X~ 73

Programming languages

languages in certain applications.Over the years, we have worked

with users and manufacturers toimprove the ease with which lineprograms can be applied to build-ing control. We promote the con-cept of output-oriented programswherein every calculation and com-mand that directly affects an out-put is installed in one (and only

provided updates that simplify theoperation of these languages byconsolidating and simplifying theirrules of application. These im-provements are making line pro-grams simpler to apply withoutcompromising the power and flex-ibility that are inherent in their ar-chitecture.

Meanwhile, to improve their

3 Typical graphic program expression.

one) section of the program. Withoutput-oriented programming, theoperator can quickly trace any con-trol path and adjust the programeasily when a point or mechanicalsystem is not operating as desired.

In 1988, our firm released aguideline for line-based programlanguages called the operators’control language (OCL). The OCLguide was intended as a functionalspecification for features our firmand our clients desired in line-pro-gram languages. A discussion of themerits of good operators’ controllanguages appeared in the Septem-ber 1990 issue of HPAC. “ O C LSpells Freedom,” by Ken Sinclair,concluded that advanced line pro-gram-based DDC systems offergreater flexibility and are easier touse than other approaches.

Our firm’s experience agrees withSinclair’s conclusion. Building op-erators are easily trained in theoperation of most line programsand are able to use many of the ad-vanced features to make theirmaintenance and troubleshootingduties easier to perform. This ispossible because most line pro-gram-based systems have recently

range of applications, functionblock-based systems continue toadd to their libraries of blocks. Butlarge libraries reduce their sim-plicity, the primary advantage offunction-block programs. Thesetrends are eroding the few advan-tages that function-block programscan offer when compared to themore advanced line programs.

New approachesThe inevitable demand for sys-

tems that operate as Capt. Kirk ex-pected continues to drive manufac-tu re r s to deve lop innova t iveapproaches to building-control ap-plications programming. SeveralDDC system manufacturers havecommitted themselves to releasingDDC systems that employ a newapproach to the applications pro-gramming language-graphics pro-gramming. This approach utilizesthe powerful graphics capabilitiesof modern PCs to permit the pro-grammer to sketch out a flow chartfor the control sequence desired. Aprogram in the PC is then em-ployed to translate this sketch intoa program (usually written in ahigh-level line language), which is

down-loaded to the selected DDCstand-alone panel.

An example of a screen contain-ing a graphics program is shown inFig. 3. Note that the entire figurewould be built by the operator withsystem points, variables, and a li-brary of mathematical and logicalfunctions. If the operator desiredan additional space temperaturefor the calculation in Fig. 3, hecould easily add it by choosing theappropr ia te sys tem po in t andsketching a connection to the cal-culation block.

The idea behind graphics pro-gramming is to find a way to pro-vide the power and flexibility ofline programming with the sim-plicity of function-block program-ming. Essentially, the programmercan deve lop cus tom func t ionblocks to meet the needs of anyparticular project application. Theidea is enticing, but there are somepotential problems with the con-cept, including:

0 Program execution uncer-tainty--Once a programmer hasdeveloped a graphic, a translatorprogram has to be employed toconvert the picture into an exe-cutable program to be down-loadedto the appropriate stand-alonepanel. Experienced programmershave long realized that there aresome special considerations re-quired when writing programs toensure that they execute as ex-pected. For example, assume thefollowing simple sequence of oper-ation is desired to start supply andreturn fans.

If the weekly schedule is on, startthe return fan, and after a 30-secdelay, start the supply fan.

Fig. 4a shows how a line programcan be written to execute that logicsequence. If the data point RE-TURN_FAN is turned on only af-ter the entire block is executed, theprogram will execute properly.However, if RETURN_FAN isturned on as soon as that line is ex-ecuted, it is clear that SUPPLY_FAN may be started an instant af-ter RETURN_FAN is started. Abetter way to write the program is

74 DECEhlf3EH 1!2!Kl n H~,~Tl~~;;l’ll’lNC/All~CONI~ITIONlS~~

ICONTROL SEQUENCE: If the weekly schedule Is on, start the return fan, and after a 30 second delaystart the supply fan

FIGURE 4a: UNCERTAlN EXECUTION ORDER

IF WEEKLY_SCHEDULE = ON THEN BEGIN

DOEVERY. SO SECS T A R T RETURN_FAN ,_/ I F RETURN_FAN O N T H E N START SUPPLY_FAN

ENDDO

E N D .

FIGURE 4b: RELIABLE EXECUTlON ORDER

IF WEEKLY-SCHEDULE = ON THEN BEGIN DOEVERY 30 SEC

IF RETURN_FAN ON THEN START SUPPLY-FAN START RETURN_FAN

ENDDOE N D

L E G E N D

WEEKLY_SCHEDUlE WEEKLY SCHEDULE (SET UP ELSE WHERE)

DOEVERY 30 SEC DO LOOP THAT IS EXECUTED ONCE EVERY 30 SECONDS RETURN-FAN DlGITAL OUTPUT THAT STARTS THE RETURN FANSUPPLY-FAN DlGITAL OUTPUT THAT OPERATES THE SUPPLY FAN

NOTE: The underline Character Is used to join words to make single terms for point names or

variables. This follows normal programming convention.

.4 Line programs to execute simple tan start sequence.

shown in Fig. 4b. Although the or-der of items in the program is re-versed from what we might expect,it is clear that SUPPLY-FAN willnever be turned on less than 30 secafter RETURN-FAN.

While programmers can expecttranslators to offer some degree ofprotection against such translationerrors, they can never be entirelycertain that the translator is not re-sponsible for operational errors.What is a programmer to do if aprogram does not appear to be exe-cuting as pictured in the graphic?The programmer will inevitably berequired to inspect and trouble-shoot the translated program if theerror cannot be found by reviewingthe graphic screens. The line pro-gram that is developed by thetranslator is likely to be very diffi-cult to review because such pro-grams do not have the form andlogical flow that programmers typi-cally provide in their line programs.Nor are such programs likely to

have comments or formatting de-vices that make them easy to read.

Graphic programming may seemto be much more straightforwardthan line-based software, but anydebugging effort can become muchmore complicated, especially if theoperator desires to write more com-plex control algorithms that makethe fullest possible use of the en-ergy and comfort-enhancing capa-bilities of DDC.l Display limitations-An-

other problem with graphic repre-sentation of control programs isthey are bulky to display. Note thatthe averaging calculation of Fig. 3requires only a single line to repre-sent in the line program of Fig. 1.When several pages or more ofgraphics are required to represent acontrol sequence, the sequence canbecome very difficult to review be-cause the operator cannot look atthe whole program at once.

Recently, a client of ours experi-mented with a prototype of a new

graphic program-based DDC sys-tem. He translated a line DDC pro-gram used to control his building’sair systems to see if the graphicrepresentation offered any advan-tages. To his surprise, he found thegraphic program required eightscreens of graphic representationand seven pages of constants andgains to duplicate a line programthat occupied (with comments)only one and one-half pages. Theresulting program was far more dif-ficult for him to review than theoriginal line program.l Operator interface cost-

Most line-based programming lan-guages can be operated directly orover phone lines with a simple PCand low-cost software. However,the complex hardware and soft-ware required to create, test, trans-late, and compile graphics pro-grams can add substantially to thecost of each such terminal. Manyusers have developed system-sup-port mechanisms that include off-site access to the system via tele-phone modem by the engineer orseveral operators (at night). Thesemul t ip le - t e rmina l opera t iona lschemes can be much more costlyto implement because simple PCsmay not be capable of the perfor-mance needed to accompl i shgraphic programming.

Furthermore, the extensive pro-prietary software required forgraphics programming can becomecostly, and it is possible a copy willhave to be purchased for everycomputer that may be used. Suchcosts could substantially impactt h e f l e x i b l e D D C o p e r a t i n gschemes employed by many users.

Graphic programming is anotherserious attempt to provide DDCsystem operators, whose primarytraining and knowledge are in me-chanical systems, with the abilityto write and adjust high-perfor-mance custom DDC applicationsprograms. Today, DDC system us-ers and manufacturers alike under-stand the need to implement DDCsystems that are both functionaland easy to use. This represents animportant change in operations

Programming languages

philosophy from a few years agowhen most manufacturers (andmany users) believed DDC systemoperators should not be permitteddirect access to control programs.

Looking to the futureWhile Capt. Kirk might not be

impressed with the improvementsin DDC programming capabilitiesthat have been made in the last fewyears , DDC sys tem opera to r sshould be. Comparing today’s DDCprogramming features with thoseonly a few years old makes one real-ize the enormous strides that theindustry has made. Recently, I dis-cussed program specifics for a re-placement DDC system with a very

knowledgeable operator of an oldersystem. His ideas and concernssuggested to me that however bril-liantly this operator had employedhis old system to provide comfortand energy efficiency, his mode ofthinking was now limited by theoperational capabilities of that sys-tem. It is likely he will be able toutilize fully the capabilities of thenew DDC system only after he hasdeveloped an understanding ofthem through experience once thenew system is installed.

This operator’s problem is prob-ably universal to the building de-sign industry-and to other indus-tries that utilize digital technol-ogies as well. None of us is par-

“FAN ON/OFF CONTROL’

DOEVERY 1 MINIF OCCUPlED_MODE = ON OR COOLING_PURGE = ON ORWARMUP-MODE = ON THEN START SUPPLY-FAN ELSE STOP SUPPLY-FAN

ENDDO

“HEATING VALVE CONTROL”

IF WARMUP-MODE = ON OR HEATING_REQD = ON THEN BEGINPID_HTG:SETPOINT = SUPPLY_SETPOlNT_CALCHEAT-VALVE = PlD_HTG

ENDELSE HEAT-VALVE = 0

“COOLING VALVE CONTROL”

IF MECH_CLG = ON THEN BEGINPID_CLG:SETPOINT = SUPPLY_SETPOINT_CALCCOOL-VALVE = PID_CLG

ENDELSE COOL-VALVE = 0

“MIXED AIR DAMPER CONTROL’

IF SUPPLY FAN = OFF THEN MIXED-DAMPER = 0 ELSEIF MECH_CLG = ON THEN IF ENTHALPY_RA = ON THEN MIXED-DAMPER = 5

ELSE MIXED DAMPER = 100 ELSEIF COOLING_PURGE = ON THEN MIXED-DAMPER = 100 ELSE BEGIN

IF MINIMUM_OA < SUPPLY_SETPOINT_CALC THENPID_MAD:SETPOINT = MINIMUM_OA ELSE BEGIN

PID_MAD:SETPOINT = SUPPLY_SETPOlNT_CALCMIXED-DAMPER = PID_MAD

ENDEND

NOTE: OCCUPIED-MODE, COOLlNG_PURGE, WARMUP_MODE. HEATING_REOD, MECH_CLG, SUPP-LY_SETPOINY_CALC, ENTHALPY _RA, MINIMUM_OA are all variables representing logical decisions or cal-culations that are not shown in the program. The program represents the logic flow of a simple fansystem that is represented graphically In figure 6. The underline character Is used lo join words tomake single terms for point names or variables. This follows normal programming convention.

5 Line program showing control logic for Simple fan System.

ticularly adept at understandinghow effective new digital tech-nologies can be until we have someexperience working with them. Thismakes it difficult to look very far inthe future with clarity. However, wecan look at the issues that need to beresolved to continue to improve thesuccess of DDC programming lan-guages, and this may provide someanswers for the most likely next im-provements.

Combined line and graphicsOur firm, and many of our

clients, continue to believe thatline-based programming languagesprovide the best method to developDDC programs that execute effec-tive energy and comfort-enhancingcontrol strategies. However, we be-lieve it is possible that combiningcertain features of both line-pro-gramming and graphics-program-ming techniques may produce aformat for control programmingthat has advantages over the all-programming techniques generallyavailable today.

Earlier in this article, it was illus-trated that line-based programscan be the most effective way torepresent many kinds of mathe-matical and logical expressions be-cause they can make such expres-s i o n s c l e a r l y a n d c o n c i s e l y .However, a problem with line pro-grams is the lack of clarity in repre-senting major logic sequences. Fig.5 shows the system-level logic thatmight be employed to control asimple fan system.

In Fig. 5, the calculations andlower-level logic can be consideredto have been made elsewhere andare represented by their resultingvariables. Note that the logic con-trolling the supply fan is very read-able in this format. However, thelogic for the mixed air dampers isnot so simple and therefore some-what difficult to follow even thoughit is concise.

Function-block and graphicsprogramming as they exist todayare also weak in representing sys-tem-level logic because they are notconcise. In these programs, logic

5%

Fan systemstart/stop

Heatingvalve

Mixed airdamper

/ Occ;w;fim$de t Line program module-accessible for viewing or editing Off = Real-time logical or calculationby placing the cursor on the module and clicking the mouse result from line program module

-Lines of calculated values

+5+~/@ Graphic program modules whose programs are written in line program -Logic flow line-real-time FALSE indicationand accessible the same as line-program modules -Logic flow line-real-time TRUE indication

6 System logic for simple fan system.

paths are provided, but each com- Note that the graphic overview is areas of logic in effect when prob-. . ponent represents a very small por-tion of the overall program. There-fore, the overall system logic can berepresented only after one haspieced together a number of indi-vidual blocks. The DDC industryhas not yet found an effectivemeans to represent system-levellogic. This is a shortcoming of everyprogramming format in wide usetoday.

With the features now generallyavailable in PCs, it may soon bepossible to combine line-programblocks with graphic representa-tions of system logic to provide aformat for improved DDC controlrepresentations. Fig. 6 shows thelogic for the simple fan-control pro-gram of Fig. 5 in graphic form.

effective in representing the systemlogic. The circumstances underwhich the dampers and heating orcooling valves are operated can bereasonably deduced from the dia-gram. The labeled rectangularblocks are line-program modules.

Imagine that this graphic repre-sentation can display real-time re-sult(s) for each line-program mod-ule with the current lines of controllogic highlighted in special colorsfor true or false. Assume furtherthat the contents of any of the line-program modules can be called upfor review or editing simply byclicking on the chosen module.With such a programming tech-nique, the operator could quicklyisolate and troubleshoot the exact

lems develop. Such a programmingscheme as represented in Fig. 6may provide advantages over bothline and graphics programmingwhile mitigating many of their dis-advantages.

Future languagesThe p rogramming concep t s

shown above may be a natural con-tinuation for recent improvementsin DDC system programming lan-guages, but a wide variety of otheroptions are possible as well. What-ever the next steps in programminglanguages may be, they will be suc-cessful only if they work towardsolving the following current prob-lems:l Concise representation of ef-

Programming languages

fective DDC control strategies-The key to success with DDC con-trols is not to emulate traditionalpneumatic controls but to use thepower and flexibility of DDC sys-tems to provide new, in-depthmodes of control that result in en-hanced comfort and energy-effi-cient operation.

Such modes of control can besupported only if they are providedin programs that can be under-stood and diagnosed by system op-erators. It is not enough to breakthe program into such small piecesthat the overall concept is difficultto determine because the operatorhas trouble assembling all thepieces together at once. Any lan-guage must include representa-tion(s) that show very clearly boththe overall concept and the calcu-lation/logic pieces that constitutethat concept.l Real-time indications of pro-

gram calculations and logic

paths-One of the most powerfultools available to a DDC system op-erator is the ability to watch pro-grams as they execute and see theresults as they are calculated. Thisprogramming tool is typicallyavailable only with interpretivelanguages. However it is accom-plished, languages must be devel-oped that enhance the operator’sability to view real-time calcu-lations and logic while the DDCsystem is operating. This featureallows the operator to check pro-grams easily when their operationis suspect.l Few and simple rules to gov-

ern the language-The more rulesthat govern how a programminglanguage can be applied, the moredifficulties the operator has tryingto support the programs. Early ap-plications program languages hadmany rules governing everythingfrom the use of integers and float-ing point numbers to the use of

math in conditional statements.Manufacturers have done a goodjob issuing revisions that have sim-plified language rules for many ex-isting DDC languages. More needsto be done, however, and new lan-guages, whether graphic- or line-based, should be as free of re-strictive rules as possible.l Low cost- Improvements in

programming languages cannot bepermitted to reverse the trend to-ward lower-cost DDC systems.Manufacturers should consider theenormous market potential fortheir products when they havecombined sufficient function andease of operation in a package thatcompetes with pneumatics on firstcost. Full DDC systems are usually10 to 30 times larger (in terms ofsystem points) than the DDC sys-tems that go into many buildingstoday. Retrofit opportunities areeven greater.

Energy cos t s a re now h igh

Solutions for Control of Multiple Stages

The HVAC World looks to HEAT-TIMER forState-of-the-Art SEQUENCING CONTROLS!

Heat-Timer Offers Three DistinctiveSequencer Systems for Multiple Devices

Name a sequencer problem and Heat-Timer hasthe solution! Systems so precise, so flexible and soadaptible, they are unmatched anywhere!

l The HWR-Q reset series is for 2 to 8 stagesof hot water boilers; the MPC-Q series is for 2 to 8stages of modular steam boilers. Endless featuresoffer nearly every conceivable operating option!

l The MOD-4 is designed for full modulation boilersor chillers. It integrates all key functions of up to 4proportional stages, replacing L-91 type controls and

. “lead-lag” boxes. It unifies all functions in one state-

MOD-4 Sequencing Control for FullModulation Boilers & Chillers OnlyControl of its Capability in the World!

of-The-art computerized control!

l The DiGi-SPAN SQ series is for control of upto 8 stages of such equipment as compressorsand hot water generators at a constant set point.

Come to the world leader in precisionsequencing equipment. Call or writeHeat-Timer-today for specific solutions!

@’ PRECISION CONTROLS THAT SLASH ENERGY CONSUMPTION!10 Dwight Place, Fairfield, NJ 07004 l 2011575-4004 l Fax: 2011575-4052

78 DECE:MHEH 1990 n HEATINC/I’I~~IN(;/AIKCONI)ITIONIN(;Circle 335 on Reader Service Card

enough that most building ownerscan find a very attractive rate of re-turn in an investment of between$1 and $2 per sq ft for completeHVAC and lighting control retro-fit. If full DDC systems that pro-vide the comfort enhancement andenergy reductions of advanced con-trol strategies can be implementedat these costs, the industry will ex-perience an enormous growth involume over the next few years thatwill help pay for some of the devel-opment requirements.

SummaryThe DDC industry has made

substantial steps over the last fewyears to improve the power andflexibility of the control languagessupplied with their systems. This isa primary reason DDC systems arebetter accepted today than ever be-fore.

The types of control languagescommonly available for program-

ming DDC systems today includel ine -based p rogramming l an -guages, function-block program-ming languages, and now graphicsprogramming. Each of these ap-proaches has certain advantageswhen compared to the others forspecific applications, but it is clearfurther improvements are st i l lneeded, particularly improved rep-resentations of system logic.

When considering programminglanguage improvements, manufac-turers should work toward ap-proaches that permit the more in-depth strategies possible with DDCcontrol to be represented clearlyand concisely and provide methodswhereby real-time logic paths andcalculation results can be displayedand reviewed as the program isoperating.

As discussed in last month’s arti-cle, the acceptance of full DDC sys-tems has become a reality by userswho wish to have their buildings

perform more effectively than theycan with traditional controls. How-ever, to ensure that this higher levelof performance can be installedand will be maintained, better per-forming and more easily supportedapplications languages need to bedeveloped. As the power, flex-ibility, and ease of implementingDDC applications programs grow,the demand for DDC products willgrow also. fi

The Hartman Co. intends to updateits operators’ control language (OCL)guide to DDC system programmingfunction in 1991. We request com-ments from all in the industry who areexperienced in high-performanceDDC systems on the subject of DDCcontrol languages and related items.We will be happy to make a copy ofthe current OCL guide available tosuch individuals at no cost. Pleasewrite to: OCL Guide, The HartmanCo., 1016 North 36th St., Seattle, WA98103.

-

mxlderelimmtes needfor preheal coils

BlenderPradudis - Inli .FORMERLY R-M PRODUCTS

5010 COOK ST. DENVER, CO. 60216PHONE: l-800-523-5705 or (303) 295-6111

A static mixing device engineeredto thoroughly 81 accurately blendstratified air streams.

STOPCoil freeze-ups, freeze stattrips, and sensor errors dueto cold air stratification.

For information on thecomplete line of BlenderProducts equipment or tohave a stratification prob-lem analyzed please call:

l-800 5234705

“lnno va tors in des tra tifica tion sys terns ”

Circle 336 on Reader Service Card~~~~~I~G~PI~~~N~;/AIKCON~~TIONIN(; l DECEMBER 1990 79

TE R M I N ALOUTPUT

INPUT ANDDEVICES

A discussion of the selection and specificationof input and output devices as well as some new,inexpensive approaches to l/O device configuration

By THOMAS HARTMAN, PE,The Hartman Co.,Seattle, Wash.

II

n the two previous articles ofthis series, I have discussedcurrent trends in DDC sys-tem architecture toward

what many in the industry call fullDDC systems. Full DDC systemsare those that incorporate terminalcontrol as well as system control.To develop successful high-perfor-mance configurations of full DDCsystems requires special attentionto some of the features of the DDCsystem.

The previous articles focused onevolving capabilities of terminalcontrollers and programming lan-guages. They suggested ideas forimprovements in those areas thatmay be beneficial for this currenttrend toward full DDC systems.

The purpose of these articles is toencourage a dialog within the in-dustry aimed at defining featuresuseful in high-performance appli-cations of full DDC systems. The

view of many who are successfulwith DDC systems is that the in-dustry is flying blind. Manufactur-ers are developing products basedon criteria that may be flawed orlack relevance. Many A/E firmscontinue to employ DDC specifica-tions and procurement proceduresthat display a profound lack ofknowledge about the features thatare most crucial to a successfulDDC system.

Recent research conducted bytrade associations on DDC seemsto have reached at least as manywrong conclusions as correct onesabout what ingredients are neces-sary for successful DDC implemen-tation. .i the DDC revolution con-tinues, our industry needs toattract and make better use of theideas from those associated withsuccessful high performance DDCsystems.

This final article will focus on in-put and output devices employedfor terminal control. We will startby considering how the evolution tofull DDC is placing enormous im-portance on how the designer

chooses terminal input/output de-vices. Then, we will look at some ofthe newer hardware that may re-duce the cost of full DDC systemswithout compromising their qual-ity. Finally, we will explore someapproaches DDC system manufac-turers could incorporate in theirproducts and interface schemes toprovide more functional and eco-nomical terminal control.

Terminal control devicesDDC system inputs and outputs

(commonly referred to as points)have increased enormously in sig-nificance as terminal control op-portunities have encouraged theuse of full DDC system config-urations. DDC systems that con-trol only fans, chillers, and centralplant equipment often require 200points or less to control a large sizebuilding effectively.

Full DDC configurations that in-corporate lighting and temperaturecontrol for every space may require5000 or more points. This 25-foldincrease in points has changed therules by which DDC systems are

HEATING/PIPING~AIRCONDIT~ONING n JANUARY 1 9 9 1 I191

I/O devices

configured to be cost effective andperform successfully.

The role of input/output devicecost has changed from a minor itemto the leading factor in determiningDDC system cost effectiveness. Arule of thumb is that the installedcost of a full DDC system must bekept under $200 per point to have achance of being cost effective whencompared to other approachesfrom a purely economic stand-po in t . Th i s ob jec t ive can beachieved only if the system de-signer pays close attention to theselection of input and output de-vices employed for terminal con-trol.

A strong emphasis needs to beplaced on the terminal control in-put/output devices because theseoften represent more than 90 per-cent of the total input/output de-vices in a full DDC system. The de-signer’s task is to select com-ponents that provide desired sen-sing or control characteristics atthe lowest possible cost. To do this,the designer must first determinewhat characteristics are realistic.For example, precision RTD tem-perature sensors are still commonlyspecified for space temperaturesensors. Such devices can cost up to$100 or more apiece. But the limit-ing factor in space temperaturesensing accuracy is almost neverthe sensing device. The end-to-endaccuracy of space temperaturesensing for DDC systems involves anumber of considerations, the mostimportant of which is usually thespace itself.

Consider that typical occupiedspaces in buildings are suppliedwith air that is 20 F warmer orcooler than the space. As this air iscirculated to mix with the space airto add or reject heat, temperaturegradients within spaces are oftensignificant. If several independentinstruments are placed within a fewinches of a space temperature sen-sor, their readings will frequentlydiffer by 0.2 to 0.5 F. Stable therm-istor-based space temperature sen-sors cost only a fraction of pre-cision RTD sensors yet provide

stable operation and precision thatis far better than the 0.2 to 0.5 Fprecision that can be realisticallyachieved in space sensing applica-tions.

In selecting input/output de-vices, designers should remindthemselves that if their full DDCsystems are configured such thatthey cannot be economically justi-fied over pneumatic alternates,their clients and the building oc-cupants are the big losers. Comparethe precision and stability of pneu-matic thermostats and humidistatsto those of the most economicalelectronic devices. If DDC systemscannot be configured to meet bud-getary and/or cost benefit require-ments, the lower cost alternative isusually pneumatic controls. Butpneumatic devices provide controlthat is several orders of magnitudeless precise and also require con-stant calibration to maintain eventhat low level of precision.

Such an understanding of therole and importance of cost consid-erations is required for a designerto make reasonable choices for ter-minal input/output devices. Termi-nal output devices such as damperactuators and reheat valves haveoften been considered to be perfectapplications for pneumatic actu-ators. However, when the applica-tions are scrutinized more closely,pneumatic devices are often not agood choice at all for terminal con-trol applications. Furthermore,pneumatic devices do not interfacewell with electronic control sys-tems. As a result, a number of neweconomical actuation devices havebeen growing in popularity over thelast few years.

Electric actuatorsModulating electric actuators

have been manufactured for manyyears, but they have not beenwidely employed in North Ameri-can HVAC applications. Until re-cently, these devices were generallycomparatively expensive and notnoted for durability. Electric con-trols of different manufacturerswere often incompatible with one

another. The increasing demandfor electric control devices that canbe interfaced directly with DDCsystems appears to be improvingthe situation.

Electric actuators are falling inprice and are becoming more stan-dardized on the 0 to 10 v DC controlsignal to modulate their position.An electric modulating actuator ofthe type employed to control aVAV box damper is shown in Fig.1. This type of actuator is now be-coming available for $100 to $200.Electric actuators have several fea-tures that make them desirable forthe control of small dampers. Gen-erally, a unit modulates by com-paring the modulating voltage fromthe DDC system with a voltage thatdepends on the position of the ac-tuator.

Unl ike pneumat ic ac tua to rswhose force is related to the outputand position, most electric actu-ators provide constant force (ortorque) at all outputs and stop onlywhen the position corresponding tothe DDC voltage is reached. Thiseliminates the uneven movementthat is often associated with pneu-matic actuation. Some electric ac-tuators also maintain constanttorque whether running or stalled.This feature is particularly usefulin applications that require torqueto be applied at the end stops tohold dampers fully open or fullyclosed.

Two-motor actuatorsA more economical electric zone

damper ac tua tor employs twosmall constant-speed clock motorsconnected together. These, as dothe previously described actuators,rotate the shaft collar or clampthrough a gear drive. The two-mo-tor type actuator is interfaced withthe DDC system through two digi-tal output points. The two motorsrun in opposite directions. One mo-tor drives the damper open, theother drives the damper closed.This device is commonly called afloating control actuator.

An example of a two-motor actu-ator is shown in Fig. 2. This type of

1 Electric actuator 0 to 10 v DC modulating signal, 24 v (AC or D C ) power.

actuator typically costs under $50. times resulted when the controlThe speed of rotation of each mo- system provided continuous powertor is constant and generally geared to the controlling motor at the fullyto be slow-l to 3 min to rotate open or fully closed position. Thefully each direction. resulting locked rotor condition ap-

In the past, reliability problems pears to have led to abnormal fail-with the two-motor actuator some- ures of the motor or gear train. Re-

liability problems also may haveresulted from the mechanicalswitching mechanism that operatesthe motors. Switches or relays inthe control circuit are subject to agreat deal of cycling, and unreliablecontact closure may have alsocaused some problems in the past.

Although many of the two-motoractuators are built much sturdierthan they were a few years ago,proper operation by DDC control-lers have also contributed to re-duced reliability problems. BecauseDDC systems today typically drivetheir outputs with solid state triacs,there are no mechanical compo-nents involved in running or switch-ing the motors.

To eliminate locked rotor prob-lems and to provide improved con-trol, several DDC manufacturershave provided special interfaces forthe two-motor actuator so that itlooks like any other analog point tothe operator. The interface auto-matically calculates the position ofthe actuator by keeping track ofthe time each motor has operated.When the actuator reaches one ex-treme or the other, the position isautomatically reset, and the powerto the motor is shut off. This ap-proach works well and usually per-mits two-motor actuators to oper-ate by standard PID controllersjust like other analog outputs. Be-cause each motor is shut off at ex-treme positions, locked rotor prob-lems are eliminated.

The two-motor controller pro-vides economical modulating con-trol that is adequate for most ter-minal control applications_ Thesedevices are typically configuredwith low-speed gear trains becausehunting problems can result whenthe devices are controlled by simplefloating-point controllers. The lowspeeds may not be adequate foropening or closing in certain cir-cumstances. However, with outputcontrollers that permit PID controlof the two-motor actuators, muchhigher speeds can be controlledwith ease, and some two-motor ac-tuators are now becoming availablethat operate at greater speeds.

HEATINO/PIPINO/AIRC~NO~T~~N~NO # JANUARY 1991 93

I/O devices

Air velocity sensorsOne of the most noticeable ad-

vantages of DDC terminal VAVcontrol is the ability to provide airbalancing of terminal boxes withthe DDC system. DDC air bal-ancing is much easier than tradi-tional methods. The maximum andminimum air flows for each box areset by the DDC system, and eachVAV box damper operates to pro-vide a specific air flow within thatrange based on space conditions.Duct pressure compensation is au-tomatic.

To achieve all these features re-quires an air flow sensor at eachVAV box, connected to the DDCsystem. At present there are twoapproaches employed by theHVAC industry to measure air flowat terminal boxes. Both actuallymeasure air velocity, which is mul-

I I

2 Two-motor electric actuator, 24 v AC.

tiplied by a constant (the constantincludes factors for duct area andflow profile compensation) to ob-tain air flow.

One approach employed to de-termine air velocity at terminalunits is to measure the velocitypressure of the air through the useof a pitot tube arrangement. Someof these velocity pressure devicesactually measure the average veloc-ity pressure with a series of up-stream orifices to help compensatefor the irregular velocity profiles

that occur in many ducts.The velocity pressure air flow

measurement technique has oneserious limitation. At low air veloci-ties, the velocity pressure of air isvery small. Much literature recom-mends the velocity pressure tech-nique be limited to air velocitiesabove 300 fpm. Tests conducted byour firm found reasonable accuracysomewhat below that figure undercertain conditions, but the bottomlimit of accurate flow measurementis still above the lowest velocitiesencountered at VAV terminalboxes.

Another problem with velocitypressure type sensors is that thedifferential pressure/electric trans-ducer required to convert the veloc-ity pressure into an electric signaland the sensing equipment can becostly. Some DDC manufacturersinclude the pressure/electric trans-ducer with their box controller,which seems to reduce transducercost dramatically. Controllers in-corporating the transducers areusually competitively priced withthose not including transducers.

A second approach to measuringair flow is to determine the air ve-locity by measuring its capacity toprovide cooling. Methods that em-ploy this approach are sometimesreferred to as “hot wire” or “heatedthermistor” techniques. There are anumber of variations in operationfrom manufacturer to manufac-turer and depending on the exactdevices incorporated in the design.Fig. 3 shows one type of heatedthermistor air velocity sensor.

Essentially, one device is heatedwhile some mechanism determinesthe cooling effect of air passing overit by measuring the actual temper-ature of the device (or the currentrequired to maintain the device at aconstant temperature).

The cooling effect approach hasone big advantage over the velocitypressure technique: it can accurate-ly measure air velocities as low as50 fpm.

The most economical of the coo-ing effect velocity sensors employsseveral thermistors. Exact oper-

ation varies, but a popular tech-nique is to employ one thermistorto measure the ambient tempera-ture of the air stream. The secondthermistor (slightly downstream ofthe first) has a voltage applied to it,and the current required to main-tain the voltage is measured. Be-cause the resistance of the therm-istors varies with temperature, thetemperature of the heated therm-istor can be calculated. The combi-nation of ambient temperature,heated thermistor temperature,and energy to the heated therm-istor can be used to calculate the airvelocity in smoothly flowing air.

This sequence may seem compli-cated, but these comparisons areusually made in a small IC chip.The cost of thermistor-based cool-ing effect air velocity sensors isabout $25 or less, and some of themautomatically provide leads thatcan be connected to another inputof the DDC system to provide theair temperature as well. Most donot provide a linear output. Farmore expensive devices are re-quired for that. But most DDC sys-tems offered today have the abilityto linearize such signals with cali-bration tables that are included intheir input point databases.

A disadvantage of the cooling ef-fect air flow measurement tech-nique is that it provides a readingof air flow at only one point in theair stream. To read more points foraverage air flow (which is easily ac-complished with the velocity pres-sure technique) requires an array ofa number of individual sensors.Thus, a sensor designed to measureseveral points can become ex-pensive.

Our firm has also conducted per-formance tests of air flow measur-ing devices. The tests concludedthat by following some basic rulesfor sensor location, we can employa single-point velocity sensor tomeasure air flow in the duct sizestypically employed for VAV termi-nal units with adequate precision.In addition to sensor location,these tests found that how the ve-locity-to-flow factor is determined

is important in achieving accuratemeasurements.

Because they generally offer thelowest cost air flow measurementoption, thermistor-based air flowmeasurement has a jump on othertechniques. However, thermistor-based velocity sensors have pitfallsthat designers should considerwhen selecting air flow devices. Thelow-cost thermistors employed inHVAC applications are generallystable and drift-free in applicationsup to about 300 F. Above this tem-perature many are subject to drift.

Because the heated thermistor inan air flow sensor usually operatesat about 300 F, some in the indus-try have expressed concern that pe-riodic recalibration of air flow sen-sors may be required. I have seenno conclusive evidence on the sub-ject of drift for heated thermistorsensors. However, most of the in-stallations we are familiar with areonly several years old. Meanwhile,manufacturers are very short onfacts that might prove drift will notbe a problem.

One solution to the drift problemis to employ sensors that use stableRTD sensors in hot wire config-urations. Another is to select higherquality thermistors that are notsubject to drift at elevated temper-atures. Glass-encapsulated therm-istors are generally very stable athigher temperatures. A recententry into the air velocity sensormarket utilizes glass-encapsulatedthermistors that are still priced atabout $20 apiece in quantity.

Other sensing devicesTemperature sensors, air flow

sensors, relays, and damper actu-ators currently represent the vastmajority of terminal end devicestypically associated with DDC ter-minal units. A number of other de-vices will become widely used ashigh-performance DDC systemsgain in popularity.

Occupancy sensing is becomingstandard fare for high-performanceDDC systems as lighting control isintegrated with HVAC terminalcontrol strategies. The simplest

_ sensor

3 The photo on the left shows a heated thermistor-type velocity sensor andintegrated circuit chip. The upper diagram shows dimensions in millimeters.The diagram on the right shows connection to IC chip.

type of occupancy sensing device isa button on the space temperaturesensor that occupants push whenthey arrive at the zone. When thebutton is pushed, the DDC systemturns on the lights and establishesan HVAC occupancy mode for thezone. Typically, the zone remainsin the occupied mode for the re-mainder of the day if the buttonhas been pushed during normalworking hours, or for a prede-termined time if it is activated dur-ing off hours.

Many manufacturers now incor-porate such buttons in versions oftheir temperature sensors. Thetemperature sensor input is alsoused by some to transmit the oc-cupancy signal to the DDC system.The result is that this type of oc-

cupancy device can be very inex-pensive to provide-adding only afew dollars per zone to the cost ofthe system.

A more effective, but also morecostly, occupancy sensing device isan infrared or ultrasonic sensor.Such sensors-usually mounted onthe ceiling or high on the wall-de-tect movement and can thus be em-ployed to establish occupancy au-tomatically as people arrive at eachzone. This type of occupancy sen-sor costs about $50 to $100. Theadded cost over pushbuttons canoften be justified because the mo-tion type sensor offers greater en-ergy savings and less inconvenienceto building occupants.

It is a nuisance for zones to lapseout of occupancy while still oc-

rorrlrfllllY~orr pug‘. I(X)

HEATING/PIPING/AIRCONDITIONING n JANUARY lew 95

I /O devices

cupied, so most pushbutton oc-cupancy strategies employ long pe-riods of occupancy each time thebutton is pushed. If someone comesinto an office to pick up some pa-pers, the zone may remain in theoccupied mode for several hours orall day. But an occupancy sensorcan automatically return the zoneto the unoccupied mode after just afew minutes.

Other devices that are growing inpopularity are humidity sensorsand various kinds of air qualitysensors. Such sensors are not re-quired in every zone. But sensorsthat provide VOC (volatile organiccompounds) and CO, readings intypical and critical areas can pro-vide useful information to thebuilding control system. The trendis clearly toward considering airquality control as important astemperature control. Most sensorsthat measure the quality of thebu i ld ing env i ronment a re be -coming very economical as new in-expensive technologies are devel-oped to accompl i sh the mea-surements.

What’s needed nextThe quality of typical input/out-

put devices for DDC terminal con-trol is improving while costs arefalling. These are good trends, butfurther improvements are possible,and cooperation with DDC systemmanufacturers is required to seethem through. There remain anumber of possible improvementsin terminal input/output devicesthat can increase the cost effec-tiveness of DDC systems in theshort term.

Modulating digital outputsA variety of electrical actuation

approaches are offered for reheatvalve operation. Those that involvemodulation are often more ex-pensive than pneumatic alternates.However, several manufacturersoffer small electric two- and three-way valves that employ a very inex-pensive method of actuation thatcould be easily modulated by aDDC terminal controller. These

valves are actuated by thermostaticelements similar to the bimetal de-vice that regulates the water tem-perature in automobiles. But in-stead of water temperature, theseactuators are energized by a smallresistance wire that is heated byjust a few watts of power. Thepower requirement is small enoughthat the valves can be powered di-rectly from the triacs that drivedigital output points in most DDCterminal controllers. While thesevalves are generally used as two-po-sition valves, they can be modu-lated by pulse-width modulation ofthe digital output points to whichthey are connected. These valvescost less than many pneumatic ac-tuated reheat valves and have agood record for reliability.

Similarly, electric reheat coilscan be modulated by pulse-widthmodulation of a digital output sig-nal to solid-state power relays. Fig.4 shows a typical connection forsuch modulation devices connectedto a single DDC digital output.

To make effective use of suchmodulation techniques, DDC prod-ucts require the capacity for opera-tors to configure digital outputpoints to look like analog points.This is already done by manufac-turers who configure two-motorelectric actuators to look like singleanalog outputs. Unfortunately,these configurations are not nowvery flexible. Rather, they are typi-cally incorporated into dedicatedcontroller outputs that are gener-ally unusable for any other pur-pose.

As was pointed out in the articleon terminal controllers, developingproducts to be more flexible shouldbe the goal of every DDC systemmanufacturer today. Flexibility inpoints configuration is extremelyhelpful to operators. With a config-uration option that permits a sys-tem to be easily configured withdigital outputs operating as analogpoints, the devices shown in Fig. 4could be operated with standardPID loop control to provide effec-tive modulating terminal controldespite the fact that they are digi-

tal output points.DDC system suppliers and man-

ufacturers appear at times to bepoorly informed about interfaceoptions for their systems. This isunderstandable since they gener-ally do not manufacture many ofthe input/output devices employedby their systems. However, if agreater effort were made to seek outeffective and economical I/O deviceoptions for their DDC systems,controls companies would find asubstantially greater demand fortheir terminal control products.

Those manufacturers who arewilling to spend the effort to pro-vide effective and flexible inter-faces to operate special I/O devicesin a manner that is consistent withmore standard devices are the ones

DDCsystem

Thermo-electricactuated valve

r - ir - i

Valvecontrolcurrent

Modulating valve on digital output

DDCDDCsystem

0

Digitaloutput(Triac)

IJ

Controlcurrent

-I-Solidstaterelay

I Electric resistanceduct heater

1 f==Powercurrent

Modulating electric duct heater

4 Typical connection for a modulatingvalve on digital output and modulatingelectric duct heater connected to asingle DDC digital output.

that will dominate the full DDCmarket in the years ahead.

Terminal DDC input/output de-vices are becoming more effectiveand lower in cost. Unfortunately,many DDC systems presently donot fully exploit low cost terminalcontrol concepts. As a result, manyDDC systems cannot provide cost-effective full DDC for high-perfor-mance applications. Such DDCsystems are losing out in a very im-portant new DDC market.

In these articles, I have suggestedthree areas that need the attentionof manufacturers, engineers, andusers to exploit the growing fullDDC market fully. These are DDCterminal controllers, programminglanguages, and terminal control in-put/output devices.

The views I have expressed havebeen developed from our firm’s ex-perience and include ideas we havegleaned from our more successfulclients. There are many other view-points and ideas about the best wayto configure full DDC systems forhigh performance applications. Allwho have success fu l ly imple-mented cost effective DDC systemsshould be heard.

Important upcoming forums todiscuss high-performance DDCsystems will be conducted at AEE’sWORLDCON’90 this spring andASHRAE’s annual meeting thissummer. It’s time those who are as-sociated with successful high-per-formance DDC systems get to-gether and share ideas so manufac-turers and designers can get a betterpicture of what users need to con-tinue their successful trends intothe era of full DDC systems.

The primary purpose of this arti-cle series was to stimulate muchneeded discussion within the in-dustry about full DDC systemfunction. Everyone stands to bene-fit from such an exchange as theHVAC controls industry attemptsto take the two giant steps to fullDDC system configuration andhigh-performance DDC systemoperation. Q

CONTROL ENERGYCONSUMPTION WTHGRAHAM ULTRAFLEX

PMTE HEAT EXCHANGERS?NO SVVEAJl

Get all the cost-saving benefits ofGraham ULTRAFLEX Plate HeatExchangers for your HVACapplications-efficiency, reliability, easymaintenance, economy For freecooling, water source heat pumpisolation, district heating, thermalstorage, geothermal heat isolation, orother comfort control requirements,Graham units can help reduce yourbuilding energy demands.

Our units are computer selected tobalance capital and operating costs, ensuring rapid return on yourinvestment. Units are compact, easily expandable, and easy toclean for simplified maintenance and reduced downtime.

Many models in inventory for quick assembly and shipment.For fast quotations, call or write: IGraham‘Manufacturing Co. Inc., 20 FlorenceAvenue, Batavia, NY 14020. 716) 343-2216.

\Telefax (716) 343-1097. Telex ( 16) 854-4142.

Circle 380 on Reader Service Card~~EA~INC/PII~ING/AIHCOSDITIONIN~~ n JA~LARY IHRI 101101

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..~..........................HIGH-TECH SYSTEMS

PROCURlNG HIGH-TECHMECHANICAL SYSTEMS

Alternative purchasing procedures forengineers and owners who are considering

procuring advanced technology componentsfor new or existing building mechanical sys terns

By THOMAS HARTMAN, PE,The Hartman Co.,Seattle, Wash.

I3

ow do engineers ensurethat the advanced tech-nology building systemsand components incor-

porated in modern buildings arepurchased effectively? Our firm de-signs high performance DDC sys-tems for building controls. We havefor many years worked with own-ers, engineers, and control vendorsto develop more effective pro-curement techniques with somesignificant successes-the cost perpoint of systems employing im-proved procurement techniques iswell below the industry average,and the achievements of these sys-tems are far above the industrynorm.

Recently, however, I realizedhow unpolished the process still isin many building construction pro-jects. In a discussion of requests forproposal procurement for ad-vanced technology mechanical sys-tems, I was explaining to a projectengineer that typical bidding pro-cesses were not effective becauseequals simply do not exist amongmost high-tech building systems.

The faraway look in the projectengineer’s eyes let me know it wastime to stop talking, so I asked hisviews. He stated, “We name names.When we look at a project, we de-

termine which of the local con-tractors can be trusted to deliver onthe size and level of complexity in-volved and limit the bid to thosefirms. It simplifies the process forus.”

In another recent discussion, adesign engineer unabashedly dis-closed that he viewed his role as aspecifier of the advanced technol-ogy portion of the project and im-plied that the actual design wasbest accomplished by the supplierwho won the bid.

Indeed, these approaches do sim-plify the process for the design en-

gineer. But my experience is thatsuch techniques often result in con-trol or other high-tech systems thatare not the best value and often donot meet the functional expecta-t i o n s o f t h e c l i e n t . F u r t h e r -more, the problems with pro-curement concepts such as theseare increasing dramatically as theuse and level of technology in-volved increase. Fig. 1 shows the in-creasing role of high-tech systemsin our firm’s projects. We believethe use of advanced technologies inbuilding mechanical systems willbe skyrocketing by the turn of the

1 Historic and estimated trends for high-tech percent of total mechanicalbudget for The Hartman Co.

HEAT~NG~P~P~NG~A~RCOND~T~ON~NGHEAT~NG~P~P~NG~A~RCOND~T~ON~NG n NovEbwmI99iNovEbwmI99i 37

High-tech systems

century.My criticisms of procuring ad-

vanced technology systems withthe traditional specification-bid-ding process fit two basic catego-ries. First, purchasing equipmentfor which exact equals do not existis not genuinely competitive andoften results in poor value for theclient. The cost penalty is poten-tially enormous. High-tech HVACcontrol systems that are procuredthrough more competitive meansusually meet the applications re-quirements more effectively andmay cost half as much as similar in-stallations (or even less) purchasedby the traditional bidding process.

My second basic criticism of pur-chasing high-tech systems by bid-ding is that the resulting specifica-tions are generally not clear andrequire substantial additional de-sign efforts by the successful con-tractor. By not accepting respon-sibility for the complete appli-cation design of an advancedtechnology system, the engineertends to lose touch with these in-creasingly important advancedtechnology systems. This can beseen with HVAC control systems,which are now as removed frompneumatic controls as office com-puters are from the typewriters anddesktop calculators they have re-placed. Still, a large number ofDDC specifications describe func-tion in pneumatic terms.

An engineer who loses touch withcertain technologies involved in adesign hurts the client becausesuch a designer seldom utilizes thattechnology to its full potential. Ad-ditional problems stem from theambiguity that results from the de-signer’s lack of understanding ofthe technology. Many high-techsystem specifications actually dis-courage contractors from respond-ing with low-cost or optimallyconfigured systems.

In this article, I will suggest ideasfor alternative procurement pro-cedures to engineers and ownerswho are contemplating advancedtechnology components for new orexisting building mechanical sys-

tems. The techniques I will discussare not new or the only possible op-tions. They are techniques our firmand others have used very success-fully for a number of years. Asbuildings continue to employ ad-vanced technology systems, thetechniques engineers and ownersdevelop to purchase, install, andoperate their advanced technologycomponents will increasingly influ-ence the level of success these sys-tems attain.

I recommend that engineers andowners consider establishing a sep-

Step 1 - What’s neededThe first step in conducting an

effective procurement process foradvanced technology systems is de-termining what features are neededand desired for the system to oper-ate successfully as part of thebuilding’s mechanical system.l Function-Building owners,

including speculative developers,almost always want more than justthe lowest cost product. They de-sire systems that will provide suit-able levels of performance at themost reasonable cost. The engineer

Ron Anderson (left) of The Hartman Co. fields some suggested changes in theDDC system for a terminal regulated air volume (TRAV) mechanical systemwith Bob Cuti and Terry Egnor of Microgrid, the project’s management firm.Plan a discussion session to explain options and alternatives so that everyoneinvolved in the procurement is working toward obtaining the best value for thehigh-tech system.

arate category at the start of eachbuilding construction project forsystems and equipment that utilizeadvanced technologies and, be-cause of a lack of equal products,require a special evaluation processfor selection. The owner and engi-neer should then work together toensure that these systems and com-ponents are designed to meet thefunctional requirements of the pro-ject with as high a degree of cer-tainty as possible and that they arepurchased as competitively as pos-sible. The focus of my experiencewith advanced technology systemsis controls, but the following three-step process is suitable for a num-ber of other advanced technologysystems and components that maybe integrated into HVAC systemdesigns.

must determine a part of what con-stitutes a “suitable level of perfor-mance” by determining what isnecessary to meet the rigors of theoverall mechanical design. How-ever, the owner’s concerns shouldbe an important factor in deter-mining what is suitable as well. De-sign engineers are generally notwell acquainted with issues of oper-ab i l i ty o r main ta inab i l i ty -es -pecially as they relate to the rapidlyadvancing technologies that are in-corporated into DDC systems andother energy management devices.

Modern mechanical designs arebecoming increasingly complex be-cause they must meet increasinglycomplex demands for space and en-ergy efficiency, comfort, air qual-ity, and fire and life safety. An engi-neer may spend months developing

38 NOVEMBER~~~~NOVEMBER~~~~ n HEATING~PIPING~AIRCONDITIONINGHEATING~PIPING~AIRCONDITIONING

and refining the overall operatingstrategy for an HVAC system, andhe must ensure that the level ofperformance of each component ofthe system is adequate to meet theoverall system performance re-quirement.

On the other hand, owners andoperators usually consider oper-ational issues of primary im-portance and generally have somestrong opinions about what qual-ities are necessary to achieve thedesired performance. The best wayto develop basic performance levelstandards is through discussionsbetween the engineer and owner’srepresentatives. In these dis-cussions, the engineer can explainhis design from an operationsstandpoint and solicit the owner’sviews concerning operating andmaintenance features of the sys-tem.

The engineer should act as a re-source for the owner to answerquestions or concerns with up-to-date information about the ad-vanced technological system beingconsidered. In this way, the basicstandard of acceptable system per-formance can be developed for thedesign. Our firm’s approach is toprepare draft specifications forthese meetings and explain the pro-visions and why each has been in-cluded. As realistic changes or im-provements are agreed upon, wesimply adjust the specification toincorporate those changes. Wehave found that the process workswell because it prepares the ownerfor participating in the process ofselecting a system from those thatwill eventually be proposed.l Maintenance and expan-

sion- Most advanced technologyenergy-related systems include asubstantial amount of equipmentthat has been designed and con-structed by only one manufacturer.To repair, expand, or replace thisequipment generally entails non-compe t i t ive so le - source p ro -curement. While this is not a majorportion of the business of mostmanufacturers, some local repre-sentatives cannot resist the chance

A large number of high-tech products are now entering mechanical systemdesigns. This influx demands improvements in procurement procedures toobtain the best value for the client. Here, Ron Anderson of The Hartman Co.inspects the variable frequency drives installed as part of a recent TRAVretrofit.

to take advantage of the ownerwhen such sole-source purchasingis required.

By making unit pricing and ex-tended warranty offers a requiredpart of the procurement process,the engineer and owner can elimi-nate these potential problems.They have an opportunity to dis-cuss options for supporting, main-taining, or perhaps expanding theadvanced technology system beforethe system is purchased and en-suring that the ongoing needs ofthe owner are addressed in the pro-curement process.

It should be noted that if it is de-cided that unit pricing or extendedwarranty offers are desired formore than several years past thedate of completion, potential ven-dors should be given wide latitudeto tie their pricing to price indexessuch as the consumers price index,interest rate, or other industrypricing indices. If maintenance ofthe system is expected to be sub-stantial, or if the owner intendssubstantial expansion of the sys-tem, then the evaluation of thewarranty offer and/or unit pricingschedule may be an important partof the evaluation process.

Our clients have found the ex-tended warranty offer requirementto be very valuable in comparing

competing systems. Because the of-fer is made in a competitive pro-cess, each extended warranty offerusually represents the contractor’sbest estimate of the true annualcosts of maintaining the system.Whether or not the owner intendsto contract with the vendor tomaintain the system after the one-year construction warranty ex-pires, the vendor’s estimate of an-nual warranty cost is very tellingabout its confidence in the integrityof the equipment to be supplied. Arequirement that the extendedwarranty agreement remain openuntil the expiration of the con-struction warranty period permitsthe owner a year of operating ex-perience before deciding whetheror not to contract emergency repairservices or provide them internally.

Step 2 - ProcurementOnce the decision to procure a

project’s advanced technologycomponents competit ively hasbeen determined, the owner andengineer must work together to en-sure that a competitive selectionprocess is developed to meet theowner’s procurement rules. If theowner represents a private organi-zation, there are likely rules andguidelines within the organizationthat may limit the flexibility of the

HEATINGIPIPINGIAIRCONDITIONING n Novmmm 1991 39

High-tech systems

procurement process. If the owneris a government agency, then lawsmay limit the procurement. Thebest rule to follow is to make no as-sumptions regarding which ruleswill actually apply.

I sometimes think I would ownthe U.S. Mint if I had a nickel forevery client who started the processby telling me his organization hasto use the traditional bidding pro-cess in its projects. To date, I havenot had one client that in the endactually had to do so. What hastranspired between initial rebuffand the successful procurementprocess provides enough materialto write a book of humor andseveral mystery tales.

The goal of the procurement pro-cess should be to break the ad-vanced technology system(s) out ofthe basic construction bid docu-ments and conduct a separate pro-curement, generally with the planto reattach it to the constructioncontract once that procurement

be. Once the procurement is com-pleted, the actual subcontractamount is substituted for the al-lowance, and the general con-tractor assumes responsibility formanaging the entire constructioneffort.

Fig. 2 illustrates how the cash al-lowance process permits the sepa-ration of advanced technology sys-tem procurement from the generalcontract. There are other ways toaccomplish the advanced technol-ogy installation, but if the systemor component is an integral portionof the overall building operation, acash allowance is recommended be-cause it does not compromise thegeneral contractor’s control or re-sponsibility for the construction ef-fort.

To ensure that an effective pro-curement process is established forhigh-tech systems, I suggest thefollowing guidelines be followedwhen developing the process:l Make the case-There is an

2 Using the cash allowance feature to conduct separate selection process forhigh-tech systems.

process is complete. The cash al- important reason you are likely tolowance feature of construction hear negative responses if you askcontracts is an excellent vehicle for about procurement flexibility inaccomplishing a separate pro- large bureaucracies. People like tocurement. The use of cash allow- do things the easiest way possible.ances is widespread in construction Conducting construction pro-contracts today. curement with a traditional bid-

The presence of the allowance ding process is usually the easiestalerts the general contractor to the way. However, if the issue is pre-fact that management resources sented candidly, I have found thewill be required to coordinate this problems associated with garneringelement of the construction effort, support within the bureaucracy’sbut the general contractor does not purchasing establishment are gen-decide who the subcontractor will erally greatly reduced.

40 NovEMBERI~~INovEMBERI~~I n HEATINGIPIPINGIAIRCONDlTlONlNGHEATINGIPIPINGIAIRCONDlTlONlNG

To make the case, the engineershould be prepared to discuss theadvantages of a proposal processrealistically along with the prob-lems of a strictly bidding pro-curement. Having as many of thedifferences expressed in bottomline dollar figures is the best way Iknow to get the attention (andsympathies) of purchasing people.

In making the case, provide aspecific procurement plan that youbelieve will meet the owner’s orga-nizational rules. The most powerfulsupport you can garner is pat-terning your plan after a similarprocess that has been recently em-ployed by the organization with asuccessful outcome. For this, someresearch and investigating is neces-sary.

When working with governmentprojects, I have had success con-tacting the offices of representa-tives that chair committees in-volved with energy to find laws thatpermit (and sometimes encourage)nontraditional procurement pro-cedures. State level energy de-partments are good resources whenworking with a state agency as aclient. Whatever resources areavailable, use them to sell the ideaof nontraditional procurementrather than ask about it.l The evaluation process-

One of the reasons organizationsare leery of nontraditional pro-curement processes is the fear ofconflict-of-interest situations. In-clude a well thought-out evaluationprocess to allay this fear. Form anevaluation committee that includesthe building operating personneland at least one person from thedepartment in charge of pur-chasing for the project. This is agood way to eliminate worry thatsomeone may put something overon an organization.

In addition, a description of therange of features that will be underscrutiny in the evaluation processand the financial considerationsthat accompany each feature is avaluable tool in making the case fornontraditional procurement. Themore it is possible to put the evalu-

ation process into a formula, theeasier it will be to make the processfly. Of course, it is important to un-derstand that many of the itemsthat may distinguish one proposalfrom another cannot be listed be-fore the proposals are received.Still, that does not prevent the en-gineer from establishing broad cat-egories of possible differences andmaking an assignment of relativevalue to each category.

The introduction of high-tech sys-tems into mechanical designs in-creases the integration of mechanicaland other building design disciplines.Here, a new lighting interface panelhas been mounted below the lightingcircuit breaker panel to provide light-ing control through the DDC system.To be effective, the procurement ofthese high-tech systems must satisfythe requirements of each design dis-cipline.

l A realistic process-Everyclient organization has a person-ality complete with preconceptionsand prejudices that are usually theresult of its past experience-goodand bad. The interpretation ofthese experiences has often led or-ganizations to a particular way ofdoing things that can conflict withthe engineer’s vision of how to de-velop the procurement processmost effectively.

The engineer must be sensitive tothe organization’s view of the pro-curement process and help in thedevelopment of a procurementplan that will accommodate the or-ganization’s way of doing business.Only if harmony would compro-mise a satisfactory outcome shouldthe engineer challenge establishedprocedures, and if such challengesare necessary, they should be mini-mized so as not to cause the wholeprocess to be discarded.

The cornerstone of successfulprocurement of advanced technol-ogy systems is breaking those com-ponents out of the traditional bid-ding process. Depending on thetechnological system to be pur-chased and the organizational con-straints and views by which theowners’ procurement proceduresare governed, a successful pro-curement may be as simple as aprequalification of acceptable sys-tem configurations and compo-nents. There is no need to rock theorganization’s boat unduly if a verysimple procurement process that iscompatible with the organization’sgeneral business attitudes will suf-fice.

Step 3 - Develop specsWhen the process by which the

procurement will take place hasbeen determined, the engineermust translate the design of the ad-vanced technology system into aspecification document that meetsthe requirements of the chosen pro-cess. The following basic rulesshould apply to the specificationdevelopment process:l Design the system only

once-Earlier, I discussed the need

for the engineer to be as specific aspossible in selecting the advancedtechnology systems. This requiresthe engineer to be knowledgeableabout the technology to be em-ployed and provide a complete de-sign of the application (with alter-nates if necessary).

If the design is not complete, thecontractor will have to provide adesign as a part of his contract. Thecontractor’s design will require fullknowledge of the way the systemmust integrate into the other me-chanical system components, butthat information may never bemade completely available to thecontractor. Therefore, the second(contractor’s) design almost alwayshas defects that usually don’t cometo light until startup, at which timethey can cause enormous head-aches for the owner and operators.

To eliminate this problem, theengineer should assume completeresponsibility for each advancedtechnology system that is a part ofthe overall mechanical design. Ifthe engineer does not have the ex-pertise to provide the design, thenan independent outside sourceshould be found to help with thedesign, the specifications, and theprocurement.

It is important to remember thatthe design of the overall mechani-cal system is often the result ofmany months of information ex-changed among the eng ineer ,owner, and others in the designteam. It is not reasonable to expectthe system subcontractor to pickup all this information and assumeresponsibility for the design of thesystem.

0 Competitive flexibility-It isimportant for the engineer to pro-vide a complete design so that thepotential contractors know exactlywhat will be expected of them.Also, specifications should not beso specific as to favor one con-tractor’s equipment over anotherwithout valid reasons.

For example, DDC system pan-els have varying point capacitiesamong the various manufacturers.When specifying for a competitive

H EATING/ PIPING/ AIR C ONDITIONING H NOVEMBER 1991 41

High-tech systems

DDC procurement, our firm tries todetail each point and its interfaceand establish the level of control,communication, and operator in-terface required. But we do notspecify a point connection list foreach panel. If we were to do so,worthy systems might be elimi-nated simply because they incorpo-rate lower point densities at eachpanel. Or systems with high panelpoint capacities might be placed ata competitive disadvantage by re-quiring a low point density. Unlessthe point layout is important tosystem operation, it is left up toeach contractor to match his equip-ment to the points list most cost ef-fectively. Then the configurationbecomes an item of the evaluationprocess.

We try to develop specificationsso that as many products as possi-ble can be considered. Obviously,the more versatile the procurementprocess, the more flexible the speci-fications can be. But flexibilityshould not be confused with ambi-guity. No matter how flexible thespecifications are, they must veryclearly describe what is required ofthe system or component.

UncertaintiesAn owner once told me that in

discussions with his engineer on aproject, he questioned the engineerabout the lack of a points list forthe DDC system. The engineer re-plied, “Oh, we never include pointslists in our specifications. This waythe vendor must supply all pointsthat are necessary to make the sys-tem work. We never have to worryabout leaving any out.”

That revealing statement shouldcause engineers and owners alike topause and rethink the interests andduties in construction projects. Weall know that there are some uncer-tainties in most building systemdesigns. Those that are most ag-gressive in achieving high levels ofenergy efficiency, comfort, and airquality probably encompass some-what larger uncertainties becausemany of them utilize technologiesor strategies that have not been

I. I . System designn Does design meet functional requirements?

Does the design ensure the necessary system function will beachieved?

e Can the owner or user operate the system?

Are Drovisians! for otieratina: f6atureti i

3 High-tech system procurement checklist.

widely employed before. Pre-sumably owners who encourage orpermit higher performance designshave some understanding of theuncertainties involved.

However, the high-tech con-tractor is the least prepared to eval-uate and accept responsibility for aproject’s uncertainties. The con-tractor has not been privy to all thediscussions and decisions through-out the design process and verylikely does not understand thecomplete mechanical system de-sign. When the specifications arereleased, each contractor is (orshould be) wholly engrossed in agood faith effort to apply his prod-uct in the most suitable config-uration possible.

Putting the uncertainties of theproject on the contractor’s back re-sults in a costly redundant engi-

neering effort. At times it seemsthat the use of ambiguous pro-curement procedures is aimed atenticing contractors to under-estimate the scope of work and ob-tain lower prices. More frequently,however, I find ambiguous specifi-cations are the product of un-knowledgeable designers.

Whatever the reason for a speci-fication that does not clearly definethe scope of work for an advancedtechnology system or component, itis a risky procurement strategy be-cause the project and the owner arealmost always the real losers. Thestaffs of most contractors havebeen around long enough to makeprovisions for risk. Their bids orproposals respond to ambiguitywith ambiguity. I have heard ven-dors state that if the engineer re-quires them to provide Part A,

42 HEATING/PIPING/AIR CONDITIONINGVC a

it to the parties to resolve pay is-sues, under the guidelines of thefacts found in the ‘i l lustrativecases,’ with the parties retaining theright to seek additional arbitration‘if resolution is not reached.’

“Based on the arbitrator’s opin-ions and the provisional nature ofthe remedy framed by him, thejudge properly could have con-cluded that the availability offunds was not directly an issue inthe arbitration beyond a generaland unsupported allegation thatfunds might not exist. Thus, thejudge could have decided that iflack of money had been a critical is-sue , the a rb i t r a to r ’ s op in ionsplaced the burden on the secretary

, . . . . . . . . . .

to show that judicial proceedingswere necessa ry to de te rminewhether the award would not beenforced in any respect because oflack of funds. . . . The secretarymade no such showing. In the cir-cumstances, the judge acted cor-rectly in allowing the union’s mo-tion for summary judgment.”

Accordingly, the judgment wasaffirmed. Thus, the engineers andscientists were entitled to addi-tional compensation even thoughthere might not be funds availableimmediately to pay for it. The bur-den, however, would be on the stateto show that it had no availablefunds and probably would lead tofurther litigation between the par-ties. D

High- tech sys terns

which they believe was not requiredby the specifications, they intendto respond by downscaling or omit-ting Parts B, C, or D, whose exactrequirements are obscure by theambiguous nature of the specifica-tions.

What those conducting the pro-cess must consider is that so long asthe requirements of the advancedtechnology system are realistic,then the more complete the specifi-cations are in describing level ofperformance and other require-ments of the system to be pur-chased, the more likely supplierswill provide competitive proposals.The specifications must also pro-vide sufficient latitude so thatseveral potential suppliers can offercost-effective proposals. The engi-neer must stay up to date with eachtechnology of the design and pos-sess sufficient knowledge of therange of products available to en-sure that the balance between de-sired features and opportunity formultiple vendors is maintained.

While a great many engineersstill use simple bidding proceduresto purchase controls and other ad-vanced technology building sys-tems, the bidding process can re-sult in costlier systems that fail tooperate up to the owner’s ex-pectat ions. Placing a project’shigh-tech systems and componentsin a special category and procuringthem separately from the generalcontract is an approach that hasproved useful to the owners and en-gineers who use it regularly.

Fig. 3 illustrates the three basicsteps that our firm has found to beeffective for successful high-techsystem procurement. The ques-tions are used as a checklist to en-sure that the resulting systemmeets the demands of the projectand will he procured effectively.Seeing that these questions are an-swered positively and attention isp rov ided to the spec ia l p ro -curement process will result in im-proved performance and cost ofeach project’s high-tech systems. 11

. . . . . . . . . . . . . . . . . . . . .

HVAC ENGINEER IImmediate position for project engineerwith BS degree and 4-5 years experiencein HVAC. Plumbing and Fire Protection.Send resume to: SSR Engineers. P.O. Box471, East Grand Forks, MN 56721. I

ENERGY ENGINEER. Dynamic, fast growing. pub-licly traded energy services company seeks highlyskllkd engwers and technicians for all phases of en-ergy work, from proposal wutmg to engra?ring designand construction management. Must be Independent.resourcelul. innovatwe and determlned CandIdalesreply to Engvwering Manager, P.0 Box 41156. Sacra-mento. CA 95841.

375 Rantan Center Pkwy, Edison, NJ 08837

. Gamer, Trane. York

. Sale, Rent. Lease-New and Recondltloned ’ Absorbers Centr,fugals

3346s. Pulaskl.Chlcaao. IL60623

QUALIFIED CANDIDATESRecruit the Best andthe Brightest Talent

with Penton Classifieds.

86 NOVEMBER 1991 w AI R HEATING/PIP CONDITIONING

.,........................................................NEW GENERATION DDC

PROMISING CONTROLINNOVATIONS WITH NEW

G ENERATION DDCThe new generation of DDC systems now available fulfills past promises of

enormous improvements in building comfort and operating efficiency

By THOMAS HARTMAN, PE,The Hartman Co.,Seattle, Wash.

0

romises don’t usuallyD mean much in an elec-

tion year. But for our in-dustry, this may be the

year that some long-standingpromises of the newest direct digi-tal control (DDC) systems finallybegin to be realized. After years oftalk about an impending revolu-tion in the way buildings will op-erate, the new generation of DDCsystems that was to have sparkedthe revolution has arrived. Butthe accompanying control revolu-tion has been noticeably absent.Instead, the new DDC systemsare being applied to operate in thesame manner as systems of thepast.

Now, developments are start-ing to paint a very encouragingpicture of realistic opportunitiespossible with these new, morefunctional DDC systems. Experi-ence has now been gained withsome exciting new control con-cepts and approaches. It is timeto consider these new approachesas candidates for current pro-jects. The following is my shortlist of design innovations madepossible by the recent advances incontrol and related technologies

that we should all be considering.l Keep buildings running-

With proper control algorithms,keeping the air moving in build-ings at low flow rates during unoc-cupied hours (using variable fre-quency drives) can, in manybuildings, save far more energythan the small extra fan energy itcosts. Reducing infiltration anddistributing warm core air to theperimeter reduces heating re-quirements during cold weather.

Purging the building with coolnight air saves cooling energy inwarm weather. In addition to en-ergy savings, keeping air movingwithout a large night setback hasshown remarkable improvementin the comfort and quality of thebuilding environment.l Integrate lighting with HVAC

control-Integrating the lightingcontrol into the DDC system per-mits the establishment of space-by-space occupancy conditions so

gg 8080

$$a a 7070tt3 3 605:.gl 60

+i 405 30

20

10

Inlet vane

Variable speedl .

drive TRAVl . . _

l *-4

-* 0.--._

Obh-100

-1 Fan power vs. air flow for a centrifugal fan with three methods of VAV con-trol. See text for definition of TRAV.

HEATING/PIPIN GHEATING/PIPING /AIR /AIR CONDITIONINGMCONDITIONING - N NOVEMBER OVEMBER 19921992 55

New generation DDC

that temperature control and out-side air ventilation can be sup-plied only to the zones that are ac-tually occupied. First costs forintegrated HVAC and lightingcontrol are often less than those ofdesigns incorporating separatecontrol systems.l Install a space temperature

sensor in every office - When a ter-minal box supplies multiple of-fices or areas, using multiplespace temperature sensors for boxcontrol can provide an enormousimprovement in comfort over thesingle sensor concept. Simple av-eraging logic combined with occu-pancy signals can increase com-fort in each office and improve theenergy performance of the me-chanical system.l Eliminate VAV box maxi-

mum and minimum air flows-Comfort, air quality, and energyefficiency can all be improved bybetter air flow control strategiesat terminal boxes. Controlling theair flow at each terminal boxbased on actual needs for ventila-tion air and cooling is a more ef-fective control technique thanstrategies that simply regulatezone air flow between preset min-imum and maximum box airflows.l Use the DDC operators’ con-

sole as an electronic drawingfile-As-built CAD drawings incolor are far easier to follow thanblueprints. Installing a CAD sys-tem in the DDC host computer asa reference, and for updating ten-ant changes of mechanical draw-ings, offers an effective manage-ment tool to operate the buildingmore effectively and keep thebuilding record documents cur-rent.

Each of these innovations rep-resents a substantial break withtraditional practices. My experi-ence with initial implementationsof each of these strategies is verypositive. All appear to have greatpromise when incorporated aspart of a careful controls design.What follows is a more detaileddiscussion of each.

Continuous fan operationTraditional thinking concern-

ing energy conservation in build-ings has focused on keeping theHVAC system off when the build-ing is unoccupied. This widely ac-cepted strategy compromises thecomfort and air quality for thosewho occupy buildings beyond reg-ular business hours. In today’scommercial and institutionalbuildings, the concept of buildingshutdown is rapidly losing its at-tractiveness because of the occu-pants’ demands for more flexibil-ity, better comfort, and improvedair quality. This trend is wellknown to building operators. Atrend that is not immediately ap-parent is that building shutdownis also becoming obsolete as a re-sult of more efficient building en-velope technologies.

In older buildings with poorlyinsulated envelopes, turningHVAC systems off was about alloperators could do to reduce en-ergy use. But the development ofmore effective building envelopesis changing the pattern of energyuse in buildings. Today’s build-ings have a much higher ratio ofbuilding mass to envelope lossthan older buildings. This devel-opment has made the thermal in-ertia of buildings a significantcomponent in mechanical systemsizing and building operation. Atthe same time, the higher resis-tance to heat flow of modern en-velopes has decreased the impactof conduction heat flow on theoverall building energy use.

For many buildings, it has beenshown to be more economical tomaintain indoor comfort condi-tions during low or unoccupiedhours than to shut the systemcompletely down. Maintainingmore constant space temperatureconditions reduces peak buildingstartup energy requirements.Overall fan energy can be reducedwith 24-hr operation because fansoperate at high speeds for fewerhours. Operating a fan for 24 hrpermits energy exchange withinthe building or with other systems

at significantly higher heat trans-fer efficiencies. In similar fashion,carefully designed heating andcooling equipment can also bemade to operate at higher efficien-cies during low load conditions.

The key to achieving energy re-ductions by implementing 24-hrHVAC operation is to develop anefficient building envelope systemand high part-load efficiencies forall the HVAC transfer and energyconversion systems. Fig. 1 showsthat approximately 2 percent ofthe design energy is required tooperate a VAV fan at 30 percent ofdesign air flow when controlled bya variable frequency drive (VFD)under a scheme that incorporatesterminal-regulated air volume(TRAV).* Supply fan output isgoverned by real-time terminalbox air flow requirements ratherthan the need to meet a ductstatic pressure set point. Continu-ous air system operation can beused to reduce infiltration, purgethe building with cool outside aird u r i n g a n t i c i p a t e d w a r mweather, or transfer the heat inthe building core to the perimeterareas in cold weather. Continuousoperation strategies also permitdown-sizing heating and coolingequipment. The continuous opera-tion strategy has an enormouslypositive impact on comfort and airquality in buildings during nor-mal work hours and an evengreater one during the off hoursthat building occupants increas-ingly wish to work.

Lighting and HVAC controlIn the past, building controls

have required the building opera-tor or mechanic to establish oneoccupied hour schedule for thebuilding’s HVAC system and asecond for lighting control be-cause these control functions wereusually provided by separate sys-tems. Lighting control has tradi-tionally been accomplished with

*See “TRAV-A New HVAC Con-cept,” by T. Hartman, HPAC, July1989, pp. 69- 73.

5656 NNOVEMBER OVEMBER PIPING / AIR CONDITIONING

2 Program for averaging multiple space temperature sensors.

simple time sweeps. When theDDC system is extended through-out the building to each terminalVAV box, lighting control can eas-ily be added to the system at lowadditional cost. Lighting andHVAC control integration offersexcellent opportunities throughsynergism to improve the perfor-mance and efficiencies of both sys-tems. The expanded logic capabil-ities of new technology DDCsystems can offer lighting controlschemes that are more effective inaccommodating occupant needs,yet result in fewer lighting operat-ing hours than other strategies.Furthermore, monitoring occu-

pancy on a zone-by-zone basis pro-vides an opportunity for improvedcomfort and HVAC energy sav-ings by allowing the system to fo-cus heating, cooling, and ventila-tion on those areas of the buildingthat are occupied.

The most obvious cost benefit ofcombining lighting and HVACcontrol is that the lights indicatethe occupancy status of eachzone-no other indication is re-quired. Savings benefits stemfrom the fact that modern build-ings are rarely 100 percent occu-pied. Establishing office-by-officeoccupancy conditions can limitcomfort conditioning to the areas

that are actually occupied.Distributed DDC control of

lights can also be more effectivethan traditional lighting controlstrategies. Most lighting controlschemes use centralized lightingpanels with “lighting sweeps” thatturn off large blocks of lights atspecific times. Under DDC opera-tion, the lighting zones are typi-cally much smaller (a single en-closed office is one zone), and acontrol relay is located at eachzone. Lighting sweeps duringnonbusiness hours are absent. In-stead, each zone’s lights operateindividually.

Pushbuttons or occupancy sen-sors, or both, can be employed todetermine building occupancy ona zone-by-zone basis. Occupancysensor technology has advancedrapidly in the last few years. Theperformance of these devices hasimproved substantially while themanufacturing costs have contin-ued to decrease. The two majordetection methods, infrared andultrasonic, are both effective indetecting occupancy at the workplace, and each has certain ad-vantages in particular applica-tions. The combination of ad-v a n c e d D D C s y s t e m l o g i ccapabilities and occupancy sensorimprovements means that fewersensors may be adequate to moni-tor open office applications. How-ever, it is very important that oc-cupancy sensing designs beconservatively developed becausesmall difficulties can result invery negative perceptions by theoccupants.

The most economical occupancysensing method is a low-voltagepushbutton similar to those em-ployed in low-voltage lighting sys-tems. The pushbutton may be in-tegral with each spacetemperature sensor, located sepa-rately, or a combination of both.When the occupant arrives andpushes the button, the DDC sys-tem is alerted to the occupied con-dition. The lights in the area areturned on, and the HVAC termi-nal unit(s) is switched to the occu-

PIPING/AIRCONDITIONING~NOVEMBER HE AT ING PIPING/AIR CONDITIONING - NOVEMBER 1992 1992 57

New generation DDC

pied mode, assuring outside venti-lation air is delivered to the zone.

When the zone is occupied, thespace temperature is controlledwithin tighter limits. The logic ofthe DDC system determines howlong the space will assume occu-pancy, depending on the time ofday and whether it is a weekday,weekend, or holiday. Generally,

3 VAV terminal control strategy em-ploying proportional relationship be-tween temperature and air flow.

the conclusion of occupancy is pre-ceded by a brief flash of the lightsthat reminds the occupant it isnecessary to press the button tocontinue the occupied state. Apress of the button at that time re-sets the occupancy timer to thepreprogrammed override time.Pressing the occupancy buttonany time before the end of occu-pancy with the lights on alerts thesystem that the occupant is leav-ing. The lights shut off, and theHVAC system reverts to the unoc-cupied control state until thepushbutton is pressed again.

Multiple space sensorsFor some years, the industry

has accepted the limitation of us-ing only one temperature devicefor each terminal box no matterhow many separate offices orspaces that box served. This mayhave been an acceptable practicein the past, but it is now out ofdate. The comfort of all areasserved by a single terminal boxcan be improved by installing aninexpensive electronic tempera-ture sensor in each area servedand using the math and logic ca-

pabilities of the DDC system tocontrol the box.

I recommend the control logicemploy a “weighted-average”technique. The weighted-averageapproach controls the box accord-ing to the average space tempera-ture of the occupied offices. How-ever, if the temperature of anyoffice or area served by the boxrises or falls beyond the heating orcooling set points, the weightingof those offices in the averagingcalculation is increased. Fig. 2 il-lustrates a simple weighted-aver-age program for a VAV box thatserves three offices. The resultingweighted-average temperature(called BOX_TEMP in Fig. 2) isused to establish the air flow setpoint for the box. This simple buteffective strategy can signifi-cantly improve the comfort of amultiple office arrangement. It isnot costly because space tempera-ture sensors are among the leastcostly devices available for DDCsystems.

VAV box maximum/minimumUntil recently, nearly all VAV

box control strategies employed aproportional relationship betweenspace temperature and air flow,limited by a box minimum air flowand a box maximum air flow. Thisrelationship is shown in Fig. 3.VAV systems have been correctlycriticized for their failure to guar-antee minimum outside ventila-tion air. If the box is at the mini-mum flow, the outside airavailable for ventilation dependson the percent of outside air in thesupply air, which is a variablewith systems that incorporate air-side economizers. The amount ofoutside air required for ventila-tion may also be variable.

On the other hand, the boxmaximum design air flow (whichis usually entered as the box max-imum) is not properly used inmost applications. If a space iswell above the space temperatureset point, there is rarely any com-pelling reason to limit the air flow

4 Improved VAV terminal control strategy possible with DDC control.

5858 NOVEMBERNOVEMBER 1992 - HEATING / PIPING / AIR CONDITIONING

to this calculated value if more airis available.

With new DDC technologies,adding separate set points forheating and cooling along with aweather-calculated air flow forthe range between the two setpoints is now possible and pro-vides benefits in comfort, energyperformance, and air quality. Irecommend the VAV box controlscheme shown in Fig. 4 be used toreplace the old pneumatic controlconcept of Fig. 3.

In Fig. 4, there is no fixed boxminimum or maximum air flow.The minimum air flow at spacetemperatures below the heatingset point is what at the time pro-vides the correct amount of out-side air ventilation to the zone(s)served by box based on current oc-cupancy conditions. The actual airflow is calculated from the currentoccupancy conditions (number ofpeople in the zones served by thebox) and the percent of outside airin the supply air stream.

Between the heating and cool-ing space temperature set points,the air flow set point for the box isbased on the space temperatureand the projected outside weatherconditions. In warm weather, theair flow set point is high duringeconomizer operation to maintainthe space temperature at thelower end of the range in anticipa-tion of warmer weather ahead. Incold weather, the air flow setpoint remains low while the spacetemperature is between the heat-ing and cooling set points to allowthe space temperature to rise inanticipation of heating require-ments.

Above the cooling set point, thebox air flow set point rises rapidlyto the design cooling air flow,where it remains level for an ap-proximate 0.3 F space tempera-ture increase. At about 0.5 Fabove the cooling set point, the airflow set point is raised to ensurethe damper is fully open to pro-vide all possible cooling and pre-vent further temperature in-creases.

This newly developed box con-trol strategy has been very effec-tive. It provides required mini-mum ventilation at all timesduring occupancy and maintainsmore constant comfort conditions.

CAD and the DDC consoleOne of the widely recognized

commissioning issues for build-ings today is the need to get gooddocumentation and otherdatabase management tools intothe hands of the operations staff.The term database is used here toinclude as-built drawings, build-ing system information, controlsoperation strategies, and docu-mentation for the thousands ofDDC points that make up theDDC system. Current buildingdatabase management tech-niques already include the use ofgraphic menus and text file accessthat permits operators to accesscontrol sequence programs anddescriptions of operating intentfrom the DDC system operator’sconsole screen.

The concept of access to docu-mentation through the DDC sys-tem operator’s console can now beextended to include the mechani-cal and control drawings as well.In this concept, the building’s me-chanical, electrical, and controlCAD drawing files are included onthe DDC system operator’s con-sole disk. A file management pro-gram that permits the operator topage through the drawings muchthe same as if the drawings werestacked on the operator’s desk isinstalled. In addition, I recom-mend a support concept that givesthe operations staff the responsi-bility of maintaining up-to-dateas-built drawing and databasematerials. I also recommend theinstallation of the CAD editor pro-gram in one of the DDC systemoperator’s consoles so that the in-evitable ongoing changes in parti-tions, temperature sensors, dif-f u s e r s , l i g h t s , a n d ( m o s timportant) heat loads can benoted on drawings and listings.The operator is responsible for up-

dating these changes and seeingthat the required balancingchanges are made to maintaincomfortable and efficient systemoperation.

Getting startedThe new generation of DDC sys-

tems now available holds thepromise of enormous improve-ments in building comfort and op-erating efficiency. But like politi-cal promises, an overabundance ofpast pronouncements about DDCsystems has already dulled the re-ceptivity of our industry to suchpromises.

Fortunately, improving thecomfort and efficiency standardsof a building’s operation is a loteasier than resolving nationalproblems. Concerned buildingowners and operators can them-selves pursue these opportunitieswithout depending on manufac-turers’ promises. The industryhas started to learn that effectivebuilding operation has more to dowith how well mechanical, electri-cal, and control systems are all in-tegrated into a sound design con-cept than with specifics of thesystems themselves. Well-trainedbuilding operators together withan experienced and knowledge-able engineering group form theteam that can make these new,ambitious opportunities cometrue.

Successful building owners andoperators already know that theoptimum strategy for developing,constructing, and operating build-ings requires a motivated teamwith strong leadership. Those areprecisely the management toolsrequired to implement the kindsof innovations listed above. Wecan make buildings that are eas-ier to operate and operate themmore effectively. To do so requireshard work, a solid team-orientedproject organization, and a strongcommitment to technical excel-lence. But the result will be worththe effort, and we don’t have to de-pend on promises to make it allhappen. 0

HEATING /PIPING/PIPING /AIR/AIR CONDITIONING~NOVEMBERCONDITIONING~NOVEMBER 1992 1992 59

. . ..L.. . . . . . . . . . . . . . . . . . . . . . . . . . . . ..*......................

INDIVIDUAL TERMINAL CONTROL

NEW ZONE CONTROLSHELP ACHIEVE TOTAL

ENVIRONMENTAL QUALITY

Increasing demandsfor total environmental

quality by buildingoccupants are

changing HVAC designcriteria from uniform toindividually adjustable

zone conditions

By THOMAS HARTMAN, PE,The Hartman Co.,Marysville, Wash.

veloping regarding lightingcontrol. Our project managerwas explaining to the clientthat dimming lighting 1ballasts allowed us toinclude lighting controlas part of our comfortcontrol strategy. In addi-tion to compensating forchanges in outdoor light, he ex-plained that our lighting con-trol strategy includes furtheradjustment of lighting levelsdepending on room space tem-peratures and weather. In-creasing the lighting levels in a

cool room in cold weather mayimprove the occupant’s senseof comfort. At the same time,the heat generated by thelights is helpful to the heatingeffort. Similarly, decreasingthe lighting level under warmconditions provides help forthe sense of comfort and thecooling effort of the space.

The client’s engineer tooksome exception to this ap-proach even though it wassimply a supplemental stepthat was not intended to af-fect traditional temperaturecontrol strategies directly.We did note, however, that itcould evolve into a strategythat permitted a slightlywider range between heatingand cooling set points.

As I listened to the discus-sion, I realized that the

client feared this ap-proach would reduce

. the building opera-

r tor’s ability to illus-trate to occupants that

the building is comfortablebecause the simple fixed cri-terion for comfort-spacetemperature-would quitepossibly no longer be the ab-solute determination of com-fort. I thought about a scene Ihave experienced many timesin which a building operator

HEATING/PIPING/AIR HEATING/PIPING/AIR CONDITIONING - NOVEMBER 1993 43

Individual terminal control

or engineer argues with an occu-pant that he or she must be com-fortable because the space tem-perature is right on set point! Iwas reminded that despite thelimited understanding our build-ing design community has for theelements of human comfort, wecontinue to try to express comfortin simplified (and unrealistic)terms so we can get on with whatmany believe to be the real job ofHVAC engineers-designing/op-erating/maintaining chillers, airhandlers, and boilers.

Those who continue to view ourindustry in these traditionalterms need to understand that anenormous change is taking placein the minds of individuals whoown and occupy the buildings wedesign and construct. Thischange is being propelled by theisolated but highly visible un-healthy buildings we all haveheard about. Building ownersand occupants are coming to be-lieve that the industry’s empha-sis on very simple comfort and airquality criteria is insufficient tomeet the needs of building occu-pants. Occupants no longer be-lieve that being shown that theirspace is within acceptable tem-perature limits and the outsideair dampers at some minimumposition are acceptable assur-ances of a comfortable andhealthy building. Furthermore,owners understand that suchsimplified building design crite-ria do not adequately addresstheir concerns over liability.

Fortunately, as the individualswho operate and occupy buildingsare beginning to hold our industrymore accountable for keepingthem comfortable and healthy, anastounding array of economicalnew comfort control techniquesare finding their way to the mar-ketplace. What is needed now isan understanding of how theserapidly evolving changes are af-fecting the basic criteria for build-ing design and how to applyequipment now becoming avail-

able to satisfy these changes mosteffectively and most economically.

New challengesFor years, building design engi-

neers have assumed a simple setof criteria to establish comfort; asingle and uniform set point forinterior space temperature forcomfort and a minimum outsideair damper position or opening forair quality. In particularly humidor cold climates, the criteria oftenincluded limits on humidity aswell. To develop an HVAC systemto meet these simple criteria, de-signers assumed maximum andminimum outdoor temperatureand humidity conditions and max-imum solar conditions. That’sabout all there was to sizing thecomponents of an HVAC system.

The industry has long beenaware that space temperaturesensing by itself is not an effectivemeans for accurately assessingcomfort. Even if it were, typicalair-based HVAC systems operatewith large air temperature gradi-ents within each space that makeit very difficult to assess overallthermal comfort conditions from apoint temperature within thespace. Compounding the problemhas been the well-known fact thatin typical office environments,comfort is substantially affectedby radiant heat exchange betweenbuilding occupants and surround-ing surfaces as well as the amountof air movement.

The practical aspects of ensur-ing air quality have also beenlargely ignored even as the poten-tial liability for doing so has sky-rocketed over the last few years.In low-rise constant volume sys-

TABLE TABLE 1 - Total environmental quality(TEQ) factors.

Space temperatureWall, window, and ceiling surface temperaturesSpace humidityAir movementOutside air ventilation rate (dilution)Light level

terns, a prebalanced damper posi-tion or inlet is a reasonablemethod of ensuring a certainamount of outside air enters thebuilding, but this approach hasbeen extended to high-rise build-ings and VAV systems eventhough it is very clear that it can-not be so applied. Furthermore,the integrity of outside air distri-bution to individual zones inbuildings has been severely un-dermined in many VAV designs.

Now, building designers arelearning that a number of otherenvironmental factors also needto be considered when developingcomfort and air quality criteria.The term total environmentalquality (TEQ) has been coined todevelop comfort and air qualitystandards based on the largernumber of factors known to con-tribute to an occupant’s sense ofwell being. My list of TEQ itemsdesigners should be consideringin their designs is provided inTable 1.

It is reasonably well known thatthese (and other less definitive orcontrollable factors) interact to es-tablish comfort and air quality forbuilding occupants. Exactly howall these factors interact is notwell understood. Furthermore,the occupant’s sense of well beingis a very subjective determinationthat depends not only on clothing,building materials, and levels ofexertion but also on a number ofcultural and individual criteriathat, at least for the present, arefar too nebulous to define withprecision.

Meeting the challengeThe large number of factors

that affect TEQ, together with ourgrowing knowledge about thevariations among individuals, cre-ates a complex web of informationthat is not likely to be adequatelyunderstood anytime soon enoughfor even the most sophisticatedcontrol system to anticipate whatis required to provide TEQ to eachbuilding occupant.

4 4 G

2 Combined VAV box controller and box damper actuator.

The natural conclusion fromthese facts is clear: The standardfor establishing building comfortmust change from a set of uniformbuilding conditions to a range ofconditions within which each zonecan be individually adjusted towhatever is desired by the occu-pant(s).

In other words, it is not thebuilding owner, the design engi-neer, or regulations or standardsthat can define comfort and asense of well being. It is the build-ing occupant. Fig. 1 illustrates theITC concept. In this figure, abuilding occupant is able to com-municate desires for changes inlighting, temperature, and out-door air ventilation levels from hisor her workstation. Equipmentand devices are now becomingavailable that afford this type ofoperator interaction with theHVAC system. To satisfy buildingoccupants, designers must beginto incorporate individual terminalcontrol (ITC) features in their me-chanical system designs, This un-avoidable conclusion is a reversalof the direction the industry hasgenerally been traveling for thelast several decades. The occu-pant has too often been seen as

the culprit in comfort, air quality,and energy use problems. Remov-ing operable windows and in-stalling locking thermostat covershave been very popular in thistime, and the advent of digitalcontrols at the terminal box hasoften eliminated any capacity foroccupant adjustment altogether.

In some instances, there havebeen valid reasons for disablingdirect occupant control, but theprospect of true individual controlwith a high-performance DDCsystem provides a formidablechallenge to the assumptions thathave guided such decisions in thepast. For example, building opera-tors are frustrated when the ad-justment of a thermostat in oneoffice leaves those in adjacent of-fices (connected to the same ter-minal unit) uncomfortable. Trueindividual control eliminates thisconcern by offering each and ev-ery office or work area the capac-ity for an individually regulatedenvironment.

Another concern of building op-erators is the possibility of exces-sive energy use or the inability toexert control over energy use withITC schemes. The mechanisms bywhich high-performance control

systems can address this concerndeserve some discussion. Simplystated, in a high-performance con-trol environment, building occu-pants express their desiredchanges to the building controlsystem rather than directly over-riding its programmed function.This permits the building controlsystem to determine the most ap-propriate (and efficient) means inwhich to act upon each request.

Applying ITC principlesHow successful a building de-

sign team is in meeting the newchallenges of building total envi-ronmental quality dependslargely on whether this challengeis viewed as an opportunity or aburden. For building owners whodo view these changes as an op-portunity, the benefits includemore satisfied (and perhaps more)tenants, less time required bybuilding operations staff to re-spond to complaints, and signifi-cantly lower energy usage.

Providing individual terminalcontrol has a number of pitfallsdesigners need to avoid, and if thedesigner is not careful, some ex-isting terminal control productsmay lead the design in the wrongdirection. The following are sev-eral key concepts that must be ap-plied in developing ITC systems.l The ITC system must have the

capability to reset occupant ad-justments automatically to a neu-tral or near neutral state fromtime to time.

A number of problems with per-mitting adjustment by occupantscan be traced to a temporary ad-justment that is forgotten by theoccupant. For example, imaginethat a portion of the cooling planthas failed in a building on a warmafternoon. If an ITC system is pre-sent, it is likely that many occu-pants will call for cooler space con-ditions because of the failure.However, if the failure is cor-rected that evening, and no auto-matic reset of the adjustments ismade, the occupants will likely ex-

Individual terminal control

perience discomfort thefollowing day becausemany will have forgottentheir adjustment the daybefore.l The ITC system

should not provide atten-tion-diverting informa-tion to the occupant.

Since the advent of di-rect digital controls, Ihave too often heard ar-guments between build-ing technicians and occu-pants about the accuracyof the space temperaturesensors employed by thesystem. Rather than en-courage the occupant toset specific space temper-ature set points and pro-vide the means for him orher to check the system’sability to meet that tar-get, our experience is thatmore effective ITC resultsfrom permitting the oper-ator to reduce less quanti-fied adjustment requests.Warmer/cooler, brighter/darker, and ventilate/re-circulate are probablymore effective ITC crite-ria than requesting spe-c i f i c set po ints fortemperature, light, and

360 degoccupancy sensor

7

Pendant mountedtemperature sensor

3 Combined space temperature and occupancy sensors.

Space temperature,occupancy, and

adjustment parametersto/from DDC system

-7

occupancy sensorLight display indicating

adjustment mode

Pendant mountedtemperature sensor remote control

sensor/decoder

4 Comfort control module combining space temperatureand occupancy sensing with infrared wireless remote re-ceptor/decoder and lighted display.

ventilation levels from the occu- must also be a prompt perceptiblepant. response.

Space temperature and

l When an individual makes anadjustment, some acknowledg-ment of the action must be pro-vided immediately.

When an occupant adjustmentis made, some signal must ac-knowledge the adjustment. Sometype of display to show that thesystem has received the adjust-ment request is essential.l Each adjustment must lead to

a perceptible change in the direc-tion of the intended adjustment.

Exactly how the system will re-act to each adjustment requestwill depend on the circumstancesat the time of the request. How-ever, it is imperative that in addi-tion to an acknowledgment, there

Putting the elements describedabove into a simple and effectiveITC design is becoming easier andless expensive with the advent ofnew products appearing on themarket today. The Fig. 1 illustra-tion of an individual adjustinglighting, temperature, or outsideair ventilation levels from his orher workstation is rapidly becom-ing a reality. The key to theserapid development efforts ap-pears to be the economics of inte-gration. A number of manufactur-ers are combining several devicesor components into single unitsthat cost no more than single de-

vices alone. In addition toproduct cost savings, theimpact on the total pro-ject cost is often dramati-cally improved because ofthe substantial installa-tion cost savings fromfewer units to install.Figs. 2 through 4 illus-trate the trend toward in-tegrated DDC devices.

Fig. 2 is a damper mo-tor upon which the manu-facturer has mounted aVAV box controller mod-ule. In addition to thededicated air velocity in-put and the damper mo-tor modulating output,the module includes anadditional three inputsand three outputs, eachof which can be pro-grammed as either a digi-tal or analog type. Thecontroller also includesfully programmable con-trol language, PID con-troller blocks, and sched-ule and display features.This motor/controller de-vice is now available andwill soon be released byseveral additional manu-facturers. Where pricingsheets have been made

available, the price of this unit isless than the price of VAV con-troller modules alone severalyears ago.

Fig. 3 is an occupancy sensorthat has been integrated with apendant type temperature sensor.This unit has recently becomeavailable and provides two impor-tant zone inputs to a DDC systemwith a very simple and low-costinstallation, The method of occu-pancy sensing for this particularunit is passive infrared, thoughultrasonic may soon be available.The cost of the unit is about thesame as the cost of an occupancysensor alone.

Fig. 4 is an enhanced version ofthe Fig. 3 device with several addi-

continued on page 92

46 NovEMBER~~~~~HEATING/PIPING/AIRNOVEMBER 1993 - HEATING/PIPING/AIR C CONDITIONINONDITIONINGG

Individual terminal control

continuedfrompage46continued from page 46

tional features. This device is nowunder development by at least onemanufacturer and will soon beavailable. In addition to the tem-perature and occupancy sensor,this device also has an infraredsensor that receives and decodeswireless remote transmitters ofthe same type used to adjust televi-sion and VCR units, For this prod-uct, the hand-held remote unit per-mits the occupant to requestwarmer or cooler space conditions,brighter or dimmer lighting, andmore or less outside air ventila-tion. The integration of this unitwith standard VAV components toeffect economical individual termi-nal control is shown in Fig. 5.

Notice that all components inFig. 5 are ceiling mounted, whichsubstantially reduces the installa-tion costs and increases the easeof adjustment to meet tenant

space changes. Our firm’s esti-mate is that the cost for the Fig. 5configuration will be substan-tially less than the present cost ofa DDC box control installation us-ing standard VAV control tech-niques. However, this does notmean that ITC will be less expen-sive to implement than currentDDC VAV control because ITCgenerally requires substantiallymore zones than conventionalVAV systems.

Under an ITC concept, each of-fice and open area (forming a“cluster” of occupants) requires aseparate VAV terminal unit. Intypical office buildings, this repre-sents an increase in VAV boxcount of between two and threetimes. Even though the per unitcost of each VAV zone can be re-duced with ITC concepts, the totalcost in all but entirely open officeconfigurations will be greater.

However, the benefits of this ap-proach include increased occu-pant productivity. A recent studyat the West Bend Mutual Insur-ance Co. showed that an individ-ual control concept improvedworker productivity at levels ofbetween 2 and 6 percent. Theseeconomies alone justify the costfor incorporating individual con-trol into VAV systems.

New ITC conceptsWe have discussed the transfor-

mation of typical VAV system con-figuration to ITC, but it is impor-tant to realize that as old HVACdesign concepts based on uniformfixed building environmentsevolve over the next few years tothe individually controllablebuilding environment, the ITCconcept will exert a strong influ-ence on the remainder of theHVAC system. New and more effi-

Asmuchhotwaterasyouneed-upto4,800gal/hr.Instantly. Armstrong’snew Flo-Rite-Temp instantaneous steam water heater makes itsimple. Ji~tAddWter: Becausewe'vedoneabout everythingelse.Like designing outallthethingsthateatup space, jackup costs and giveyour maintenance crew fits.The Flo-Rite-Temp heatsitas youneeditwith__ ^

nobulky,leakystoragetank.Nothermostaticcontroltorupture.No retrofithassles.Andit

92 92 NOVEMBERI~~~~HEAT~NG~P~P~NG/AIRNOVEMBERI~~~~HEAT~NG~P~P~NG/AIR C CONDITIONINGONDITIONING

DDC zone controller

Comfort control module

cient mechanisms for deliveringair to individual zones are being

surface temperature, humidity,

developed, and traditional VAVventilation air, air movement,and light intensity. When an occu-

concepts of zone temperature con-trol are likely to be replaced with

pant requests a change, the con-trol system will be able to make

new strategies that include a con-nection between air temperature,

complete and prompt response byadjusting the matrix of compo-

5 Individual terminal control (ITC)employing standard VAV and a com-fort control module.

nents rather than a single iso-lated factor.

Summary and conclusionThe increasing demands for to-

tal environmental quality bybuilding occupants are transform-ing HVAC design criteria fromproviding uniform building condi-tions to offering individually ad-justable zone conditions. DDCcomponents becoming availableare integrating previously sepa-rate functions and offering capa-bilities suitable for individualizedcontrol at low cost. Designersshould begin to develop ITC con-figurations for their projects asthese products are becomingavailable. n

fails closed (cold) so there’s no danger of scald-ing. Lower installed cost. Minimal maintenanceand very very little space. So circle the number belowor ask your Armstrong Representative for a free

High-maintenancefeedback systemswith bulky tankunits like this oneoften leak, corrodeor rupture athermostaticcontrol,

The ArmstrongF/o-Rite- Temp caneasily do the workof a storage tankunit many times itssize -at lowerinstalled cost andwith minimummaintenance.

Circle 366 on Reader Service CardHEATING / PIPING /AIR CONDITIONING w NOVEMBER 1993 93

By THOMAS HARTMAN, PE,The Hartman Co.,Marysville, Wash.

scar Wilde, a cynicalobserver of modernmankind’s follies, onceobserved “When the

gods wish to punish us, they an-swer our prayers.” I fear this in-sightful commentary is soon to bevisited upon the building con-struction industry. For someyears, I have heard my colleaguescarp about the lack of progress to-ward a standard protocol for DDCsystems. Well, soon such stan-dards will exist, and for many itmay well become more of a pun-ishment than a blessing!

Those with hands-on DDC sys-tems experience must shudder atthe thought of having multipleDDC systems on a single network.While a standard in communica-tion may lead to a uniform mecha-nism of displaying data and issu-ing commands to individualsystem points, operators will stillhave to cope with the differencesamong the various systems inchanging programs, displayingtrends, and point database opera-tions, not to mention dealing withthe idiosyncrasies of each differ-ent system. Furthermore, knowl-edgeable operators must surelywonder to whom they will turn ifthe network performance does notmeet expectations. What if con-

trollers are not receiving neces-sary information from others ofdifferent manufacture, or what ifit takes too long for operator dis-plays to be updated?

On the other hand, designersand operators are beginning tounderstand the need for improvedcommunication capabilitiesamong components of differentmanufacturers. Virtually everypiece of mechanical and electricalequipment used in buildings to-day comes factory assembled withsome type of microprocessor con-

trol module. Requiring each man-ufacturer to be compatible with aparticular control system to inte-grate the control of all compo-nents is becoming increasingly ex-pensive and slowing the pace oftechnological advancement that isso important to our industry to-day.

Importance of controlsSome in the industry question if

all this emphasis on controls is re-ally necessary. The answer shouldbe obvious. For years, studies

HEATING / PIPING / AIR CONDITIONING ■ AUGUST 1994 45

PRACTICAL CONSIDERATIONS FORPROTOCOL STANDARDS

A realistic look at how open/standard protocols will meetdesigners’ and operators’ needs for improved interoperability

among HVAC equipment and control systems of different manufacture

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .OPEN/STANDARD PROTOCOLS

O

Entering condenserwater temperature

(AI)

T

T

T

T

Leaving condenserwater temperature

(AI)

Return chilledwater temperature

(AI)Condenser

Supply chilledwater temperature

(AI)

Evaporator

R R RPS CS

DDCinterfacemodule

Run status (DI)Alarm status (AI)

Start/stop (DO)Chiller power (AI)

Head pressure (AI) Vane position (AO)

Transducer interface legendT = temperature sensorR = relayPS = pressure sensorCS = current sensor

Point type legendDI = digital inputDO = digital outputAI = analog inputAO = analog output

1 Typical DDC interface to chiller.

have shown large numbers of of-fice workers are dissatisfied withthe workplace comfort and airquality. The underlying problemsonly rarely involve a lack of capac-ity of HVAC components to pro-vide necessary heating, cooling, orventilation. Rather it is almost

entirely the lack of proper con-trol—particularly zone control—that breeds this dissatisfaction.

A major roadblock to HVAC de-signers’ understanding of theneed for improved controls stemsfrom our industry’s historic as-sumption that uniform tempera-ture and ventilation conditionswithin each building will ade-quately satisfy all occupants, andthat under normal occupied condi-tions, the internal loads through-out a building are reasonably uniform. Acceptance of this as-sumption has reduced the re-quirement for effective controls inthe minds of many designers.However, the notions that uni-form load conditions actually existin buildings today and that uni-form comfort conditions are ac-ceptable to building occupantsneed to be reconsidered. Studieshave found that the perception ofideal comfort and air quality mayvary substantially among individ-uals. More recent studies are alsoshowing a strong link betweenworker productivity and ability tocontrol space conditions. Compli-cating the picture are the increas-ingly large variations of heatingand cooling loads within areas ofbuildings today.

Architects and interior design-ers have traditionally worked todesign buildings and spaces thatare visually attractive and invit-ing. HVAC designers are now be-ginning to realize they must design mechanical/electrical sys-tems that are environmentally at-tractive and inviting. As buildingowners and managers are learn-ing to exploit their tenants desiresfor improved comfort and air qual-ity, our industry is having to learnhow to design HVAC systems thatprovide improved individualizedterminal control. The major dif-ference between such systems andthe systems of today is the highlevel of integrated controls re-quired to provide such conditionseconomically.

Pitfalls of protocol standardsSome designers and building

operators believe the develop-ment of communication protocolstandards will solve our indus-try’s problems with controls bysimplifying the process by whichcontrol systems are designed andprocured. These individuals maystill believe a standard protocolwill permit designers to treatDDC components as they remem-ber treating pneumatic controlscomponents—as interchangeablecomponents. This is an unrealisticexpectation for DDC-based sys-tems. The utter simplicity ofpneumatic controls, along withlong-standing competitive forces,required each pneumatic controlmanufacturer to manufacturepneumatic devices that were func-tionally identical to other prod-ucts. DDC controls by contrast areorders of magnitude more com-plex than pneumatic-based con-trols, and manufacturers havechosen substantially different de-velopment paths for their prod-ucts. While many in the industrybelieve communication standardsmay foster a competitive environ-ment in which certain functionalcharacteristics are encouraged,even an optimistic scenario putswidespread interchangeable DDC

products well beyond the horizon.The introduction of standard

communication protocols will notmake controls design easier.Rather, it will add significant pit-falls in the path of such designwork. Controls designs that relyon substantial intercommunica-tion via standardprotocols amongproducts from dif-ferent manufactur-ers will require spe-cial consideration.Such designs willve ry l i ke ly en -counter startup andongoing operationalproblems unless theexact nature of theintercommunica-tion requirementsare well understoodby the designer andclearly expressed inthe specifications.

The persistingidea that simplystating that eachcontrol componentshall be compatiblewith a particularcommunication pro-tocol standard willensure the opera-tional integrity of a DDC system net-work needs to be rebutted firmly. A

Open/standard protocols

46 AUGUST 1994 ■ HEATING / PIPING / AIR CONDITIONING

DDCinterfaceconnect

plug Connect to DDCsystem trunk

Evaporator

Condenser

2 DDC interface to trunk compatiblechiller.

DDC SYSTEMCOMPONENT LEVEL

Operator's console

DDC proprietary network

Stand-alone panels

Proprietary subnetwork

Unit controllers (UC)VAV boxes and other

terminal control

I/O point device

DDC SYSTEMCOMPONENT LEVEL

OperationsLocal area network

DDC proprietary network

DDC system controllersAir handler controllers

VAV zone controllersDual duct zone controllers

Misc. controllers

I/O point device

Key

Modem

Host

Stand-

DDC

computer

alone panels

Printer

controllers

standard communication protocolmay ensure that devices coexist ona network but not ensure that theywill operate together. DDC con-trollers that are connected by astandard communication protocolnetwork require the designer toprovide a complete description of

virtually all aspects of the connec-tion and operating criteria in thespecifications so that each supplierunderstands not only the commu-nication requirements but also theinteroperability requirements ofthe product to be supplied. Such de-scriptions are well beyond the level

of specificity typically seen in build-ing control specifications today.

OpportunitiesWhat communication stan-

dards can do is aid those design-ers who are motivated to improveoccupants’ perception of comfort

HEATING / PIPING / AIR CONDITIONING ■ AUGUST 1994 47

Input/outputpoints

Input/outputpoints

Coolingtower

controller

Dual ductsystem

controller

Auxiliarycooling

loopcontroller

zone zonezoneVAV zone

VAV VAV VAV zone zone

DD zone

DD zone

DD zone

DD zone

DD zone

Stand-alone panel

(general controller)VAV

systemcontroller

VAV VAV

4 Emerging DDC system architecture.

SAP1

SAP2

SAP3

UC1 UC2 UC3 UC4 UC5 UC6 UC7 UC8 UC9 UC10

Input/outputpoints

Input/outputpoints

Input/outputpoints

TEMVEL

ionswork

work

lerslerslersllerslers

vice

3 Typical present day DDC system architecture.

and air quality by economicallyintegrating high-performancecontrols into HVAC and electricalsystems. Knowledgeable design-ers will have the ability to employstandard protocols to developmore effective HVAC controlstrategies with economy. To showhow this is possible, consider theintegration of a standard centrifu-gal chiller with a building controlsystem that is of different manu-facture than the chiller. Tradi-tionally, many such chillers haveoperated quite independently ofthe building controls. Suchchillers often start and stop basedon outside air temperature ratherthan demand for cooling and usu-ally operate to a fixed presetchilled water temperature ratherthan what is required to meet cur-rent conditions.

Experienced designers knowthat such operation is often coun-terproductive to both comfort andeconomy. The chiller plant mayoperate many more hours thannecessary, and the fixed lowchilled water temperature setpoint may make it uneconomicalto reset supply air temperaturesupward in cool weather for im-proved comfort.

To achieve improvements incomfort and economy, many de-signers now look to integrate thechiller operation into the overallbuilding control strategy. Fig. 1 il-lustrates how a typical DDC sys-tem instrumentation configura-tion can accomplish this. Thisinterface includes a total of 10DDC system points and wouldlikely cost from $3000 to $6000 toaccomplish. While the exact num-ber of points and associated costwill vary depending on the inter-face requirements, the introduc-tion of a standard communicationprotocol between the two systemswill almost always reduce the in-terfacing cost. Fig. 2 shows howsimple it may be to connect such anetwork.

However, simply having the ca-pacity to connect a piece of equip-ment to a network does not ensure

the equipment can be made to op-erate as desired. The chiller oper-ating program may not permit theunit to be started or stopped overthe communication network, or itmay not permit the chilled waterset point or the demand limit to beadjusted over the network. Thesefeatures are operational and arenot included in communicationstandards, but they are of criticalimportance to ensuring the suc-cess of the overall HVAC systemoperation.

To be certain a communicationnetwork that relies on a standardprotocol operates effectively, sys-tem designers must very carefullyselect all operational featuresthat must pass across the networkto components of different manu-facture and then be certain thatthese features are provided by thevendor. As time passes and our in-dustry gains more experiencewith protocol standards, a num-ber of operational features arelikely to become universal. Butinitially, a great deal of diligenceon the part of designers will be re-quired to ensure we are not pun-ished as our prayers are finallybeing answered!

State of standardsCurrently, there are two major

efforts underway within the indus-try to introduce communicationsstandards, both of which are likelyto have significant impacts overthe next few years. The ASHRAEStandard 135P committee hasbeen working diligently now forseven years to develop a communi-cation standard. This standard isfocused on HVAC controls and isexpected to be released within thenext year under the name BACnet.While the BACnet effort is to beapplauded as a true “grassroots”movement, it should be ap-proached with caution because itwill likely be released without adefinitive compliance testing for-mat. Because compliance testingmay initially be voluntary, manu-facturers can claim compliancewithout demonstrating that such

compliance actuallyexists. In fact, man-ufacturers have al-ready claimed com-pliance to BACnetfeatures in theircommunication pro-tocol. Since the finalstandard does notyet exist, theseclaims are obviouslyexaggerated. De-signers should notreact negatively tosuch claims becausethey may be demon-strating manufac-turers’ good faith effort to support in-dustry standards.But while we wel-come such bold be-havior by equipmentmanufacturers, wemust also under-stand that in theshort term, a highdegree of diligencewill be required onthe part of design-ers to interconnectcomponents in theBACnet environ-ment successfully.

The second majorprotocol standardeffort in the indus-try has been devel-oped by Echelon, anindependent com-

Open/standard protocols

48 AUGUST 1994 ■ HEATING / PIPING / AIR CONDITIONING

DDC SYSTEMCOMPONENT LEVEL

Operations

Local area network

DDC proprietary network

DDC system controllersAir handler controllers

VAV zone controllersDual duct zone controllers

Misc. controllers

I/O point device

Protocol standard network

Standard compliantDevices from other

manufacturers

I/O point device

DDC SYSTEMCOMPONENT LEVEL

Operations

Local area network

DDC proprietary network

DDC system controllersAir handler controllers

VAV zone controllersDual duct zone controllers

Misc. controllers

I/O point device

Network gateways

Standard compliantDevices from other

manufacturers

I/O point device

Modem

Host

Stand-

computer

alone panels

Printer

Networkgateway

KeyKey

HEATING / PIPING / AIR CONDITIONING ■ AUGUST 1994 49

Input/output points Input/output points

Coolingtower

controller

Dual ductsystem

controller

Auxiliarycooling

loopcontroller

VAV zone

VAV zone

VAV zone

VAV zone

VAV zone

VAV zone

DD zone

DD zone

DD zone

DD zone

DD zone VAV

systemcontroller

Boilercontroller

Chillercontroller

Motorcontrolcenter

Input/outputpoints

Input/outputpoints

Input/outputpoints

Networkgateway

Input/output points Input/output points

Coolingtower

controller

Dual ductsystem

controller

Auxiliarycooling

loopcontroller

VAV zone

VAV zone

VAV zone

VAV zone

VAV zone

VAV zone

DD zone

DD zone

DD zone

DD zone

DD zone VAV

systemcontroller

Boilercontroller

Chillercontroller

Motorcontrolcenter

Input/outputpoints

Input/outputpoints

Input/outputpoints

Networkgateway

Networkgateway

Networkgateway

TEMVEL

ions

work

work

lerslerslersllerslers

vice

work

ianttherrers

vice

TEMVEL

ions

work

work

llersllersllersllersllers

vice

ways

liantother

rers

vice

6 DDC architecture with gateways to standard protocol devices.

5 DDC architecture with gateway to standard protocol network.

puter communications firm. Thisfirm hopes to be the supplier ofchoice for communications chipsin a fashion similar to the way In-tel supplies microprocessor chipsto the computing industry. Eche-lon’s product is called LonWorks.The concept behind LonWorks isthat computer manufacturerswhose products require communi-cation among processors can buythese chips, which are prepro-grammed to provide the networkservices. Such a scheme may beattractive to manufacturers be-cause they could concentrate theirefforts on the performance of theirproducts and not on communica-tions issues.

Echelon is in the business of sell-ing chips and development pack-ages and is not really concernedwith what protocol is employed inits chips, so long as it works. Inthat sense, Echelon is not reallycompeting with BACnet. In fact,Echelon asked the ASHRAE Stan-dard 135P committee to approve itas one of the environments for op-erating BACnet, but the commit-tee turned Echelon down. Echelonhas already developed its own pro-prietary protocol and has aggres-sively entered the DDC system in-dustry marketing its chips tooperate with that protocol.

Protocol warsHVAC designers are soon to

find that BACnet, LonWorks, anda number of open protocols thatare being offered by some manu-facturers will be competing overthe next few years to set the stan-dard for building control commu-nications. For LonWorks, the mo-tive is financial; for BACnet, themotive is more altruistic. If BAC-net is found to be a workable stan-dard, and if our industry has theenergy and finds the resources tosupport, improve, and extend thestandards, then it may very wellflourish. Otherwise, BACnet willlikely wither on the vine and belimited to a few esoteric applica-tions noteworthy only for theirdistance from the beaten path of

DDC technologies.The key to the success of Lon-

Works will be cost. DDC manufac-turers incur costs both in purchas-ing each LonWorks chip as well asthe expense of developing inter-faces to their controllers. The cur-rent explosion in distributed pro-cessing for DDC systems todaymeans that a typical building ofthe future may employ hundredsor thousands of separate DDCcontrollers communicating to-gether. What looks like a smallmarginal cost for an individualcontroller may soon become alarge cost in an increasingly com-petitive environment. For Lon-Works to be widely employed, itwill have to compete with thecosts of employing BACnet or pro-prietary communications net-works in such installations.

Implementing standardsWhile the rapidly approaching

competition for the hearts andminds of communication networksmay become interesting, HVACdesigners will do better if theytake a very practical and conser-vative approach over the next fewyears. Designers should under-stand that a far more importantdevelopment is now underway inthe DDC industry. Fig. 3 showsDDC system architecture as itcommonly exists today. The basisof many DDC systems today is thestand-alone panel (SAP). Gener-ally, unitary controllers that oper-ate terminal boxes or other uni-tary devices exist on subnetworksat the SAP level. However, net-works of the future are more likelyto connect all controllers at a sin-gle network level. Such architec-tures have been shown to improvethe consistency of controller oper-ation as well as integration capac-ity. Such future control networksmay more closely resemble thatshown in Fig. 4. Additional devicessuch as communication controllersand/or routers may be employed todirect messages along this busynetwork.

Many manufacturers are now

developing networks similar tothat shown in Fig. 4. This networkconfiguration makes every con-troller more like the stand-alonepanel with which we are familiar.As this trend toward further dis-tribution of processing capabili-ties continues, the demands uponthe communication network areincreasing rapidly; and fast, effi-cient communication networksare becoming necessary.

It is within this environmentthat manufacturers are now beingasked to develop standard compli-ant networks. These new de-mands on throughput brought onby the increasing distribution ofprocessing resources make controlmanufacturers nervous (andshould make designers nervoustoo!), especially when they are be-ing asked to connect to equipmentmanufactured by others as well.

Designers need to be very care-ful about their designs that de-pend on standard networks toconnect components of differentmanufacturers to operate to-gether. To be certain adequate ac-countability exists when theequipment is installed, the follow-ing rules should be followed in de-veloping such designs.

◆ Focus control system specifi-cations on a complete system thatwill be the responsibility of a sin-gle vendor.

◆ Require that the control sys-tem vendor supply the gateway tothe protocol standard network(s)that will be employed to connectequipment of different manufac-ture.

◆ Include precise descriptionsof information that must be ex-changed between the control sys-tem and equipment that may be ofdiffering manufacture. Such de-scriptions must be included in thecontrol specifications and theequipment specifications as well.

◆ Include a procedure for as-sessing and resolving networkproblems that may appear. Re-member, startup failure is onlyone possible problem and proba-bly the easiest to solve. Occasional

Open/standard protocols

50 AUGUST 1994 ■ HEATING / PIPING / AIR CONDITIONING

malfunctions or sporadic losses ofcommunications may be morecommon and certainly are moretroublesome types of networkingproblems.

Until the industry has more ex-perience with the emerging proto-col standards and their use inDDC networks, I recommend thatdesigners develop their designsaround gateways. Fig. 5 illus-trates a DDC communicationsgateway connecting a chiller,boiler, and motor control center toa DDC system network. The gate-way may or may not be an actualphysical device, but it is a point ofresponsibility transfer. The DDCsystem vendor is responsible formaintaining the DDC system net-work and gateway. The chiller,boiler, and motor control centermanufacturers have the responsi-bility of providing and acceptinginformation as selected in a for-mat compatible with the stan-

dard network.Initially, I recommend design-

ers consider a separate gatewayfor each different manufacturer’sequipment that is intended to beinterfaced to the DDC system.Such a network is shown in Fig. 6.This approach limits the fingerpointing in the event of an inter-face problem to no more than twoseparate entities. It also may pro-vide clues to the culprit. For ex-ample, if the chiller and boiler arecommunicating well with theDDC network but the motor con-trol center is not, then likely theproblem is in the motor controlcenter controller. However, if eachis experiencing similar faults,then the problem likely resides inthe DDC gateway module.

Things to rememberThe imminent arrival of com-

munication standards in thebuilding control industry can be

an enormous opportunity or apunishment, depending on howthey are employed. Building oper-ators and designers who look tocommunication standards as ameans of simplifying the procure-ment and operation of buildingcontrol systems may be very dis-appointed with the results. Thosewho rise to the challenge and putforward the extra effort to provideeffective controls designs will findthat protocol standards providemore economical interface capa-bilities, resulting in integratedcontrol strategies never beforepractical. At least for the shortterm, designers should be verycautious with the application ofnetworks employing standardprotocols. Selection should clearlyand completely describe all com-munication and interoperating re-quirements and establish ac-countability in case of problemsassociated with such networks. V

HEATING / PIPING / AIR CONDITIONING ■ AUGUST 1994 51

By THOMAS HARTMAN, PE,Principal,The Hartman Co.,Marysville, Wash.

ver since the energy cri-sis, when digital con-trols (then called EMCSfor energy management

and control systems) were un-ceremoniously ushered intowidespread use for HVAC control,the industry has tried to makethem look and act like the pneu-matic controls they have super-seded. Only occasionally are someof the profoundly expanded oppor-tunities available with digital

controls applied effectively. Fur-thermore, terms like reset sched-ule and direct acting, relevantonly to pneumatic systems, arestill commonly employed in whatis now the digital controls era.

While the process of transitionto digital control technologies tol-erates this mixed bag, a multitudeof new demands are requiring ourindustry to move ahead and real-ize the full potential of digital con-trol technologies. Building occu-pants are demanding morecomfortable and higher qualityenvironments. Building ownerscontinue to press for greatereconomies in construction, opera-

tion, and maintenance. Finally, avariety of pressures are upon us toprovide more precise control anddocumentation that standards fortemperature, ventilation, and in-door air quality are being met.

In this article, I will discusshow DDC technologies permit anew flexibility in the traditionalrules concerning the need for lin-ear signals and responses with in-put and output devices. Whenproperly applied, this new flexibil-ity can reduce the cost of DDCtechnologies. Next month, I willshow how, by combining thesefundamentals with emerging in-ter-manufacturer controls inte-gration, designers can achievenew horizons in performance andenergy efficiency.

Why linear devices?When pneumatic controls domi-

nated our industry, building own-ers paid a high price for modulat-i n g l o o p p e r f o r m a n c e a n dstability. One of the prices paidwas the requirement that inputand output devices be linear withrespect to the system variablethey sensed or controlled. Thisneed for linear response was es-sential to match the limited con-trol capabilities of pneumatic con-trollers. A number of rules andconventions were establishedwithin our industry that madeachieving this linear response re-quirement easier. Among thesewere the development of the equalpercentage valve, which included

HEATING / PIPING / AIR CONDITIONING n FEBRUARY 1995 63

DIRECT DIGITAL CONTROLFUNDAMENTALS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .DDC TECHNOLOGIES

E

DDC technologies are permitting a new flexibility in the traditional rules concerning the need for linear signals

and responses with input/output devices

Resi

stan

ce, o

hms

(Tho

usan

ds)

Temperature, F

3634323028262422201816141210 8 6 4

12011010090807060504030

1 Resistance curve for thermistor temperature sensor.

the seemingly backwards rule ofthumb that called for sizing con-trol valves smaller than the pipesize. Similarly, mechanical sens-ing devices were constructed toprovide linear change in controlair pressure over their entiresensing range.

While these conventions andrules of thumb served the days ofpneumatics, they now need to berethought. Requiring what I callexternal linearization in digitalcontrol designs adds costs in twoways. Linear devices are oftenmore expensive than nonlineardevices that may offer improvedlevels of performance in DDC ap-plications. Further, linear outputconventions, such as designing ahigh pressure drop throughvalves or dampers, carry a sub-stantial continuous operating en-ergy penalty. By developing newrules and conventions, the knowl-edgeable designer can produce de-signs that have lower first and op-erating costs and may operatemore reliably as well.

Linear devices in the DDC eraThe need for linear response in

modulating control loops has notbeen eliminated by the introduc-tion of digital controls. While digi-

tal controls offer improved modu-lating control capabilities, includ-ing proportional/integral/deriva-tive (PID) controllers, these controlloops continue to be based on theprinciple of linear response, atleast over certain ranges. How-ever, in most typical applications,digital controls can easily inter-nally linearize both input signalsand output control functions.

Internal linearization of inputsOne way to reduce the cost of

some DDC configurations is topermit nonlinear input devicesand use the DDC system for scal-ing to achieve the correct readingover the range required for the ap-plication. I continue to see DDCspecifications that limit the selec-tion of input devices to those thatprovide a linear signal to the DDCsystem over a wide range of val-ues. Except in special cases, thisis an unnecessary requirementthat adds costs and may causeother problems. Consider temper-ature sensors. Fig. 1 shows a re-sistance curve for an inexpensivethermistor type temperature sen-sor that may be employed forroom temperature sensing. Ther-mistors are excellent choices forHVAC applications. They are in-

expensive, have excellent accu-racy and very low hysteresis, andrespond quickly to temperaturechanges. Furthermore, at temper-atures normally involved inHVAC applications, thermistorshave excellent long-term stability(some care should be taken inchoosing thermistors when tem-perature may rise above 240 F).Finally, because thermistors aretypically high resistance (10,000ohms is typical), they are not af-fected by variations in wiring dis-tances. However, some designerscontinue to exclude thermistorsbecause the input signal is not lin-ear with temperature over widetemperature ranges. Instead, lowimpedance RTD type sensors areoften specified. This type of sensortypically requires an electric cir-cuit at the sensor that linearizesand transmits the signal in a waythat it will not be affected bywiring resistance (usually a cur-rent loop signal is used).

Employing low resistance RTDsensors with additional electron-ics presents a number of potentialproblems in DDC applications.First is the matter of accuracy.While the RTD sensors them-selves provide excellent accuracy,it is not uncommon to find end-to-end accuracies (I use end-to-endas the comparison of the valueread by a precision thermometerat the device compared with theactual reading at the DDC systemoperator’s terminal) out of toler-ance. Calibration of the currentloop input may be more difficultthan that of a simple resistancetype thermistor.

Other potential problems withRTDs range from the additionalelectronics (usually located at thedevice) that may complicate relia-bility issues all the way to how the

DDC fundamentals

64 FEBRUARY 1995 n HEATING / PIPING / AIR CONDITIONING

Resi

stan

ce, o

hms

(Tho

usan

ds)

Measured variable

11

10

9

8

7

6

5

12

80

75

70

65

60

DDC linearized curve

Sensor curve 2 Nonlinear sensor resistance

curve. The sensor performancecurve is a smooth curve over thesensor’s operating pressure. TheDDC linearized curve is a series ofstraight lines that closely approxi-mates the sensor’s performance.

continued on page 66

sensor and electronics are config-ured, which on occasion has beenfound to affect adversely the sen-sor signal.

Table functions that are nowreadily available with DDC prod-ucts can be employed to scalethermistors and other nonlineardevices over a wide range of val-ues. Fig. 2 shows how a DDC sys-tem can linearize a continuous,nonlinear sensor input curve witha table function. A number ofstraight line curves are estab-lished in the table function to ap-proximate closely the nonlinearfunction of the device. As long assimple, inexpensive devices canmeet the repeatability, hysteresis,and stability requirements for anHVAC application, such devicesshould not be rejected becausetheir signals are not linear.

Is linear output required?Once it is understood that input

devices need not be linear, it is nota great leap to recognize that theresponse from output devices con-trolled by analog outputs simi-larly need not be linear. However,the issues here are more complexand more ingrained in the rules ofthumb that engineers frequentlyapply automatically, so some in-depth discussion is required.

Because of the pneumatic back-ground, valve design manualscommonly stress the need to selectcoil/valve combinations for whichequal increments in valve positionwill effect equal increments inheat transfer of a typical heatingor cooling coil throughout thestroke of the valve actuator. Fig. 3shows how traditional designpractice seeks to linearize theoverall performance of valve andcooling coil. Carefully selecting acoil and valve combination canprovide nearly linear performanceover the entire range of load possi-bilities. Such selection is done be-cause it is assumed that the valvewill be operated by a controllerwith a fixed proportional gain.Though this design principle isstill widely employed, it is no

longer applicable in many modernHVAC applications. In VAV cool-ing coil applications, the varia-tions of air flow and air/chilled wa-ter temperature characteristicsact to change dynamically the heattransfer characteristics of thevalve/coil arrangement as theseparameters change. This makes itvery difficult to select a valve/coilcombination that will be linearthrough the variety of conditionsthat may accompany its operation.

The higher performance of DDCsystems permits designers muchgreater flexibility in the design of mod-ulating controls without establishing

static (and therefore unrealistic) de-sign criteria. Fig. 4 shows a valve andcoil combination that does not providea linear response of valve position tocoil capacity. However, modern DDCsystems permit scaling tables to be ap-plied to analog outputs as well as theinputs. Output scaling permits an in-herently nonlinear device combinationto respond in a linear fashion to signalsfrom the DDC system. In this exam-ple, the valve and coil combination pro-vides about 70 percent of the designcooling capacity at about 20 percentvalve travel. The DDC output to thevalve can be adjusted with the scalingtable to position the valve at 20 percent

DDC fundamentals100

90

80

70

60

50

40

30

20

10

0

1009080706050403020100

Valve travel, precent (Coil design flow, percent)Ca

paci

ty, p

rece

nt (V

alve

flow

, per

cent

)

Coil characteristic

Combined valve and coil

Valve characteristic

3 Design of chilled water valve and coil combinations for proportional control.

100

90

80

70

60

50

40

30

20

10

0

1009080706050403020100

DDC analog output, percent (Valve travel, percent)

Capa

city

, per

cent

Combined valve and coil

Scaled DDC output to valve

4 Use of scaling to linearize control for DDC control.

continued from page 64

continued on page 68

66 FEBRUARY 1995 n HEATING / PIPING / AIR CONDITIONING

travel at a 70 percent output signalfrom the DDC system. The scaling fac-tor allows standard PID control to op-erate the valve effectively because of asoftware linearization of the valve/coilcombination.

However, the chilled water flowand heat transfer performance as-sumed for Fig. 4 is valid only forconstant load-side flows and inlettemperatures and for constantchilled water supply tempera-tures. Whether inherent in thesystem design or for optimizationreasons, rarely in real HVAC ap-plications do these other variablesremain constant as control loopsoperate. As previously discussed,the issue of linear output combi-nations has therefore been onlyweakly resolved in the past by at-tempting to linearize componentsat one set of system conditions.Obtaining good control over wideranges of system conditions canbe resolved far more completelyand effectively with the higherperformance capabilities of DDCsystems. The proportional, inte-gral, and derivative gains can betied to algorithms that adjusttheir values as the variables suchas load-side flow, temperatures,and chilled water temperaturechange. Even more impressive isthe emergence of self-tuning con-trollers. These controllers contin-ually re-establish the variousgains associated with a controlloop to provide continuously pre-cise control without hunting. Thebenefits of self-tuning are espe-cially important because vari-ables beyond the immediate con-trol loop can have profound andwidely varying effects on eachcontrol loop. Self-tuning featuresare becoming widely availablewith DDC systems and are enor-mously effective in adjusting con-trol loops to continue stable opera-tion as other system variableschange.

ControllabilityAs previously discussed, select-

ing equipment for linear responseshould not be an overriding con-

sideration for designers in this eraof digital controls. However, thisdoes not mean designers can beimprecise in their designs or inthe selection of control loop com-ponents. The issue of controllabil-ity is one that will continue toplay a prominent role both in thedesign of systems and the selec-tion of individual components.Controllability remains largely asizing issue. If a valve is oversizedfor given conditions such that thesmallest increment possible fromthe control loop will substantiallyovershoot the desired control con-ditions, the loop has become un-controllable. This is a problemthat typically emerges during pe-riods of low load. Fully under-standing the issue of controllabil-ity and applying DDC capabilitiescorrectly allows designers to solvesuch problems and at the sametime vastly improve the efficiencyand performance of these sys-tems.

Selecting a control valve with alower pressure drop will reducethe pumping power required tomeet the load conditions. Tradi-tional practice strongly condemnsthe idea of employing large valveswith lower pressure drops becauseof the nonlinear response and thelack of controllability at low loads.Fig. 5 illustrates the dilemma. Thevalve/coil combination with ValveA may be selected according to tra-ditional design practice because itis reasonably controllable at low

loads. The vertical axis interceptrepresents the smallest incremen-tal cooling transfer possible as thevalve is cracked open. Note that itis small—only about 10 percent ofthe design maximum cooling rate.The coil combination with Valve Bhas a much lower pressure dropbecause Valve B is a larger sizevalve. While valve/coil Combina-tion B would require less pumpingpower, the Y-axis intercept ismuch higher than that for Combi-nation A. Traditional design crite-ria typically declare Valve B un-suitable for the applicationbecause it is uncontrollable atlower loads and the valve posi-tion/cooling capacity relationshipis nonlinear. But when it is inte-grated with a high-performancecontrol system that can adjustboth the chilled water tempera-ture and the loop head pressure,will linearity and controllability ofCombination B really be a prob-lem?

System dynamicsTo see how this question can be

answered, consider the graphs inFigs. 6 and 7. Fig. 6 shows the op-eration curves for valve/coil Com-bination B at a number of differ-ent approach (chilled watersupply less air temperature leav-ing coil) temperature conditions.It is clear that increasing thechilled water temperature rela-tive to the leaving air tempera-ture markedly improves the con-

DDC fundamentals

68 FEBRUARY 1995 n HEATING / PIPING / AIR CONDITIONING

100

90

80

70

60

50

40

30

20

10

0

1009080706050403020100

Valve travel, percent

Capa

city

, per

cent

Valve A designed for 30 ft pressure drop and 13 F approach Valve B designed for 5 ft pressure drop and 13 F approach

Valve and coil design B

Valve and coil design A

5 Design of valve and coil combinations for proportional control.

continued from page 66

continued on page 70

trollability at low loads. Similarly,Fig. 7 illustrates that the decreasein pressure across the valve/coilcombination also improves thecontrollability at low loads.

Designers can use these rela-tionships to reduce substantiallythe problem of controllability. Atperiods of uniform low loads, theDDC system can reduce the headpressure across a valve and in-crease the chilled water tempera-ture to improve controllability. Ifall valves on a common chilled wa-ter loop experience similar de-creases in load concurrently, as istypical in many HVAC applica-tions, this parameter adjustmentis a great help in improving con-trollability at low loads.

It is apparent from the two fig-ures that larger rangeability andlow load controllability areachieved by controlling the chilledwater temperature for load ad-justment. Raising the chilled wa-ter temperature provides a bonusof chiller efficiency increases, butchilled water adjustment reducespumping savings because ahigher chilled water temperatureincreases the water flow neces-sary to meet loads. Additionally,under certain circumstances de-humidification requirements maylimit the permissible chilled wa-ter adjustment.

Exploiting the integrated con-trol capabilities of DDC systemsand controlling chilled water tem-perature and hydronic loop pres-sure in coordination with the con-trol valves allows valve/coilCombination B to perform verywell in many HVAC applications.

Next month we will focus on thelevel of integration required tomake valve/coil Configuration Boperate effectively. We will dis-cuss integrating the operation ofthe various equipment involved inproviding comfort, possible nowthrough the industry moves toprovide communication bridgesamong manufacturers. By concen-trating on selecting the most cost-effective input/output devices andby utilizing the emerging commu-

nications pathways betweenequipment from various suppli-ers, we will see that new horizonsof performance and energy effi-ciency can be attained with simpleand economical controls configu-rations.

Summary and conclusionDesigners must exploit the ben-

efits of higher performing DDC

systems to develop an under-standing of the fundamentals ofinterfacing hardware points toDDC systems. In so doing, a morein-depth look into total system op-eration must be evaluated beforesolutions are selected. Simply fol-lowing traditional rules of thumbregarding linear input and outputdevices is a poor design practice inthis digital controls era. V

DDC fundamentals

100

90

80

70

60

50

40

30

20

10

0

1009080706050403020100

Valve travel, percent

Valve and coil B performance at various approach temperatures (assuming constant differential pressure and inlet temperature)

Desi

gn c

apac

ity, p

erce

nt

13 F approach

10 F approach

7 F approach

4 F approach

100

90

80

70

60

50

40

30

20

10

0

1009080706050403020100

Valve travel, percent

Desi

gn c

apac

ity, p

erce

nt

Valve and coil B performance at various loop head pressures (assuming constant chilled water and air inlet temperatures)

4 ft head

20 ft head

16 ft head

12 ft head

8 ft head

7 Heat transfer vs. valve travel for various loop head pressures.

6 Heat transfer vs. valve travel for various approaches (leaving air tempera-ture minus entering water temperature).

continued from page 68

70 FEBRUARY 1995 n HEATING / PIPING / AIR CONDITIONING

By THOMAS HARTMAN, PE,Principal,The Hartman Co.,Marysville, Wash.

n last month’s article,we saw how the firstcosts and operatingcosts associated with

modern DDC systems can be re-duced by rethinking our selectionand application of input and out-put devices. This month, I intendto carry the idea of new thinkinga step further and show how in-stallation and operating costs ofHVAC systems can be even moredramatically reduced by furtherextending the opportunities af-forded by controls integration.

For example, to see the benefitsof more rigorous design ap-proaches and challenge our accep-tance of traditional design prac-tice, let’s consider the operation ofa variable flow chilled water loop.

Chilled water loopFig. 1 is a schematic of a chilled

water system employing variableflow on the secondary (load) cir-cuit. The loads are cooling coils ina building’s various air systems.In this case, the designer is con-fronted with a chiller that servesan extensive piping array withmultiple chilled water coils—areasonably common occurrence.Assume all the cooling loads varyfairly uniformly with outdoor con-ditions and the coils operate atlow load conditions much of thetime, which is typical of HVACcooling systems. To reduce pump-

ing energy, the designer decidesto incorporate a variable speedpump on the secondary loop. Todecouple the constant chilled wa-ter flow requirement of the chillerevaporator from the variable flowthrough the loads, the designerdecides upon a two-pump ar-rangement with a two-way modu-lating valve for each load and avariable speed drive to operatethe secondary chilled waterpump. The primary chilled waterpump operates continuously withthe chiller.

For the secondary loop, tradi-tional design manuals suggestsizing control valves for a full-flowpressure drop of at least 30 per-cent of the total system drop. Thisfactor is often called valve author-ity, and rules of thumb dictatethat it should be equal to orgreater than the pressure dropthrough the load, which in thiscase is the cooling coils. As dis-cussed in last month’s article, the substantial pressure dropthrough each valve is an effort toattain some measure of linearitybetween valve action and coolingeffect delivered to the load. Thehigh pressure drop also serves toisolate the operation of eachvalve/coil arrangement from oth-ers.

Let’s start by calculating the to-tal head across the secondarypump at full flow, then calculatethe pumping power required. De-pending on the physical locationand piping lengths, a 20 ft dropacross the loads, 30 ft drop acrossthe valves, and perhaps 40 ft head

drop across the secondary pipingat full flow conditions would betypical for a design using pub-lished rules of thumb. Assuming atotal full-flow requirement of allthe load coils of 1000 gpm, we cancalculate theoretical pumpinghorsepower at full flow as follows:

Pump hp = [gpm 3 8.35 lb pergal 3 pump head]/33,000 ft-lb permin-hp

Pump hp = [1000 3 8.35 3 (30 +20 + 40)]/33,000 = 22.8 hp

A traditional design would in-volve one or more differentialpressure transducers in the loop

HEATING / PIPING / AIR CONDITIONING MARCH 1995 75

NEW HORIZONS FOR HVAC CONTROL

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .DDC TECHNOLOGIES

I

How installation and operating costs of HVAC systems can be dramatically reduced by extending the opportunities afforded by controls integration

Chiller

Load 1

Load 2

Load 3

Load 4

Constant speed primary pump

Variable speed secondary pump

1 Primary/secondary chilled waterloop.

as shown in Fig. 2 to ensure thepump maintains a 50 ft headacross the valve(s) and coil(s). Ifall the coils spend large amountsof time operating at less than de-sign capacity, the piping head losswould drop with the square of thewater flow decrease. The pressuredrop through the coil would simi-larly decrease, but the savingswould be excluded because thecontrols keep a constant pressureacross the valve/coils—the pres-sure drop across the valves actu-ally increases as the flow de-creases. If reasonably even loadprofiles for all coils are assumed,the pipe head pressure drop willincrease with the square of theflow. At 75 percent flow the pipinghead loss is:

40 3 0.752 = 22.5 ftUsing this new pressure drop

figure for the piping, we see thatthe theoretical horsepower re-quirement at 75 percent flow is:

Pump hp = [(1000 3 0.75) 38.35 3 (50 + 22.5 )]/33,000 = 13.8hp

At 50 percent of design flow, the

piping head loss is:40 3 0.52 = 10 ftThe horsepower requirement at

50 percent flow is:Pump hp = [(1000 3 0.5) 3 8.35

3 (50 + 10)]/33,000 = 7.5 hpThe designer would then apply

appropriate pump efficiencies andsimulate or estimate the amountof time the pump and coils wouldspend at various conditions andcalculate the power requirementsas above. If the system operateslong hours at low loads, the sav-ings from a variable flow sec-ondary loop will be substantial.

High-performance designThe simplified analysis above

shows how substantial energy usereductions are possible by the ap-plication of variable flow to typi-cal hydronic loops in HVAC sys-tems. However, designers with agood understanding of high-per-formance DDC controls shouldask themselves why they allowtheir designs to be limited by tra-ditional design methodology. De-sign teams often lose track of thereasons for the rules of thumbthey apply so regularly. Whenthis happens, designers run therisk of applying outdated designtechniques—an important reasonmany designs fail to meet ex-pected levels of performance.

As we saw in last month’s dis-cussion, a control scheme that in-tegrates chilled water tempera-ture control with secondarychilled water loop control can pro-vide effective cooling coil controlwith considerably reduced controlvalve pressure drop. In such a de-sign, the DDC system operatesthe integrated system by increas-ing the chilled water temperatureand decreasing the differentialpressure of the loop as the loadson the coils decrease. Let’s seehow applying the design tech-niques discussed last monthwould affect this example.

Further improvementsObviously, it is wise to keep the

chilled water temperature adjust-ment at levels that permit flow re-

ductions to maximize pump powersavings, but adjusting the chilledwater temperature upward offerssavings too. The DDC systemmust be chosen so that it offersthe use of advanced algorithms toadjust the chilled water tempera-ture and secondary pump speedas load conditions change. Fur-thermore, the DDC system chosenshould provide the ability to self-tune modulating loops.

The question remaining is howthe system pressure and watertemperature changes will affectthe other valve/coil combinationson the cooling circuit. The as-sumption for this example is thatall coils will experience similarloads, as is often the case withHVAC systems. However, the de-signer must be very careful to en-sure this is actually the case. Ifone of the coils serves an interiorcomputer room that has high con-stant year-round cooling loads orif coils serve different perimeterzones of a building subject to highsolar gains, these unusual zonesmay have to be specially accom-modated either by a booster pumpor a separate chilled water circuitas shown in Figs. 3 and 4.

In Fig. 3, a small booster pumpis added to increase the differen-tial pressure for Load 2, which thedesigner has determined will notfall as quickly as the others on theloop. In Fig. 4, two entirely sepa-rate loops have been configured topermit the separation of loadsinto groups that will have similarpart-load patterns. The configura-tion in Fig. 4 may be a cost-effec-tive configuration if the loadgroupings are in different loca-tions and do not require extensiveadditional piping.

Performance control programThinking back to last month’s

article, consider how the chilledwater loop of Fig. 1 might achieveimproved energy performancewhen operated by an integratedcontrol program. The designermay develop a high-performancecontrol program for the loop withseveral basic elements as follows:

DDC fundamentals

76 MARCH 1995 HEATING / PIPING / AIR CONDITIONING

Chiller

Load 1

Load 2

Load 3

Load 4

Constant speed primary pump

Variable speed secondary pump

Control

VFD

DP

2 Primary/secondary chilled waterloop with differential pressure sen-sor (DP) and controller.

continued on page 80

● The pump speed will be de-creased and the chilled watertemperature will be increased toensure at least one of the controlvalves serving the loads served bythe loop is fully opened at alltimes during operation. The sys-tem will further ensure that allloads are met at all times.

● Moving from a full-load topart-load situation, the pumpspeed will first be reduced down to90 percent of maximum speedwhile the chilled water tempera-ture remains constant. There-after, the chilled water tempera-ture will be increased along withpump speed reductions at a rela-tive reduction ratio that main-tains a minimum valve opening of20 percent.

Note that the control programoutline does not anticipate theneed for a loop differential pres-sure input.

Evaluating improvementsAs discussed last month, a high-

performance control system per-mits the control valves to be se-lected for lower pressure dropswhen valve control is integrated

with chiller and pump control asabove to handle the issues of lin-earity and controllability. Assumethe same piping and coils are se-lected as in the earlier example,but the control valves are sized fora 5 ft pressure drop at design flow.Now the total head for the sec-ondary loop at full load conditionsis the sum of 20 ft drop across theloads, 5 ft drop across the valves,and 40 ft drop across the sec-ondary piping. At the full flow of1000 gpm, the theoretical full-load pumping horsepower can becalculated as:

Pump hp = [gpm 3 8.35 lb pergal 3 pump head]/33,000 ft-lb permin-hp

Pump hp = [1000 3 8.35 3 (5 +20 + 40)]/33,000 = 16.4 hp

This compares to the 22.8 hp forthe traditional loop design.

Although this design requireslarger valves to achieve a lowerpressure drop at full flow, thepump, motor, and variable fre-quency drive components are allonly about 70 percent as large asrequired for a traditional design,which means that the overall costof the mechanical components ofthe design is likely to be the sameor less than the initial design.

At lower loads, the pumping en-ergy calculations are more compli-cated because both the chilled wa-ter temperature and the looppressure will be adjusted by thehigh-performance DDC system tomeet the specific requirements ofthe various loads. Raising thechilled water temperature raisesthe chilled water flow required tomeet the loads but also increasesthe efficiency of the cooling plantand results in additional energyuse reductions. Reducing the totalloop pressure as the load de-creases will enhance the energysavings beyond the savings fromflow reduction. Using the perfor-mance graphs for the valves inlast month’s article, we can as-sume that at 75 percent averageload the system operates at an

11.5 F approach. At this operatingpoint the flow would be:

1000 3 (0.75 3 13/11.5) = 848gpm

The piping and coil head lossesare, respectively:

40 3 (848/1000)2 = 28.8 ft20 3 (848/1000)2 = 14.4 ftThe assumed head loss through

the valves depends on the controlstrategy and the piping arrange-ment to be employed. Generally,the head loss through the valvescan be assumed to decrease at thesame ratio as the piping and coilhead losses, but this is not alwaysthe case. Conservatively, it is as-sumed for this example that thevalve head loss remains at 5 ft.

Based on these assumptions,the horsepower requirement at 75percent flow is:

Pump hp = [848 3 8.35 3 (5 +14.4 + 28.8)]/33,000 = 10.3 hp

This design clearly offers someadditional part-load pumping sav-ings over the original example, butmore important is that it also of-fers savings from a 1.5 F increasein chilled water temperature. Achiller saving of 0.02 KW per ton-Fenergy reductions per degree in-crease in chilled water tempera-ture yields an additional savings ofmore than 9 hp at the chiller for atotal saving of more than 15 hp.

DDC fundamentals

Chiller

Load 1

Load 2

Load 3

Load 4

Constant speed primary pump

Variable speed secondary pump

Check valve

Booster pump

Chiller

Load 1

Load 2

Load 3

Load 4

Constant speed primary pump

Variable speed secondary Pump 2

Variable speed secondary Pump 1

3 Primary/secondary chilled waterloop with booster pump on one load.

4 Primary/secondary chilled waterloop with two secondary loops.

80 MARCH 1995 HEATING / PIPING / AIR CONDITIONING

continued on page 83

continued from page 76

At 50 percent of design flow, itcan be assumed that the systemoperates at a 10 F approach. Atthis operating point the flowwould be:

1000 3 (0.5 3 13/10) = 650 gpmThe piping and coil head losses

are, respectively:40 3 (650/1000)2 = 16.9 ft20 3 (650/1000)2 = 8.5 ftTo be conservative in advance of

a certain piping layout, we as-sume the head loss of the valvesremains constant at 5 ft.

With these assumptions, thehorsepower requirement at 50percent flow is:

Pump hp = [650 3 8.35 3 (5 +8.5 + 16.9)]/33,000 = 5.0 hp

As before, the part-load opera-tion pumping savings are againovershadowed by the 3 F increasein chilled water temperature. Achiller saving of 0.02 KW per ton-Fenergy reductions per degree in-crease in chilled water tempera-ture yields an additional saving ofabout 7 hp at the chiller for a totalsaving of approximately 18 hp.Clearly, the use of integrated con-trol and low pressure drop valvesresults in substantial additionalenergy savings for this chilled wa-ter system.

New horizons for designersSo far, our design thinking has

dealt only with the controls aspectof a traditional chilled water looplayout. Now that we understandthe possibilities of integrated oper-ation, let’s rethink the physicalconfiguration of Fig. 1. The strikingfeature of Fig. 1 is the necessity ofhaving two pumps. Why? We knowthat below certain velocities of flowthrough chiller heat exchangers,the heat transfer capacity drops offrapidly. The purpose of the pri-mary pump is to ensure that condi-tion never troubles our system.However, with an integrated sys-tem, we can tell when the waterflow is too low for chiller conditionsbecause evaporator and water tem-peratures can both be continuouslymonitored. So, instead of Fig. 1,let’s consider the simpler pipingconfiguration in Fig. 5.

In Fig. 5, the chiller is a vari-able speed unit that offers a highturndown ratio and high coeffi-cient of performance at low loads.The required rate of flow throughthe chiller obviously depends onthe cooling load. This is a good de-sign fit because the flow throughthe loads also varies with theloading. However, this does notmean that such a system willwork adequately under all loadconditions without specific atten-tion to the chiller flow.

Single-pump systemsTwo basic requirements must be

met to ensure effective and effi-cient operation of the Fig. 5 config-uration. First, the system must notbe operated unless the cooling re-quirement is above a minimumthreshold load. Second, the waterflow through the chiller evaporatorheat exchanger must be sufficientto maintain less than a predeter-mined maximum temperature dif-ference between the refrigerantand chilled water under all condi-tions.

In most HVAC applications,these requirements can be met aslong as the DDC system has the ca-pacity to integrate the control ofthe chiller, pump, and valves andthe capacity to operate these ele-ments with high-performance con-trol algorithms. The operating se-

quence for a single-pump circuitprovides a continuous calculationof cooling load requirements foreach load. The cooling loop wouldbe shut down until the sum of thecalculated load requirementsreaches a specific value that de-pends in part on the anticipatedupcoming conditions. Once en-abled, the cooling plant will operateas described above in the high-per-formance sequence except that thechilled water temperature wouldbe raised any time the evaporatorheat exchanger temperature differ-ential reaches the maximum differ-ential for the load conditions. Rais-ing the chilled water temperaturewill require higher flows to satisfyloads and will automatically returnthe heat exchanger to optimal per-formance levels.

It is clear that the simpler con-figuration of Fig. 5 reduces thefirst cost of the system and alsoprovides some further reductionof operating costs over the Fig. 1configuration.

Benefits of high performanceThe above example is indicative

of the opportunities available tothose designers with the expertisenecessary to apply high-perfor-mance DDC to HVAC systems.The expected results for mechani-cal designs that are carefully inte-grated with high-performancecontrols are:

● Smaller (and usually sim-pler) equipment sizes with accom-panying savings that can reducethe system costs or be invested inhigher quality, longer lastingcomponents.

● Lower total system energyuse than what is possible withnonintegrated design and controlstrategies.

● Control precision that is su-perior to that of traditional con-trol approaches.

DDC cost considerationsControl system costs can be a

factor in the feasibility of a high-performance design, but general-izations should be avoided untilthe exact control point require-

DDC fundamentals

5 Single-pump chilled water loop.

Load 1

Load 2

Load 3

Load 4

Variable speed chiller

Variable speed chilled water pump

HEATING / PIPING / AIR CONDITIONING MARCH 1995 83

continued from page 80

continued on page 85

DDC fundamentals

ments are fully considered. Nu-merous economies are possible byintegrating the microprocessorcontrol units of various stand-alone elements. Modern, high-performance DDC control optionsmay cost less than nonintegratedelectronic control based alter-nates. The designer needs to con-sider that integrated, high-perfor-mance control layouts can usuallybe configured with fewer instru-mentation points than traditionalcontrol approaches. Communica-tion bridges between componentsof different manufacture are nowbecoming available. Furthermore,the intelligence of high-perfor-mance DDC systems can increas-ingly be used to replace instru-mentation.

Let’s consider the chilled waterloop. The traditional control of thesecondary pump requires one ormore differential pressure sensorsand a controller to operate thevariable frequency drive that setsthe pump speed. Also required arecontrol points to operate eachvalve and operate the chiller. Ifthe chiller and air handlers arepackaged units with microproces-sor control, the only costs incurredfor the high-performance controlis the capacity to integrate the op-eration of this equipment. A high-performance DDC configurationcan be simpler because the systemuses the existing monitoring andcontrol instrumentation. With mi-croprocessor control at each thatcan be integrated in its operation,no additional I/O instrumentationis required.

Summary and conclusionTo fully exploit the benefits of

high-performance DDC systems,designers must engineer designsbefore solutions are selected.Rules of thumb have no place in ahigh-performance design process.The benefits in improved energyperformance, and in many in-stances lower first costs, are com-pelling reasons for designers toacquire the tools and expertise re-quired to apply high-performancedesign approaches more widely. V

continued from page 83

By THOMAS HARTMAN, PE,The Hartman Co.,Marysville, Wash.

Nowhere are new HVACtechnologies more dramat-ically showcased for their

practical application than inbuildings such as museums andlibraries. These buildings requirecontinuous operation and precisecontrol under widely varyingloads for visitors and contentsalike, and providing these im-provements with economy is theideal calling for recently intro-duced HVAC technologies. Today,the knowledgeable designer candevelop facilities with perfor-mance features far better thanever before while working withinlimited first and operating costbudget constraints.

In Part 1 of this article, I willdiscuss how new technologies insuch buildings can significantlyimprove part-load operation. The

energy, maintenance, and controlperformance offered by these newtechnologies make them very at-tractive not only for new buildingconstruction but as retrofit oppor-tunities as well.

Continuous HVAC operationOne of the most important but

least recognized characteristics ofHVAC design is the long hoursmost buildings operate at low-loadconditions. Low-load operation isfurther increased in archivalbuildings where uninterruptedcontrol of conditions, essential tothe preservation of the materialsthey contain, requires some formof continuous HVAC system oper-ation. Our industry’s failure tocapitalize fully on the benefits ofnew technologies has led opera-tors of such facilities to believethat energy costs will be signifi-cantly higher because of the longhours of operation. It has beendemonstrated in certain cases that

the opposite is true. Integratedcontrol and variable-speed tech-nologies, properly applied, can ac-tually reduce energy use whensystems are kept running!

To see how this can be true, let’sfirst be certain we understand theenergy implications of part-loadoperation. According to fan andcentrifugal pump laws:

▼ Fluid flow quantity varies di-rectly with speed of fan or pump.

▼ Static pressure deliveredvaries as the square of the speed.

▼ Power required varies withthe cube of the speed.

These laws apply to axial andcentrifugal fans as well as centrifu-gal pumps. They dictate that re-ducing fan or pump speed reducesflow in proportion to the speed butreduces power required by thecube of the speed reduction. Thismeans that it is theoretically possi-ble to provide 50 percent flow (orcapacity) for 13 percent (0.53) of thefull capacity power, an efficiency

NEW HVAC TECHNOLOGIES

April 1996 HPAC Heating/Piping/AirConditioning 57

Library and Museum HVAC:New Technologies/New Opportunities

— PART 1How new technologies can dramatically improveperformance and economy of part-loadoperation in continuously operating systems

The large variation in occupancy condi-tions typical for library and museum facil-

ities requires designers to establish ameans of providing real-time occupancy

data to the control system.

increase of almost 400 percent.Theoretically, power for 25 percentof the flow requires about 2 percentof full capacity power, an efficiencyincrease of over 1000 percent!

The importance of these high

part-load operating efficiencies isillustrated in Figs. 1 and 2, whichdepict simulated chiller load pro-files for a continuously operatingbuilding in four U.S. cities. Notethat despite the wide variations inchiller plant operating hours as de-picted in Fig. 1, the chiller plantload profiles for cities as shown inFig. 2 are very similar and mostcertainly skewed toward low-loadconditions. Note also that theseprofiles are based on a perfectlysized chiller plant. In practice, de-signers tend to oversize chillerplant elements, resulting in pro-files even more skewed toward partloads.

A designer working with tradi-tional technologies might developan indirect evaporative cooling cir-cuit to maximize chiller plant en-ergy efficiency at all the variousload conditions. This configurationuses tower water directly for cool-ing during low-load, low outsideambient conditions. The designmight also employ multiple chillersof various sizes to permit combina-tions of chillers operating at maxi-

mum efficiencies (usually nearpeak load) to meet the various in-termediate-load conditions. The re-sult could be a complicated me-chanical room and even morecomplicated operating sequence

that is costly to install anddifficult to support.

Part-load operationIf we consider all the en-

ergy consuming compo-nents employed to providecooling in typical buildings,we find that all of them, thefans, pumps, and centrifu-gal chillers, are subject tothe basic fan and pumplaws. While we know thatdelivery efficiencies, thevariance of system loadsand outdoor conditions, andother factors make it unre-alistic to achieve theoreti-cal part-load operating effi-ciencies, we should expectto develop cooling systemsthat can operate muchmore efficiently at part-

load conditions than at relativelyinfrequent peak conditions.

Designers often believe the solu-tion to saving cooling plant energyat part-load conditions is simply toemploy a variable-flow cooling dis-tribution system with variable-speed drives operating distribution

pumps. Unfortunately, present de-signs of these systems fail to opti-mize the energy reduction poten-tial of such systems. For example,consider the chilled water distribu-tion circuit in Fig. 3. In this widelyemployed design, the variable-speed distribution pump is con-trolled to maintain a fixed differen-tial pressure set point. Thelocation of the differential pressuresensor is usually at the end of thepiping run. A controller operateseach of the load control valves.Even though such systems arelikely to be connected together in aDDC system network, the opera-tion of each controller is almost al-ways entirely independent of allothers as shown in the figure.

In Fig. 3, each individual controlvalve is typically sized for a sub-stantial portion of the total systempressure drop at full load (usually30 to 40 percent). The differentialpressure set point is the sum of thefull-flow valve and load pressuredrop, which is in the range of 75percent of the total full-load pumphead, the remainder being the full-flow distribution piping pressuredrop.

The system operating curve for asystem of this design is shown asthe bold line in Fig. 4. Fig. 4 as-sumes the design full-flow head is100 ft at 1000 gpm, and the differ-

New HVAC technologies

58 HPAC Heating/Piping/AirConditioning April 1996

Chill

er p

lant

ope

ratin

g ho

urs

4000

3500

3000

2500

2000

1500

500

1000

Annual chiller operating hours

0Seattle

Los Angeles Helena New York

1 24-hr building chiller operation.

Perc

ent o

f tot

al a

nnua

l ope

ratin

g ho

urs

60

50

40

30

20

10

0

Percent of design maximum load<20

20 to 40 40 to 60

60 to 80 >80

Seattle Helena

Los Angeles New York

2 24-hr building chiller operation based on percentage of operating hours atvarious loads.

ential set point for the distributionpump controller is 75 ft. The Y axisintercept is the differential headset point because even at zero flowthe pump must operate to main-tain that pressure. Let’s see howthis design performs at variousload conditions.

At full-load the power curveshows that the pump will drawabout 33 hp. As flow is reduced to50 percent, the pump speed is re-duced to slightly more than 1450rpm, and the pump draws about 14hp. As flow drops further to aboutone-third, the pump speed remainsapproximately constant, and thepower drops to approximately 11hp. Further reductions see littlefurther reduction in power.

Although a reduction from 33 to11 hp may seem to be substantial,

it is actually a stark display of thefailure of this design. The operat-ing efficiency (flow per hp) of thesystem remains almost constantdespite the part-load conditions;one-third of the design flow re-quires exactly one-third of the de-sign power. At lower flow require-ments, the efficiency begins todecrease further.

Better part-load efficiencyThe challenge for designers to-

day is to design systems that can

better exploit the benefits of newvariable-speed and control tech-nologies. Because HVAC systems,especially for continuously oper-ating public buildings such as mu-seums and libraries, spend longhours at part-load conditions andbecause new technologies enablesuch enormous improvements inpart-load efficiency, the industry’scurrent focus on full-load perfor-mance criteria is missing themark entirely. Instead, designersmust come to view full-load opera-tion as primarily a sizing issue. Itis essential that each design ade-quately address full-load condi-

tions. But let’s be clear that atpresent, operating costs underthese infrequent conditions areonly a very minor portion of an-nual operating costs.

The failure of the widely em-ployed variable-flow distributionschematic of Fig. 3 is that it doesnot attempt to mitigate problemsassociated with the second powerrelationship between pump headand speed. Because of pressure setpoint constraints, the pump in thisexample cannot operate at lessthan 1450 rpm. This severely limitspower reduction as flow require-ments are reduced. Now imagine a

April 1996 HPAC Heating/Piping/AirConditioning 59

Chilled water return to

chiller plant

Chilled water supply from chiller plant

Differential pressure sensor

and controller

Bypass

Variable-speed distribution pump

Load 1

Load 2

Load 3

Load 4

C

C

C

C

DP

3 Schematic of typical variable-flowchilled water distribution system.

Head

, ft

120 110 100 90 80 70 60 50 40 30 20 10 0

0 100

200

300

400

500 600

T00

800

900

1000 1100 1200Flow, gpm

Brak

e ho

rsep

ower

40

32

24

16

8

0

1750 rpm pressure curve

System curve

1450 rpm

pressure curve

1750 rpm power curve

1150 rpm pressure curve 1450 rpm power curve

900 rpm pressure curve

600 rpm pressure curve

1150 rpm

power curve

900 rpm power curve 600 rpm power curve

Head

, ft

120

100

110

90

80

70

60

50

40

30

20

10

Flow, gpm

40

32

24

16

8

0

Brak

e ho

rsep

ower

00 100

200

300

400

500

600

700

800

900

1000

1100

1200

1750 rpm pressure curve

1450 rpm pressure curve

1150 rpm pressure curve

900 rpm pressure curve

600 rpm pressure curve

1750 rpm power curve

System

curve

1450 rpm power curve

1150 rpm power curve

900 rpm power curve 600 rpm power curve

4 Operating curve for system shown in Fig. 3 assumes design full-flow headof 100 ft at 1000 gpm with the differential set point for the distribution pumpat 75 ft.

5 Operating curve for a system design that permits the pressure differentialacross the valves and loads to fall as flow requirements are reduced, pro-ducing dramatic increases in pumping efficiency.

design that permits the pressuredifferential across the valves andloads to fall as flow requirementsare reduced. Such a system curvemay look more like the one in Fig.5. Note in Fig. 5 that as the flowrequirement falls to one-third, thepump power drops to less than 2hp, a nearly 600 percent increasein the pumping efficiency com-pared to full-flow conditions.

The question posed by Fig. 5 ishow to accomplish such part-loadefficiencies. Some answers areshown in Fig. 6. In Fig. 6, the con-trol of the variable-speed distribu-tion pump is accomplished bytransmitting the conditions ateach of the loads served to thepump controller rather than em-ploying a differential pressuresensor as in the Fig. 3 schematic.Also, the load control valveswould typically be line sized with-out a substantial pressure drop.

As the cooling load requirementsat each of the zones fall, the corre-sponding valve begins to close. Ifnone of the valves are fully openand the loads are satisfied, thedistribution pump slowly reducesits speed. If one or more valves isfully open and its load is not satis-

fied, the distribution pump gradu-ally speeds up. Such a simple, net-work-based control scheme is veryeffective in substantially improv-ing part-load efficiencies of achilled water distribution systemif the load profiles for the loadsserved (Loads 1 to 4 in this exam-ple) are similar.

Note that the control configura-tion in Fig. 6 is actually simplerthan that of Fig. 3 since it employsno differential pressure sensor.Note also that the basic designphilosophy employed in Fig. 6 canbe extended to other componentsof a building cooling system, par-ticularly the supply air fans andair distribution system. Operat-ing an HVAC system suitably con-figured with these componentscontinuously at low loads withcorresponding high delivery effi-ciencies may use less energy thanshutting them off and forcing

them to start each day athigh loads (to bring thebuilding back under con-trol) and correspondinglylow operating efficiency.

Part-load design emphasisBecause HVAC design

emphasis has not tradi-tionally been placed onpart-load operating char-acteristics, the fan, hy-dronic, and cooling sys-tems in a great manybuildings operate at lowerefficiencies at part-loadconditions than at fullload. For buildings thatmust operate continu-ously, this is an especiallyenticing design opportu-nity because the opportu-nities for energy reductionare enormous. It is not toodifficult to see how imple-menting or upgrading

HVAC systems that operate athigh part-load efficiencies can of-fer vastly improved performanceand reduced energy costs.

A final note on part-load designemphasis regards assessing thelevel of occupancy to provide ade-quate ventilation air. For build-

ings that may experience substan-tial variations in occupants or vis-itors, and museums and librariesfit this category, it is an absoluterequirement that building occu-pancy levels be determined con-tinuously from real-time informa-tion. Virtually all museums havesome mechanism for tracking visi-tors; often it is a turnstile count.Nowadays such data can be di-rected with relative ease to theHVAC control system.

Because the occupant density inlibraries is usually low, occupancysensors can be located throughoutand connected to the HVAC con-trol system to operate the lightsas well as HVAC terminal units.Information from the occupancysensors may then be suitable toassess occupancy of the buildingfor a determination of the outsideair requirement at all times.Whatever mechanism is em-ployed, it is essential both for pro-viding good environmental con-trol and economy that the HVACsystem design incorporate ameans to assess real-time occu-pancy conditions accurately. Thisinformation is used by the controlsystem to establish and maintainventilation air flow set points asrequired throughout the facility.

Next month I will continue mydiscussion on library and museumfacilities focusing on humidity,dehumidification, and buildingpressure control. HPAC

New HVAC technologies

60 HPAC Heating/Piping/AirConditioning April 1996

Chilled water return to

chiller plant

Chilled water supply from chiller plant

Integrated DDC controllers and network

Bypass

Variable-speed distribution pump

To other DDC

controllers

DDC DDC

DDC

DDC

DDC

Load 1

Load 2

Load 3

Load 4

6 Schematic of variable-flow chilled waterdistribution system designed to maximizepart-load energy savings.

The wide open areas typical in librariesand museums require special attentionto ensure that suitable conditions canbe maintained throughout the wide vari-ations of indoor and outdoor loads andconditions they will experience.

By THOMAS HARTMAN, PE,The Hartman Co.,Marysville, Wash.

Last month, I discussed hownew technologies are espe-cially suited for museums

and libraries, which require closecontrol of indoor environmentsyear-round, both for contents andoccupants. The focus was on thedramatic improvements in theperformance and economy of part-load operation that these technolo-gies permit. This month I will fo-cus on year-round humiditycontrol and building pressure con-trol and their role in controllingthe indoor environment of theseand other institutional buildings.

Humidity controlThe HVAC industry has been

deluged with a great deal of mis-information about humidificationin the last few years. Severalyears ago the indoor air qualityconsultant on a new office build-ing on which we were working rec-ommended against a humidifica-

tion system on the basis of IAQconcerns!

Certainly very high humidityand water penetrating duct insula-tion have been shown to be sub-stantial contributors in manybuildings that have experiencedIAQ problems, but low humidity isa serious IAQ issue as well. The po-tentially serious problems that canlead to high humidity and watercarryover are not difficult to re-solve by the disciplined designerwhereas the idea of omitting hu-midity control altogether is nolonger a reasonable response.

In museums and libraries thatcontain valuable artifacts, thequestion of humidity control isgenerally not one that can be easilyducked. Constant changes in hu-midity have been shown to con-tribute to the deterioration of paper, wood fiber, and other water-pervious materials. Maintainingconstant humidity as well as tem-perature is important to preserv-ing the valuable contents of thesefacilities. The question of how con-stant the humidity must be re-

mains one that must be deter-mined on a building-by-buildingbasis. Designers should rememberthat stability and diffusion charac-teristics of humidity, which isbased on vapor pressure, are verydifferent from air temperature,which is based on molecular activ-ity. Still air can be a thermal insu-lator but is far less effective in in-hibiting the transmission of watervapor. It is much more difficult tomaintain differing levels of specifichumidity within a building than itis to maintain space temperaturedifferences. When curators desiredifferent levels of humidity in dif-ferent areas, I recommend tryingfirst to compromise on a level thatwould work throughout the facil-ity and next seeing if the samespecific humidity with a variationin temperature to alter the rela-

LIBRARY/MUSEUM HVAC

May 1996 HPAC Heating/Piping/AirConditioning 63

Library andMuseum HVAC:New Technologies/New Opportunities

— Part 2

New technologies candramatically improveperformance andeconomy of library andmuseum HVAC systems,which must operatecontinuously underwidely fluctuating loadsfor occupant comfortand contents protection

Museums often include exhibitareas that require specialtemperature and humidityconditions. Designers must takecare to ensure sufficient isolation isprovided for such special areas.

tive humidity would provide thedesired effects.

Humidification in cold climatesHumidification of buildings in

cold climates requires special at-tention to the building envelope.To protect the building envelope, Irecommend operating the buildingat a static pressure that is slightlynegative with respect to the out-side anytime the outside air tem-perature is below freezing and thespecific humidity inside is higherthan outside. (How to controlbuilding static pressure is dis-cussed in greater detail in the nextsection.) This reduces the potentialfor building envelope degradationfrom water vapor migrating intowalls and freezing. It should benoted that the high rate of vapordiffusion in air makes it impossibleto prevent the migration of mois-

ture into the walls altogether justby maintaining a slight pressuredifference. Therefore, strong atten-tion must be paid to achieving anoccupancy-side vapor barrier asabsolutely impenetrable as possi-ble. This requires the coordinatedefforts of the architect and the con-tractor. Very tight infiltration

specifications are in order, and arigorous design and testing proce-dure to ensure they are met mustbe accepted.

Still, the idea of operating build-ings at 50 percent relative humid-ity in subzero weather is not at allan attractive one. The size of thehumidification plant required isvery large. Puddles of water con-densed from the air might formregularly in entrance lobbies. Thisand the potential formation of icejust outside the building pose haz-ards. To mitigate such problems,our firm has successfully employedan annual humidity set pointschedule for museums and li-braries located in cold climates.Such a schedule is shown in Fig. 1.Although humidity is not abso-lutely constant, there is only onegradual cycle each year. For mostbooks or artifacts, this annual hu-

midity cycle is not seen as a threatto their longevity.

Controlling humidityTechnologies for the measure-

ment and control of humidity inbuildings have improved enor-mously in the last few years. Fif-teen years ago, human hair was

commonly employed with mechan-ical amplification to provide hu-midity control. Calibration and re-liability were serious concerns.Today, sensitive membranes pro-vide accuracy and long-term sta-bility not possible just a few yearsago without incurring enormousexpense.

Unless the facility demands un-usually rigorous humidity controland/or verification, I recommendinstalling good commercial qualityrelative humidity sensors that of-fer readings well within severalpercent RH and are very reasonablypriced. Installing a number ofthese inexpensive sensors in vari-ous spaces and the return air ductof each air handling system allowsthe building staff to target thosethat may be out of calibration (bycomparing readings during moder-ate weather and low buildingload). At the same time, this con-figuration ensures humidity con-trol is being provided uniformlythroughout the building. Becausethe relationship between relativehumidity and specific humidity istemperature dependent, I also rec-ommend locating each RH sensorwithin, or immediately adjacent to,a space temperature sensor so thattemperature compensation can beperformed accurately.

Dehumidifying strategiesFor many North American cli-

mates, warm weather humiditycontrol can be accomplished by ad-justing primary supply air temper-ature or cold deck temperature. Irecommend that designers con-sider a VAV system with the ca-pacity to provide supply air tem-perature a few degrees less than 55F, depending on the climate. In cli-mates that are not prone to warmhumid weather, the operation of aVAV system to control humidity aswell as temperature entails adjust-ing the supply air temperaturedownward below its normal setpoint any time building humidityreaches the high limit. As the sup-ply temperature is reduced, the airflow required for the cooling load is

Library/museum HVAC

64 HPAC Heating/Piping/AirConditioning May 1996

RH s

et p

oint

, per

cent

Day of yearJan Jan

60

55

50

45

40

35

30

25

20Feb Mar Apr May Jun Jul Aug Sep DecNovOct

1 Set point schedule for cold climate space humidity.

continued on page 67

also reduced, providing a higherratio of latent-to-sensible coolingand reducing the relative humidityin the building. For many NorthAmerican climates, this approachworks very well, and the control issimple and stable.

In climates that experiencemild but humid weather, thebuilding can at times require almost entirely latent cooling.Supplying the low temperature airto provide the required latent cool-ing in very small amounts becausesensible cooling is not requireddoes not provide satisfactory circu-lation, and it can lead to seriousconsequences from localized con-densation as supply air is intro-duced at low velocities below thedew point of the space. Before au-tomatically considering a systemwith summer reheat in these ap-plications, I suggest that designersconsider a method first describedto me by fellow engineer and fre-quent HPAC author Ken Gill. Thisapproach employs splitting thecooling coil into two or more sepa-rately controlled sections. So longas building humidity is at or belowset point, the sections operate inunison. However, as space humid-ity rises above set point, the coilcontrol reverts to a staging se-quence wherein the cooling valvefor one stage is fully opened beforethe next stage valve begins open-ing. The resulting mix of air thathas been cooled well below its dew

point and that bypassed throughnonoperating coil(s) provides thewarmer, drier air required for sucha load. If the designer takes care tolocate coils and mixing plenumsuch that the highest humidity airstream (usually outside air) ismost directed at the lead coil, andconfiguration ensures condensatewill not be evaporated back intothe bypassed air, this approachcan be every bit as effective as a re-heat scheme in warm weather.And it is certainly more economi-cal. (See also the discussion under“Coils” in Ken Gill’s article “IAQand Air Handling Unit Design” inthe January 1996 issue of HPAC.)

Humidifying alternatesProviding humidity during the

cold months of the year can be amuch more substantial designchallenge. Objectivity has been no-tably absent from recent guidesand proposed standards. However,discussions by the technical ex-perts representing various inter-

ests invariably conclude there is noreason to provide some inherentpreference of one means of humidi-fication over another. Each methodhas advantages and disadvantageswhen compared to others. Theknowledgeable designer canquickly sort out salient issues re-garding the choice of a humidifica-tion system for each project depending on the characteristics of the application, climate, and,certainly not to be forgotten, thepreference of those who will be op-erating and maintaining theequipment. Let’s remember that ahumidification system is an IAQfriend, not a foe, as long as it is de-signed and maintained properly.The advent of vastly improved hu-midity sensors and advanced con-trols has made this fact more eas-ily proven than ever before. Theaccompanying sidebar provides abrief comparison of potential ad-vantages and disadvantages of thethree major types of humidifica-tion systems for buildings. In de-

Library/museum HVAC

May 1996 HPAC Heating/Piping/AirConditioning 67

How major types of humidification systems compare

Wetted media evaporative humidifier—ADVANTAGES:◆ Self-regulating, proper sizing results in limited control requirements.◆ Control failure cannot result in water discharge to the ductwork.◆ In dry or moderate climates, building energy use can be reduced.◆ Extensive water treatment is not required because impurities do not

readily enter air stream.◆ System is inexpensive and easy to control with no concern of

oversaturating air stream.Wetted media evaporative humidifier—DISADVANTAGES:◆ Requires an external source of heat to evaporate water in cold weather.◆ Requires more room in duct, must be accessible, and medium requires

cleaning and periodic replacement.

Water atomization injection humidifier—ADVANTAGES:◆ System is low in cost and easy to control◆ In dry or moderate climates, building energy use can be reduced.Water atomization injection humidifier—DISADVANTAGES:◆ Control failures can cause water deposition in ductwork.◆ Water treatment is required to prevent deposits on artifacts and/or

IAQ problems.◆ In cold weather, external source of heat to evaporate water is required.

Steam injection humidifier—ADVANTAGES:◆ In cold weather, system provides its own heat of vaporization.Steam injection humidifier—DISADVANTAGES:◆ Control failures can cause water deposition in ductwork.◆ Extensive water treatment is required to prevent deposits on artifacts

and/or IAQ problems.◆ Water treatment and steam boiler are expensive to buy, operate, and

maintain.

continued from page 64

Museums located in harsh climatesrequire coordination between the ar-chitect and engineer to ensure theenvelope will perform as required.This is the Royal Tyrrell Museum/Al-berta Community Development inDrumheller, Alberta, Canada.

veloping the humidification de-sign, the engineer should keep thefollowing in mind:

▲ Provide a separate, indepen-dent control loop to disable humid-ification operation and alert staff ifever any failure could cause mois-ture to be deposited on the duct

walls from oversaturation of theair or excessive carryover.

▲ Make certain the locations ofall humidification hardware in-clude suitable access for easy in-spection and maintenance.

▲ If a steam or water injectionsystem of any kind is to be em-

ployed, the water must be treatedto be pure. The buildup of residueon artifacts or odors from injec-tion-type humidity systems is al-ways a concern.

The beauty of humidificationsystems today is how well theycan be controlled with modernsensors that are accurate and reli-able, and modern control systemscan regulate the operation of suchsystems in accordance with multi-ple variables.

Building pressure controlI urge the HVAC designers of

museums and libraries to considerimplementing variable volumeHVAC systems for these facilities.The HVAC systems are typicallysized to meet large peak occupantloads. With systems designed tooperate at improved efficienciesduring part-load conditions, theinitial investment for the larger ca-pacity can be earned back throughreduced operating costs. However,whatever system is finally chosen,if the air volume does vary withload, the problem of how to controlbuilding pressure looms large. Incold weather climates, a positivebuilding pressure can force watervapor from the interior into thestructure where it can freeze andcause structural problems. In hu-mid climates, a negative buildingpressure can draw in high-humid-ity outside air that can containthree times or more the moisture ofthe interior air and overload thedehumidification system.

Fortunately, technological ad-vances have a solution to theseproblems. In the early days ofVAV, fan flow tracking was com-monly employed to provide a bal-ance in outside air and exhaust air.More recently, methods that in-volve measuring the actual air en-tering the building and that exit-ing the building have beenproposed. However, technology isnow available to measure buildingpressure directly. This is a muchpreferred method from my point ofview, and it has been employedwith success.

Library/museum HVAC

68 HPAC Heating/Piping/AirConditioning May 1996

continued on page 72

Measuring building pressurePressure sensors are now avail-

able that measure ranges of +0.1 to–0.1 in. WG and less with 2 percentor better accuracy over that range.Such devices are entirely suitablefor measuring building static pres-sure directly to provide return fancontrol. Some of these instrumentsemploy a self-zero feature that pe-riodically connects the two portstogether and resets itself to ensurethe integrity of the zero reading.This is a most attractive featurebecause it is a near-zero set pointthat is desired in such buildings.

My recommendation is to con-sider a building as a single entityno matter how many supplyand return/exhaust fans it con-tains. (However, buildings withkitchens containing large exhausthoods and makeup that are some-thing other than constant volumemay constitute an exception to thisrule.) Multiple space static pres-

sure sensors are essential with atleast one on every floor and at leastone per 20,000 sq ft on a singlefloor. A common outside air refer-ence is employed for all the sen-sors. Standard 3/8 in. pneumatictubing can be employed for refer-

ence piping if the runs are short.For longer runs (100 ft or greater),the size should be increased corre-spondingly. I recommend the spacereference be at ceiling level. Hav-ing the reference tube exposed ap-

Library/museum HVAC

2 Algorithm used to operate the return air fan in medium-rise buildings.

“CALCULATE AVERAGE BUILDING STATIC PRESSURE”“NOTE: CERTAIN SENSORS MAY CARRY MORE WEIGHT THAN OTHERS”

DOEVERY 1 MINUTESTATIC-AVE = (STATIC-1 + STATIC-2 + STATIC-3 . . . STATIC-N)/N

“CALCULATE SUPPLY/RETURN FAN SPEED OFFSET TO BE APPLIED BY ALL FANS”

CALC1 = (STATIC-AVE – STATIC-SP) * 100IF CALC1 > 3 THEN CALC1 = 3IF CALC1 < –3 THEN CALC1 = –3

SUP/RET-OFFSET = SUP/RET-OFFSET + CALC1IF SUP/RET-OFFSET > 10 THEN SUP/RET-OFFSET = 10IF SUP/RET-OFFSET < –10 THEN SUP/RET-OFFSET = –10

“CONTROL ALL RETURN FANS ACCORDING TO SUPPLY FAN SPEED AND OFFSET”

RET-FAN1 = SUP-FAN1 + SUP/RET-OFFSETRET-FAN2 = SUP-FAN2 + SUP/RET-OFFSETRET-FAN3 = SUP-FAN3 + SUP/RET-OFFSETRET-FAN4 = SUP-FAN4 + SUP/RET-OFFSETRET-FAN5 = SUP-FAN5 + SUP/RET-OFFSET. . .

continued on page 103

continued from page 68

72 HPAC Heating/Piping/AirConditioning May 1996

proximately 1 in. below the ceilingtile away from doors, supply dif-fusers, and return grilles hasworked well. Be certain it is welllabeled.

The outside reference must con-nect to at least two separate pointsof outside air reference on at leasttwo opposite sides of the building.The piping connecting outside ref-erences must be larger than that tothe sensors. Each outlet to outsideair must be protected from directwind. Even with such precautions,I find the building static pressurereadings are subject to continuousvariation as wind blows, peoplemove through the building, doorsopen and close, etc. However, sta-ble building pressure can be main-tained as long as sensed buildingpressures are not permitted tomake drastic changes in return fanspeed. Fig. 2 shows the algorithmour firm has employed with suc-cess to operate the return air fan inmedium-rise buildings. Note thatthe intermediate calculation(CALC1) is limited in magnitudebut can gradually increase themagnitude of the offset. Our firmtypically limits the return fanspeed offset to no more than 20percent of the supply fan speed.However, the exact magnitudeand range of the offset are deter-mined empirically based on theoperational performance of thetwo fans and actual operation ofthe building.

Summary and conclusionThanks to new technologies

available to the HVAC industry,providing constant environmentalconditions in libraries and muse-ums to protect valuable books, ar-tifacts, and art is a challenge withincreasingly effective and economi-cal solutions. When designers con-figure systems employing thesetechnologies properly, a suitableenvironment can be maintainedfor first and operating costs thatare only a slight premium overwhat is required to maintain com-fort conditions in typical commer-cial buildings. HPAC

Library/museum HVAC

continued from page 72

BUILDINGS AS HABITATS

By THOMAS HARTMAN, PEPrincipal,The Hartman Co.,Marysville, Wash.

One very cold winter morn-ing, I was touring an officebuilding with its operating

engineer when a tenant ap-proached to ask if it would be pos-sible to raise the temperature inher office for a short time eachmorning during this cold weather.The building engineer patientlyexplained how costly it would be,and the tenant departed disap-pointed but accepting of the expla-nation. I was reminded that yearsago, when my parents remodeledour home, my mother had beenconvinced that her new all-electrickitchen would be so expensive tooperate that she stopped much ofher baking for a time. When we fi-nally did an analysis to determinethe costs, we were all surprised athow little baking actually cost.This tenant’s request was in truthmuch less costly than suggestedby the engineer. I mused to myselfthat an hour’s extra warmth wascertainly far less than the pre-mium she had paid for the latteshe carried with her. So why, Iwondered, hasn’t our industry—like the coffee industry—begun ca-tering to individual desires?

Sophisticated occupantsHVAC designers are beginning

to realize that building occupants

have become much more demand-ing and sophisticated. In recentyears, a lot of industry discussionhas revolved around the sophisti-cated building owner, but ownersare just trying to keep up withtheir tenants’ demands. Buildingowners are faced with demandingenvironmental requirements fromprospective tenants. And when atenant’s employee has a com-plaint about comfort or air qual-ity, the building owner often findshim- or her-self in a face-to-facediscussion with a whole troupe of“experts” retained by the tenantto represent just such issues.

It is not difficult to understandwhy tenants are so concernedabout comfort and air quality is-sues. Employers are interested increating an accommodating work-place to hold valued employees,build or improve company morale,and improve worker performance.Employers have heard of the dis-ruptive horrors that air qualitycomplaints cause. Meanwhile,building occupants are not wait-ing for their employers to act.Publicity over the last few years

has begun to raise occupants’ con-cerns about the building environ-ment in which they work. Build-ing engineers are learning to takeoccupant complaints seriously asthey are increasingly challengedwith probing questions aimed atensuring that the building’s envi-ronment will not contribute tolong-term health problems.

Renewed focus on comfortThis renewed focus on building

environments forces engineers toalso face the nagging issue of occupant comfort. Studies are be-ginning to show a strong link be-tween comfort and worker perfor-mance and an even stronger linkbetween individual control ofbuilding environments andworker performance. Severalstudies have shown us that work-place performance does improvewhen comfort is improved or whenoccupants perceive control overtheir environment. Further, it issuggested by a recent survey thatsick leave may be dramatically re-duced when occupants are af-forded control over their work-place environments.

The building industry has fortoo long adhered to the dated con-cept that a building environmentacceptable to at least 80 percent ofthe occupants is adequate. The in-dustry must evolve from this ruleof thumb because it is well knownthat a single-space thermal condi-tion leaves a lot of building occu-

February 1997 HPAC Heating/Piping/AirConditioning 63

Trends Toward More User-Friendly

Building EnvironmentsHVAC designers arebeginning to realize thatbuilding occupants aredemanding more fromtheir environment

USER-FRIENDLY BUILDINGS

pants unsatisfied. There is a mis-conception among many design-ers and building operators abouthuman comfort. A widely heldview by occupants of commercialbuildings is that they demandvery precise space temperatureconditions that are best providedby a sophisticated centralized con-trol system. Those who viewbuilding operation in these termsoften applaud the trend awayfrom local thermostat control as apositive development that will, inthe long run, lead to fewer comfortcomplaints.

Individual comfort criteriaThough I am sympathetic to the

plight of building operators, I alsoknow that people have difficultydistinguishing any thermal differ-ence at all when space tempera-tures are changed by up to 1 F. Sothe problem of satisfying morebuilding occupants is not solvedby tighter space temperature con-trol but rather by accepting andaccommodating the variations inthermal environments requiredby the variety of building occu-pants in each building. Let us con-sider the interaction of individualcomfort requirements in popula-tions to see why many buildingoperators have been sidetrackedby the “tight control” solution anddiscover a more effective solutionto raising comfort levels in build-ings.

Generally, it is accepted thatwhen holding other contributingfactors constant, an individual’sresponse to space temperatureconditions is represented by acurve similar to that in Fig. 1. Onthe X-axis is space temperature.The Y-axis records the individ-ual’s perception of comfort at eachspace temperature. A boundaryseparates levels of perceived com-fort for which the individual is un-likely to complain from the levelsat which the individual is likely tocomplain. The range for which anindividual is unlikely to complainabout comfort is about 4 to 6 F.The individual whose comfort

curve is represented in Fig. 1 isunlikely to complain as long asthe temperature range remainsbetween 72 and 76 F. The centerpoint of this individual comfortcurve is 74 F. Generally, it is as-sumed that differences such asclothing, physical size, condition-ing, metabolism, gender, culture,and perhaps a number of otherfactors do not significantly changethe shape of the curve in Fig. 1 butmay alter where on the curve thespace temperature spectrum falls.So, for example, if the individualupon whom Fig. 1 is based were toput on a sweater or jacket, wewould expect the curve to move tothe left so it is centered on a lowerspace temperature. If the individ-ual were to change into light sum-mer clothing, we would expect thecurve to move to the right.

For a building inhabited by avariety of occupants with differ-ences in clothing and other fac-tors, one would expect a variety ofindividual comfort curves cen-tered around different space tem-peratures. From these, it would beeasy to calculate the number ofexpected complaints if one knewthe center point of each individ-ual’s comfort curve. Fig. 2 showswhat percentage of occupantswould be likely to complain at var-ious space temperatures if therewere a normal distribution of indi-vidual temperature curves forbuilding occupants. In Fig. 2, theline plot shows the number of peo-ple whose individual comfort

curves are centered on thevarious space tempera-tures, and the bar graphshows the percent of occu-pants who would be likelyto complain when thebuilding is operated atthat space temperature.Surprisingly, the curve inFig. 2 does not correlatewell with the experience ofmany building operatorsbecause it does not predictthe large increase of com-fort complaints originatingfrom very small changes in

building space temperature.If, however, it is assumed that

the occupants of buildings gener-ally fall into two groups, one thatlikes the space warmer than thecurrent set point and one thatlikes the space cooler than thatset point, the resulting satisfac-tion curve takes on the appear-ance of Fig. 3. Note that in Fig. 3,nearly two-thirds of the buildingoccupants’ individual comfortcurves are assumed to be centeredon two space temperatures that,rather than being next to one an-other, are about 4 F apart (72 and76 F). The temperature/complaintgraph in Fig. 3 looks much morelike what one would expect fromthe experience of building engi-neers. Comfort complaints are lowas long as building space tempera-ture is maintained within an ex-tremely narrow temperaturerange. If the building tempera-ture varies only slightly, the num-ber of occupants likely to com-plain rises rapidly. So despiteeach individual’s tolerance for arather wide variation of spacetemperatures, a group of peoplecan require a very narrow spacetemperature range.

Satisfying all occupantsMany explanations have been

presented for a polarized distribu-tion of comfort curves in build-ings. I suggest that it is the uni-form conditions themselves thatour industry struggles to main-tain in buildings that tends to po-

64 HPAC Heating/Piping/AirConditioning February 1997

Occu

pant

sat

isfa

ctio

n ra

ting High

Low

70 71 72 73 74 75 76 77 78Space temperature

Unlikely to complain

Likely to complain

1 Typical occupant comfort curve.

USER-FRIENDLY BUILDINGS

larize individual groupings. Byunderstanding the dynamic na-ture of comfort curves (e.g., theychange with activity and/orstress), one does not need to makea great leap in faith to see thatsubjecting these dynamic individ-ual curves to a static environmentcan act to exaggerate their com-fort curves away from that staticset point condition.

Whatever the true distributionof comfort curves in buildings, theindustry has begun to realizewe’re nearing the end of the era ofuniform conditions inbuildings for many rea-sons. First among them isthat despite improvementsin control technologies, oc-cupants do not appear tocomplain less about com-fort than they did decadesago!

Another problem withuniform building condi-tions is that the work manyin the industry have donewith dynamic control hasshown that operat ingbuildings at constant ther-mal conditions is not themost energy-efficient modeof operation. And finally,polarization as representedin Fig. 3 requires a spacetemperature that providesonly marginal satisfactionfor most of those who do notcomplain.

Rather than pursue tech-nologies that are aimed atmaintaining uniform fixedspace thermal conditions inbuildings, it is far moreprudent now to seize theopportunity to developtechnologies and HVAC de-signs that offer individualenvironmental control forbuilding occupants. Indi-vidual control concepts pro-vide the opportunity of op-erating buildings morecomfortably with lower en-ergy use. The first cost ofindividual control systemconcepts is improving as

the economy of the controls tech-nologies upon which they arebased improve. In my view, indi-vidual environmental control is in-deed a concept whose time hascome.

Development of individual control Our firm first started working

with individual control conceptsnearly two decades ago. At thattime, we designed and tested asmall air conditioning unit to bemounted in the modesty panel ofoffice desks with outlets on the

front edge of the desk. The unitdrew air from the floor and dis-charged the condenser cooling airvertically from the rear of the desk.For heating, it employed a smallradiant panel on the modestypanel. The occupant could adjustcooling air flow and set the radiantpanel to any power setting. Maxi-mum power drawn off the unit wasa mere 175 watts, and tests deter-mined that the unit could provide aperceived thermal sensation of ±3to 5 F from the ambient space tem-

February 1997 HPAC Heating/Piping/AirConditioning 65

Percent of occupants likely to complain at this temperaturePercent of occupants comfort curves centered at this temperature

100%

80%

60%

40%

20%

0%70 71 72 73 74 75 76 77 78

Space temperature, F

100%

80%

60%

40%

20%

0%70 71 72 73 74 75 76 77 78

Percent of occupants likely to complain at this temperaturePercent of occupants comfort curves centered at this temperature

Space temperature, F

2 Predicted comfort complaint chart assuming a normal distribution of individualcomfort curves within the building.

3 Predicted comfort complaint chart assuming a polarized distribution of individualcomfort curves within the building.

continued on page 67

February 1997 HPAC Heating/Piping/AirConditioning 67

USER-FRIENDLY BUILDINGS

perature. We were pleased withthe result, but the office furnituremanufacturer with whom we wereworking lost interest in the con-cept, and it was never integratedinto their office furniture.

By the mid 1980s, workstation-based comfort control systemsthat connected directly to thebuilding’s air distribution systemwere becoming available. Today,the workstation-based individualcontrol system still offers the bestcapacity for satisfying individualenvironmental requirements, es-pecially in open office areas. Butthis type of system also has somedrawbacks. The requirement thatthe workstation be connected tothe air system may limit the spacedesigner’s flexibility. Such sys-tems may also negatively affectthe esthetics of the space andmake even minor rearrangementscostly. Furthermore, addingHVAC components to and aroundworkstations adds clutter to spacethe occupants would often like touse to store their clutter!

In the last few years, a numberof new approaches to individualcontrol have been advanced bymanufacturers and design engi-neers. Workstation-focused indi-vidual control products availabletoday range from systems thatprovide a totally stand-aloneHVAC system for each worksta-tion to those that simply employfans in the workstations and per-mit the occupants to direct moreor less ambient air over andaround them. Other approaches,such as underfloor air distribu-tion systems that afford occu-pants the ability to adjust the flowof air around their workstation,have also been successfully em-ployed.

Personal diffuser technologiesAbout the same time the work-

station-based individual controlconcepts were being introduced,work began on other paths towardmeeting the growing demand formore individual control within

buildings. Most air-based HVACsystems employ a single terminalunit for every two to four offices. Asingle controlling thermostat isplaced in just one of the offices,and the rest operate with essen-tially no control at all. This loosezone control is based on the prem-ise that offices sharing a commonbuilding exposure experiencenearly identical heat load charac-teristics. However, improvementsin building envelope technologiesover the last several decades to-gether with variations of internalheat loads now make the heat bal-ance equations of offices widelyvariable even when they share thesame perimeter exposure.

To deal with heat load variationsin different spaces served by a sin-gle terminal unit, engineers begandeveloping designs that couldchange the balance of air flow fromthe terminal unit among the indi-vidual offices or areas it servedbased on the thermal conditions ofeach area. For example, if one of-fice was at its thermal set point, adamper at the diffuser supplyingair to that room (or area) wouldclose, directing most of the termi-nal unit’s supply air to the other of-fices or areas served by that termi-nal unit. This concept has evolvedto what is today called personal

diffuser technolo-gies.

Originally, per-sona l d i f fuserswere connec ted to standard VAVboxes and employed to redistributeair among the areas served by thebox. However, the technology hasevolved to one where the VAV boxnow regulates the downstreamstatic pressure and switches fromheating to cooling, depending oncommands from the personal dif-fusers it serves. Typically, whenconfigured with low pressurerooftop air supply systems, per-sonal diffusers require no interme-diate control boxes or devices atall. As personal diffuser technolo-gies have matured, a number ofapproaches have been taken to im-prove control capabilities. One im-portant improvement is the use ofwireless infrared remote controlsthat allow occupants to make tem-perature adjustments with ease.Fig. 4 shows a personal diffuser inan office application. Here, twopersonal diffusers are employed ina master/subordinate configura-tion. An in-depth explanation ofthis system concept appears in theMarch 1995 issue of HPAC. A moretypical personal diffuser, whichusually fits in a 2 by 2 ft ceiling

4 Personal diffuser installed in an office. This style providesair diffusion by injecting a stream of air into the room. Insetphoto is a closeup of the diffuser. Photos courtesy of Zero Complaints.

continued from page 65

USER-FRIENDLY BUILDINGS

grid, is shown in Fig. 5.Most personal diffuser designs

do not offer complete individualcontrol because personal dif-fusers are typically connected toan air supply system serving sev-eral offices or spaces and cannotsimultaneously provide heatingto one diffuser and cooling to an-other. But personal diffuser tech-nology does go a long way towardimproving building environmentsfrom the days of one thermostatfor every three or four offices.Each personal diffuser employsan integrated space temperaturesensor to provide local tempera-ture control. Furthermore, thepersonal diffuser technology canbe easily added to any type ofstandard overhead air dis-tribution system commonin office buildings today.And it is an economicaltechnology, often addingonly modestly, if at all, tothe overall cost of a tradi-tional HVAC system. Insome ways, the personaldiffuser is a true cross-roads product. Althoughits name suggests individ-ual control, it really cannotprovide distinct thermalenvironments, especiallyin open office environ-ments. What the technol-ogy can do is provide amore precise, uniformspace temperaturethroughout the building.The question is, will thistechnology be enough tosatisfy building occupants,or will they be captivatedby the possibilities it suggestsand want more?

Industry developmentsThere is by no means a consen-

sus on the future of individual en-vironmental control in buildings.As indicated earlier in this article,some still believe it is more usefulto apply advanced control tech-nologies to maintain rigidly uni-form thermal conditions in build-ings. Others are concerned that

the cost premium for individualcontrol will keep it out of main-stream designs for the foreseeablefuture. Those involved in the indi-vidual control market are experi-encing an expanding demand forsuch systems, but the marketshare for these products is stillquite small. Few anywhere in theindustry will go on record withpredictions for the role of individ-ual control.

Since our firm’s experienceswith an individually controllabledesk-based comfort system, I havebeen convinced that a confluenceof trends would some day make in-dividual control-based HVAC/lighting the obvious choice forcommercial buildings. Emerging

digital control and network tech-nologies have continuously re-duced controls costs for individualcontrol concepts while at the sametime, building occupants havebeen increasing their demands forbetter comfort and control overtheir work environments. Also,building owners and managershave been looking for amenities toattract potential tenants who areincreasingly working at theirhomes. These trends increase the

attractiveness of individual envi-ronmental control. Now, the emer-gence of deregulated electric utili-ties, some of which are anxious toinvest in high-quality buildingcomfort systems, may radicallyshift the way engineers viewHVAC system design.

Buildings in the next centuryThe term human habitat is used

to describe the places where wespend our time because they arethe places where we feel most com-fortable. Office buildings have his-torically not been seen as con-formable environments. Instead,people flock to them simply be-cause their jobs require it. Now,modern technology offers us

choices. We can now benearly as “connected”from our homes as at ouroffices. This trend to for-sake the office environ-ment and, despite theproblems of isolation anddisruptions, have employ-ees work primarily athome is growing. If officebuildings are to retaintheir economic viabilityand flourish into the nextcentury, we engineersmust sharpen our focuson comfort. To do so, wemust reach beyond a de-sign methodology thatsees comfort as a simplematter of maintainingheat balances. Instead,we must implement de-signs that make buildingsmore environmentally at-tractive. Individual com-

fort control is an obvious vehicle forturning unfriendly offices into de-sirable worker habitats. HPAC

Background information on technolo-gies discussed in this article are avail-able at www.hartmanco.com/engr.Any questions or comments about thisarticle may be addressed to Mr. Hart-man at tomh@hartmanco.com orCompuserve 104067,3463. ThomasHartman is a member of HPAC ’sBoard of Consulting and ContributingEditors.

68 HPAC Heating/Piping/AirConditioning February 1997

5 Technician installing a lay-in style of personal dif-fuser. This unit mixes supply air with room air by main-taining a fixed outlet velocity at reduced supply airflows. Photo courtesy of Warren Technology.

ADVANCED CONTROLS CONCEPT

By THOMAS HARTMAN, PE,Principal,The Hartman Co.,Marysville, Wash.

Two of the greatest recentt e c h n o l o g i c a l b r e a k -throughs in our industry

have been variable frequencydrives (VFDs) for AC motors anddirect digital control (DDC) net-works. But if there were a lawagainst under utilizing either ofthese technologies, I fear allHVAC engineers would be in jailtoday; VFDs and DDC networksare the least effectively appliedtechnological tools in our indus-try. Difficult as it may seem,many designers are still ques-tioning whether these two tech-nologies are even useful in pro-j e c t s w h e r e t h e i r p r o p e rapplication could reduce energy

use and improve overall systemperformance by an order of mag-nitude or more.

Variable speed drives and net-work-based DDC controls can beapplied to nearly all of the majorHVAC components—from fans topumps to chillers. In fact, an in-creasing number of these are be-ing installed with VFD speed reg-ulation and DDC control. But,make no mistake about it, eventhough variable speed drives andDDC controls are becomingwidely employed, the technologyis overwhelmingly underutilized.To show why and offer sugges-tions for improving the situation,this article uses the example of achilled water distribution pump-ing system in which a variableflow chilled water pump servesmultiple loads with modulating,two-way control valves (Fig. 1).

Here, a variable speed pump isoperated to maintain a differen-tial pressure setpoint for a distri-bution system that serves a num-ber of loads. Each load has amodulating valve operated by aDDC controller. Fig. 1 also showsthe DDC network that connectsall controllers together such thatthe value and status of all con-nected points are accessible froman operator workstation. Notethat the network is used to trans-mit the differential pressurevalue to the controller operatingthe pump. So, we have a variablespeed pump and networked con-trols operating the system.

W h a t ’ s w r o n g w i t h t h i s picture?

Pump and fan lawsFan and pump laws dictate

that a centrifugal pump can sup-

80 HPAC Heating/Piping/AirConditioning November 1998

Packaging DDC Networks withVariable Speed Drives

By designating complexnetwork-based controlsequences as packaged

products withresponsibility for theirperformance vested tomanufacturers, it maybe possible to simplifydesigns and increase

the utilization of VFDsand other advancedcontrol technologies

in the field

VFD

VFD

Network to othercontrollers andoperator workstation

Chilled watersupply

Chilled waterreturn

C

C

C C C C

DDC point connection

DP

DP

T

T

T T T

DDC controllerTemperature sensor

Variable speed drive

Differential pressure sensor

DDC communications network

1 Typical variable flow chilled water distribution system.

ply 50 percent of design flow at50 percent speed and requireonly 121⁄2 percent (0.5 cu) of thefull flow power, but only if thepressure at which the fluid issupplied is permitted to fall to 25percent (0.5 sq) of the full flowpressure. If the supply pressuredoes not fall with the square ofthe flow requirements, then thepump speed cannot be reducedas the law permits. This resultsin a reduction in pump effi-ciency. These relationships areshown in Fig. 2.

A zone of highest pumping effi-ciency throughout the speedrange of the pump is highlightedin blue. If the system curve ap-proximates that of a simple circu-lating system (e.g., the head pres-sure requirements fall with thesquare of the flow), the systemcurve is approximated by CurveA. Such a system curve ensuresthat the pump operates at thehighest pumping e f f i c iencythrough all flows and that thepower required falls with thecube of the flow requirements.The system curve for a typicalvariable flow distribution sys-tem, as represented in Fig. 1, isshown by Curve B. In Curve B,the pumping efficiency quicklyfalls as flow requirements are re-duced.

Part load operationThere are actually two costly

energy penalties for systems thatfollow Curve B. Loss of pump effi-ciency is one of them. The secondis the high pump-head pressurerequired at decreased loads. Be-cause the majority of variable flowpumping systems are operated tomaintain a fixed differential pres-sure between the supply and re-turn distribution lines, the pump-ing pressure is not permitted tofall significantly as the load de-creases, and a substantial energypenalty is the result. This combi-nation of reduced pump efficiencyand a higher head pressure limitsopportunities for reducing pumppower at part-load conditions.

Regarding chillers, manufac-turers try to remedy this loss byoffering screw compressors orother positive displacement-typecompressors that do not lose effi-ciency under such conditions.

Multiple units are often em-ployed to reduce part-load ineffi-ciencies. But these approaches donot attack the second penalty,which is the extra power require-ment due to the high head re-quirements (that are unneces-sary) at low flows.

Centrifugal fans, pumps, andcompressors are simple devicesthat are very efficient at theirsweet spots (the high-efficiencyzone in Fig. 2). It is reasonable tofind ways to accommodate limita-tions of these devices with sys-tem concepts permitting thepressure to decrease as flow de-creases to operate at the highestpossible efficiencies throughoutall flow conditions.

The value of such an approach

November 1998 HPAC Heating/Piping/AirConditioning 81

Desi

gn m

axim

um p

ump

head

cap

acity

, per

cent

Design maximun flow, percent0

02010 30 40 50 60 70 80 90 100 110 120 130 140

20

40

60

80

100

120

140

1750 rpm pump speed

1450 rpm pump speed

1150 rpm pump speed

900 rpm pump speed

600 rpm pump speed

Curve "B" for systemsoperating with fixeddifferential pressure control

Highefficiency

area

Curve "A" for systemsoperating accordingto a "natural" system curve

2 Pump/system curves for pressure and network pump controls.

3 Cooling load profile for comfort conditioning in two major climates.

Oper

atin

g ho

urs

at lo

ad, p

erce

nt

Design maximum cooling load, percent9585756555453525155

25

20

15

10

5

0

Cooling load profile – tropical climatesCooling load profile – temperate climates

continued on page 83

ADVANCED CONTROLS CONCEPT

November 1998 HPAC Heating/Piping/AirConditioning 82

is illustrated in Fig. 3, whichshows the expected load profilesfor chilled water plants employedfor comfort conditioning. Fig. 3also shows the percent of totaloperating hours the systemspends at various load capacitiesfor two basic climates. Assumingthe chilled water flow is propor-tional to the load, Fig. 3 approxi-mates the flow requirements fora distribution system to the loadsserved. Note that the overwhelm-ing majority of pumping hours isspent at reduced flows in both cli-mate types. Systems operatingaccording to Curve A in Fig. 2 canachieve enormous energy savingsover Curve B systems at theseflow conditions.

Change is neededRules for sizing valves and cer-

tain pump control features pre-date the availability of variablespeed drives. To achieve goodcontrol, it is still recommendedthat control valves be sized sothat 25 to 50 percent of the fullflow system pressure drop occursacross the valve.

Rules of thumb in hydronicsystems vary, but it is common tosee systems configured so that atfull flow the control valve pres-sure drop is one-third of the totalsystem head; the pressure dropthrough each load is one-third ofthe total head; and the remainingone-third consists of piping andall other pressure drops. This isthe criterion employed for CurveB in Fig. 2 wherein the differen-tial pressure setpoint is abouttwo-thirds of the total full flowsystem head. While these rulesdid provide successful control formany applications back in thedays of simple pneumatic con-trols, the solution is energy in-tensive and also has negativeside effects.

For example, consider the Fig.1 system operating at low loads.Because the flow through the coilis low, the pressure drop acrossthe coil approaches zero. How-

ever, the pump operates to main-tain a constant pressure acrossthe valve and coil, so the pres-sure drop across the valve in-creases. This increase in pres-sure across the valve at low flowsreduces “controllability” and isone reason why control instabil-ity is more likely to occur at lowload conditions.

New rules are neededThe old valve-sizing rule of

thumb served an important pur-pose when linear pneumatic controllers were prevalent, but ithas always had significant limita-tions, and it is not an energy-efficient approach in the era ofvariable speed pumping. Knowl-edgeable designers recognize immediately thetwo questions thathave to be answeredto move beyond thisreliable, but less ef-ficient, approach.The first questioni s “What i s themost effective de-sign direction totake?” and the sec-ond is “How can Ibe sure it will bei m p l e m e n t e dsmoothly withoutwreaking havoc onmy projects?”

The answer tothe first questionis to employ theD D C n e t w o r krather than differ-ential pressure tooperate the pumpspeed. Such an approach wouldpermit the pump to slow, so longas all the loads are satisfied,without regard to a differentialpressure. Such control is nowpossible because DDC technolo-gies are maturing and developingmore sophisticated software con-trollers.

Control manufacturers havehad sufficient experience withsoftware loop control to add fea-tures like variable gains, anti-

windup, and self-tuning capabili-ties. Perhaps the most importantfeature that is evolving with DDCsystems is their ability to “net-work” points among various con-trollers. It may be time to considerputting these features to work.

The two primary reasons therule of thumb concerning valveselection seeks a relatively highpressure drop across controlvalves are 1) to ensure linear con-trol response, and 2) to ensurethat the operation of each valveis undisturbed by the action ofadjacent valves. If the full flowpressure drop across the valve islow, then linear response is lost.One valve suddenly opening orclosing as its load changes statecould change conditions at an ad-

jacent load.It is not difficult

to imag ine that instabilities couldo c c u r a s t w o o rm o r e a d j a c e n tvalves each correctthemselves in re-sponse to changesof the other. Thistype of instabilitywould have beenenormously trou-b l e s o m e i n t h edays of pneumaticcontro l lers , butnetworked DDCsystems can ac-commodate suchchanging condi-tions smoothly andeasily.

Imagine that thecommunications

network in Fig. 1 is used to con-trol the variable speed pump andvalves instead of the differentialpressure sensor. Each valvecould be modulated not only ac-cording to local load changes butalso in response to systemchanges as communicated overthe network. The pump could beoperated to meet flow require-ments as efficiently as possible. Asystem schematic that providessuch control is shown in Fig. 1—

While the industryemploys both

variable speed drives and

networked DDCtechnologies with

increasingregularity, the fullenergy savings and

performancecapacities of these

remarkabledevelopments arerarely achieved.

continued from page 81

▼ ▼ ▼

ADVANCED CONTROLS CONCEPT

just remove the differential pres-sure sensor. Properly pro-grammed, such a configurationcan permit the chilled water dis-tribution pump to operate in thehigh-efficiency zone of Fig. 2 atnearly all conditions. The controlvalves for each load can be sizedfor a significantly lower pressuredrop, thus reducing the total sys-tem head at all loads and reduc-ing the size of the pump.

In simulations and field tests,such chilled water distributionsystem configurations have beenshown to cut the required annualpumping power by one-half totwo-thirds below that of a tradi-tional system. This is exciting.These designs provide savings inboth first costs and annual en-ergy costs, resulting in energysavings in the range of $1 to morethan $3 annually per gpm ofchilled water pump capacity.

A change worth pursuingSuch an advance in pumping

control technology is an excitingprospect, but the second questionstill begs an answer. How can thedesigner be sure the new ap-proach will be implemented effec-tively and without hassles? It isnot too difficult to envision thebasic changes in controls and op-eration required to capture thesesubstantial pumping energy sav-ings; however, designers cringeat the prospect of trying to de-scribe networked solutions in asequence of operations. Most de-signers will dismiss entirely thechances that a controls contrac-tor will provide a trouble-free im-plementation of a network-basedsequence no matter how well it isdescribed.

The unhappy truth seems to bethat the process by which controlsequences are implemented is anenormous part of the problem inraising the level of DDC technol-ogy applied in the HVAC industrytoday. Most designers realize thatthe valve sizing rules result in apremium for pump-power costs,but they are reluctant to change a

design approach they know can beimplemented without trouble-some startup problems.

ProductizationTo try to develop a solution to

this dilemma, imagine that thepump applications technologyoutlined in this article was aproduct that designers couldspecify, such as air handlers orVAV boxes. In these products, itis the product supplier and notthe designer who assumes ulti-mate responsibility for the per-formance of the product. Thatperformance is based on sizingand other general informationprovided by the designer in theconstruction documents.

Software “products” are widelyused today but not as elements indesign documents for the build-ing construction industry. How-ever, by designating certain morecomplex ne twork -based se -quences as products whose basicfeatures are generally under-stood by a wide segment of ourindustry, it may be possible to de-velop an implementation paththat is similar to that of otherproducts. This process of “pro-ductization” is a promising modelfor successfully implementingnew network-based applicationcontrols.

Manufac turers o f re la tedHVAC components and controlsmanufacturers are both logicalchoices for delivering network-based software applications. Mycompany is working with inter-ested firms in these two cate-gories to try to define and modu-larize reasonable networkingtechnology “packages” such asthe example of low power pump-ing outlined in this article.

Other network application tech-nologies that can be modularizedin similar fashion are chiller plantcontrol, VAV system control, andcomfort controls operated by indi-vidual tenants. The improvementin performance and reduction inenergy available by implementingsuch networked controls are sub-

stantial. We in the industryshould consider working togetherto package specific networkedtechnologies so they can be em-ployed efficiently and effectivelyas applications arise.

To get the benefit of all voices inour industry, I urge interestedreaders to visit our Website fre-quently over the next year to get abetter understanding of and pro-vide comments to a developing vi-sion as to how networking tech-nologies can be packaged mosteffectively. I also urge readers towrite to HPAC to voice your opin-ions on this important topic.

Summary and conclusionWhile the industry employs

both variable speed drives andnetworked DDC technologieswith increasing regularity, thefull energy savings and perfor-mance capacities of these re-markable developments arerarely achieved. The present de-sign and implementation processthat places complete responsibil-ity for the implementation ofsuch technologies on the designeris effectively blocking such ad-vancements in our industry.

A promising prospect thatcould change this situation is thedevelopment of DDC network“products.” To undertake thischallenge, the industry needs tohear all interested voices to de-termine if and how such pack-aged products can be developedand implemented in systems toachieve higher performance. HPAC

Mr. Hartman is a member ofHPAC’s Board of Consulting andContributing Editors. Additionalinformation on technologies dis-cussed in this article is available atwww.hartmanco.com. Any ques-tions or comments about the articlemay be addressed to Mr. Hartmanat tomh@hartmanco.com. See pg. 7for contact information.

84 HPAC Heating/Piping/AirConditioning November 1998

Circle 504 on reader service card if this article was useful; circle 505if it was not.

THE HARTMAN LOOP

ALL-VARIABLE SPEED CHILLER PLANT DESIGN ANDOPERATING TECHNOLOGIES

FREQUENTLY ASKED QUESTIONS

Last UpdatedSeptember 7, 2001

The Hartman Companywww.hartmanco.com

755 County Road 247Georgetown, Texas 78628Phone: 254-793-0120FAX : 254-793-0121

E-mail: thartman@hartmanco.com

©The Hartman Company September 7, 2001

Table of Contents

12SUPPORT FOR LOOP DESIGN & OPERATION . . . . . . . . . . . . . . . . . . . . . .

11LOOP EQUIPMENT REQUIREMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8LOOP PLANT OPERATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7LOOP RETROFITS OF EXISTING PLANTS . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6CONSTRUCTION COST ISSUES FOR LOOP PLANTS . . . . . . . . . . . . . . .

5LOOP PLANT DESIGN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4LOOP CHILLER PLANT ENERGY USE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2GENERAL QUESTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001

INTRODUCTIONHartman “LOOP” all-variable speed chiller plant technologies offer revolutionary reductions inchiller plant energy use with only modest alterations in plant equipment configurations. Thisdocument is intended as a supplement to other technical information with answers to practicalquestions that arise when owners, operators and designers consider employing LOOPtechnologies in specific applications. If additional technical information is desired, please visit TheHartman Company website at http://www.hartmanco.com or contact Hartman by e-mail for theinformation you desire.

Hartman LOOP chiller plant design and operating technologies are straightforward to apply, butcertain aspects of these technologies such as chiller loading and sequencing control are quitedifferent from traditional chiller plant operating concepts. Owners, operators and designerscontemplating the use of these exciting cost reducing technologies often have questions aboutwhether their application is well suited to LOOP technologies.

As a part of developing LOOP technologies, Hartman has initiated testing of LOOP features, andhas discussed aspects of LOOP operations with a wide range of industry experts that includeowners, designers, and equipment manufacturers. LOOP technologies have been evaluated formany climate regions and facility types. Hartman has vigorously pursued all issues that have beenraised in these tests and discussions such that an owner can be confident a LOOP plantimplemented anywhere in the world will achieve the level of performance projected by theperformance calculations without unanticipated costs or additional work.

It is our firm's intent to keep LOOP technologies completely open to all in the industry. This FAQdocument is intended to provide information to answer questions being asked when LOOPtechnology is first considered for a project. Detailed design and operating information is alsoavailable. To recover our considerable development and support costs, LOOP technologies arepatented and a low cost site license is required for each application. We have tried to make thelicensing process as simple and easy to use as possible. We are always eager to hear our users'comments about improving the information and licensing procedures. If you have any ideas, Iwould like to hear from you!

I hope this information helps to answer your questions about the LOOP chiller plant technologies,and I trust you will not hesitate to contact us if you have additional questions or comments. Letus know if you have further questions or need more information so you can better evaluate thecost and energy impact of LOOP technologies on your chiller plant.

Tom Hartman

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 1

GENERAL QUESTIONS

1. Why is this called “LOOP” technology?ANSWER: During the development process Hartman named these “LOOP” technologies becausethey employ integrated “closed LOOP” controls for the entire plant whereby the operation of allchillers, pumps and towers is coordinated in order to optimize total plant efficiency under allconditions. Most equipment in conventional plants operates in stand alone fashion, respondingonly to certain temperature setpoints rather than operating in coordination with relatedequipment.

“LOOP” is also used to describe the chilled water distribution system in this technology whichinvolves a fully determinant, single circuit chilled water loop instead of the common but lessefficient indeterminate primary/secondary systems.

2. What is new about LOOP technologies?ANSWER: The most significant improvement of LOOP all-variable speed chiller plant anddistribution systems over conventional plants is the significantly reduced energy use in comfortconditioning applications. Under LOOP operation, the entire chiller plant annual energy useusually averages about 0.6 kW/ton or less for most comfort cooling applications. This representsan annual energy reduction of 25% to 50% (depending on climate and application) below themost highly optimized conventional configurations of components of the same operatingefficiencies. In LOOP chiller plants, all components are variable speed and chiller sequencingendeavors to keep chillers and towers operating at lower loads and flows rather than sheddingthem to keep the on-line equipment at high loading as in conventional plant operating strategies.

To achieve these higher levels of performance, an entirely new approach to operating theequipment in chiller plants has been developed. Aside from the use of variable speed drives for allpumps, chillers and tower fans, LOOP plant configurations are very similar to conventional chillerplants. It is how the equipment is sequenced and operated with simple, straightforward and stablenetwork based controls that is really new.

3. In what applications are LOOP technologies most effective?ANSWER: LOOP technologies have been developed specifically for chiller plants that servecomfort conditioning loads. Industrial process loads may be suitable for LOOP technologies if theprocess loads are variable because LOOP technologies reduce energy use only at part loadconditions. Warm, dry climates usually offer the best savings opportunities, but LOOPtechnologies offer huge savings in comfort conditioning applications all over the world. Theenergy savings calculator at The Hartman Company website can estimate energy savings for achiller plant that is employed for comfort conditioning in any of more than 200 different climates

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 2

worldwide. To estimate the potential reduction for an industrial process cooling plant, contactThe Hartman Company with the plant’s estimated load profile information.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 3

LOOP CHILLER PLANT ENERGY USE

1. How are the savings estimates calculated that are employed in the web site“Energy Savings Calculator” and how accurate are they?

ANSWER: A combination of hourly simulation and spreadsheet calculations are employed for thiscalculator and designers can see the comparison results for any application they wish to considerusing the Savings Calculator at the Hartman website. This on-line program has been thoroughlytested and provides a good first cut estimate; savings are likely within 10% to 30% of the truevalue. Care has been taken to ensure the savings estimates in the savings calculator are notinflated. More accurate estimates can be easily developed when additional information about theload being served is available. The most useful additional information is the actual cooling loadprofile (e.g. the percent of time the plant spends operating at various loads) and/or total chillerplant operating hours for a typical year. At present, this additional information cannot be inputdirectly into the on-line calculator, but if you have this information, you can e-mail it to us usingthe “contact us” page, and we will quickly send you back a corrected savings calculation that isadjusted by this additional information.

2. What efficiency equipment is used to calculate energy savings and what would bethe effect on savings if chillers of different efficiencies or towers of differentapproaches were employed?

ANSWER: Savings comparisons between LOOP plants and optimized conventional plants arecalculated by simulating chillers with nominal full load ARI rated efficiency of 0.62 kW/ton andtowers with 8 to 10oF approach. However, the percentage savings between LOOP andconventional plants is independent of chiller efficiency or tower approach temperatures. Thus, asefficiencies of chillers and related equipment are improved, LOOP configuration efficiencies willalso improve offering approximately the same percentage of energy savings over theconventionally configured and controlled chiller plant.

3. Is water side economizing used in the LOOP system to achieve reduced energyuse?

ANSWER: No, the savings estimates do not consider water side economizing but a direct towercooling cycle could be incorporated into a LOOP chiller plant just as it can into conventionalplants. However, the energy saving comparisons for LOOP plants are not based on varyingequipment configurations or equipment efficiencies, but on straight-across comparisonsemploying identical equipment efficiencies, approach temperatures, and weather and load data.The only change made to compare LOOP performance with conventional plants is that theequipment in the LOOP plant is operated by variable speed drives and employs network basedLOOP control technologies for sequencing and equipment speed control.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 4

LOOP PLANT DESIGN

1. How does a LOOP chiller plant layout differ from a conventional chiller plant?ANSWER: On the chilled water generating side, there is usually little or no difference in layoutbetween a conventional chiller plant and LOOP chiller plant. Designers may decide to employcommon headers or dedicated chilled and condenser water pumps with either system, althoughLOOP control considerations may influence which approach to choose. Though LOOP plantlayouts are the same or very similar to conventional plants, for LOOP configurations employingmultiple chillers, it is recommended that all chillers be the same in size and have the sameoperating characteristics, and it is helpful, though not necessary, to have the same number oftowers or tower cells as chillers.

2. How does LOOP chilled water distribution differ from standardPrimary/Secondary systems?

ANSWER: LOOP chilled water distribution technologies employ a single circuit chilled waterdistribution system based on “Low Power Pumping” technologies that are a subset of the LOOPchiller plant technologies. In these recommended distribution systems, there are no decouplerlines. For single building and small distribution systems Hartman recommends that the same set ofpumps that pump the chillers also pump the distribution system and the loads. In large systems,primary pumps pump the chillers and distribution system (maintaining a neutral pressure in thedistribution system), and “booster” pumps (that are connected in series with the primary pumpswithout decoupling lines) pump each load, major aggregate of loads, or building.

Note that instead of a primary/secondary pumping system, LOOP plants employ primary only orprimary/booster arrangements with the booster pumps in series with the primary pumps. A bypassvalve may be installed at the end of each main to ensure a minimum flow is maintained at all times.Operation of this simple configuration is optimized with network based control sequences.Enormous efficiency improvements come from network optimization of the pumping, load flowcontrol, and the elimination of direct mixing of supply and return chilled water.

3. How can LOOP technologies be applied; must the chiller plant and distributionsystem both employ LOOP technologies for proper operation?

ANSWER: No, LOOP chiller plant technologies are modular. The two major parts are the LOOPchilled water generation technology and the LOOP chilled water distribution technology. LOOPtechnologies can be applied to each independent of the other. It is possible to further modularizeLOOP technologies such that only the heat rejection circuits employ LOOP technologies. This isbeing done cost-effectively for existing chilled water plants that employ constant speed chillersand cannot justify the expense of changing them at this time. However, energy reductionopportunities are substantially increased by implementing a complete “LOOP” network basedsystem to the entire chilled water plant and distribution systems.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 5

CONSTRUCTION COST ISSUES FOR LOOP PLANTS

1. Are construction costs higher for LOOP plants?ANSWER: No, not necessarily. It is sometimes assumed that a cost premium equal to the cost ofthe variable speed drives less the cost of the across-the-line starters for the chillers, pumps andfans will be necessary. However, in many applications constructing a LOOP all-variable speedchiller plant in place of a constant speed plant of the same size and nominal efficiency costs aboutthe same. Below about 80% loading, a LOOP all-variable speed chiller plant configurationincorporating variable speed chillers that are somewhat less efficient at full load, but cost the sameas more efficient constant speed chiller of the same capacity will begin to operate more efficientlythan a conventional plant with the more efficient constant speed chillers. This means that whenchiller plants are sized with a 20% or greater margin of excess capacity, the operating efficiencyof a LOOP all-variable speed chiller plant incorporating equipment of about the same cost willoperate more efficiently even at peak load conditions than a conventional constant speed plantwhich loses efficiency when the equipment is oversized.

So, anytime a chiller plant is oversized for failure or standby protection the nominal efficiency ofthe chiller plant should be based on the actual peak load served by the plant rather than the totalcapacity of the plant. Doing so reduces the nominal full load efficiency requirements of variablespeed plant components and therefore lowers their cost. This cost reduction, along with furtherreductions from effective network control connections offsets the extra cost for the variable speeddrives and allows all-variable speed chiller plants to provide substantial annual energy savingswhile costing about the same to implement as an optimized constant speed alternative of the samecapacity.

2. Are the network controls required for LOOP plants more expensive orcomplicated to operate than standard controls?

ANSWER: No. these controls are usually the same DDC controls that are employed inconventional plants. Most modern DDC systems have the capacity for network control, but it isseldom employed. See the section on LOOP equipment requirements for more information onLOOP control system requirements.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 6

LOOP RETROFITS OF EXISTING PLANTS

1. Is it cost-effective to retrofit an existing constant speed chiller plant to LOOPtechnologies, and if so, how is that accomplished?

ANSWER: Many existing chilled water plants can be cost-effectively retrofitted to LOOP plants.Chiller plants that are located in warm climates and do not employ effective tower optimizationstrategies are the very best candidates as they will provide the greatest annual energy savings.Because LOOP chiller plant technologies are modular, it is possible to upgrade a plant in stages,or to limit the upgrade to only certain elements of LOOP technologies. For example, applyingLOOP technologies to the heat rejection circuits (condenser pumps and tower fans) is almostalways a cost effective measure and therefore a good first step in an upgrade program. A heatrejection LOOP upgrade can be done without any configuration changes to the existing constantspeed chillers. For chiller plants that serve large distribution systems, it is usually cost effective toconvert the distribution system from primary/secondary to primary/booster and upgrade to LOOPlow power pumping technologies. Whether or not it is beneficial to convert the existing constantspeed chillers to variable speed chillers and apply LOOP operation technologies depends on theage, efficiency and configuration of the current chiller plant. The Hartman Company will gladlyprovide preliminary guidance on the costs and energy reductions associated with various upgradeoptions. Simply e-mail or mail the basic plant equipment schedule, configuration schematic, andload profile or description of loads served. You will receive a good first cut estimate of what thevarious upgrade options will cost and save.

2. Can any existing constant speed centrifugal chiller be upgraded to variable speedfor LOOP operation?

ANSWER: Like all existing motors, those that drive centrifugal chillers can be retrofitted withvariable speed drives. Like other motors, if the compressor motor is in good condition, manyyears of excellent operation can be expected. However, there are some limits that may makevariable speed conversion uneconomical. Reasonably priced variable speed drives are generallynot available for high voltage motors (over 600 volts), and for motor sizes on chillers rated atover 2000 tons. Furthermore, some constant speed chillers cannot be significantly slowed due tominimum refrigerant pressure requirements or certain mechanical limitation. Finally, chillers morethan ten years old may employ refrigerants that are being phased out, and may operate at muchlower efficiencies than modern chillers. Chiller plants with older, less efficient chillers may not begood candidates for variable speed upgrades. Because of low efficiency and lack of useful servicelife, it may be better to replace older chillers outright with new variable speed chillers, or focusimmediate upgrade efforts on the heat rejection and distribution circuits. New variable speedchillers can then be retrofitted when the existing chillers reach the end of their useful life, or whenscheduled as part of a refrigerant phase-out program.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 7

LOOP PLANT OPERATION

1. How are the chillers staged in a LOOP chiller plant?ANSWER: Hartman has developed an entirely new method of chiller sequencing called the"Natural Curve" method of sequencing. The "Natural Curve" is a term coined to describe themost efficient operating load point of a chiller at various head (condenser and evaporatortemperature) conditions. Typically, the Natural Curve for a constant speed chiller is at or verynear full load at all head conditions, but for a variable speed chiller the most efficient operatingpoint is at much lower loads and varies with the head conditions. Thus, a curve can be developedfor variable speed chillers that plots their most efficient operating point at various headconditions, and this curve is called the "Natural Curve" of the variable speed chiller.

In this easily applied Natural Curve sequencing method, chillers (and towers) are staged in LOOPchiller plants such that chillers operate at all times closest to their Natural Curve. Typically,LOOP plant chiller shedding occurs at much lower loads than in conventionally operated plants.As the load falls from full load, all equipment is operated at reduced speed until the Natural Curvealgorithm calculates that the plant can operate more closely to the remaining chillers' NaturalCurves if a chiller is shed. The same is true when a chiller is staged on. The exact points of thisstaged operation depend on the characteristics of the variable speed chillers and towers employed,as well as the current characteristics of the load served and the chiller head conditions.

2. Is control of chiller capacity included with speed control in LOOP operation?ANSWER: Yes. Currently, a DDC controller is configured and programmed to operate chillers ina LOOP plant just as DDC controllers typically operate chillers in a conventional plant. Thus, theoperation of variable speed chillers in LOOP chiller plants is very similar to the operation ofconventional plants. A LOOP DDC controller controls both chiller sequencing (on/off control) aswell as the amount of capacity (demand limit) of each chiller. All factory built variable speedchillers include internal logic that is intended to continuously optimize vane and speed control tomeet current conditions and variable speed chillers can also be configured to accept “demand”commands from the LOOP plant controller. Thus it is not difficult to establish LOOP plantoperations with any variable speed chillers. In plants that are retrofit from constant speed chillers,the speed/vane control logic may be externally applied in some circumstances.

3. How are the condenser water pumps and tower fans controlled in a LOOP chillerplant?

ANSWER: In LOOP chiller plants, the condenser water pumps and tower fans are variable speedas are all other components. LOOP control of these components is accomplished with very simplealgorithms that tie the pump and fan operation directly to the power input to the chiller(s). Somespecial care must be taken in the choice of the towers for LOOP operation such that each tower isable to handle a range of flows and still achieve full coverage of its fill and provide efficient

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 8

air/water surface exposure. It is also important that the fixed head requirement of the tower beconsidered when selecting the condenser pump(s). There are, however, many tower andcondenser pump configurations that easily provide the variations in flow required for efficientLOOP operation. While condenser pump speed and tower fan speed are both adjusted inaccordance with chiller loading, the control also employs limits on this preset relationship toensure that maximum efficiency for the plant as a whole is achieved, and that certain temperatureand flow operating limits are not exceeded.

4. How realistic are water flow, temperature values and power requirements that areemployed in sample LOOP energy savings models?

ANSWER: The values for outdoor dry and wet bulb temperatures come for actual weather datafor the specific location chosen. Operating flows and temperatures are all within manufacturers’limits for the chiller or tower to which they are applied. Pump flows and power requirements arecalculated from standard manufacturers’ pump performance curves. Chiller and tower operatingtemperatures and power requirements at each load condition come directly from the ARI andCTI performance data for the equipment involved. This data has been determined from testing atthose specific flows, loads, and temperatures. Therefore the values and power requirements areconsidered to be very realistic.

5. If chillers run longer at lower loads in LOOP plants, does that mean chillermaintenance costs will rise?

ANSWER: No, but this is a very important point. This question was first raised when the LOOPtechnology was being developed. We have discussed maintenance issues with chillermanufacturers and others who agree that chiller maintenance based on component wear will likelybe reduced from the application of LOOP operation. The reason for reduced maintenance despitethe longer operating hours of each chiller are 1) fewer starts, 2) softer starts, and 3) loweraverage loading on each machine. While there is not yet sufficient data to show conclusively thatmaintenance costs are reduced in a LOOP plant, there is strong agreement among those expert invariable speed operations, including chiller manufacturers, that maintenance for wear and tearcertainly does not increase in LOOP plants.

Currently, much periodic chiller maintenance is triggered by runtime hours. It is generally agreedthat for a LOOP plant these should be adjusted or replaced with new PM guides that recognizethe reduced wear per operating hour that LOOP plants achieve. Also, the newer studies that showmechanical failures are not generally reduced by periodic maintenance based only on run time, butonly when maintenance is triggered by vibration, power or other operating anomalies.

6. Are there any new operating or maintenance issues that should be consideredwith a LOOP chiller plant?

ANSWER: Yes. Because LOOP plants operate at reduced condenser water flows at low loads,there is the possibility of a greater rate of condenser tube fouling with a LOOP plant than with a

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 9

conventional plant in some configurations. Because the flow is variable, it is difficult to preciselyplan the frequency of required tube cleaning of a LOOP plant. However, if specific tube cleaningintervals are essential, there is a great deal of flexibility in LOOP plant design and operations thatshould be considered by the designer or plant operations manager. To reduce the frequency oftube cleaning, the designer may decide to employ a three pass condenser bundle. This design canraise the flow rate in the chiller and eliminate any potential problem altogether. Once the plant isoperating, the minimum condenser flow can be adjusted by the plant operators at any time toestablish any tube cleaning intervals that are required. These steps may have a small effect onconstruction or operating costs for a LOOP plant, but they can keep the tube cleaning at presentlevels, or even reduce the frequency of cleaning . Furthermore, there are now several differentapproaches to automatic tube cleaning that can be implemented to ensure that chillers operate atall times with the highest possible condenser heat transfer.

7. How does the owner know if LOOP technologies are performing as projected?ANSWER: The Hartman LOOP Design Guide outlines a simple and low cost means ofintegrating real-time chiller plant efficiency monitoring into the plant controls. This added featurecosts very little, but it is of enormous assistance in operating the plant and managing maintenanceactivities. The energy performance instrumentation provides a continuous readout of the currenttotal chiller plant operation effectiveness in kW/ton and also accumulates data that can becompared with previous periods during which the plant operated under similar load and weatherconditions. This information helps operations staff and management know very quickly when theplant operations stray from projected and historic energy use patterns, and it helps providedirection for getting the operation back on track.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 10

LOOP EQUIPMENT REQUIREMENTS

1. What type of chiller and cooling tower are required by LOOP plants, and are theseproducts readily available?

ANSWER: LOOP technologies are specifically developed for centrifugal chillers. Hartman hasworked with the major chiller and cooling tower manufacturers during the development of theLOOP chiller plant technologies. Optimum LOOP plant configurations require that the centrifugalchiller be variable speed; all major chiller manufacturers make such variable speed chillers suitablefor LOOP operation. LOOP plants also require that the cooling tower be a low head type withgravity or rotating sprinkler hot water distribution. Towers must also be constructed such that thetower works effectively with a condenser water flow turndown ratio of approximately 2.5:1.Many US and international manufacturers of cooling towers make such towers that are suitablefor LOOP plants.

2. Are manufacturers concerned about applying their equipment in LOOP plants,and does LOOP operation have any effect on equipment warranty?

ANSWER: No. LOOP designs never exceed the operating limits for the equipment selected.There is no effect on warranty, and manufacturers are generally pleased to have their equipmentchosen for this ultra-efficient application. While in some locations the local manufacturers’representatives for chillers and cooling towers may not fully understand LOOP technologies orenvision widespread applications, the major manufacturers are supportive of LOOPconfigurations for their equipment.

3. What type of control system is required by LOOP plants to provide networkcontrol?

ANSWER: A LOOP chiller plant can be implemented with any of a number of direct digitalcontrol (DDC) systems to operate the chiller plant and/or distribution equipment. Thefundamental requirements are that:

The DDC system has a functional and flexible programming language that uses floating pointmath and allows multiple layers of custom mathematical calculations and logic statements.You can find more about this requirement by viewing our Operators' Control Language(OCL) guide to DDC manufacturers on our web site.

The DDC system has extensive and automatic network management features such that eachcontroller can employ point & variable data from any other controller in its control programs.

The DDC system employs standard or gateway protocol features such that it can beconnected to communicate with chiller and variable speed drive equipment.

Not all, but many of the existing DDC systems on the market today incorporate these featuressuch that they are suitable for LOOP chiller plant control.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 11

SUPPORT FOR LOOP DESIGN & OPERATION

1. What is the history and present status of LOOP technologies?ANSWER: Hartman first began investigating the benefits of all-variable speed chiller plants in1992. At that time Hartman encouraged manufacturers and industry organizations to develop thispromising technology. However, none elected to do so. Because even at that early date, theimpact on plant energy use appeared to be very beneficial, Hartman determined to make theinvestment to develop the technologies internally. Along with the development, Hartman wasencouraged by industry members to develop a mechanism for ongoing support since concernsarose that plant operators would not be able to tap their normal sources of operations informationto keep all-variable speed plants operating at peak efficiency levels.

Since 1996 when development began in a large scale, Hartman has invested nearly a milliondollars in the development of LOOP technologies and plans to continue to invest at this rate forongoing development and support for LOOP plants that are now being implemented. LOOPtechnologies are protected by three patents (US Patent No. 5,946,926, US Patent No. 6,185,946,and US Patent No. 6,257,007) and one other pending.

In early 2000, the LOOP technologies were released to the public for use in new chiller plantdesigns as well as retrofits of existing plants. Projects incorporating LOOP technologies are nowunder design and construction.

2. What is The Hartman Company’s role in LOOP projects? ANSWER: Hartman continues to be an advanced technology engineering firm and does notrepresent any manufacturer or sell any equipment. Nor does Hartman specifically recommend anyone manufacturer for chillers, towers, pumps or DDC systems. While Hartman can act as thedesign engineer when desired by the owner, the firm’s main focus is on providing implementationsupport of LOOP technologies to design and contracting teams and ongoing operations supportas required to the plant’s operations staff.

3. How are LOOP technologies implemented and what is the cost?ANSWER: Although LOOP technologies are protected by patents, Hartman endeavors to makethem all open technologies subject to industry discussion and testing as any other new technology.There is no need to employ “black box” controls or other secretive devices or products in aLOOP chiller plant. LOOP technologies are implemented with standard mechanical systems andcontrols through standard plan/spec or design-build contract procedures. To recover developmentcosts, Hartman issues a "Site License" for each installation. The fee for each site license is aone-time payment of $5.00 per installed ton of the chiller plant employing LOOP technologies.The site license entitles the licensee to employ all Hartman LOOP technologies at the site.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 12

For additional support, a “Support License” provides the site license; an updated “HartmanLOOP Design Guide,” and a two year Ongoing Operations Support Agreement. The HartmanLOOP Design Guide includes suggested plant configurations, equipment specifications, LOOPcontrol sequences and other important design considerations and guides. The Ongoing SupportAgreement provides support for plant operations staff with on-line information, Q&A resources,and membership in the Internet based LOOP users group. The ongoing support services alsoinclude direct contact channels to LOOP operations specialists for assistance in troubleshootingfor an initial two year period which can be easily extended to the entire life of the plant if desired.The cost for the Support License option is a one-time payment of $10.00 per installed ton.

However, experienced designers know that in order for a newly developed technology to beimplemented effectively, the design team should have access to special expertise in thattechnology throughout the design and implementation process. In LOOP projects, this can beaccomplished with a LOOP “Engineering Agreement” which includes all the features of the siteand support Licenses plus direct engineering support from The Hartman Company to the designand construction team. In Engineering Agreements, Hartman acts as a LOOP technologyspecialist to the design and construction team(s) and provides one or more of the following asdeemed necessary by the design and construction team:

A. Engineering and peer review services to the Owner, designer or contractor to assist indeveloping the configuration of chillers, towers, pumps, piping and controls that is most suitablefor the plant needs and ensures optimal LOOP plant operation.

B. Specifications for critical elements of the plant that include the chillers, towers, and DDCsystem, along with procurement services in order to ensure that the purchase of major equipmentemploys value analysis principles in a “pre-purchase” or similar process that fits the needs of theOwner’s organization.

C. Construction services to review and support the DDC control contractor’s software andhardware design, and assistance to the startup, test and balance, or commissioning agencies inorder to ensure that LOOP hardware and control sequences are implemented correctly and thatoptimum operational efficiency of the plant is achieved.

The cost of an Engineering Agreement varies depending on the amount and areas of support thatare identified as necessary. However, an Engineering Agreement option which includes sitelicense, design guide and ongoing support in addition to this direct engineering support will costapproximately $20.00 per installed ton. Such an agreement is adequate to fully support a designteam new to LOOP technologies but otherwise experienced in chiller plant design. The annualenergy reduction achieved by employing LOOP technologies usually makes the payback for thisone-time fee less than a year.

4. How do I start evaluating LOOP technologies to see if they fit my application?

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 13

To start the process of implementing LOOP technologies into a chiller plant project, the first stepis to see if LOOP technologies energy savings can meet the financial threshold required for yourspecific application. The Savings Calculator on The Hartman Company web site -http://www.hartmanco.com/innovate/savecalc/index.htm - be employed for this purpose bygiving a quick first cut estimate of the savings that can be achieved by implementing the LOOPtechnologies in a specific application. The savings calculator also provides a range of first costdifferentials for implementing a LOOP plant in place of a conventional plant for new, upgrade, orretrofit applications. These broad ranges can be refined by your design team or with our help.

If the decision is made to employ LOOP technologies, the project engineer or owner shouldcontact Hartman to purchase a Site License, Support License, or an Engineering Agreementdepending on the level of support required. With the license or agreement in place, any change inplant capacity requirement that may be made during the design process can be easilyaccommodated by a simple adjustment to the license or agreement. Such flexibility for plant sizeadjustments is built into all licenses and agreements.

HARTMAN LOOP CHILLER PLANTS FREQUENTLY ASKED QUESTIONS

©The Hartman Company September 7, 2001 Page 14

equipment manufacturers for factoryand, ultimately, field installation. Asthe project nears completion, the con-trols contractor goes through the build-ing, adding equipment and connectingthe devices previously supplied to theothers.

Though still common, this practiceof single-sourcing controls is not par-ticularly efficient because it can lead toconstruction delays and/or cost in-creases attributed to delivery and othercoordination problems and can resultin substandard operation or reliabilitywhen components do not fit togetherproperly and fail to perform as ex-pected. In recent years, this approachalso has resulted in an increasingamount of control redundancy, includ-ing duplication of controls instrumen-tation. This is particularly true in thecase of chillers, which now are supplied

from the factory with a complete di-rect-digital-control (DDC) and moni-toring package.

The redundancy in chiller-plantcontrol systems is illustrated in Table1, which shows the monitoring andcontrol points typically included in achiller package and those points thatgenerally are field-installed separatelyin the DDC controls installation.

A TREND TOWARDGATEWAYS

Before the mid-1990s, many manu-facturers of digital HVAC controls andequipment utilizing digital control weresecretive to the point that the commu-nications details of their products oftenwere not made available to purchasers.Today, the trend is for manufacturers toopen communications architecture andencourage (or at least not discourage)

By THOMAS HARTMAN, PE,The Hartman Co.,Marysville, Wash.

In the building-construction indus-try, a design that reduces operatingcosts is appealing, but only when itscost fits the construction budget.

This truism was clear in our minds aswe began implementing a new all-vari-able-speed-chiller-plant design thatprovides significant reductions in an-nual cooling-energy costs. Adopting anapproach employed for many new tech-nologies, we aimed for both lower in-stallation and lower operating costs.However, because the variable-fre-quency drives (VFD) of chillers, pumps,and tower fans in an all-variable-speed-chiller plant add to equipment costs, webegan looking for ways to reduce the to-tal capital cost of this new plant. Wefound at least part of the answer in an-other emerging industry technology—control gateways, which are eliminatingthe need for duplicating sensing andcontrol points in an increasing numberof HVAC applications.

REDUNDANT CONTROLS ANDINSTRUMENTATION

Generally, commercial-building con-struction projects involve a single man-ufacturer and contractor for controls.The job of this contractor is to make allHVAC equipment operate according toa sequence of operations spelled out inthe project specifications. In doing this,the controls contractor applies the con-trol system he represents to all elementsof the project. The contractor suppliescontrollers and actuators, modulatingvalves, dampers, and other equipmentto the mechanical contractor(s) and

Chiller-Plant ControlUSING GATEWAY TECHNOLOGIES

Practice eliminates need for duplicating sensing and control points

Heating/Piping/AirConditioning • January 2000 81HPACENGINEERING

TYPICAL CHILLER-PACKAGE POINTS TYPICAL DDC POINTSChilled-water supply temperature Chilled-water supply temperatureChilled-water return temperature Chilled-water return temperature

Chilled-water flow statusEntering condenser-water temperature Entering condenser-water temperatureLeaving condenser-water temperature Leaving condenser-water temperature

Condenser-water flow statusChiller power (amperes) Chiller power (amperes)

Evaporator refrigerant pressureEvaporator refrigerant temperature

Condenser refrigerant pressureCondenser refrigerant temperature

Compressor discharge refrigerant temperatureOil pressure

Oil temperatureChiller status Chiller statusAlarm status Alarm status

Chiller start/stop command Chiller start/stop commandChilled-water set-point control Chilled-water set-point control

Chiller maximum-demand control Chiller maximum-demand controlTABLE 1. Points included in chiller controllers and chiller-plant DDC controls.

others to communicate with their sys-tems through published interface infor-mation using communication “gateways.”

Many manufacturers of HVAC con-trols and equipment have an in-houseinterface standard that can be employedby others for connection to and commu-nication with their equipment, whileothers are embracing one or more of theemerging industry communication pro-tocol standards. Currently, the primarymethod of connecting different systemsinvolves gateway-type connections, butin the future, systems may employ in-dustry-standard architecture through-out or have gateways embedded in allcontrollers so that connection to othersystems can be made throughout theirnetwork architecture.

The result of this trend is that con-trols and equipment from differentmanufacturers usually can be connected

and interoperated to some degree. Thisdevelopment is being embraced because:

n Manufacturers of equipment suchas air handlers, cooling towers, andVAV boxes, who typically either haveshipped components without controlsor had to factory-install controls and de-vices supplied by others, realize that de-signing their equipment around theirown controls and devices enables themto better control manufacturing costs,streamline production, and capturemore value.

n Design engineers realize that it maysimplify and speed the construction pro-cess and help them satisfy their clients’desire not to be tied to a single control-system vendor.

n Building owners and managers be-lieve that it will facilitate integrationwith other building and computing sys-

tems and improve the “enterprise”-levelperformance of their facility-manage-ment system with features that reach farbeyond the simple HVAC monitoringand control that is the limit of mostbuilding-control systems.

GATEWAYS FOR CHILLERSAND VARIABLE-FREQUENCYDRIVES

As the trend toward gateways devel-oped, the first two types of equipmentto become available with gateway-typecommunication interfaces werechillers and VFDs. For this reason, anall-variable-speed-chiller plant is anexcellent starting point for employingthis new approach.

A comparison of a standard chiller-plant DDC connection and a DDC in-terface with a gateway-type connec-tion can be made by considering figures1 and 2. Figure 1 shows how a conven-tional chiller plant that employs con-stant-speed chillers, pumps, and fanstypically would be instrumented usingdiscrete devices for each DDC I/Opoint. The power- and flow-meteringequipment is included to providechilled-water production and plant-ef-ficiency information to the owner andoperations staff. The DDC interfaceshown in Figure 1 requires 45 separatepoints. An all-variable-speed plantwould be expected to have the same oreven a larger number of discrete points,because each VFD must be operatedboth digitally (start/stop) and via ana-log control (motor speed).

However, by employing a gatewayapproach as diagrammed in Figure 2,the number of discretely connectedpoints required in an all-variable-speed-plant design is reduced to nine.All other input and output points areconnected through the networks. Thislarge reduction in discrete hardwareconnections is made possible by severalnetworking and interconnecting fea-tures of VFDs. Note that the VFD op-erating the chiller compressors in Fig-ure 1 are connected to the chillercontroller, but in the connectionscheme of Figure 2, each also is con-nected to the chiller-plant network.Depending on the chiller connectionto the VFD, this is useful because itmakes information accessible via a net-

CO N TROL GATEWAYS

HPACENGINEERING

82 January 2000 • Heating/Piping/AirConditioning

What Is a ‘Gateway’?

M any digital systems used in building operation today employ rules for com-municating among their units that are unique to their system. That is why

even when two different systems employ the same RS485 or Ethernet local-area-network architecture, it usually still is not possible for them to exchange informa-tion. Only when both systems also employ the same communication protocol,such as BACnetE or LonWorks, can intercommunication be counted upon.

For systems that do not employ the same network architecture and communica-tion rules, a “gateway” (sometimes called “protocol interface”) is required. A commu-nication gateway is a device that connects two networks and converts dissimilarprotocols. Some gateways employed in large wide-area-network and Internet appli-cations are complex and expensive, but those required for direct-digital-control(DDC) systems usually are quite simple. Typical communication gateways consist ofthree elements:

● A communication link that connects the two (or more) systems.● One or more translators (or drivers) that understand the communication rules of

each system being connected so that data streams can be converted into useful in-formation for the system or systems that require the interconnection.

● A simple operating system and section of memory (or buffer) where required in-tercommunication data is mapped so that relevant data is captured and stored until itcan be communicated to the other system(s).

Today, most DDC field panels are built with gateways. Gateways enable simplevariable-air-volume terminal-control networks to communicate on a higher-speedfield-panel communications network. Many such panels also include configurablegateways. Depending on the requirement for intercommunication, such controlpanels can be shipped from the factory with the communications link and driversrequired for intercommunication or these features can be field-installed and pro-grammed. Thus, the cost of gateway integration of systems has fallen dramatically inrecent years and can be expected to approach zero for common systems in the near future.

Many in the industry see gateways as a temporary fix as the move towardstandard communication protocols takes place. However, whether the entirebuilding-construction industry ever will embrace a single communication standardremains an open question.

work connection that otherwise wouldnot be accessible through the chillercontroller. While such a dual-networkconnection may be unusual, it will notcompromise effective chiller operationbecause all operating commands to theVFD still would be coordinatedthrough the chiller controls. The net-work connection to the VFD may beused to report only the power (KW)draw of the chiller.

With pump and fan VFD, the net-work connection is used to turn theVFD on and off, set its speed, deter-mine its status, and read the instanta-

neous KW use of the motor controlledby the VFD. The points that are com-municated to the chiller-plant con-troller by the network are shown inTable 2. We estimate that the reduc-tion in devices and field work attrib-uted to these network connections willcut controls costs by 25 to 30 percent,which is sufficient to pay the premiumfor VFD for all fans and pumps in theplant. With this network-based con-trol scheme and other economies, it ispossible to implement all-variable-speed-chiller plants that cost the sameas or less than and that significantly

outperform conventional high-effi-ciency constant-speed-chiller plants.

PITFALLS OF GATEWAYCONNECTIONS

However, before implementing acontrols design that employs gateways,a designer should be aware of the com-mon pitfalls of gateway connectionsand take steps to avoid them. Thethree most common problems withgateway connections we have seen are:

n Information is not accessibleacross the network as desired. A com-mandable point such as a digital output

HPACENGINEERING

CT1 SS DORelay

CT1 STS DI

AICurrent switch

CT1 KWkw transducer

DDC Ppoint Legend

C

C

C

CT1 STEMP

CT1 ALRM DI

Wtr temp sensor

CT1 HTR DORelay

AI

CDWP1 SS DORelay

CDWP1STS DICurrent switch

CDWP1 KWkw transducer

AI

Wtr temp sensorAI CH1ECDWT

DO CH1 SSRelay

DI CH1Dry contact

AO CH1CWTSP

4-20ma signal

4-20ma signalAO

Al

CH1DMDSP

CH1 KWkw transducer

COND

COND

EVAP

Chiller 1

PCWP1 SS DORelay

PCWP1STS DICurrent switchPCWP1 KW AIkw transducer

AI

AI

AI

CH1 ECDWT

Wtr temp sensor

Wtr temp sensor

Wtr temp sensor

AI CH2 ECDWTWtr temp sensor

Wtr temp sensorCH1 RCWT

CH1 SCWT CH2 SCWTAI

AI

Wtr temp sensor

Wtr temp sensorCH2 RCWT

AIWtr temp sensor

RCWT

CWFLOWAI

AI

Flow meter

SCWT

Chilled-water supply and return to and from the distribution system

DO CT2 SSRelay

DI CT2 STSCurrent switchAI CT2 KWkw transducer

C

CDI CT2 ALRM

Dry contactDry contact

DO CT2 HTRRelay

AI CT2 STEMPWtr temp sensor

DO CDWP2 SSRelay

DI CDWP2STSCurrent switchAI CDWP2 KW kw transducer

AI CH2ECDWTWtr temp sensor

COMP

Chiller 2

COND

EVAP

DO CH2 SSRelay

DI CH2Dry contact

CH2CWTSP4-20ma signal

AO

AO

CH2DMDSP4-20ma signalAI CH2 KWkw transducer

CDO PCWP2 SS

RelayDI PCWP2STSCurrent switchAI PCWP2 KWkw transducer

Point typeDI — Digital input

DO — Digital outputAI — Analog input

AO — Analog output

Point name

DO CT2 HTRRelay

I/O device

kw

kw

kw kw

kw

kw

kw kw

Heating/Piping/AirConditioning • January 2000 83

FIGURE 1. Conventional constant-speed-chiller-plant DDC interface.

(DO) or analog output (AO) may beunable to receive a command across thenetwork because operating capabilitiesacross gateways usually are limited byequipment manufacturers—unless, thatis, requirements are specified clearly.

n Timing and precision issuesacross the network hamper smoothoperation. The network interface maybe so slow that information exchangestake minutes to perform or a gatewaymay have such a large change-of-value

(COV) limit that an analog input (AI)must change significantly before thechange will be reported across the net-work. Both of these limitations maymake cross-network control unstableor erratic if timing and COV reporting-limit requirements are not made clearin the specifications.

n Point value or status becomesunstable under certain conditions. If apoint value exceeds the range that isexpected by the gateway or if commu-

nication across the network is lost, thereceiving controller may not retain adefault value or status. The designermust be certain that these possibilitieshave been carefully considered anddealt with effectively to ensure the op-erational integrity of a network-basedcontrol system.

These potential problems and othersthat can crop up easily make it impera-tive that a designer use great care whenemploying gateway data-transfer de-

CO N TROL GATEWAYS

HPACENGINEERING

84 January 2000 • Heating/Piping/AirConditioning

DDC Point Legend

Dry contact

CT1 STEMP

CT1 ALRM DI

Wtr temp sensor

CT1 HTR DORelay

AI

COND

COND

EVAP

Chiller 1

Dry contact

CT2 STEMP

CT2 ALRMDI

Wtr temp sensor

CT2 HTRDORelay

COND

COND

EVAP

Chiller 2

Wtr temp sensor

Wtr temp sensor AI

AI

AI RCWT AIWtr temp sensor

RCWT

CWFLOWFlow meter

SCWT

Network connection

Network connection

Network connection

Network connection

Network connection

Network connection

Network connection

Network connection

VFDVFD

VFD VFD

VFD

VFD

VFD

Point name

I/O device

Point typeDI — Digital input

DO — Digital outputAI — Analog input

AO — Analog output

AI

VFD

Chilled-water supply and return to and from the distribution system

FIGURE 2. All-variable-speed-chiller-plant DDC interface using network connections to chillers and variable-speed drives.

TABLE 2: Networked all-variable-speed-chiller-plant DDC points.

C H I L L E R - P L A N T - N E T W O R K P O I N T S L I S TDescription Point name Type Range CommentsSystem Bldg System PointChiller plant — networked points scheduleChiller No. 1 control panel Connect via specified interface between chiller controls and chiller-plant controller.

Chilled-water supply temp CP CH1 CWST AI 20F - 70F Range is minimum.Chilled-water return temp CP CH1 CWRT AI 20F - 70FChilled-water flow status CP CH1 CWFLO DI Yes/NoEntering condenser-water temp CP CH1 ECDWT AI 50F - 120FLeaving condenser-water temp CP CH1 LCDWT AI 50F - 120FCondenser-water flow status CP CH1 CDWFLO DI Yes/NoEvaporator refrigerant pressure CP CH1 EVAPP AI See comment. Scale point as required depending on operating pressure of refrigerant employed in chiller.Evaporator refrigerant temp CP CH1 EVAPT AI 20F - 70FCondenser refrigerant pressure CP CH1 CDNP AI See comment. Scale point as required depending on operating pressure of refrigerant employed in chiller.Condenser refrigerant temp CP CH1 CDNT AI 20F - 70FCompressor-discharge temp CP CH1 DISCHT AI 50F - 120FInlet vane position signal CP CH1 VANE AI 0% - 100%Oil pressure CP CH1 OILP AI 0-50 psiOil temperature CP CH1 OILT AI 50F - 160FChiller status CP CH1 DI On/OffAlarm status CP CH1 ALRM DI Alarm/OffChiller start/stop command CP CH1 SSC DO Start/StopChilled-water set-point control CP CH1 CWSPC AO 38F - 54FChiller-demand control CP CH1 DMDC AO 10% - 100%

Chiller No. 1 variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD compressor speed signal CP CH1D RPM AI 0% - 100%VFD status CP CH1D DI On/OffVFD alarm CP CH1D ALRM DI Alarm/Off Connect to provide 0- to 100-percent outside-air damper control. Actuator by AHU mfg.VFD power CP CH1D KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 1 CW pump variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD pump speed command signal CP CWP1 RPMC AO 0% - 100%VFD status CP CWP1 DI On/OffVFD alarm CP CWP1 ALRM DI Alarm/OffVFD power CP CWP1 KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 1 CDW pump variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD pump speed command signal CP CDWP1 RPMC AO 0% - 100%VFD status CP CDWP1 DI On/OffVFD alarm CP CDWP1 ALRM DI Alarm/OffVFD power CP CDWP1 KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 1 tower-fan variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD fan speed command signal CP CT1F RPMC AO 0% - 100%VFD status CP CT1F DI On/OffVFD alarm CP CT1F ALRM DI Alarm/OffVFD power CP CT1F KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 2 control panel Connect via specified interface between chiller controls and chiller-plant controller.Chilled-water supply temp CP CH2 CWST AI 20F - 70F Range is minimum.Chilled-water return temp CP CH2 CWRT AI 20F - 70FChilled-water flow status CP CH2 CWFLO DI Yes/NoEntering condenser-water temp CP CH2 ECDWT AI 50F - 120FLeaving condenser-water temp CP CH2 LCDWT AI 50F - 120FCondenser-water flow status CP CH2 CDWFLO DI Yes/NoEvaporator refrigerant pressure CP CH2 EVAPP AI See comment. Scale point as required depending on operating pressure of refrigerant employed in chiller.Evaporator refrigerant temp CP CH2 EVAPT AI 20F -70FCondenser refrigerant pressure CP CH2 CDNP AI See comment. Scale point as required depending on operating pressure of refrigerant employed in chiller.Condenser refrigerant temp CP CH2 CDNT AI 20F -70FCompressor discharge temp CP CH2 DISCHT AI 50F - 120FInlet vane position signal CP CH2 VANE AI 0% - 100%Oil pressure CP CH2 OILP AI 0-50 psiOil temperature CP CH2 OILT AI 50F - 160FChiller status CP CH2 DI On/OffAlarm status CP CH2 ALRM DI Alarm/OffChiller start/stop command CP CH2 SSC DO Start/StopChilled-water set-point control CP CH2 CWSPC AO 38F - 54FChiller demand control CP CH2 DMDC AO 10% -100%

Chiller No. 2 variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD compressor speed signal CP CH2D RPM AI 0% - 100%VFD status CP CH2D DI On/OffVFD alarm CP CH2D ALRM DI Alarm/OffVFD power CP CH2D KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 2 CW pump variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD pump speed command signal CP CWP2 RPMC AO 0% - 100%VFD status CP CWP2 DI On/OffVFD alarm CP CWP2 ALRM DI Alarm/OffVFD power CP CWP2 KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 2 CDW pump variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD pump-speed command signal CP CDWP2 RPMC AO 0% - 100%VFD status CP CDWP2 DI On/OffVFD alarm CP CDWP2 ALRM DI Alarm/OffVFD power CP CDWP2 KW AI See comment. Scale point as required depending on operating power range of unit.

Chiller No. 2 tower-fan variable-frequency drive Connect via specified interface between VFD network and chiller-plant controller.VFD fan speed command signal CP CT2F RPMC AO 0% - 100%VFD status CP CT2F DI On/OffVFD alarm CP CT2F ALRM DI Alarm/OffVFD power CP CT2F KW AI See comment. Scale point as required depending on operating power range of unit.

HPACENGINEERING

86 January 2000 • Heating/Piping/AirConditioning

Heating/Piping/AirConditioning • Month Year 6HPACENGINEERING

sign schemes. While there is a develop-ing consensus about how gateway in-terfaces should perform, individualopinions vary widely at this early stage,with some manufacturers new to theidea of cross-network operations. Thiscan lead to misunderstandings aboutthe role of each gateway and result inunsatisfactory performance of systemsthat rely on gateway communication.To minimize such problems, I recom-mend that special consideration begiven to specifying each gateway con-nection or interface.

SPECIFYING GATEWAYSWhile much has been written about

how to specify interfaces between sys-tems, much more needs to be devel-oped. My rule for myself is never spec-ify features I do not fully understand orthat are too difficult to evaluate to beenforced. To ensure an effective inter-face, our firm believes one should:

n Base the gateway or interfacespecification on a well-known indus-try standard. We recommend neverbasing an interface specification on asingle company’s “open” standard. In-stead, we specify at least one industry-standard communication-interface op-tion. However, after all of theequipment and systems have been se-lected, it may be useful to entertainproposals for an inter-system interfaceemploying a non-standard “open” in-terface. Such a proposal could be ac-cepted if it results in reduced costs, itcomes with a guarantee that the opera-tional capabilities will not be compro-mised, and the owner agrees that suchchange will not limit the ability to im-plement desired upgrades.

n Describe exactly what data needto be transferred across the interfaceand how that data is to be employedby each system involved. For example,if a chiller must be controlled on andoff and have its maximum power set-ting set across a gateway interface, it isnecessary to call for the chillerstart/stop and demand-control pointsto be networked across the gateway. Italso is necessary to explain how andwhere in the network the commandsare to be made. Finally, it is useful todetail how the information will be em-ployed on each side of the interface,

how often data must be sent and/or re-ceived, and any other information thathelps ensure that the system operateseffectively.

n Assign a single source of respon-sibility and authority in making eachinterface work. From chiller and VFDinterfaces to DDC controls, we recom-mend that the DDC-controls vendorbe responsible for each interface and begiven the authority to determine if theinterfaces provided by the chiller orVFD vendor meet the requirements ofthe specified interface standard andother specification items. If the con-trols vendor concludes that an inter-face does not meet specified require-ments, then the provider of thatinterface must make the changes man-dated by the controls vendor or clearlyshow where the controls vendor erred.This approach greatly reduces the fin-ger pointing that can overwhelm a de-signer when a multi-vendor network isbrought on-line.

By basing gateway connections onindustry-standard communications,fully specifying the data that must betransferred and how it is to be trans-ferred, and assigning a single point ofauthority and responsibility for the in-terconnections, confusion and fingerpointing during installation andstartup is minimized. This approach is agood model for designers approaching afuture in which multi-vendor controlsnetworks likely will become standardpractice.

GATEWAYS: CHANGING HVACCONTROLS FOREVER

As the trend toward gateways andother interface standards is acceleratedby the increasing demands and expec-tations of major industry players, I be-lieve that the way our industry workswill change forever. Controls contrac-tors increasingly will work to fill a seri-ous void in construction services andincrease their sales by recasting them-selves as “system integrators.” As partof this changing role, they will losetheir loyalty to the control-systemmanufacturer to which they histori-cally were bound and start looking toexpand their product line and theirknowledge of other systems so thatthey will be able to offer the most at-

tractive combination of hardwareproducts and capture a larger piece ofthe growing equipment and data-inte-gration market.

Manufacturers of control productswill begin to find the constructionmarket for their products shrinking, asmore and more controls are factory-in-stalled under OEM agreements. Con-tractors will find the construction pro-cess to be simpler and startupeasier—but only if designers have donetheir job well. The design stage, there-fore, will become more important andrequire designers to develop digital-communication expertise that is notwidely available in the industry today.But if designers are successful in devel-oping this expertise and employing itin well-structured controls specifica-tions, then putting together an HVACsystem will, from the standpoint ofcontrols, become more like putting to-gether an office data-communicationsnetwork. It will be easier and less costlythan controls implementation is todayand result in a better-performing system.

SUMMARY AND CONCLUSIONEmploying gateways to connect sys-

tems effectively requires substantial ad-justments to the design and specifica-tion of affected systems. However, theresults are worth the effort, as construc-tion costs and startup problems are re-duced through the elimination of pointredundancy.

Because manufacturers of chillersand VFD are leaders in the trend to-ward gateway connection, employinggateways to connect their equipmentto DDC systems offers a good opportu-nity for designers to begin employingthis important new technology andprepare for a future in which multi-vendor control systems will be common.

Additional information on technolo-gies discussed in this article is availableon the Web at www.hartmanco.com.Questions and comments about the arti-cle can be addressed to the author attomh@hartmanco.com. HPAC

Circle 502 on reader service card if thisarticle was useful; circle 503 if it was not.

Heating/Piping/AirConditioning • January 2000 87HPACENGINEERING

CO N TROL GATEWAYS

A HARTMAN LOOP EXAMPLE

AN ENERGY COMPARISON OF A LOOP CHILLERPLANT WITH CONVENTIONAL CHILLER PLANTS

byTHE HARTMAN COMPANY

NOTE: This document shows a specific example of how a LOOP chiller plant reduces energy usecompared to conventional constant speed or variable speed chiller plant configurations. In order topresent an example with verifiable numbers, this example uses and identifies specific pieces ofequipment. However, any one of a number of different equipment configurations could be employedto achieve nearly identical results. This document is not intended to indicate endorsement or supportof one manufacturer’s equipment over another.

The purpose of this document is to evaluate chiller plant performance for a typical application ata specific operating point to show how LOOP technologies can reduce energy use compared toconventional chiller plants. This example is for a chiller plant that serves a load that peaks at1800 tons of cooling. The load is a commercial office building and the chiller plant consists ofthree 600 ton York centrifugal chillers, each with a nominal 0.6 kW/ton efficiency. Each chiller isconnected to a Marley NCB2A1 Tower 600 ton cross flow cooling tower. For simplicity, aconstant 42oF chilled water supply temperature is assumed at all loads (though actual LOOPprojects employ variable chilled water temperature when possible for additional energy savings).The plant is located in Detroit, and has a load profile shown in the figure below, which is typicalfor such a facility in the Detroit climate. Detroit is chosen for this specific example because theDetroit load profile is typical of cooling load profiles for chiller plants throughout much of theUS. A design tower approach temperature of 8oF establishes an 85oF entering condenser watertemperature at the design condition or 77oF wet bulb.

Page 1 of 5

%fax: 254-793-0121 (ph: 254-793-0120 7E-mail: tomh@hartmanco.com

Consulting Engineers & High Performance Building Designs

/755 County Road 247, Georgetown, Texas 78628

TheHartman Company

Using this chiller plant load profile, this example considers a specific part load operating point of1/3 total plant capacity. Note from the chart above that this chiller plant spends most of its timeoperating at and around this point. We will compare operating energy consumption at this pointfor:

1. a constant speed centrifugal chiller plant2. a conventional variable speed centrifugal chiller plant3. a “LOOP” all-variable speed centrifugal chiller plant

First, consider a constant speed plant. At 1/3 plant capacity, one chiller, one tower, and onecondenser pump are operated. Assume a 3 gpm/ton condenser water flow, and a constant speedtower fan with an optimization strategy aimed at minimizing condenser water temperature.The Marley tower head requirement is 12 ft, the chiller condenser head requirement is 16 ft, andour assumed piping loss is 32 ft. for a total pump head of 60 ft. A B&G 1531, 6BC pump yields anoperating power requirement of 34 hp at these conditions. The Marley NCB2A1 Tower achievesan 8oF approach for an 85oF entering water temperature at design conditions. This tower employsa 25 hp fan. The Marley performance data shows that at a 600 ton load and the 56oF Wet Bulbtemperature shown on the chart above for 1/3 plant capacity, the tower will deliver 69.9oF leavingTower Water temperature with the fan in constant operation. From York constant speed chiller

Page 2 of 5

%fax: 254-793-0121 (ph: 254-793-0120 7E-mail: tomh@hartmanco.com

Consulting Engineers & High Performance Building Designs

/755 County Road 247, Georgetown, Texas 78628

TheHartman Company

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Chiller Plant Loading

0%

5%

10%

15%

20%

Per

cent

of O

pera

tions

Tim

e at

Loa

d

50

55

60

65

70

75

80

Ave

rage

Wet

Bul

b T

empe

ratu

re

Cooling Load Profile Average Wet Bulb Temperature

Chiller Plant Load Profile for Detroit

power curves, the chiller power requirement is 84% of the power at design entering condenserwater conditions. So, power requirements for the entire plant are as follows:

Chiller - constant speed 600 tons x 0.6 kW/ton x .84 Demand = 302 kWCondenser pump 34 hp x 0.746 kW/hp / .92 efficiency = 28 kWTower Fan 25 hp x 0.746 kW/hp / .92 efficiency = 20 kW------------------------------------------------------------------------------------------------------------------------------------TOTAL 350 kW

Now assume a plant with variable speed chillers of the same nominal efficiency. A conventionalvariable speed plant still operates only a single chiller at this operating point. Making allowancefor the VFD losses, the York power curves show that the power requirements from the Yorkvariable speed chiller curves reduces the power from 84% to 75%. Thus the power requirements for this operating point with a variable speed chiller are:

Chiller - variable speed 600 tons x 0.6 kW/ton x .75 Demand = 270 kWCondenser pump 34 hp x 0.746 kW/hp / .92 efficiency = 28 kWTower Fan 25 hp x 0.746 kW/hp / .92 efficiency = 20 kW------------------------------------------------------------------------------------------------------------------------------------TOTAL 318 kW

Now lets look at an all-variable speed LOOP chiller plant in operation. In a LOOP chiller plant,all chillers, pumps and tower fans are variable speed. The operations calculator in the DDCcontroller that operates the plant calculates that two chillers is optimum for this operatingpoint, and sequences the operation as follows:

Chiller Capacity (each for two chillers ) = 50% (300 tons)Effective entering condenser water temperature = 64oF

Because each tower is only loaded to 300 tons, this optimized lower leaving tower watertemperature is obtained with the following operating parameters for each cooling tower:

Condenser Pump Flow (Gpm) = 1440 gpmCondenser Pump Head (feet) = 42 ft.Condenser Pump speed (RPM) = 1470 rpmCondenser Pump motor hp = 19.2 hpTower Fan speed (RPM) = 1440 rpmTower Fan motor hp = 13 hpActual tower leaving temperature = 62.9oF

Note that the 64oF “effective” entering condenser water temperature is based on 1800 gpm flow.To adjust for heat transfer changes due to flow reduction and the decrease in log mean

Page 3 of 5

%fax: 254-793-0121 (ph: 254-793-0120 7E-mail: tomh@hartmanco.com

Consulting Engineers & High Performance Building Designs

/755 County Road 247, Georgetown, Texas 78628

TheHartman Company

temperature in the bundle, the calculator shows a true tower leaving temperature requirement of 62.9oF is equivalent to the 64oF entering water temperature at full flow.

Using these operating parameters, the power requirements are obtained from York, Marley andB&G performance curves and data and adjusted for variable speed drive losses. The total powerconsumed is as follows:

Chillers - variable speed 300 tons x 0.6 kW/ton x .51 Demand x 2 = 184 kWCondenser pumps (variable speed) 19.2 hp x 0.746 kW/hp / .88 efficiency x 2 = 33 kWTower Fans (variable speed) 13.0 hp x 0.746 kW/hp / .88 efficiency x 2 = 22 kW-----------------------------------------------------------------------------------------------------------------------------------------------------------TOTAL 239 kW

The constant speed plant uses 46% more power and the conventionally operated variable speedplant approach uses 33% more energy than the all-variable speed LOOP chiller plant at thisoperating point. By making the same calculation for each of the intervals shown in the abovechart, we find that in this application on an annual basis, the LOOP plant offers more thandouble the energy reduction of a conventional variable speed chiller plant compared to aconstant speed plant. These results are shown in the chart below:

Page 4 of 5

%fax: 254-793-0121 (ph: 254-793-0120 7E-mail: tomh@hartmanco.com

Consulting Engineers & High Performance Building Designs

/755 County Road 247, Georgetown, Texas 78628

TheHartman Company

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Plant Capacity Requirement

0.00

0.25

0.50

0.75

1.00

Tot

al C

hille

r P

lant

Ene

rgy

Use

(K

W/to

n)

0%

5%

10%

15%

20%

25%

Per

cent

Of T

ime

Pla

nt O

pera

tes

At L

oad

Performance Map of Conventional VS Chiller PlantPerformance Map of Hartman LOOP with constant supply CWtemperature

Performance Map of Conventional CS Chiller PlantLoad Profile of Chiller Plant

Comparison of Hartman LOOP vs Traditional Chiller System PerformanceAt various load conditions in Detroit application

This improved energy savings coupled with a simpler plant configuration and other features make LOOP chiller plants cost about the same, or in some cases less than conventional plants toconstruct. Also, in many climates, LOOP technologies makes changing out chillers that employphased out refrigerants a very attractive investment. While it may appear that the patentedcontrol sequences that constitute LOOP chiller plant technologies result in a more complexplant, they do not. In fact, the sequence of operations for a LOOP plant is actually simpler than aconventional optimized chiller plant because LOOP technologies include a new simple set ofparameters that provide much more straightforward control of all plant equipment.

Please review this example to see if the concepts employed hold true for the plant you areconsidering constructing or upgrading. If you have any questions or comments please call, writeor e-mail us, and we will be happy to respond promptly with whatever information is required tocomplete your analysis of this exciting new technology for your facility!

March, 2000

Page 5 of 5

%fax: 254-793-0121 (ph: 254-793-0120 7E-mail: tomh@hartmanco.com

Consulting Engineers & High Performance Building Designs

/755 County Road 247, Georgetown, Texas 78628

TheHartman Company

A market worth tens of billions of dollars awaits

those courageous and creative enough to give

building occupants what they want most

41HPAC Engineering • January 2001

THE COMFORTINDUSTRY:

In designing HVAC systems for commercialbuildings, many engineers believe that comfortand ventilation will be suitable as long as a

minimum outside-air flow is maintained and heatflows are balanced in each zone. What they fail toconsider is that building zones often encompasslarge areas, multiple offices, and a number of occupants, whose perception of comfort and environmental quality is influenced by thermalvariances, radiant energy, the movement and distribution of air, and a host of other factors.

Consider Figure 1, which shows the elements ofa typical HVAC system in a commercial building.When I present this to industry professionals, mostcan cite rules of thumb for sizing and selectingchillers, boilers, air systems, and piping. Few, onthe other hand, can recall basic facts about the influence of humidity, air movement, and radiantenergy on occupant comfort.

What this shows is that while the purpose ofan HVAC system is to keep building occupantscomfortable, environmental quality for individuals seldom is a significant factor in design. Usually, occupants have no direct control

A member of HPAC Engineering’s Editorial AdvisoryBoard, Thomas Hartman, PE, is principal of The Hartman Co., a Marysville, Wash.-based firm that employs new technologies in offering enhanced, individu-ally adjustable, and accountable comfort systems for commercial buildings. More information about thesetechnologies can be obtained at the firm’s Website(www.hartmanco.com) or by contacting Mr. Hartmanat tomh@hartmanco.com.

By THOMAS HARTMAN, PEPrincipal, The Hartman Co.Marysville, Wash.

a 21st-Century Opportunity

Background: Stone/Terry Vine; hand with remote control: Photodisc/Nancy R. Cohen.Composited by: HPAC Engineering.

42 January 2001 • HPAC Engineering

over their thermal environment. Indeed, most do not even have a temperature sensor or thermostat located near them.

This article discusses the emerging“comfort industry,” the forces drivingit, and ways designers, manufacturers,and system integrators can benefitfrom it.

THE ENEMY IS USFor decades, surveys have shown that

building occupants’ biggest complaintabout their workplace is lack of thermalcomfort. A recent Building Owners andManagers Association (BOMA) surveyreaffirmed this, showing that lack of occupant control also is of utmost concern. Yet despite these long-standingcomplaints, low-cost sensing and processing capabilities and other technology enhancements have not beenwidely employed in products and rarelyare sought for HVAC-system designs.

In an effort to understand why occupant desires for improvement havenot been heeded as enabling technologieshave advanced, my firm co-sponsoredand conducted a survey of attendees atthe 1997 BOMA Convention and Office Building Show in Minneapolis.We were interested in seeing if buildingowners and operators were impeding theimplementation of better comfort systems. The results of the survey areshown in Figure 2.

What we discovered was that comfortranks first in importance to both occupants and building owners, whichmeans the HVAC industry cannot reasonably blame developers and build-ing owners for the failure to implementmore comfortable systems. Increasingly,it looks as if, “We have seen the enemy,and it is us.” While buildings may bemore comfortable than they weredecades ago, it is clear that the industry is

not meeting building occupants’ risingexpectations for comfort and control.

INSTITUTIONAL IMPEDIMENTSSo, if technology has made possible

more comfortable buildings, and bothoccupants and building owners wantthem, why are they not being designedand built? The answer is that the building-construction industry has institutional impediments to the type ofchange required to make systems morecomfortable. Such impediments are:

• Innovation is discouraged.The HVACindustry is one of the few technology-

focused industries in which innovationin product design is discouraged. Manufacturers know that if they developa product with features that significantlyset it apart, a substantial portion of themarket will be closed to that product because designers are trained to specifyonly those components that have equalsso that any one of several (usually at leastthree) products can be applied. This isbecause the number of products required

for a typical HVAC system is large, andcurrent construction procedures do notprovide designers with the capacity to negotiate and enforce pricing.

• Pressure on first costs. This is not newto anyone in the building-constructionindustry. Many designers have seen com-fort-enhancing features such as improvedcontrols, smaller zones, and better glaz-ing minimized and even eliminated as final budgets were developed. This hap-pens with surprising frequency and dis-courages the industry from focusingmore on comfort, even when doing so requires only a very small cost premium.

The HVAC industry is one of the few technology-focused industries in which innovation in

product design is discouraged.

T H E C O M F O R T I N D U S T R Y

Building-control network

Tower

DDC

DDC

DDC DDC

DDC

DDC

DDC

DDC

Chiller

Cond.

CC

Fan

Boiler

box

HWR

HWS

CDW

SCD

WR

CWS

CWR

FIGURE 1. Elements ofa typical HVACsystem.

44 January 2001 • HPAC Engineering

• Operations-support issues. The rela-tionships between designers and betweenmanufacturers and building-operationsstaffs never have been strong. All too often, we hear, “The operator will reduceany system to his/her level of understand-ing.” True of all systems in all industries,this should challenge designers and manufacturers to put more emphasis onoperations-support issues. But in theHVAC industry, it often is used as an excuse to not work harder to achievemore effective systems, particularly at thezone (comfort) level.

• Lack of integration. Because theHVAC industry continues to considercontrols and hardware independently ofone another, the functionality of zoneproducts is compromised. Comfort-enhancing features such as integrated occupancy sensing and individual occupant control could be implementedin a zone system easily and inexpensively;however, terminal hardware is crude, andterminal-controller program capacity isaimed at hardware adaptability. Thus,the cost of each terminal unit with separate controls and hardware is highwhile function is low. This lack of controls integration is one of the mostsubstantial impediments to improved occupant comfort.

• Lack of vision. Perhaps the greatestimpediment of all is the profound lack of vision on the part of the HVAC industry as a whole. Although engineers,manufacturers, contractors, developers,

and building operators seem to under-stand the issue of comfort, only a few seevalue in working to provide better comfort solutions. Many are dismissiveof criticisms from occupants, whom theyconsider to be “complainers.”

NEW IMPLEMENTATION PATHSRemoving these impediments would

require a restructuring of the building-

construction process, which is unlikely inthe short term. However, with bothbuilding owners and tenants beginningto view occupants as “profit centers,”change is coming—although it likely willoccur very differently than most in theindustry are anticipating.

Consider that individual-comfortdevices that can be employed with underfloor or ceiling air-supply systemsor incorporated into work stations arebeing developed. Consider also that recently developed technologies that allow occupants seated adjacent to oneanother in open office areas to perceivea difference in thermal comfort of 2-3 Fcan be integrated into low-cost termi-nal comfort devices and that these devices can provide individualizedcomfort billing, just as the phone com-pany provides monthly bills for eachoffice phone. Because these technolo-gies show such great promise, it is rea-sonable to envision that truly effectivepersonal-comfort devices soon can beconnected to standard HVAC systems.

Consider for a moment that suchproducts are available. Because they employ specially developed, integratedfeatures designed for particular applica-tions, the market does not offer equals.Cost depends on a device’s applicationfocus, which varies from, say, a large individual office to a densely populatedopen office area.

That such products could be imple-mented in a building intended for multi-ple tenants through a change in thebuilding-construction process is notlikely for several reasons. First, becausethe products do not have equals, thespecifier would lose the ability to controlthe cost of the products. Second, becausebuilding owners often have little idea

T H E C O M F O R T I N D U S T R Y

PERC

ENT

OF A

TTEN

DEES

60

50

40

30

20

10

0ExtremeSubstantialDon’t know SomeNone Little

Superior comfortFlexible and easily configuredEfficient and cost-effective servicingTemperature and occupancy sensingIncreased worker productivityDetailed monitoring of energy usageAutomatic self-diagnostics

FIGURE 2. Perceived value of improved zone-control features.

While buildings may bemore comfortable than

they were decades ago,it is clear that the

industry is not meetingbuilding occupants’

rising expectations forcomfort and control.

45HPAC Engineering • January 2001

about how space in a new building willbe utilized, the types and number of eachtype of products would not be known atthe time the building was built.

It is becoming increasingly clear that abuilding’s tenants—not its owner—should determine the types of comfortsystems to employ in offices. This is important because if a mechanism tomarket end-use comfort systems directlyto tenants were developed, many of theinstitutional impediments listed abovecould be avoided. Consider that employ-ers normally spend $8,000 to $12,000per employee on workspace fixtures andequipment to enable each employee toperform his or her tasks effectively. In thiscontext, the cost of a personal-comfortdevice does not seem large, especiallywhen one considers that, according tothe aforementioned BOMA survey,comfort is regarded as not only the mostimportant aspect of a workspace, but thearea most in need of improvement. Marketing directly to tenants wouldleave value issues to be weighed by employers, who would benefit from theinvestment in improved comfort fortheir employees, and eliminate the needfor equals, as tenants would appreciatedifferentiating features and have the abil-ity to negotiate for products best-suitedfor their particular work environment.

A NEW PARADIGMMarketing end-use terminal comfort

components more or less independentlyof an HVAC system would not be difficult because, as shown in Figure 1,HVAC systems typically are quite discon-nected from the distribution system ateach occupant’s workspace. Furthermore,building projects increasingly havemoved the purchase and installation ofterminal distribution equipment to thetenant build-outs. Often, a credit is ap-plied, and the tenant is required to adhereto certain building standards regardingthe selection and layout of equipment.

The ease of implementing individualterminal control is further reinforced byFigure 3, in which an individually controllable terminal comfort device hasreplaced a standard diffuser at the workstation. Such a personal-comfort devicecould be ceiling-mounted as shown,work-station-based, or part of an under-floor distribution system. Functional individually controllable terminal com-fort devices are becoming available fornearly every type of delivery system.

Figure 3 shows that the most impor-tant network connection for an end-usedevice is not to the building-control network, but to a network that allows occupants to request comfort adjust-ments from their personal computers and

enables the exchange of data necessary toprovide a record of activity for end-use accounting and/or billing purposes. Withthe communication standards availabletoday, it would not be difficult to connectthe two networks and achieve certain optimization functions. It also would notbe difficult to make individual-comfortproducts completely independent of anexisting HVAC system so that such acomfort-system network could be imple-mented in an older building lacking amodern building-system network.

To be cost-effective, an integrated,individually controllable terminalcomfort product must enhance employee performance by about one-quarter of 1 percent or reduce sickleave by about a half-day per year per employee. Studies generally showboth performance gains of 5-15 per-cent and sick-leave reductions of atleast several days.

THE POTENTIAL OF THECOMFORT INDUSTRY

As noted earlier, a recent BOMAsurvey found that lack of comfort andlack of individual control are buildingoccupants’ two biggest complaints abouttheir workplaces. Integrated technologiesthat permit microclimates at each workstation through the use of radiant thermal

Building-control network

DDC

DDC

DDC DDC

DDC

DDC

DDC

DDC

Chiller

Cond.

CC

Fan

Boiler

HWR

HWS

CDW

SCD

WR

CWS

CWR

Individually controlled

terminal unit

Office LAN cor Internetn

Optional connection

networks

FIGURE 3. Building- and comfort-control networks.

46 January 2001 • HPAC Engineering

exchange and local air-movement patterns have been developed and areready for implementation into morefunctional individual control products.Such products will provide exactlywhat occupants desire: an immediateand noticeable (but not excessive) response to requests for thermalchanges. Our firm estimates that ter-minal units with individual comfortcontrols could be applied for some 20percent of the approximately 60 billionsq ft of commercial floor space in theUnited States without modificationsbeing made to the remainder of thebuilding system. This amounts to ap-proximately 50 million workspacesand represents an immediate potentialmarket of more than $50 billion. Weestimate that as the focus of HVAC de-sign shifts to comfort, this element ofthe building-construction industry willattain a size comparable to that of theentire HVAC industry today.

A LOOK TO THE FUTUREThe future of the HVAC industry

will involve not an evolutionarychange in the way buildings are de-signed and constructed, but the devel-opment of a parallel comfort industrythat is network-based and looks morelike the PC and peripherals industrythan it does the building-constructionindustry. This new industry will mar-ket to tenants, and products will be in-stalled by office-furnishing installers,not building trades. Those who standto benefit most from this new industryare manufacturers who build productsthat integrate state-of-the-art controlswith terminal hardware in a singlepackage that is designed to capture theeconomies of high-volume produc-tion. Designers and system integratorswill be able to capture value by apply-ing these new products effectively andintegrating them into functional con-trol and accounting networks.

SUMMARYA lack of focus on occupant-comfort

issues in the HVAC industry has cre-ated an enormous gap between termi-nal systems that are being installed inbuildings today and those that couldbe installed to enhance comfort andprovide the individual-control featuresbuilding occupants want. This gap hascreated an enormous opportunity thatis unlikely to be captured by workingwithin current building-constructionprocesses and procedures. Those whowish to participate in capturing thisvalue must be willing to work to de-velop alternative implementationpaths that include direct or indirectsales to building tenants and the devel-opment of comfort systems and net-works that can operate relatively inde-pendently of building HVAC systems.Enabling procedures and technologieshave been developed. The market isready. Are we?

T H E C O M F O R T I N D U S T R Y

www.ventprod.com

9HPAC Engineering • July 2002

C O N T R O L F R E A K S

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Marysville, Wash.

PID (pro p o rtional + integral +d e r i va t i ve) control has been thefoundation of building contro l

for several generations, but with theemergence of digital networks, the drawbacks of PID control are becomingm o re apparent, especially as our industrybecomes more concerned with energy e f f i c i e n c y.

PID control has persisted as long as it has largely because of a lack of compe-tition (i.e., an alternative that can be u n i versally applied when modulatingc o n t rol is re q u i red). That is changing,h owe ve r, with the recent development ofmethods of control that take adva n t a g eof network integration to improve p e rformance and efficiency. The time hascome, then, to bid farewell to PI D a n dfocus on new, more-efficient means ofc o n t ro l .

DRAWBACKS OF PID CONTROLPID contro l’s most serious drawback

is inefficient operation. Application ru l e sdictate that for effective PID operation,va l ves and dampers must be undersize dso that a substantial pre s s u re drop occursa c ross them under all operating condi-tions. When applied to chilled- or heat-ing-water distribution systems, this re c-ommended va l ve pre s s u re drop can asmuch as double the pumping power re q u i red. This loss is magnified becausePID control moves va r i a b l e - s p e e dpumps and fans away from their “n a t u r a l

c u rve s” (the curve of the pump’s or fan’shighest efficiency at various speeds), thus reducing operating efficiency whileadding substantially higher operating-p re s s u re re q u i rements. As much as thre e -f o u rths of annual power consumption in some distribution systems can be attributed solely to PI D - c o n t rol losses.

The problems of PID control do notend with inefficient operation:

• Because PID loops operate in isola-tion from each other, they cannot assurethat all loads will be satisfied at any give nt i m e .

• PID control can be costly to imple-ment and support. The goal for any distribution system employing PID c o n t rol is to have the pre s s u re differe n t i a la c ross supply and return headers at eachload be constant as flow in the systemchanges. To accomplish this, designersoften ove r s i ze distribution mains or e m p l oy re verse return-piping configura-tions. These pre s s u re - l e veling techniquesadd costs.

• PID control is a headache for opera-tions staffs because to operate effective l y,it re q u i res frequent va l ve- and damper-position readjustment (1- to 5-sec re p o s i-tion intervals often are re c o m m e n d e d ) .This nearly continuous re p o s i t i o n i n gs h o rtens actuator life, adds to mainte-nance costs, and makes control stabilityan ongoing issue.

“INTERDEPENDENT” NETWORK CONTROLAll of these problems can be mitigated

or eliminated by replacing “independ-e n t” PID control with “interd e p e n d e n t”n e t w o rk contro l .

Hi s t o r i c a l l y, PID control loops haveoperated independently of other equip-ment to maintain a stable pre s s u re ort e m p e r a t u re condition that is employe dfor control by the next system down theline. Ne t w o rk-enabled interd e p e n d e n tc o n t rol combines these subsystems and operates them as a single system. Inaddition to automatically optimizing op-eration, effective network control avo i d sthe costly and energy-wasting decou-pling re q u i red for local PID contro l .

The integrated network control of

H VAC systems is not unlike the contro la p p roach employed in the information-technology (IT) world to manage In t e r-net services in an office network. T h i n kof the loads in a distribution system asclients and the pump or fan as a server i n t e rconnected through a network. T h ed i re c t - d i g i t a l - c o n t rol system acts as a re s o u rce manager, processing client requests and directing the server accord-i n g l y. Each load is satisfied while re s o u rc eexpense is minimize d .

T h e re are several important adva n-tages of applying IT re s o u rc e - m a n a g e-ment strategies for HVAC systems.Among them are lower costs, simplerconfigurations, more-efficient and e f f e c t i ve operation, reduced mainte-nance, and greater long-term stabilityand re l i a b i l i t y. In implementing thesestrategies as a replacement for PID c o n t rol in a distribution system, the p re s s u re / t e m p e r a t u re-setpoint control of subsystems is replaced with energy-o p t i m i zed control based on the Eq u a lMarginal Pe rformance Principle (see thea u t h o r’s feature article “Ul t r a - Ef f i c i e n tCooling With De m a n d - Based Contro l , ”December 2001).

Designers understand that the pur-pose of an HVAC system is not to main-tain distribution pre s s u res or tempera-t u res, but to keep all of the people in abuilding comfortable as efficiently as possible. A network control system designed to connect eve ry occupied zo n ed i rectly to comfort - re s o u rce systems can do a far better job of maintainingc o m f o rtable conditions in all zones with less energy than can a system inwhich these components are isolated andoperated independently. Achieving sucha system, howe ve r, re q u i res a significantchange in operating strategies.

In August, Thomas Ha rtman, PE, willoutline some demand-based-control opera t -ing strategies and provide some tips for setting up effective network controls for typical distribution systems.

For previous Control Freaks columns,visit w w w. h p a c . c o m.

A member of H PAC En g i n e e r i n g’s Ed i t o-rial Ad v i s o ry Board, Thomas Ha rt m a n ,

PE, is an intern a t i o n-ally re c o g n i zed experton the use of adva n c e dbuilding controls andc o n t rol network s .Comments and ques-tions can be addre s s e dto him at t o m h @h a rt m a n c o. c o m.

Technology is obsolete in the age of digital networks

PID Control: May It Rest in Peace

9HPAC Engineering • August 2002

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Marysville, Wash.

Last month, I discussed cost- ande n e r g y - related drawbacks of continuing to employ PID (pro-

p o rtional + integral + deriva t i ve) contro lin HVAC systems and introduced theconcept of interdependent network c o n t rol, which can replace PID contro lto improve the performance and effi-ciency of many systems. In t e rd e p e n d e n tc o n t rol employs re s o u rc e - m a n a g e m e n ttechniques similar to those used in information-technology networks. T h i smonth, I will provide a brief ove rv i ew of a “d e m a n d - b a s e d - c o n t ro l” applicationto show how temperature- and pre s s u re -setpoint control can be eliminated whilethe performance and efficiency of an a l l - variable-speed HVAC system are i m p rove d .

ALL-VARIABLE-SPEED HVAC SYSTEMSApplying variable-speed AC d r i ves to

H VAC components with built-in extracapacity saves energy at all expected loadconditions. For this energy saving to bem a x i m i zed, pre s s u re differentials mustchange as the load conditions change.This is why temperature- and pre s s u re -setpoint reset commonly is used as an optimization technique in conve n t i o n a lc o n t rol systems.

DEMAND-BASED CONTROLDemand-based control is a network -

enabled method of control that modu-lates equipment to meet current loadingconditions and, at the same time, auto-

matically optimizes overall system opera-tion, thus eliminating the setpoint-opti-mization step. Demand-based contro lalso reduces construction, energy, andoperation-and-maintenance costs bysimplifying system configurations, re-ducing pre s s u re losses at all load condi-tions, and eliminating much of the we a rand tear on modulating components.

Imagine designing the controls for aH VAC system in which all cooling-towe rfans, condenser-water pumps, chillers,chilled-water distribution pumps, andsupply fans employ electric va r i a b l e -speed operation. Because of mechanical-and electrical-efficiency considerations,each of these components operates mostefficiently within a capacity span thatvaries for each system element and maydepend on more than a single operatingp a r a m e t e r. Demand-based control re c-o g n i zes that the operation of each pieceof equipment influences the operation of eve ry other piece of equipment in a system. In a demand-based-control system, network controls coordinate the operation of all equipment as a singlesystem to meet all of the cooling loads e f f e c t i vely and efficiently.

Demand-based control at the zone level.Va r i a b l e - a i r - volume (VAV) boxes in a d e m a n d - b a s e d - c o n t rol system employ“c o o l i n g - e f f e c t” operation rather thanjust airf l ow control. Cooling effect is acombination of airf l ow and temperatureeffect and is calculated approx i m a t e l yonce a minute in each VAV zone. T h ecalculation considers airf l ow, supply-airt e m p e r a t u re, and zone temperature. T h ecooling effect is compared to the devia-tion from the space-temperature setpointin each zone, and a damper-position ad-justment is made to increase or decre a s ethe cooling effect as re q u i red in eachzone. A self-balancing calculation alsomay be included so that the cooling-effect parameters for each zone are auto-matically adjusted. Each zone communi-cates certain information to the fan sys-tem serving it.

Demand-based control at the fan-systeml e v e l . With demand-based control, a fansystem can respond to changes in zo n e -

cooling demand by adjusting the speedof the fan(s) and/or the flow of chilledwater through the cooling coil. Su p p l y -air temperature and pre s s u re are not d i rectly controlled, acting only as limit-ing values to ensure equipment operateswithin manufacturer and system designparameters. This approach offers a gre a ta d vantage in meeting zone loads. If oneor more zones is not obtaining adequatecooling effect, the system determineswhether it would be more efficient to i n c rease airf l ow or reduce air tempera-t u re. If the system chooses to reduce airt e m p e r a t u re, it will slightly incre a s echilled-water flow to the air-handling-unit (AHU) cooling coil. VAV zones thata l ready are satisfied will automatically reduce their airf l ow, resulting in both ani n c reased air volume and lowe r - t e m p e r a-t u re air being immediately available tothe zones that re q u i re additional cooling.Such control substantially improves system re s p o n s i veness and operating e f f i c i e n c y. The readjustment takes placeabout once a minute.

CHILLED-WATER-DISTRIBUTION DESIGNAdjusting chilled-water va l ves once a

minute simplifies control and makesc o n t rol stability much less of a pro b l e mthan it is with PID control. The less-stringent incremental re q u i rements ofdemand-based control afford designersan opportunity to improve efficiency.

Va l ves can be sized for lower pre s s u relosses, as linear response to actuation is not re q u i red. Fu rt h e r m o re, with net-w o rk-management techniques, there isno re q u i rement that each va l ve operateindependently of the others. De m a n d -b a s e d - c o n t rol va l ve-selection rules callfor a full flow pre s s u re drop of 1 to 2 ft of head in worst-case configurations. The rules also re q u i re that some methodof automatic coil-ove rf l ow protection be provided. This feature may be incor-porated into the va l ve itself or the AHU controls. Simple ball or butterf l ymodulating va l ves work well in demand-b a s e d - c o n t rol applications. Mo d u l a t i n gva l ve actuators should be slow-acting. Although 120- to 300-sec actuators are

C O N T R O L F R E A K S

A member of H PAC En g i n e e r i n g’s Ed i t o-rial Ad v i s o ry Board ,Thomas Ha rt m a n ,PE, is an intern a t i o n-ally re c o g n i zed experton the use of adva n c e dbuilding controls andc o n t rol networks. Con-tact him at t o m h @h a rt m a n c o. c o m.

Replacing PID control with demand-based control

Out With the Old, In With the New

11HPAC Engineering • August 2002

ideal, faster-acting actuators will workfine as long as their total movement during each adjustment period is limited.

With demand-based control, there isno need to have identical pre s s u re differ-entials at each load, so re verse return configurations are not necessary. Pi p e sshould be sized to keep velocities andp re s s u re losses within recommended limits, with some system analysis p e rformed to select the distributionpump(s). Slightly oversizing the pump-ing capacity in a variable-speed systemcarries no energy penalty; howe ve r, tomaintain good electrical efficiency atl ower flows, pump motors should bes i zed as small as possible. Operating thepump motor into its service factor at peak conditions is what the factor isdesigned for. T h e re is no concern aboutmotor overload because the va r i a b l e -f requency drive can automatically limitthe power to its rated maximum.

DISTRIBUTION-PUMP CONTROLT h e re are several methods of applying

demand-based control to chilled-water-distribution pumping systems. The one Ip refer is the orifice-area method, bywhich pump speed is regulated accord i n gto the percentage of total va l ve orificea rea opened. For ball and butterfly va l ve s ,fractional va l ve opening can be calcu-lated by raising the fractional actuatorposition to a power of about 2.7. In single-pump systems, multiplying thisfractional opening by the va l ve’s full flowas a percentage of total-system maximumf l ow and summing the value for all va l ve sis used to directly set pump speed as ap e rcentage of the maximum speed whenthe pump is sized to meet full-flow re-q u i rements. The advantage of this tech-nique is that it allows the pump to oper-ate at its highest efficiency at all flow s .Experience shows that making this va l ve -a rea calculation and using it to set pumpspeed once eve ry minute works ve ry we l lre g a rdless of the size of the pump and distribution system. When multiplepumps are employed, it is not difficult todetermine the optimum orifice area at which a pump should be added or subtracted and to incorporate theseswitching points into the pump contro l .

ELIMINATION OF SETPOINTSWith demand-based control, there is

no direct temperature- or pre s s u re -setpoint control at the fan or pumpingsystem. This frees the system to re s p o n dto changes in cooling re q u i rements withsimple network-management techniquesthat automatically optimize the ove r a l lsystem efficiency.

When demand-based control is ap-

plied to fan systems and chillers, as well asto chilled-water distribution systems, ithas the ability to drastically improve thec o m f o rt of buildings and cut totalH VAC electricity use in half.

For previous Control Freaks columns,visit w w w. h p a c . c o m.

Circle 309

C O N T R O L F R E A K S

decouple) one system element from another. Fi g u re 1 shows basic control modules for a typicalH VAC system. The cooling towers, chillers, distribution pumps, and supply fans are contro l l e d

independently with temperature orp re s s u re setpoints that ensure thes u r rounding equipment also canoperate independently over a widerange of loading re q u i rements.

Although a network-capable dire c t - d i g i t a l - c o n t ro l(DDC) system may be employed for control, then e t w o rk typically is used only to collect informa-tion for operations. In many systems, additionalisolation is provided with primary - s e c o n d a rypumping, bypasses, decoupling lines, and va l ves ordampers that have large pre s s u re dro p s .

This focus on independent equipment operationwastes energy. No r m a l l y, the chillers in Fi g u re 1would be operated at a fixed chilled-water tempera-t u re. At low-load conditions, the chiller compre s s o rwould operate at higher-than-needed head (and reduced efficiency) to provide colder-than-re q u i re dchilled water, which the distribution pump woulddistribute at a higher-than-necessary pre s s u re. Toe n s u re stable and independent chilled-water coiloperation under these conditions, the va l ves wouldbe selected for high pre s s u re drops. All of theseequipment-isolation measures would reduce ove r a l lsystem efficiency.

Imagine operating a chilled-water plant with-out controlling the chilled-water temperatureand operating a va r i a b l e - f l ow chilled-water

distribution network without a differe n t i a l - p re s s u resetpoint. Why would you operate acooling system this way? The re a-sons may surprise you. First, operat-ing plants without directly contro l-ling the temperature or pre s s u re oftheir outputs can be much more efficient. Se c o n d ,t e m p e r a t u re and pre s s u re control is considerablym o re complex and less stable than are control strategies that combine more direct control and o p t i m i z a t i o n .

T h rough a new approach to HVAC contro lcalled “demand-based control,” building systemsa re operated using the network capacity of modernbuilding control systems. Combining va r i a b l e -s p e e d - d r i ve equipment with network-enabled d e m a n d - b a s e d - c o n t rol technologies can make s i m p l e r, smaller, and lower-cost building energysystems operate as much as 30- to 50-percent moreefficiently than conventional system configurationswith the same basic HVAC-component efficienciescan. Also, demand-based control enhances thec o m f o rt of buildings and provides a platform for individual control and other valuable occupanta m e n i t i e s .

HOW CONVENTIONAL CONTROL WASTES ENERGYC o n ventional HVAC controls employ pre s s u re -

or temperature-setpoint control to isolate (or

29HPAC Engineering • December 2001

The value and methods of applying

direct-coupled network control

to building-energy-system design

B y T H O M A S HARTMAN, PEThe Hartman Co.

Marysville, Wash.

A member of H PAC En g i n e e r i n g’s Editorial Ad v i s o ry Board, Thomas Ha rtman, PE, is an i n t e rnationally re c o g n i zed expert on the use of a d vanced building controls and control network s .Comments and questions can be addressed to him att o m h @ h a rt m a n c o. c o m.

Direct-coupled control canreduce the size and

complexity of both controlsand system components.

U LT R A - E F F I C I E N T

C O O L I N Gwith Demand-Based Contro l

30 December 2001 • HPAC Engineering

If current HVAC - c o n t rol practiceswe re applied to automobile operation,d r i vers would be taught to control the accelerator to maintain a fixed enginerpm, operate the clutch to provide a specific torque, and use the brake to c o n t rol the ve h i c l e’s speed. Considerwhat this would do to a car’s ave r a g emiles per gallon!

Traditional optimization appro a c h e scannot ove rcome the built-in inefficien-cies of conventional HVAC control. Instead, they typically create one or moreadditional layers of software that reset thevarious setpoints to marginally improveoperating efficiency as conditionschange. These optimization appro a c h e sadd complexity to the controls and arelimited in their ability to reduce energyc o n s u m p t i o n .

Di rect-coupled control, on the otherhand, can reduce the size and complexityof both controls and system compo-nents. Howe ve r, a basic change in how

the system is operated must be madef i r s t .

DEMAND-BASED CONTROL: A NETWORK-CONTROL SOLUTION

Demand-based control is a method of applying direct-coupled network c o n t rol. It is based on the idea that abuilding HVAC system is a single systemthe energy efficiency and comfort p e rformance of which are optimize d

when the operation of all components isc o o rdinated to meet actual needs in thespaces serve d .

Demand-based control is intended tofill the vacuum in controls technologiesc reated by the development of va r i a b l e -f requency drives (VFDs) for HVACequipment. Prior to the introduction of VFDs, coordinating the operation of HVAC equipment mattered little because equipment efficiency re m a i n e dalmost constant over ranges in loading.Now, with VFD modulation, the efficiency of HVAC components canchange dramatically as load conditionsva ry. Although coordinated operation is essential to maximizing overall energyefficiency today, most conventional con-t rol schemes operate VFD equipment as they do mechanically modulatedequipment. Demand-based control cans o l ve this problem; howe ve r, applying ite f f e c t i vely re q u i res new thinking abouth ow VFD equipment operates most e f f i c i e n t l y.

HOW DEMAND-BASED CONTROL WORKSWith conventional control strategies,

little if any information concerning up-s t ream or dow n s t ream loading/operatingconditions is employed to adjust the operation of equipment. Ty p i c a l l y,H VAC components operate to maintaina single temperature or pre s s u re setpoint.If a number of spaces in a building we re to begin to overheat, most centralsystems would not self-adjust to prov i d em o re cooling and, once the spaces we resatisfied, readjust to meet the re d u c e dload with greater efficiency. Ac c o m p l i s h-ing this would re q u i re a network - c o n t ro lscheme that communicates with theloads being serve d .

Ne t w o rk-enabled demand-based con-t rol is ve ry cost-effective because, in most

D E M A N D - B A S E D C O N T R O L

FIGURE 1. A conventional HVAC system in which each piece of equipment is operatedwith an independent control loop that maintains a temperature or pressure setpointregardless of the end-use requirements.

If current HVAC-control practices were applied toautomobile operation, drivers would control the

accelerator to maintain a fixed engine rpm,operate the clutch to provide a specific torque,

and use the brake to control speed.

31HPAC Engineering • December 2001

cases, it does not re q u i re additionalequipment or controls. In fact, whenp roperly applied, demand-based contro lre q u i res less equipment and often em-p l oys a simpler configuration than thec o n ventional system and control it re p l a c e s .

Although demand-based control ties the operation of all equipment toend-use re q u i rements—actual space re q u i rements in single-building HVACapplications—this does not mean that chillers and cooling towers operated i rectly from space-temperature sensors.R a t h e r, as in automobile operation, demand-based control optimizes anH VAC configuration by directly cou-pling the operation of all components so they operate as a single system in meeting the actual needs of spaces. T h i ssingle-system approach generally is notc o n s i d e red because HVAC designers a re too used to ensuring that their designs isolate the operation of HVACcomponents to give an alternative athought. For example, while low delta Tis a serious problem in many chillerplants, designers continue to employ decoupling or bypass lines that permit d i rect mixing of supply and re t u r nchilled water.

With variable-speed equipment andn e t w o rk - c o n t rol capabilities, the long-standing dictum that equipment must be decoupled to operate effectively hasbeen re versed. Di rect coupling leads tos i m p l e r, more-efficient operation. It is i n t u i t i ve that coordinating the operationof a chiller plant and chilled-water distribution network is re q u i red toa c h i e ve the highest overall cooling-system efficiency. The question is howdoes one control these together so theycan be operated most efficiently.

The answer is to consider all equip-ment invo l ved in cooling a building as a single system instead of a series of

systems. When cooling needs to be adjusted in response to space conditions,demand-based control coordinates theoperation of all elements to provide cooling where it is needed according top redefined efficiency re l a t i o n s h i p s .

We know that coordinating the operation of two identical pumps in a d i rect-coupled parallel or series circ u i tbased on power or speed is the mosts t r a i g h t f o rw a rd method of optimizingthe pumps’ overall efficiency. But what if the pumps (or other type of equip-ment) are not identical? In a similar fashion, one or more power-based relationships could be developed usingthe Equal Marginal Pe rformance Pr i n c i-ple (see sidebar), which would optimizethe pumps’ operation under all loadingconditions. A circuit consisting of cool-ing towers, chillers, pumps, and condi-tioning fans with variable-speed drive scould be optimized in this manner. Like

D E M A N D - B A S E D C O N T R O L

Equal Marginal Performance

Formulated more than a decade ago, when variable-speed drives were first appliedto pumps and fans, the Equal Marginal Performance Principle (EMPP) states that

the operation of a system comprised of multiple modulating components (in series orparallel) is optimized when the marginal system output divided by the marginal systeminput is the same for all components. To better visualize how the EMPP is applied to anentire cooling system, consider the schematic below.

Imagine that the knobs below each piece of equipment in this schematic adjust thecapacity of that element by changing its speed or by some other means. Assume thatinstrumentation for measuring the system output and system input is provided. Howwould you optimize this system at its current point of operation? According to theEMPP, it could be done by making small adjustments to the capacity of each elementand noting the changes in system output and input. Then, system efficiency could beimproved by: (1) reducing the capacity setting of elements that show relatively smallmarginal-system capacity change per unit input change and (2) increasing thecapacity setting of elements that show larger changes in capacity per unit powerchange so that the total system output remains at its current point of operation. Theseprocesses of testing each element and resetting the system would be repeated until allelements had exactly the same marginal output per unit change in power input. At thatpoint, the system would be optimized.

With variable-speed equipment and network-control capabilities, the long-standing

dictum that equipment must be decoupled tooperate effectively has been reversed.

c o m f o rtable building and a higher leve lof indoor environmental quality.

CASE STUDYBellevue Corporate Plaza is a multi-

tenant mid-rise office building in Bellevue, Wash. It is an all-electric building that employs va r i a b l e - a i r -volume (VAV) air systems with fan-p owe red perimeter boxes and electric reheat. The chiller plant consists of twoair-cooled centrifugal chillers and a c o n s t a n t - f l ow primary-only distributionsystem. In 1993, an integrated lightingand terminal-re g u l a t e d - a i r - vo l u m e(T R AV) controls re t rofit was ord e re d .T R AV was the first demand-based-c o n t rol strategy developed, and its application reduced total building energy use from about 80,000 Btu per sqft annually to about 50,000 Btu per sq ft.Se ven years later, it was time to upgradethe chiller plant. The plant employed aphased-out refrigerant, and the chillerswe re near the end of their useful life, asevidenced by a dramatic increase inmaintenance and numerous failure s .

designed to use less than 50,000 Btu persq ft annually. Note that despite the initial low-energy design, the network -based control further reduced the electric-energy use of the chiller plantand HVAC distribution system by nearly50 percent. At the same time, the n e t w o rked control resulted in a more

32 December 2001 • HPAC Engineering

FIGURE 3. The new Bellevue Corporate Plaza cooling system. The Loop demand-based-control network is shown schematically. Its purpose is to connect equipment to thenetwork for coordinated operation rather than to maintain temperature or pressuresetpoints.

dissimilar pumps, these componentscould be directly coupled and contro l l e dusing demand-based control to operatethe circuit as a single system and prov i d ethe cooling capacity re q u i red. Such c o o rdinated control is the simplestmethod of obtaining the highest overall operating efficiency. Thus, withdemand-based control, equipment is c o o rdinated to operate according top ower (kilowatt) setpoints, which is s i m p l e r, more stable, and much more efficient than the use of temperature or pre s s u re setpoints is. Fu rt h e r m o re, because demand-based control is n e t w o rk-based, enhancing system per-formance by making system adjustmentsthat focus on meeting exceptional space-conditioning re q u i rements that occurf rom time to time is ve ry easy.

ENERGY-REDUCTION IMPLICATIONS OFDEMAND-BASED CONTROL

By directly coupling HVAC compo-nents into a single system with networkcommunication and acting on end-usere q u i rements in setting system capacity,d e m a n d - b a s e d - c o n t rol strategies can i m p rove building comfort, enviro n m e n-tal quality, and energy perf o r m a n c e .While the assessment of comfort and e n v i ronmental quality usually is subjec-t i ve, the energy savings attributable to demand-based control can be accuratelydetermined through hourly simulation.The simulation results for the electricp o rtion of the HVAC system in a newmid-rise office building in De n ver ares h own in Fi g u re 2. The building was

D E M A N D - B A S E D C O N T R O L

FIGURE 2. A comparison of the annual energy use with conventional optimized HVACsystems and equipment of the same basic configuration and efficiencies exceptoperated with demand-based control of a new mid-rise office building in Denver.

34 December 2001 • HPAC Engineering

Fu rt h e r m o re, the plant lacked the capacity to meet growing tenant internalheat loads at peak conditions.

After exploring several re p l a c e m e n toptions, the building ow n e r, Ha l l w o o dC o m m e rcial Real Estate, adopted an upgrade approach that employed then ew Loop demand-based-contro l -

sequence package.The Ha rtman Co. helped implement

Loop ultra-efficient demand-based-c o n t rol sequences and bring the new system on line. The Loop control sequences we re installed in the building’sexisting DDC system without the needfor additional control equipment, save

some input/output devices. The re s u l t i n gcooling system provided additional capacity with a lower design chilled-water temperature, allowing the existingcooling coils to be used.

The Loop demand-based-contro ltechnologies significantly improved thechiller plant’s operating efficiency by using the control network to coord i n a t ecooling supply with the actual coolingdemand. All balancing va l ves we re re m oved, and the three-way modulating-f l ow control va l ves on the cooling coils

we re changed to two-way line-size dmodulating va l ves. Variable-speed drive swe re added to the pumps. The distribu-tion system was re c o n f i g u red to va r i a b l ep r i m a ry flow without a bypass va l ve .

The chiller-plant components are nowc o n t rolled at optimal levels of re l a t i vep ower use rather than to maintain a specific chilled-water temperature or d i s t r i b u t i o n - p re s s u re setpoint. W h e n-e ver possible, when cooling is called for,the chiller plant, distribution pumps,and fans are optimally operated together.Under normal operations, there is no d i rect control of chilled-water tempera-t u re or distribution pre s s u re. At low - l o a dconditions, chilled-water temperaturefloats upw a rd, and pump-head re q u i re-ments for the distribution system fall, resulting in ve ry stable operation and lowp ower re q u i rements. Even at full loading,the chilled-water temperature often remains above the design minimum, andpumping head is only half of what is wasb e f o re the re t ro f i t .

The operation of the chillers, distribu-tion pumps, and main supply/return fans is coordinated for the highest ove r a l lcooling and distribution efficiency at alltimes. This automatic network optimiza-tion is especially useful during periods of demand limiting. In conventional systems, demand limiting may be ap-plied only to chillers. The chilled-water-distribution pumps and air-distributionfans speed up during demand limiting

Circle 000

D E M A N D - B A S E D C O N T R O L

Direct coupling leads tosimpler, more-efficient

operation.

35HPAC Engineering • December 2001

because their loads are not met when the chillers reduce their loading. This results in greater power use by pumpsand fans, which reduces the demand-limiting function of the chillers. At Bellevue Corporate Plaza, the Loop d e m a n d - b a s e d - c o n t rol sequences coord i-nate the operation of all equipment and provide more - e f f e c t i ve demand limiting with less impact on occupantsbecause system efficiency increases during demand limiting so that the cooling effect remains high.

Despite a larger cooling system andl ower design operating temperatures, the building’s summer peak electrical demand has been reduced by about 20 percent. And as a result of extendingdemand-based control to the cooling system, the building’s daily energy use

during cooling-system operation hasbeen reduced by about 22 perc e n t .

The building is using 50-cents-p e r - s q u a re-foot less energy annually thanit would be if the demand-based-contro lp rogram, which began with the T R AVre t rofit, had not been undertaken. In addition, the building’s occupants re p o rtbeing much more comfort a b l e .

SUMMARY AND CONCLUSIONEnergy efficiency and comfort can be

i m p roved substantially by applying n e t w o rk-enabled demand-based-contro lstrategies to operate building HVACequipment. The enormous electric-energy reductions possible from the application of demand-based-contro l

strategies are such that this change in building-control technology should be considered as part of national energy policies. When demand-basedc o n t rol is installed throughout most typical HVAC systems, one can expect a30- to 50-percent annual electrical-energy reduction. Equipment configura-

tions re q u i re only slight modifications tobe operated in accordance with demand-based control in place of conve n t i o n a lo p t i m i zed control. Fu rt h e r m o re, a system specifically configured for demand-based control can be less e x p e n s i ve to design, install, and maintainthan a conventional system.

Circle 000

D E M A N D - B A S E D C O N T R O L

When demand-basedcontrol is installed

throughout most typicalHVAC systems, one can

expect a 30- to 50-percent annual

electrical-energyreduction.

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Marysville, Wash.

In a December 2001 article,1 a newnetwork control approach that canbe used in place of PID control for

the operation of much of the equipmentthat constitutes a building’s HVAC system was introduced. Called “demand-based control,” it avoids the extra energyrequirements, control-stability problems,and frequent repositioning of PID con-trol. Furthermore, it is simpler to applybecause it provides stable, optimized control in a single step, while PID controlrequires two separate steps: (1) setting upa control loop to maintain a setpoint and(2) continuously adjusting the setpoint to optimize operation.

Using the equal-marginal-perform-ance principle, the 2001 article showedthat demand-based control enables sta-ble, optimized control of building-com-fort-system components in a single stepand does not require any of the energy-wasting design rules that PID controldoes. That is why energy costs with de-mand-based control are 30 to 50 percentlower than they are with PID control.

However, demand-based control typically uses direct power relationships,rather than temperature and pressure setpoints, to control equipment. In some instances, temperature- or pressure-setpoint control is required to achieve effective HVAC-system operation. Some consider this to be a reason to continue using PID control. But with the availability of digital network controls, there is a far more efficient andeffective alternative.

REPLACING PID CONTROL WITH INTELLIGENT ITERATIVE CONTROL

When temperature- or pressure-setpoint control is necessary, designersshould consider new approaches thatmake better use of modern networkeddigital controls before relying on PIDcontrol. As indicated earlier, the use ofPID control usually wastes energy by requiring 25 percent to more than 90percent of the fluid total dynamic head to be dissipated through the control valveunder various operating conditions. This

makes the power requirements of the system far greater than should be neces-sary to distribute air or water to the loadsserved. Furthermore, a number of otherfactors in modern HVAC systems, including the application of optimizationsetpoint reset, cause PID control loops tobe non-proportional much of the time.This requires modulation readjustmentsat short intervals, which reduce equip-ment life and add to maintenance costs.

For applications requiring tempera-ture- or pressure-setpoint control, a newapproach called “intelligent iterative control” can avoid these PID pitfalls,providing more efficient and stable sys-tems with longer-lasting devices requir-ing less maintenance. Intelligent iterativecontrol is a method of control based onthe iterative problem-solving techniquesused by computers since the beginning of digital processing. To show how an intelligent-iterative-control procedurecan be developed to operate more effi-ciently and effectively than PID controlin a temperature-setpoint-control appli-

C O N T R O L F R E A K S

Presenting Intelligent Iterative Control: PID Replacement for Setpoint ControlNew approach makes better use of digital controls

A member of HPAC Engineering’s Editor-ial Advisory Board,Thomas Hartman, PE,is an internationallyrecognized expert onthe use of advancedbuilding controls andcontrol networks. Con-tact him at tomh@hartmanco.com.

13HPAC Engineering • September 2003

Editor’s note: This is the first part in athree-part series.

C

C

AOFC1CCV

VAI

FC1DAT

DOFC1SS

DIFC1

T

FIGURE 1. A simple cooling-only fan coil.

Circle 209

cation, consider the simple cooling-onlyfan coil depicted in Figure 1.

The fan coil in Figure 1 serves a spacethe temperature of which is monitoredwith a space-temperature sensor. To en-sure even comfort, control of the fan coilinvolves maintenance of a discharge-air-temperature setpoint in response tospace-temperature conditions. The con-trol we are interested in for this discus-sion is that of the cooling coil in main-taining a specific fan-coil discharge-air-temperature setpoint.

If conventional PID control were to beused, the designer would face severalproblems. First, conventions dictate thatthe cooling-coil control valve be sized fora substantial pressure drop at full flow(the ASHRAE 2000 HVAC Systems andEquipment Handbook recommends thepressure drop across the valve at full flowbe 25 to 50 percent of the total pressuredrop of the chilled-water distribution

network). This radical undersizing ofcontrol valves adds dramatically to pumpsize and cost and, just as troublesome,long-term pumping-energy cost. Also, tomaintain stable PID control, the valvemay have to be repositioned every severalseconds, reducing its life and adding tomaintenance costs.

These problems can be avoided by using an iterative-control technique thatmarshals the computing and networkingcapacity of modern DDC systems moreeffectively. Instead of PID control, imag-ine the incorporation of line-sized andfull-ported control valves on all of the fan coils in an effort to reduce pump sizeand the pumping-energy requirementsof the chilled-water distribution system.We then could adjust pump speed usingdemand-based control and the valve-ori-fice-area method outlined in my August2002 column.2 That approach would reduce the pumping power from fixed

pressure-setpoint control. The questionthen becomes how can line-sized valvesbe operated to provide stable and precisecontrol to maintain the fan-coil dis-charge-air-temperature setpoint.

Next month, Part 2 in this series will discuss the implementation of an intelli-gent-iterative-control algorithm.

REFERENCES1) Hartman, T. (2001, December).

Ultra-efficient cooling with demand-based control. HPAC Engineering, pp.29-32, 34, 35.

2) Hartman, T. (2002, August). Outwith the old, in with the new. HPAC Engineering, pp. 9, 11.

For previous Control Freaks columns orto visit the Networked Controls microsite,go to www.hpac.com. Send comments andsuggestions to controlfreaks@penton.com.

C O N T R O L F R E A K S

14 September 2003 • HPAC Engineering

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Marysville, Wash.

Though widely used to solvecomplex problems, iterativelogic is well-suited to providing

modulation control. Indeed, the iterativeprocess can be easily applied to controlapplications.

For the fan-coil-valve-control applica-tion in Figure 1, assume the unit is oper-ating, but the discharge-air temperature is deviating from the setpoint value. Theiterative process of control involves: (1)estimating the change in valve positionrequired to achieve the desired change indischarge-air temperature, (2) makingthe change, (3) waiting to see how close tothe desired temperature the result is, and(4) adjusting the estimation logic (if necessary) and repeating the process. The process becomes “intelligent” whenthe estimate is enhanced by logic and real-time information derived from other elements of the fan coil and/or related systems. With an iterative-controlscheme, control for the cooling valve inFigure 1 (FC1CCV) involves estimatingand implementing a capacity-adjustment

algorithm intended to bring the loop toits current setpoint. Because this is an iterative process, the algorithm employedto make the estimate does not need to beprecise; however, it should be as accurateas possible and contain the primary datapoints needed to affect the desiredchange in valve position, along with afactor that correlates the relative impor-tance of each.

Among the substantial benefits of applying intelligent iterative control arethat the intervals between valve reposi-tionings can be much longer than thosetypically employed with PID control,and there is no need for interval times tobe fixed. This has the potential to greatlyreduce the frequency of required valverepositioning and, thus, extend valve andactuator life.

From an energy-performance perspec-tive, a significant benefit of replacingPID control with intelligent iterativecontrol is system head pressure can be reduced dramatically at all load condi-tions because valves can be line-sized and

full-ported. With intelligent iterativecontrol, there is no requirement for a linear response between valve actuationand the cooling effect it produces. Toachieve the linear relationship requiredfor PID control, valves must be under-sized with significant pressure drops,while controllability requires only thatthe relationship between actuation andcoil cooling output be a continuouscurve and that the slope be positive at all times. As shown in Figure 2, theseconditions are not difficult to meet withline-sized valves. A linear relationship asshown by the red line in Figure 2 can beestablished when the valve is undersizedand consumes 25 to 50 percent of the total system head-pressure drop. Thoughnot offering a linear relationship betweenactuation and coil cooling capacity, the line-sized valve in Figure 2 does meetthe requirements for controllability.

The iterative logic used to effectivelycontrol a line-sized valve such as the one in Figure 2 is based on informationcoming from the fan coil and across the

C O N T R O L F R E A K S

Presenting Intelligent Iterative Control: PID Replacement for Setpoint ControlImplementing an intelligent-iterative-control algorithm

A member of HPAC Engineering’s Editor-ial Advisory Board,Thomas Hartman, PE,is an internationallyrecognized expert onthe use of advancedbuilding controls andcontrol networks. Con-tact him at tomh@hartmanco.com.

9HPAC Engineering • October 2003

Editor’s note: This is the second part in athree-part series.

C

C

AOFC1CCV

VAI

FC1DAT

DOFC1SS

DIFC1

T

FIGURE 1. A simple cooling-only fan coil.

network from other system elements.During the design process, it is useful

for designers to learn what factors are involved for each modulating controlloop. A reasonably effective approxima-tion of the incremental cooling effect required to meet a change in load is not difficult to approximate. When aline-sized valve is employed, and thechilled-water temperature is fixed for the application shown in Figure 1, the iterative-control algorithm often can be:

where:FC1CCVnew = the estimated new valve

position required to meet the loadFC1CCV = the current valve positionC2 = a constant that depends on the

relative capacity of the cooling coil andthe chilled-water lines that connect it.Initially, it can be set at 0.05. From there,it generally is adjusted between 0.02 and0.10, with the lower values ensuring less potential for hunting and the higherones ensuring faster response

C1 = a constant between 0.0 andabout 20 that is adjusted depending onthe proximity of each valve to the pump(C1 should be 0.0 for the load farthest

from the pump)PumpRPM = the speed of the distri-

bution pump (100-percent speed = 1.0)FC1DAT = the current fan-coil

discharge-air temperature in degreesFahrenheit

FC1DATSP = the current fan-coil discharge-air-temperature setpoint in degrees Fahrenheit

C3 = a constant developed to compen-sate for the slope of coil capacity vs. theactuation curve in Figure 2 as a functionof valve position. The value of C3, whichmust be greater than zero, generallyranges between 0.1 and 0.5

Note that this algorithm calculatesvalve position as a fractional value (0.0 to1.0) and uses pump speed as a fractionalvalue. Some control systems providethese percentages as whole numbers (0 to 100) or other values. Constant and variable values may need to be scaled tobe compatible with the control systememployed.

As discussed earlier, one of the pur-poses of this type of control is the mini-mization of valve repositioning. For thefan-coil application shown in Figure 1,the minimum valve-repositioning inter-val typically is 30 sec. Following is the final form of a useful control algorithmfor the fan coil in Figure 1:

where:A and B = intermediate variables used

in this algorithm only

This iterative process compensates forthe fact that changes in pump speed maycause changes in pressure across the valveand, thus, change the valve movementnecessary to attain the same change incooling effect at different pump speeds.

This valve-control algorithm is simplerthan many of the PID algorithms now in use. With some experience, intelligent-iterative-control algorithms are relativelyeasy to sketch out during design and can be easily inserted into sequences ofoperation. Initial values for C1, C2, andC3 usually can be estimated based on coil characteristics and the location of thefan coil in the distribution circuit. If afan-coil supply fan is variable-speed, anda variable chilled-water-supply tempera-ture is incorporated, additional factorscan be implemented easily. Sometimes, it is helpful to incorporate into iterativealgorithms self-learning, whereby con-stants are automatically set and adjustedto reduce startup-time requirements andthe frequency of repositioning. Lastly, incritical systems, it sometimes is useful toadjust valve position for system-dynamicsfeatures, in addition to marginal cooling-load requirements. Such adjustmentsmay be made independently of valve-position changes attributed to changes inload requirements. All of these controlsare surprisingly easy to set up, as long asthe digital controller has flexible control-programming capabilities.

Next month, the third and final part inthis series will discuss limitations in design-ing with line-sized chilled-water valves andbenefits of intelligent iterative control.

C O N T R O L F R E A K S

10 October 2003 • HPAC Engineering

Valv

e ac

tuat

ion,

per

cent

100

80

60

40

20

0

Cooling-coil capacity, percent of design maximum0 10 20 30 40 50 60 70 80 90 100

FIGURE 2. Actuation vs. coil capacity for undersized valve used in PID control (red line)and line-sized valve in intelligent-iterative-control application (green line).

FC1CCV FC1CCV

C2 1 0 C1 PumpRPM 0 3

FC1DAT FC1DATSP C3 FC1CCV

new = +

÷ + × −( )( )( ) ×

−( ) × +( ). .

Do every 30 sec

A C2 1 0 C1 PumpRPM 0 3

FC1DAT FC1DATSP 0 2 FC1CCV

If absolute value B 15 0 15 then

FC1CCV A FC1CCV and B 0

otherwise B B Anew

:. .

.

% . ,

;,

= ÷ + × −( )( )( ) ×

−( ) × +( )

( ) > ( )= + == +

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Georgetown, Texas

Although designing chilled-water distribution systemswith line-sized control valves

reduces pump-head requirements and,thus, reduces both first costs and energycosts, there are restrictions and limita-tions when such a design path is taken.One of the most critical of these concernsthe pressure differential across valves. Irecommend that systems be designed so that the pressure across a control valve does not exceed double the designpressure drop through the load it serves.This means that if a cooling coil is rated for a pressure drop of 10 ft at full flow, the distribution system shouldbe designed so that the valve serving thecoil has a differential that remains belowabout 20 ft. As an example, consider the variable-primary-flow distributionsystem shown in Figure 3.

In this centralized pumping scheme,chilled water is pumped through thechillers and into the distribution net-work. If the distribution system is large,then loads near the chillers (such as loads

1-1 and 2-1) may be subjected to largesupply-header and very low return-header pressures at high-flow conditions.These high pressures may be necessary

to transport water to the farthest reachesof the system. Thus, if iterative controlwith line-sized valves were employed, theloads close to the chillers could become

C O N T R O L F R E A K S

Presenting Intelligent Iterative Control: PID Replacement for Setpoint ControlLimits of designing with line-sized chilled-water valves

A member of HPAC Engineering’s Editor-ial Advisory Board,Thomas Hartman, PE,is an internationallyrecognized expert onthe use of advancedbuilding controls andcontrol networks. Con-tact him at tomh@hartmanco.com.

9HPAC Engineering • November 2003

Editor’s note: This is the third part in athree-part series. C

Load 1-n

CLoad 1-2

CLoad 1-1

CLoad 3-n

CLoad 3-2

CLoad 3-1

CLoad 5-n

CLoad 5-2

CLoad 5-1

CLoad 2-1

CLoad 2-2

CLoad 2-n

CLoad 4-1

CLoad 4-2

CLoad 4-n

CLoad 6-1

CLoad 6-2

CLoad 6-n

Chiller

Chiller

Chiller

FIGURE 3. A variable-primary-flow distribution system.

CLoad 1-n

CLoad 1-2

CLoad 1-1

CLoad 3-n

CLoad 3-2

CLoad 3-1

CLoad 5-n

CLoad 5-2

CLoad 5-1

CLoad 2-1

CLoad 2-2

CLoad 2-n

CLoad 4-1

CLoad 4-2

CLoad 4-n

CLoad 6-1

CLoad 6-2

CLoad 6-n

Chiller

Chiller

Chiller

FIGURE 4. Alternative configuration using distributed pumps.

Circle 161

uncontrollable at high-flow conditions.To avoid this potential condition, an

alternative configuration using distrib-uted pumps (Figure 4) should be consid-ered. In this pumping scheme, pumps are

placed at the distribution mains, ratherthan centralized at the plant. In otherconfigurations, it may be advantageousto employ a primary-booster pumpingscheme. Because the pressures across thevalves in Figure 4 are lower (on average)than those in Figure 3, the configuration

in Figure 4 is more energy-efficient, inaddition to being more controllable at all flows.

Readers may notice that the variable-primary-flow configurations in figures 3

and 4 do not employ a bypass valve forlow-flow conditions. When intelligent iterative control is employed, a bypassvalve usually is unnecessary, as the intelli-gent, network-based valve control canensure a minimum level of flow throughthe loads at all times a chiller is operating.

BENEFITS OF INTELLIGENT ITERATIVECONTROL

The benefits of replacing PID controlwith the type of iterative-control schemedescribed in this series of columns havebeen shown to include a reduction in distribution-pumping-energy require-ments to less than one-half of conven-tional distribution-energy requirements.Furthermore, average valve-reposition-ing frequency can be reduced from one-tenth to a hundredth of what is required conventionally. In addition,much greater control stability undernearly all operating conditions usually is achievable. Finally, when coupled withdemand-based control in plant and dis-tribution systems, this control is far moreeffective, with the opportunity to ensurethat all loads are satisfied all of the time.

For previous Control Freaks columns,visit www.hpac.com.

C O N T R O L F R E A K S

10 November 2003 • HPAC Engineering

When coupled with demand-based control in plant and distribution systems, this control is

far more effective, with the opportunity to ensurethat all loads are satisfied all of the time.

22 March 2003 • HPAC Engineering

system when operated with variable speed.An obstacle on the road to variable speed has

been the cost associated with variable-frequencydrives, which are the most popular and efficient

method of applying variable speedto AC motors. In recent years, how-ever, that cost has fallen dramati-cally. For the most popular sizes ofmotors for HVAC fan and pump

applications, variable-frequency-drive cost is approaching $50 per horsepower. When all costsare considered, variable speed is now much lesscostly than two-speed-control alternatives. In addition, the soft start from variable speed addsenormous life to belts and other drive components.

One of the largely untapped cost-saving featuresof variable-speed drives is built-in intelligence. The internal control logic of variable-speed drives is

microprocessor-driven. Most driveshave capacity to contribute to buildingcontrol systems with networking services and processing capabilities. It ispossible to envision a system in whichmuch of the control sequencing takesplace in intelligent end devices con-nected to the system, such as variable-speed drives, rather than in dedicatedcontrol panels of the building controlsystem, where it takes place now.

Unfortunately, control-sequence ca-pabilities from separate componentsare not easily integrated into building

In recent years, there has been a trend towardusing variable speed to operate motors forequipment and systems that provide heating,

ventilation, and air conditioning for buildings.Variable speed has been shown toimprove the operating efficiency of equipment such as fans, chillers,and pumps, the capacity of whichmust be modulated for proper system operation. Furthermore, designers now understand that motors powering devices tradition-ally operated at constant speed, such as fan coils andfan-powered terminal units, can benefit enor-mously in terms of both performance and efficiencyfrom the application of variable speed. It is becom-ing clear that virtually every motor involved in a building HVAC system today has the capacity to improve the efficiency and effectiveness of the

Achieving the highest levels of system performance

and efficiency through reliance on control networks

By THOMAS HARTMAN, PEThe Hartman Co.

Marysville, Wash.

A member of HPAC Engineering’s Editorial Advisory Board and a regular contributor to the magazine’s “Control Freaks” column, Thomas Hartman, PE, is an internationally recognized expert on the use of advancedbuilding controls and control networks. Comments and questions can be addressed to him at tomh@hartmanco.com.

DIRECT NETWORK

CONNECTIONof Variable-Speed Drives

High-speed building network

DDC panel

DDC panel

DDC panel

VFD

VFD

VFD

VFD

VFD

VFDVFD

VFD VFD

VFD

RF RF

AHU AHU

FIGURE 1. Preferred direct network connection of variable-speed drives.

control systems because each buildingcontrol system employs proprietary pro-gramming functions that are not compat-ible beyond the product. The use of coor-

dinated, distributed processing withoutsome level of programming compatibilitycomplicates the operation of such a configuration. However, one area inwhich standards have greatly progressedand now permit a certain level of integra-tion and resulting economy is the incor-poration of variable-speed drives directlyinto the control network, rather thanthrough discrete point connections, as hasbeen the norm.

HISTORICAL IMPEDIMENTS TO DIRECTNETWORK CONNECTION

While the direct network connectionof variable-speed drives and other equip-ment no longer is rare, the reliance on thenetwork for control, as well as monitoringfunction, still appears to be. The reasonfor this is that design engineers continueto be somewhat hesitant to rely on thecontrol network to execute direct controlactions. This hesitance goes back to the early days of direct digital control

(DDC), when, from reliability and time-of-response standpoints, it was recom-mended that all input and output pointsof control for each system be imple-

mented in a single panel. That is whycontrol designs still are made with remotestatic-pressure sensors wired all the wayback to the panel in which the fan orpump they are controlling is connected.

Although this approach had value

when it was developed a decade or moreago, network speed and reliability haveimproved substantially since that time.Also, the industry has realized that themost effective control of any variable-speed device almost always requires theuse of multiple factors in making controldecisions. Furthermore, it now is under-stood that very frequent speed adjust-ments to large fans or pumps can wasteenergy. For these reasons, relying onbuilding control networks to operateHVAC equipment is not just acceptable,it is absolutely necessary if the highestpossible levels of system performance andefficiency are desired.

DIRECT-NETWORK-CONNECTIONCONFIGURATION CONSIDERATIONS

Once the decision to consider the useof direct network connections for vari-able-speed drives is made, the operatorneeds to determine the nature of the integration of the drives into the buildingcontrol system. Leading the list of con-siderations is the type of connection to be made. Presently, the most popularmeans of connecting variable-speeddrives to building control networks areBACnet, LonWorks, and Modbus inter-faces. The benefit of a BACnet or Lon-works connection is that many systemsemploy one of these communicationprotocols as their network backbone.Thus, it may be possible to connect

D I R E C T N E T W O R K C O N N E C T I O N

26 March 2003 • HPAC Engineering

High-speed building network

VFD gateway VFD gateway

DDC panel

DDC panel

DDC panel

Low-speed VFD network Low-speed VFD network

VFD

VFD

VFD

VFD

VFD

VFDVFD

VFD VFD

VFD

RF RF

AHU AHU

FIGURE 2. Alternate network connection of variable-speed drives.

High-speed building network

Low-speed VFD networkLow-speed VFD network

VFD

VFD

VFD

VFD

VFD

VFDVFD

VFD VFD

VFD

RF RF

AHU AHU

DDC panel and

gateway

DDC panel and

gateway

DDC panel and

gateway

FIGURE 3. Indirect network connection of variable-speed drives.

One of the largely untapped cost-saving featuresof variable-speed drives is built-in intelligence.

Circle 174

variable-speed drives directly to a building control networkwith relative ease. Such a connection is shown in Figure 1.

Figure 1 illustrates the direct connection of variable-speeddrives to a building network as nodes, just as the building DDCpanels are connected. The DDC panels connect to equipmentand devices associated with the chillers, towers, and air handlers. The variable-speed drives are connected directly to the high-speed building network. The chillers may be connected via either direct network connection or the DDCpanel, as shown. The chillers’ variable-speed drives usually arepart of that overall chiller interface.

Although the means of connection shown in Figure 1 is pre-ferred because of its simplicity, many manufacturers of variable-speed drives continue to employ older communications andhave developed gateways so that their drives can be connectedto the more-advanced building control networks. So whilethese drives may advertise BACnet or LonWorks compatibility,they actually employ a gateway to provide that compatibility,

and it may not be economically attractive to buy a gateway foreach drive. This results in many direct-connect networks thatneed to be configured as shown in Figure 2. In Figure 2, theconnections are functionally the same as in Figure 1, but thevariable-speed drives are connected to their own network(s)and then connected to the high-speed building network withspecial gateways.

While the configuration shown in Figure 2 does result in connection of the variable-speed drives to the high-speedbuilding network, it is more complicated. Also, because thevariable-speed-drive network may employ older communica-tions that are slow by today’s standards, it can result in somecommunication bottlenecks, which can lead to data-exchangedelays (the issue of timing will be discussed later). In short,while the configuration in Figure 2 is acceptable and widelyemployed, the configuration in Figure 1 is preferred.

Another connection option is shown in Figure 3. This con-figuration permits the variable-speed drives to be integratedwith the building control system, but without a direct connec-tion to the building network. This configuration usually is employed when the high-speed building network is proprietaryand the interface, therefore, must be provided by the supplier

D I R E C T N E T W O R K

27HPAC Engineering • March 2003

Virtually every motor involved in abuilding HVAC system today has thecapacity to improve the efficiency

and effectiveness of the systemwhen operated with variable speed.

Circle 167

of the DDC system. For such connec-tions, gateways often are provided as anintegral component of the control pan-els, as shown in Figure 3. The most com-mon interface technique is to connect thevariable-speed drives on their own net-work and integrate them into the controlsystem via one or more DDC panels thathave gateway features. In some systems, itis preferred that each variable-speed drivebe connected to the panel in which therelated systems are connected. This typeof connection is shown in Figure 3. Thereason for this is that many of the propri-etary building networks in use today arelegacy networks that do not employ themost modern network-managementtechnologies and, therefore, can be sub-ject to certain network-traffic limitations.

DIRECT-NETWORK-CONNECTIONOPERATING CONSIDERATIONS

There has been a great deal of discus-

28 March 2003 • HPAC Engineering

D I R E C T N E T W O R K C O N N E C T I O N

Response Time and VFDs

Some designers avoid the direct network connection of variable-speed drives (VSDs)because they are concerned about the time it takes a VSD to receive and execute a

command after a signal is issued from DDC logic. Although modern network capabilities are lightning fast, at this early point in building-system networking, manyequipment manufacturers employ crude network interfaces that result in signal-transfer delays, which can add up to several seconds in certain applications. Whileresponse time for direct network connection eventually will take care of itself, thequestion is whether such delays are detrimental to proper system operation.

The answer comes in part from looking closely at the system that each variable-frequency-drive (VFD) operates. For the most part, each system involves motors, pumpsor fans, and fluid circuits, all of which contain substantial inertia. Adjusting the speed ofthese systems at short intervals not only is ineffectual, it also usually is detrimental tothe energy performance of the system. It is like driving a car by constantly pressingdown and letting up on the accelerator. Most of us were taught in driver’s educationthat this is an inefficient method of operating a car, and for similar reasons, it also istrue for fans and pumps. I recommend that control for VFD-operated motors bedeveloped so that a speed adjustment is required at intervals no shorter than 30seconds under normal operating conditions. This constraint can be implemented easily,and when done properly, it almost always leads to more stable and more efficientoperation of the system. Furthermore, with this larger interval, response-time issuesassociated with the direct network connection of VFDs disappear.

sion about the advisability of the directnetwork connection of variable-speeddrives in certain critical applications. Ithas been argued that certain motors oper-ated with variable-speed drives requirespeed adjustments once every second orless and that network connections inmany systems may delay such signals.Depending on the network, it is possibleto see a delay in information transmissionand reception of up to a second or moreunder certain circumstances. Therefore,when a network connection is employedto turn on lights from a manual switch,occupancy sensor, or similar control, themaximum end-to-end signal transmis-sion/reception time should be evaluatedcarefully. However, implementing con-trol for a variable-speed drive that subjects a motor to speed-adjustment intervals as short as 1 second is not rec-ommended because it is not energy-effi-cient. Consider that pumps and fans havesubstantial inertia, especially when thefluid involved also is taken into account.When speed adjustments are made so frequently, energy can be wasted by thecontinuous change in inertial energy.With modern control techniques,1 speedadjustment can be much less frequentwithout compromising control stability.

The direct network connection of variable-speed drives sometimes isavoided because the designer desires tohave the controlled variable (usually apressure or temperature sensor) and con-trolled device (the variable-speed drive)

connected directly to the DDC panelcontaining the control algorithm or relationship. To employ variable speed effectively, however, the control is required (by codes, in many instances) tobe composed of more information than a single controlled variable. In some cases,

this additional information is used to adjust the setpoint of the controlled variable; in others, it is used to construct avirtual controlled variable or provideother means of control. However, inmany (if not most) cases, much of the information employed for effective

variable-speed control is not available directly from a single DDC panel; there-fore, network communication is requiredfor effective control. The idea that effec-tive and efficient control can be achievedin modern buildings without the use ofthe network simply is out of date.

ADVANTAGES OF DIRECT NETWORKCONNECTIONS

Before embarking on the direct net-work connection of variable-speed drives,it is imperative that designers develop a thorough understanding of all relevant issues. Some earlier articles focus on suchissues.2,3 The benefits of the networkconnection of variable-speed drives usually far outweigh potential problems.Photo A shows an all-variable-speedchiller plant. The direct network connec-tion of the numerous variable-speeddrives in this plant reduced control costs and simplified the control system.Furthermore, the operator can access remotely or from the plant workstationall of the variable-speed-drive parametersavailable from the individual drive panels. On this project, the direct net-work connection of variable-speed driveshas resulted in greater economy and simplicity and improved the operability

D I R E C T N E T W O R K C O N N E C T I O N

29HPAC Engineering • March 2003

RS485 Connections

RS485 is a method of transmitting and receiving serial digital data over twisted-pairwiring. Many manufacturers of equipment with internal digital controls, including

variable-frequency drives, offer an RS485 connection option or port. As a result, thequestion, “Can Manufacturer A connect to the RS485 port of the equipment supplied byManufacturer B if Manufacturer A supports RS485 communications?” often is asked.The answer is, “That depends.”

The RS485 protocol establishes the lower hardware layers of communications, butdoes not ensure communication will occur. Imagine you want to communicate withyour long-lost cousin in France. You decide to contact him by telephone, knowing thatif you dial the right number, a connection will be made, and your cousin’s phone willring. However, if you do not speak French, and he does not speak English, you may nothave a meaningful conversation.

The situation is similar when only an RS485 connection is specified or offered. It isvery difficult to ensure communications because RS485 is a very popular Institute ofElectrical and Electronics Engineers standard link for proprietary communicationprotocols. Many protocols operate on RS485 connections that cannot be understoodby others. For example, a great many variable-air-volume terminal networks employRS485 communications. Until BACnet and Lon, they all were proprietary—none couldcommunicate with anything but its own kind, even though all were RS485.

In the HVAC industry, the closest thing to an open RS485 protocol may be Modbus.Modbus is a higher-level protocol implemented with software that allows communication over the RS485 physical link. It is the language spoken over the twistedpair.

But beware. Even specifying the protocol may not end the potential for communication problems. RS485 links operate at a variety of speeds (baud rates). Mostports offer a range of speed settings, but some do not. Therefore, care must be taken toensure that a common protocol and communication speed can be established whenRS485 communication is employed.

The idea that effective and efficient control can beachieved in modern buildings without the use of

the network simply is out of date.

Circle 192Circle 187

of the central plant.

REFERENCES1) Hartman, T. (2001, December).

Ultra-efficient cooling with demand-based control. HPAC Engineering, pp. 29-32, 34, 35.

2) Hartman, T. (2001, August). Effectively integrating controls utilizinggateway connections. HPAC Engineering,pp. 11, 15.

3) Hartman, T. (2000, January).Chiller-plant control using gateway technologies. HPAC Engineering, pp. 81-84, 86, 87.

Additional information on technologiesdiscussed in this article is available atwww.hartmanco.com.

For HPAC Engineering feature articlesdating back to January 1992, visitwww.hpac.com.

32 March 2003 • HPAC Engineering

D I R E C T N E T W O R K C O N N E C T I O N

PHOTO A. All-variable-speed chiller plant configured with direct network connectionsfor the variable-speed drives (silver and black enclosures).

simplest and least-costly hard w a re .The result is that, for ve ry simplesystems, the use of detailed se-quences in specifications can re s u l tin confusion and duplication of ef-f o rt. Fu rt h e r m o re, a design intentdocument for such simple systems,though still important, can be ve rybrief and should be easily deve l-oped. At present, simple stand-alone systems for comfort contro l ,lighting, fire, and life safety may beless costly than an integrated net-w o rk-based control system inmany small buildings. A simplec o n t rol system specification anddesign intent statement that placec e rtain aspects of the control sys-tem design on the control systemsubcontractor can work in thesea p p l i c a t i o n s .

Howe ve r, in larger buildings,building complexes, and cert a i nspecial-purpose buildings, theeconomies that can accrue from awell-designed, integrated networkc o n t rol system can be substantial.Facility managers can expect ve rysignificant performance improve-ments, as well as reductions in en-

It is not a simple matter tosuccessfully implement alarge-scale integrated net-w o rk control system. Ma n y

such systems as applied to build-ing control have fallen short oftheir expected performance andre l i a b i l i t y. Howe ve r, the poten-tial benefits of network contro l sa re enormous, including gre a t e rbuilding comfort for happier oc-cupants, vastly improved energye f f i c i e n c y, more effective lifes a f e t y, and lower-cost opera-tions. In many building pro j e c t s ,these benefits add significantlyto the bottom line and make in-tegrated network control config-urations essential to the pro f-itabili ty of the building. Ani m p o rtant missing ingredient inmany of the less-successful net-w o rk building control systems ist e a m w o rk. To achieve a success-ful integrated network contro lsystem, a design-and-implemen-tation process that includes ef-f e c t i ve teamwork throughout ise ve ry bit as important as the sys-tems or technologies employe d .

F E A T U R E

FOCUS ON DESIGN INTENT

It is no secret that today’s build-ing control systems are oftenpoorly specified with enormousdeficiencies in sequences of opera-tion and that most mechanical sys-tem specifications lack a clear de-sign intent. Many suchspecifications re veal that the de-signer hasn’t given much thoughtto how elements of the va r i o u sbuilding systems should work un-der certain conditions. For somesimple applications, this appro a c hmay be acceptable. Control sys-tems subcontractors usually doh a ve extensive practical experiencein implementing control for thesimpler types of comfort systemsincorporated in many buildings,and their internally deve l o p e dc o n t rol sequences may work satis-f a c t o r i l y. Also, because the contro lroutines for each manufacture r’sapplication-specific hard w a re ared i f f e rent and lack flexibility, it canbe ve ry difficult for manufacture r sto conve rt precise, specified se-quences into code that fits their

By THOMAS HARTMAN, PEPrincipalThe Hartman Co.Marysville, Wash.

Achieving and VerifyingDesign Intent:

It Takes a

Te a m

Achieving and VerifyingDesign Intent:

It Takes a

Te a m

NE T W O R K E D CO N T R O L S

14 July 2002

Integrated

Network

Controls

Require the

Right Process

and Teamwork,

as Well as

Effective

Technologies

ergy and maintenance costs of 50cents to more than a $1 per squarefoot per year that are directly at-tributable to integrated networkc o n t rol. But achieving these costreductions and the other benefitsof network control re q u i res specialattention not only to the technolo-gies applied, but the process of ap-plying them effective l y. The key tosuccess is the development ve ryearly in the project of an effectivedesign team and a design intentthat the entire team, including thep roject ow n e r / o p e r a t o r, under-stands and agrees with. Then, fur-ther project stages must be orga-n i zed to ensure that design intentis an active participant in eve ry de-cision that follows thereafter andthat the design team will support ite f f e c t i vely throughout the pro j e c t .

DEVELOPING AND COMMUNICATING DESIGN INTENT

When designers decide to con-sider implementing integratedn e t w o rk control aimed at captur-ing the benefits listed above, it is

i m p e r a t i ve to start the process byd e veloping a design-intent docu-ment during the schematic designphase with the participation, fullunderstanding, and agreement ofthe building owner and the entiredesign team. The design intentmust include the role of the inte-grated network control system.Creation of the design intent is fol-l owed in detailed design by an in-depth specification that effective l ycommunicates design intent to po-tential contractors. These steps arei m p o rtant and often not pro p e r l ye xecuted. The specification mustreflect the design intent with re a l i s-tic technologies and provide spe-cific achievable operational se-quences and goals.

Howe ve r, producing an effec-t i ve design-intent document andwe l l - o r g a n i zed and detailed speci-fications re q u i res good teamworkamong the design disciplines andthe ow n e r. This team must re m a i nt h roughout the project because itis teamwork in applying technol-ogy effectively—as much as thetechnology itself—that determinesh ow successful an integratedbuilding control network will be.

Fi g u re 1 s h ows a flow chart that,if followed pru d e n t l y, can lead tosuccessful integrated network con-t rols. The first two steps are thoseof developing and communicatingthe design intent for the networkc o n t rol system. As discusseda b ove, this process is often poorlyp e rformed in designs for buildingp rojects. Failing to develop a uni-versally agreed-upon design intentand communicating it effective l yin the specifications can be disas-t rous to the success of integratedn e t w o rk control technologies. Andthe lack of teamwork in deve l o p-ing an effective vision for the oper-ation of the building systems is of-ten the root cause of this failure .

To protect against this earlyf l a w, design engineers need to

communicate effectively to thep roject owner and design team thepotential risks and benefits of anintegrated network control system.If it is decided that such a systemshould be considered for the pro-ject, the design team must make as t rong commitment to interd i s c i-p l i n a ry teamwork and owner in-vo l vement, and a designer experi-enced in network contro ltechnologies appropriate for thep roject should be selected to jointhe team. This designer should as-sume responsibility for network is-sues, which include deve l o p i n gp re l i m i n a ry costs and benefits sothat the cost-effectiveness of net-w o rk options can be discussed.Such a designer may be part of themechanical design team (in rarecases, the electrical design team), aseparate consulting engineer work-ing directly with the mechanicaland electrical designers, or an engi-neer from a systems integrator orc o n t rol system subcontractor (thismay limit technology options thata re considered, but can work we l lin design-build projects in whichthe contractor already has selectedthe control system subcontractor).It is up to the project owner andthe initial design team to select themembers of the group who will de-velop the schematic design, whichwill include a design-intent docu-ment and pre l i m i n a ry plans for theintegrated network control system.That group must then become thec o re of the team that will see theimplementation of these technolo-gies through. It is imperative thatthe designer experienced in net-w o rk technology join the gro u pduring the schematic design stage.Integrated network control tech-nologies have subtle but ve ry im-p o rtant influences on equipmentand configurations. If the detaileddesign of equipment or specificconfigurations has begun beforethe start of the integrated net-

NE T W O R K E D CO N T R O L S

15July 2002

NE T W O R K E D CO N T R O L S

16

this conference are not forgotten,the minutes of the meeting can beincluded in the contract docu-ments. If, as a result of the meet-ing, changes to the specificationsa re necessary, an addendum isp romptly issued after the meeting.

Once the bids or pro p o s a l sh a ve been re c e i ved, they must bee valuated to determine which willbe accepted. As noted pre v i o u s l y,an advanced-technology systemcan be effectively pro c u red with abid process, but it is absolutely es-sential that the bid be accompa-nied by some additional informa-tion as set forth in thespecifications so that the teamcan make a determinationwhether or not the lowest bidmeets the intent of the specifica-tions. An article I authored titled“ Procuring Hi g h - Tech Me c h a n i-cal Sy s t e m s” in the Nove m b e r1991 issue of H PAC m a g a z i n ep rovides an in-depth discussion

w o rk design, the potential bene-fits of the network may be limitedby a degree of incompatibility be-t ween the equipment and net-w o rk, an incompatibility thatcould easily be avoided by con-c u r rent deve l o p m e n t .

Wo rking closely with the otherengineers, the architect, and theow n e r, the designer experiencedin network control technologiesshould be a key player in deve l o p-ing the design-intent documentand have responsibility for inte-grating this narrative into thec o n t rol specifications. The net-work technology designer shouldalso have primary re s p o n s i b i l i t yfor developing the control se-quences, equipment interf a c i n g ,and system operation details suf-ficient to communicate the de-sign intent to the specific systemsand equipment that are to beconnected to the integrated net-w o rk. The network technologydesigner must include in thespecifications the methods andmanner in which the contractorsinteract with the core team tooversee the implementation andp e rformance verification of theintegrated network control sys-tem.

HIGHLIGHTING AND PRO-TECTING DESIGN INTENTDURING PROCUREMENT

Using teamwork to develop asound design intent and effectivespecifications is only a start on theroad to a successful network con-t rol project. The next major chal-lenge is effectively communicatingall aspects of the design intent andi m p l e m e n t ation process to thetechnology subcontractors, whoa re so used to seeing traditionalspecifications that they cannoteasily understand approaches thata re substantially different. Tocounter this potential derailing ofthe process in the transition fro mdesign to construction stages, oneor more pre p roposal or pre b i dc o n f e rences (and after the con-

tract has beena w a rded, pre c o n-s t ruction confer-ences) are necessaryto explain to poten-tial contractors whatis expected of them,both from the stand-point of technologyand in terms of thep rocess that will bee m p l oyed to see thatthe technology is im-plemented pro p e r l y.

While all technol-ogy items for a build-ing construction pro-ject should bep ro c u red using a pro-posal-call appro a c h ,some owners and ar-chitects continue tofeel they need to uses t a n d a rd bid pro-cesses for all buildingsystems in order tomeet certain legal re-q u i rements. Whilenot the best ap-p roach, a bid pro c e s s ,if constructed pro p e r l y, can bemade to accommodate the pro-c u rement of the technology com-ponents in a building constru c-tion project that employ sa d vanced-technology integratedn e t w o rk controls. Howe ve r, nomatter what type of pro c u re m e n tis employed, a meeting withp ro s p e c t i ve contractors is abso-lutely essential before the bids orp roposals are re c e i ved.

In the pre p roposal or pre b i dc o n f e rence, the design team high-lights the design intent insofar asthe technology systems are con-cerned and also highlights areas ofthe integrated network contro lsystem that may be unusual orchallenging. Fi n a l l y, the process ofimplementation oversight and re-v i ew (which are described later) isc a refully spelled out, and ques-tions are posed about any and allaspects of the network system(s).To ensure the issues discussed in

Figure 1. Thisflow chartillustrates a step-by-step method tosuccessfullyintegrate networkcontrols andthose who shouldparticipate ineach step.

July 2002

of what should be included inspecifications for effective eva l u a-tion of integrated network con-t rol system proposals and bidsand why.

REALIZING DESIGN INTENT THROUGHOUTC O N S T R U C T I O N

The traditional submittal/re-v i ew process is insufficient for as-suring the performance of a net-w o rk control system. Thes u b m i t t a l / re v i ew process was de-signed for hard w a re and work swell to determine whether sen-sors, actuators, and some otherequipment meet specified re-q u i rements. But ensuring a goodo p p o rtunity for success for an ad-vanced network control systemre q u i res a continuous discussionb e t ween the control system sub-contractor (now more likely to bea systems integrator) and the de-signer(s) responsible for the net-w o rk control design. It also re-q u i res that the control software aswell as the operator-interf a c es o f t w a re be re v i ewed by the de-signer as it is being deve l o p e d .This includes certain software -module and interf a c e - p e rf o r-mance testing and ve r i f i c a t i o nwell in advance of the completes o f t w a re implementation. This“c o n t i n u o u s” submittal re v i ewp rocess, which starts with hard-w a re and carries on to software ,has been shown to highlight po-tential problems so that they canbe solved before the network con-t rol system and its associated soft-w a re have been finalize d .

From an organizational stand-point, one of the greatest pro b-lems that occurs during constru c-tion is that an adve r s a r i a lrelationship between the contro lsystem designer and the systemsintegrator often develops. This ismost unfortunate, and this inabil-ity to incorporate the systems inte-grator into the core team is themost notable direct cause of fail-u re in network control projects. It

is up to the design team to fosterand maintain a spirit of teamworkas the contractors are selected.While the designer is re q u i red toa p p rove or disapprove submittals,this should be a cooperative pro-cess that is greatly enhanced by thedesign team’s understanding ofthe particular control equipmentand technologies that are beingapplied. If there is no one on thedesign team with such specific sys-tem or equipment know l e d g e ,that expertise should be added asthe contractor and systems inte-grator are brought on board .

DEMONSTRATING ANDVERIFYING DESIGN INTENT

Activities of the final steps h own in Fi g u re 1 are rarely we l le xecuted except in some perf o r-mance contracts. De m o n s t r a t i n gand verifying that the entire sys-tem is actually operating as ex-pected is of crucial importance inany advanced network contro lsystem. The activities in this stepg i ve the designers and contractorsan opportunity to discuss withthe operating staff operations is-sues that support or supplementwhat was discussed in trainingsessions, helping operators part i c-ipate better as team memberscontributing to system perf o r-mance. This step invo l ves initialtesting, commissioning, and per-formance verification of the sys-tem and development of an ongo-ing process to maintain systemp e rformance over the years of op-eration. Some designers find littlei n c e n t i ve to participate in theseactivities, but participating ac-t i vely in this step is ve ry impor-tant. The purpose of all the pro-ject events up to this point is thes a t i s f a c t o ry (or even outstanding)p e rformance of the resulting sys-tem. Especially at this s tage,t e a m w o rk is crucial because themost common problems encoun-t e red at startup are those invo l v-ing multiple disciplines and/ortrades. These problems are usu-

ally readily solved by working to-gether to determine the best re s o-lution. And resolving them beforethe trades have left the site ismuch more economical thanbringing people and equipmentback after the job is completed.

SUMMARY AND CONCLUSION

The technologies described inthe articles of this supplement of-fer improved building perf o r-mance and significantly re d u c e dbuilding owning and operatingcosts easily amounting from 50cents to $1per square foot perye a r. But successfully implement-ing these technologies re q u i re sspecial attention to the pro c e s s e se m p l oyed, from initiating the de-sign all the way to verifying itsp e rformance. To ensure the suc-cess of the processes and, there-f o re, the technologies to be em-p l oyed, it takes a special emphasison teamwork that crosses disci-plines and project stages. Wi t h o u tattention to such teamwork, ad-vanced technologies can lose theirvalue and become a liability to thep roject rather than an asset. How-e ve r, with a process that incorpo-rates effective teamwork thro u g h-out the pro j e c t ,a d va n c e d - n e t w o rk technologiescan help building projects rise fara b ove their competition in pro f-itability and attractiveness to oc-cupants! N C

Additional information on tech-nologies and issues discussed in thisa rticle is available at w w w. h a rt-m a n c o. c o m. Comments and ques-tions about the article may be ad-d ressed to Mr. Ha rtman att o m h @ h a rt m a n c o. c o m.

A member of H PAC En g i n e e r i n g’sEditorial Ad v i s o ry Board, ThomasHa rtman, PE, is principal of TheHa rtman Co., which employs newtechnologies in offering enhanced, individually adjustable, andaccountable comfort systems forc o m m e rcial buildings.

NE T W O R K E D CO N T R O L S

18 July 2002

F E A T U R E

ing. Note how each system has itsown means of operations interf a c e .The interface is usually either by adedicated computer work station,o r, as in the case of simpler sys-tems, a control panel with perhapsa small digital display and keypad.

Modern buildings have manyd i f f e rent systems in place, so it canbe difficult for operators to keeptrack of them all. When a compu-ter-based work station is prov i d e d ,the staff may have the opport u n i t yto consolidate the work stationstogether in a building operationsc e n t e r. Yet even when this is done,staying on top of operations is noteasy and often re q u i res that one orm o re people be in the operationsroom at all times to keep track ofall the systems.

With a greater industry empha-sis on open communications andcommunications standards for ef-f e c t i ve building networks, it is be-coming increasingly possible to in-tegrate all the major buildingc o n t rol/monitor systems togetherso that a single point of operations

Ne t w o rk-based buildingc o n t rol systems offermany advantages in-cluding easier and

m o re convenient monitoring, im-p roved energy efficiency, simpli-fied HVAC system maintenance,self-balancing and self-setup fea-t u res, and more .

As building professionals adopta broader view of building net-w o rks, they are beginning to seethe full range of benefits achiev-able through the use of In t e g r a t e dFacility Ne t w o rks (IFNs). Thisn ew perspective includes manyd i f f e rent electronic systems be-yond HVAC that are invo l ved inthe operation of typical commer-cial buildings. With the applica-tion of network management inother arenas as a model, our in-d u s t ry is realizing how import a n tit is to bring information andc o n t rol capacity together in as t r a i g h t f o rw a rd, uniform way toc a p t u re value by stre a m l i n i n gbuilding operations and mainte-nance (O&M) activities.

EVOLVING ROLE OF BUILD-ING NETWORKS

When direct digital contro l s(DDCs) we re first introduced tothis industry around 25 years ago,their primary purpose was to re-place pneumatic controls andelectric time clocks with a moreaccurate way to control buildingH VAC systems. Ne t w o rks we reset up mainly so informationcould be exchanged between con-t rollers and equipment operators.

Digital controls, which havebeen employed in building envi-ronmental control and fire alarmsystems for many years, are nowu n i versally applied to almost alltypes of commercial buildingequipment. To d a y, digital contro l soperate security, emergency,s t a n d by generation, uninterru p t-ible powe r, and all kinds of otherspecialty systems found in mod-ern buildings. They provide notonly control capability, but alsomonitoring and alarm capability.

Fi g u re 1 shows a digital contro lscheme for a typical office build-

F E A T U R E

System designers no longer think about digital controls solely

in terms of the mechanical portion. We are now considering

entire buildings being controlled via digital integrated net-

works. Indeed, the Integrated Facility Networks (IFNs) of the

future will produce a multitude of benefits for building owners,

managers, and operators

By THOMAS HARTMAN, PEPrincipal

The Hartman Co.Marysville, Wash.

Whole Building

N e t wo r k s —

B e yond HVAC

NE T W O R K E D CO N T R O L S

36

A member of HPAC

Engineering’s Editorial

Advisory Board,Thomas

Hartman, PE, is principal of

The Hartman Co.,w h i c h

e m p l o y s new tech-n o l o g i e s

in offeringenhanced,

i n d i v i d u a l l yadjustable, and ac-countable comfort

systems for com-mercial buildings.

NE T W O R K E D CO N T R O L S

37

i n t e rface can be created to connectbuilding operators to each digitallyc o n t rolled and monitored systemin the building (Fi g u re 2) .

While it is exciting to think thata single communication standardcould be stipulated so that allequipment could be connectedtogether seamlessly, we’re notquite there yet. There are manyd i f f e rent manufacturers invo l ve d ,and many of the products foundin commercial buildings are alsoused in ve ry different types of fa-cilities. To d a y, and ve ry likely we l linto the future, one or more “g a t e-w a y s” (shown in Fi g u re 2 as theBuilding Controls Ne t w o rk In t e r-face) are re q u i red to connect thesystems together to share neces-s a ry information and control fea-t u res, and to employ a commonoperator interf a c e .

Methods for interfacing equip-ment in today’s buildings va ry. Fi g-u re 2 shows each DDC system di-

rectly connected to the gatew a y,which is in turn connected to abuilding network. Wo rk stationscan be installed anywhere in thebuilding or be accessed remotely to

connect the operator to any of thea c t i ve building systems. Alterna-t i ve integration schemes that con-nect each system to a building net-w o rk and employ a translating

Figure 1. Digitalsystems in atypicalcommercialbuilding.

Figure 2. A typical buildingsystemintegrationnetwork scheme.

NE T W O R K E D CO N T R O L S

p ro p e rty management firm hascontracts to manage a dozen or sobuildings in a metropolitan are a .The firm may assign one or moregeneral technicians to each build-ing, or to small groups of build-ings located near one another.Howe ve r, one group of specialtytechnicians may be responsible forthe fire alarm systems, another forthe chillers and towers, and an-other for the standby generators inall the buildings.

This type of O&M arrange-ment, coupled with an integratedfacility management system thatties all the buildings together,could greatly reduce O&M costs.The opportunity to reduce main-tenance costs through network e dmanagement approaches are espe-cially attractive in light of re c e n tre s e a rch that shows equipmentsuch as chillers, fans, and pumpsa re most effectively maintainedwhen maintenance is scheduledbased on attention to certain oper-ating parameters rather thanequipment ru n t i m e .

An Integrated Fa c i l i t yNe t w o rk (IFN) connecting thebuilding systems together(Fi g u re 3) provides direct access tothe types of operating parametersn e c e s s a ry to provide more effective“fault based” maintenance pro-grams for any type equipment inthe integrated building network. Itis also not difficult to envisionoverlapping networks wherein sep-arate building management firmsand specialty O&M contractorscooperate together and with othersso that each can get informationf rom the buildings and systemsthey are responsible for to supportthem more efficiently.

In Fi g u re 3, the Internet pro-vides the inter-building communi-cations backbone. Use of the In t e r-net in place of dial-in/dial-outconnections vastly increases thespeed of data transmission andusually lowers the costs associatedwith the inter-building network .The use of the Internet for a com-munications backbone has someother potential advantages. Al-though it is not a re q u i rement, it isbecoming a significant adva n t a g e

38

g a t eway separately, usually as partof the operator workstation arepossible. In other schemes, all gate-way connections are made dire c t l yto special DDC controllers on thededicated control network and thebuilding Internet protocol (IP)n e t w o rk shown in Fi g u re 2 m a ynot be employed at all.

W h a t e ver method of interc o n-nection and integration is used totie the various building contro land monitoring systems together,the advantage of such network in-tegration is apparent. Instead ofhaving to physically go to eachsystem to check its operation, ev-e rything can be checked at one lo-cation. This provides much betterassurance that critical buildingsystems will be supported pro p-e r l y. This integration also permitsa single dial-in/dial-out modemconnection so that the operations

room need not be staffed continu-ously to provide 24/7 support ofcritical building systems.

Another benefit of network in-tegration is the need for intero p e r-ability among the various buildingsystems to minimize the disru p-tion caused by an abnormal eve n tin any of the systems, and to coor-dinate the operation of these sys-tems to minimize overall buildingoperation costs.

MULTI-BUILDING MANAGEMENT

The type of system integrations h own in Fi g u re 2 g i ves the opera-tions staff located in a stand-alonebuilding or complex improve doversight capabilities. Howe ve r,this building-by-building man-agement approach to O&M israpidly giving way to multi-build-ing approaches where integrateddigital networks are an essentialc o m p o n e n t .

Consider that nearly all largebuildings have chillers, coolingt owers, standby generators, and ahost of other equipment. The op-erations staff in a single buildingcannot afford the time or re-s o u rces to become experts on ev-e ry piece of equipment. In mostsingle-building O&M schemes,the support of this complex equip-ment is provided by a combina-tion of the building’s maintenancestaff and outside contractors.These arrangements usually costm o re and are not as effective asspecifically trained O&M staffwhose responsibilities are focusedon the equipment and who haved i rect access to it at all times.Howe ve r, such specially trainedtechnicians, whether they are onstaff or with an outside contractor,may be economically justifiedwhen a large amount of equip-ment is invo l ve d .

To achieve these economies ofO&M, some building manage-ment and contract maintenancefirms have conceived and deve l-oped multi-building managementconcepts where buildings areg rouped together to maximize thec o s t - e f f e c t i veness of providing cer-tain services. For example, say a

F E A T U R E

H VAC and Lighting Control IntegrationAn integrated HVAC and lighting system can

create cost savings, improve energy efficiency,and enhance occupant comfort. It can also makethe building operator’s job easier. The properpoint for integrating the two systems is within thebuilding environmental control system (Figure 1) .

Lighting and HVAC systems share the samebasic operating parameters. Furthermore, theyare interrelated in terms of building energy use,due to the effect of lighting on the HVAC system.Coordinating the operation of the two systems atthe zone level is an important design and controlsissue in any commercial building. Both need to re-act to occupancy and occupant preferences atthe smallest practical zone size. Zones should bethe same, or coordinated, for both systems.

Successful integration of HVAC and lighting re-quires good coordination between the mechani-cal and electrical designers. It also requires ef-fective leadership on the architect’s part. Themechanical and electrical designers must under-stand integration issues, and be anxious to se-cure the benefits of integration for their client.

Designers who have successfully integratedHVAC and lighting control should use the proce-dures as a model for integrating other buildingsystems. Those who have not yet successfullyintegrated HVAC and lighting controls shouldconsider it a first step in their building systemintegration efforts.

Figure 3.Integrated facilitymanagementsystem formultiplebuildings.

NE T W O R K E D CO N T R O L S

39

to employ certain Internet stan-d a rds such as the use of standardWeb browsers for the communica-tion and display of information toand from the systems in eachbuilding. By adopting widely em-p l oyed Internet approaches, theIFN will be less dependent on thevarious control systems ve n d o r swhose products are used thro u g h-out the buildings in order to beconnected to the network .

CHALLENGES OF MULTI-BUILDING NETWORKS

Although the potential oppor-tunities for implementing IFNsystems are substantial, a num-ber of issues persist that makeimplementing such networks aser ious challenge for the de-signer. These include:

The Role of IT Departments in Integrated Facility NetworksAlthough Information Technology (IT) departments generally, and correctly, view

building systems integration as relatively minor applications for their Intra and/or In-ter building networks, they should be consulted early and continuously while the In-tegrated Facility Network (IFN) is being designed. The main reason is so the design-ers can get information from the IT managers about the corporation or institution’sapproach to building and supporting the networks involved. The most importantquestions designers need to answer are:

1. How critical do the IT managers view this network?2. How is network security provided, and how can security for this application be

made compatible with others?3. How does the IT management intend to support this network and network

a p p l i c a t i o n s ?The answers to these questions will help the design team map out an integration

scheme that 1) maximizes the use of internal network resources without jeopardizingthis or any other application, and 2) is compatible with development and operationsstandards and procedures that are used for other network applications. Such compat-ibility is essential to ensure that the IFN can and will be readily supported along with theother applications on the network.

NE T W O R K E D CO N T R O L S

43

1. A lack of uniformity in howa point of control or monitoringis re p resented, adjusted, or ove r-ridden at the operations workstations can make it difficult foroperations staff to interact withindividual systems within anIFN. In some cases, operationsstaff is required to know and un-derstand a whole array of indi-vidual digital control system fea-t u res even though they arere p resented on a common plat-form. This increases the com-plexity of such a platform anddecreases its effectiveness.

2. Understanding the criticalcomponents of each building sys-tem when thinking in terms of in-tegration. Fi re alarm, security, andu n i n t e r ruptible power systemsneed to stand alone within eachbuilding so all critical functionswill operate through specificallya p p roved equipment and com-munication links, re g a rdless of thestate or condition of the integra-tion network components.

3. Re g u l a t o ry and administra-t i ve practices for the non-criticalfunctions of many building sys-tems. For example, some fire ju-risdictions expect non-criticalf i re alarm management tasks tobe administered entirely fro mwithin the facility served by thef i re alarm system. Remote moni-toring and operations of non-

alarm functions that may in-clude resetting trouble messagesor adjusting detector sensitivitya re new concepts for some offi-c ials, and can be difficult forthem to accept and approve .The fact that fire alarm systemscan be better supervised byg rouping them together and in-c reasing the attention and super-vision over a facility networkmay take some time for these of-ficials to acknowledge. Pa t i e n c eby the bui lding managementand operations team in deve l o p-ing such integration pro g r a m smay be re q u i re d .

4. Lack of communicationss t a n d a rds. A lack of standards isthe greatest barrier to widespreadimplementation of IFNs. Al-though many digital contro l smanufacturers claim to use openand/or standard communicationp rotocols, connecting dive r s esystems effectively to integratedn e t w o rks continues to take spe-cial attention and oversight andis often more expensive than nec-e s s a ry. The near-term goal forour industry should be to con-vince all the various manufactur-ers of building systems that havedigital controls to subscribe notjust to open communicationsp rotocols, but to adopt industryre c o g n i zed standard communi-cation protocols that allow their

equipment to be tied easily ande f f e c t i vely into larger integratedbuilding management networks.

SUMMARYOwners stand to achieve signif-

icant O&M cost savings thro u g hthe use of IFNs that connect sys-tems within buildings, and thenby aggregating multiple buildingst o g e t h e r. Howe ve r, implementinge f f e c t i ve and economical IFNsthat enable effective re m o t eO&M management continues tobe more complicated and morecostly than necessary.

Fortunately, the industry is be-ginning to understand the op-p o rtunities and impediments toe f f e c t i vely implementing IFNs.All parties should join togetherto encourage the use of industry-re c o g n i zed communication pro-tocol standards for all buildingequipment that has digital con-t rols. This will help us move to-w a rd more widespread applica-tion of IFNs, which will result ing reater economies for buildingowners. NC

T h o m a s Ha r t m a n , P E ,Fi r s t Ri ght s , The Ha rt m a nC o m p a n y,w w w. h a rt m a n c o. c o m,9905 39th Dr. NE, t o m h @h a rt m a n c o. c o m, Ma rysville, Wa s h -ington 98270, © Fe b ru a ry 22,2 0 0 1 , 3 6 0 . 6 5 8 . 1 1 6 8 .

One of the great things about integrated build-ing systems is the opportunity they present to gar-ner the benefits of interoperability among systemswhich can result in enhanced control, more flexi-bility, and impressive cost savings.

Designers have long accommodated the need forinteroperability between fire alarm and environmen-tal controls systems. However, consider the furtherbenefits of integrating security systems to providehigher levels of security without limiting egress dur-ing emergencies. Network-based integration mayalso permit standby generation to be used as an au-tomatic peak power reduction tool, and the uninter-

ruptible system to shed all but the most critical loadsas available backup power is consumed.

The opportunities from network integration alsoinclude reduced capital costs. Smaller centralheating and cooling plants can sometimes bespecified when the operation of heat generationequipment in many types of buildings is better co-ordinated with integrated network control. So,while Integrated Facility Networks (IFNs) are oftenseen primarily as information gathering tools, de-signers should never lose track of the potentialbenefits of incorporating system interoperabilityas a part of their IFN design.

Benefits of Interoperability Among Building Systems

F E A T U R E

With another energy crisis possibly looming,

HPAC Engineering dusts off and revises

a list of ways to conserve fuel and electricity

WAYS TO SAVE

ENERGY

Engineers must evaluate items on these lists toensure they are suitable and safe for their buildings.

BEST PRACTICES1. Participate in voluntary above-code perform-

ance programs, such as Building Research Estab-lishment Environmental Assess-ment Methodology (BREEAM)and the U.S. Green BuildingCouncil’s Leadership in Energyand Environmental Design

(LEED).2. Verify that the design conforms to or surpasses

the state’s energy-conservation code (if applicable)and/or ASHRAE standards.

3. Keep operation-and-maintenance (O&M)documentation up-to-date and be certain thatO&M staff are knowledgeable, well-trained, informed, and proactive in their duties.

4. Establish energy-consumption and demandgoals and track and discuss them with operationsstaff at regular intervals.

5. Work as a team to keep in mind the buildinglife cycle throughout the design, construction, andoperations phases.

6. Educate occupants on recognizing and report-ing IAQ and energy problems and reward them fordoing so.

7. Participate in public and peer reviews of design for adherence to new codes and standards.

8. Encourage participation in professional organizations, provide training in sustainable prac-tices, subscribe to quality trade publications, andbookmark useful Websites.

THE BUILDINGEnvelope

9. Design for low infiltration by employing ahigh-quality envelope and isolating elevator shafts

In May 1975, HPAC Engineering published atwo-page “data sheet” listing 133 ways to saveenergy in existing buildings. Modifications to

the building shell, HVAC and lighting systems, andoperations-and-maintenance practices were sug-gested. A lot of emphasis was placed on industrialbuildings, as well as generalHVAC considerations. Publishedat the peak of the oil embargo, thelist was very well-received.

Once again, energy conserva-tion is a priority for property owners—a trend thatis likely to last beyond the current surge in oil, gas,and electricity prices and regional supply uncertain-ties. Responding to reader inquiries for an updatedlist, we revised the 1975 data sheet.

Although the list from 1975 has many brute-force energy-conservation measures that are validtoday, there have been a lot of refinements to the artof saving energy. To list them all would be impossi-ble. The point of this list is to inspire progress inachieving energy-efficient systems, not to docu-ment every way to save a Btu.

The following is the revised list. Some old meth-ods, such as, “Fix broken windows,” were dropped,while a new category—best practices—was added.Also, the list was broken down into more subcate-gories to make it easier to use. For comparison, the1975 list is contained in the sidebar on pages 51and 52.

50 March 2001 • HPAC Engineering

Continued on page 54

Then and Now

Prior to becoming editor of HPAC Engineering,Michael G. Ivanovich was a research scientist workingon environmental projects in a variety of areas, includ-ing energy efficiency. Thomas Hartman, PE, is the principal of The Hartman Co., a Marysville, Wash.-based high-technology engineering and developmentfirm. Jack Terranova, PE, is division manager for Colliers ABR Inc. in New York. Both Hartman andTerranova are members of HPAC Engineering’s Edito-rial Advisory Board.

By MICHAEL G. IVANOVICH, Editor;THOMAS HARTMAN, PE; and

JACK TERRANOVA, PE

51HPAC Engineering • March 2001

W A Y S T O S A V E E N E R G Y

The following list originally was published in the May 1975 issue of HPAC Engineering.

STRUCTURE1. Add additional insulation to roofs, ceilings, or walls where

practical.2. When reroofing, use light-colored material to reduce solar gain

on air-conditioned structures.3. Ventilate attic spaces.4. Put solar film on windows to cut cooling loads.5. Install solar screens on windows to reduce cooling loads.6. Install double glazing in place of single glazing.7. Recaulk window and door frames.8. Reseal curtain walls.9. Eliminate excessive crackage between double-entry doors.10. Install weather stripping around windows and doors.11. Repair broken windows.12. Keep garage and warehouse doors closed as much as possible.

LIGHTING AND POWER13. Install higher-efficiency lighting systems where possible.14. Reduce overall illumination levels.15. Implement a lighting-maintenance program to obtain maximum

efficiency from existing systems.16. Use supplemental lighting for specific tasks instead of

increasing the overall illumination for a given area.17. Utilize natural lighting in perimeter office spaces.18. Utilize multiple switching for selective lighting levels in offices,

conference rooms, etc.19. Reduce lighting in areas not requiring higher levels (stockrooms,

corridors, etc.).20. When redecorating, use light colors on ceilings and walls to

achieve good illumination levels with less lighting.21. Reduce decorative and advertising lighting.22. Use timers or photocells to control outdoor lighting.23. Reduce parking-lot lighting to minimum levels required for safety.24. Use proper-sized motors. Grossly oversized motors operate at a

low power factor.25. Apply power-factor correction where applicable.26. Install demand-limiting equipment.

CONTROLS27. Recalibrate all controls.28. Lock thermostats to prevent resetting by unauthorized

personnel.29. Check room thermostats for proper location.30. Install individual room control whenever possible.31. Install temperature-control valves (self-contained) in radiators

controlled by hand valves.32. Install enthalpy controls to optimize use of outdoor air for

building cooling.33. Install building automation system if feasible.

HVAC AND MISCELLANEOUS34. Study system carefully before making changes—some changes

may increase energy usage.35. Retest, balance, and adjust systems.36. Turn off air-conditioning machinery during unoccupied hours.37. Revise cleanup schedule so lights and system can be turned off

earlier.38. Optimize system-startup times.39. Shut off outdoor air during unoccupied hours.40. Reduce outdoor-air quantity.41. Reduce system air volume.42. Reduce air-duct leakage.43. Adjust outdoor-air dampers for tight closure.44. Replace dampers with higher-quality ones whenever possible.45. When balancing or rebalancing a system, consider outdoor-air

leakage when making minimum-outdoor-air settings.46. Adjust dampers in mixing boxes and multi-zone units so that they

shut off tight to reduce leakage.47. Avoid use of preheat coils if possible.48. Raise the mixed-air temperature.49. Reset hot and cold deck temperatures in direction of reduced

heating and cooling.50. Set reheat schedule as low as can be tolerated.51. Reset chilled-water and heating-water temperature in

accordance with loads.52. Use lowest radiation temperature possible in perimeter spaces.53. Do not permit perimeter and interior systems to buck one

another.54. Optimize multiple chiller operation.55. Run heating and cooling system auxiliaries only when they are

required.56. Whenever possible, only operate return-air fans for heating

during unoccupied hours.57. Install auxiliary air risers to reduce fan horsepower.58. Convert constant-volume fan system to VAV.59. Reduce heating in unoccupied areas.60. Reduce heating in overheated spaces. Do not open the window

to cool these places!61. Shut off exhaust fans during unoccupied cycles.62. Check exhaust systems to ensure they are exhausting only the

amount of air required.63. Reduce exhaust-air quantities from toilet rooms, laboratories,

etc. when feasible.64. Convert toilet-room exhaust fans to operate only when room is

occupied.65. Reduce supply temperature of domestic hot-water systems.66. Use condenser water to preheat domestic hot water.67. Use condenser water for air-conditioning reheat.68. Retrofit solar collectors to the building to heat domestic and

process water.

133 Ways to Save Energy in Existing Buildings

Continued on page 52

52 March 2001 • HPAC Engineering

W A Y S T O S A V E E N E R G Y

69. Install heat-recovery device to reclaim heat from building, kitchen, and process exhaust.

70. Replace forced-air heaters with infrared heaters.71. Replace indirect-fired makeup-air units with direct-fired

equipment.72. Insulate piping and ductwork located in unconditioned spaces.73. Replace worn insulation on boilers, furnaces, pipes, ducts, etc.74. Reduce hot- and chilled-water flows.75. Trim pump impellers to match load.76. Convert three-way valves to two-way operation and install

variable-speed pumping.77. Clean strainer screens in pumping systems.78. Check vents in hot-water and steam systems for proper

performance.79. Check expansion-tank size. Undersized tanks can cause

excessive water loss.80. Determine whether the boiler plant can be shut down and small

boilers and water heaters can be used during the summer.81. Do not waste condensate; return it to the boiler.82. When high-pressure steam is available, consider use of steam

turbines for pump and fan drives. Turbines can operate as a PRVvalve to meet low-pressure-steam needs.

83. Repair all leaks.84. Use proper water treatment to reduce fouling of transfer

surfaces in boilers, chillers, and heat exchangers.85. Use no more water-treatment chemicals than necessary.86. Check cooling-tower bleedoff periodically to ensure that water

and chemicals are not wasted.87. Maintain cooling towers, evaporative coolers, and air-cooled

condenser for peak efficiency.88. Periodically inspect and repair faulty equipment.89. Implement a filter-maintenance program to ensure peak

efficiency.90. Clean and maintain cooking equipment to maintain peak

efficiency.

COMBUSTION EQUIPMENT91. Check buildings for negative pressure, which can reduce

combustion efficiency.92. Check flues and chimneys for blockages or improper draft

conditions.93. Clean combustion surfaces.94. Check and adjust fuel-air ratios.95. Replace atmospheric burners with power burners.96. Install pressure controls on furnaces (industrial).97. Install automatic air-gas combustion controls.98. Do not overfire equipment.99. Repair furnace linings frequently.100. Reduce production-equipment preheat times to minimum

required.101. Reduce production-equipment temperatures to holding

temperatures when production stops for relatively long periods.102. Shut off drying and curing ovens when not in use. Do not start

them until just prior to shift.

103. Seal all cracks in furnaces, ovens, etc.104. Preheat combustion air with waste heat.

INDUSTRIAL PLANTS105. Study plant heating and air-conditioning systems to determine

if they are of correct design. Many are not.106. Reschedule operations whenever possible to second and third

shift to get them off the 10-a.m.-to-2-p.m. peak electric-demand period.

107. Plan work so that the whole plant can be shut down on given weekends.

108. Shut off machinery when not in use.109. Keep covers on tanks and vats closed to reduce evaporation

losses.110. Use push-pull ventilation on open surface tanks; 50 percent or

more of the air can be saved.111. Use immersion heaters whenever possible.112. Use cold-water detergent in washers whenever possible.113. Combine operations where possible to reduce the number of

washers.114. Shroud openings of furnaces, ovens, paint booths, and washers

so that only the minimum amount of exhaust air will be required.115. Eliminate stratification of air in the plant during the winter,

thereby warming the floor. This can be done easily with fans (high) blowing down ducts terminating close to the floor.

116. Use spot heating or cooling of people when they are located farapart. Each should have control of the air direction and velocity over them.

117. Use evaporative cooling for human cooling whenever practical.118. Determine whether pressure blowers could replace some

compressed-air usage.119. Do not use compressed air at higher pressures than required.120. Do not permit compressed air to be used for “people” cooling.121. Reduce the quantity of exhaust air; use local, not general,

exhaust.122. Use low-volume, high-velocity exhaust systems whenever

possible.123. Analyze all solid waste to determine whether it can be

recycled, burned, or composted.124. Salvage all oil used in the plant. It either can be reused through

refining or burned in the boilers.125. If exhaust air is contaminated, evaluate air-cleaning devices to

determine if the air could be cleaned and recycled.126. Determine feasibility of utilizing energy in production

operations before exhausting it.127. Analyze interplant truck runs. Consolidate loads and eliminate trips.128. Shut off interplant truck engines when not in use.129. Shut off fork-lift engines when not in use.130. Replace worn-out machinery with modern, efficient equipment.131. Keep heat and smoke-relief vents closed during the winter.132. Consider using waste water for roof sprays during the summer

to reduce heat load on the plant.133. Use automatic regulators to control the volume of water used.

Continued from page 51

54 March 2001 • HPAC Engineering

and other ventilated areas. Specify tests toensure that design criteria are met.

10. Employ the highest practical insu-lation values for the roof, walls, and glaz-ing. Consider reduced mechanical-sys-tem size and improvements in comfort.

11. Use swing/revolving doors atbuilding entrances. According to theAmerican Society of Heating, Refrigerat-ing and Air-Conditioning Engineers, theuse of these doors can reduce infiltrationby 75 percent compared with standardsingle and double-leaf doors.

12. Specify Low E, double- or triple-pane argon-filled glass with a factory-applied coating (low shading coefficient).Use interior blinds and/or shades to helpfurther reduce cooling loads during peakhours.

13. Consider planting shade trees andshrubbery to reduce cooling loads.

14. Maintain and improve the enve-lope’s resistance to infiltration through

aggressive building-entrance and enve-lope-maintenance/upgrade programs.Mechanical/electrical systems—general

15. Perform an energy study or visit anearby site with the application that youare considering.

16. Periodically re-evaluate loads toensure that the HVAC system and com-ponents are properly sized and opera-tionally balanced.

17. Incorporate distributing schedul-ing so that comfort and lighting energy isexpended only when building areas areoccupied.

18. Maintain good control of outsideair to eliminate excessive outside-air flowat extreme temperature conditions whilemaintaining adequate air quality at alltimes of occupancy. Use enthalpy controlfor air economizers.Controls

19. Design controls for straightfor-ward operation and maintenance withself-diagnostics and other integrated

fault-detection capabilities.20. Check the location of sensors and

control loops and recalibrate them as required to maintain design conditionsfor comfort and energy efficiency.

21. Keep the control system operatingas designed by correcting problems thatrequire manual overrides and other per-formance-limiting fixes.

22. Upgrade to higher levels of room-or work-station-based control wheneverpossible.

23. Maintain control software to ensure that programs do not contain errors. Upgrade as required to add con-trol features such as improved schedulingor more operator-adjustment capacity.

24. Upgrade controls or the controlsystem when feasible. The replacementof pneumatic systems with direct digitalcontrol can be particularly effective.

25. Consider adding network featuresto improve alarm and maintenance response and to react to real-time energy-

W A Y S T O S A V E E N E R G Y

Continued from page 50

55HPAC Engineering • March 2001

pricing signals.26. Maintain a sound maintenance

program that includes periodic checksfor broken, stuck, or loose dampers, link-ages, control valves, and other mechani-cal control devices.

27. Consider upgrading work stationsto provide more information to operators.

28. Consider installing power sub-meters. Also, consider installing tem-porary data loggers or extending thecontrol system to detect load andscheduling anomalies.

AIR SYSTEMS29. Convert constant-volume fan

systems to variable-air-volume (VAV)systems to reduce system air volume andenergy use at part-load conditions.

30. Replace mechanical air-volume-control VAV systems with variable-fre-quency-drive (VFD) air-volume control.

31. Place fixed duct static setpointcontrol of fans with network-based ductstatic reset or direct zone control of fanvolume to reduce energy at part loads.

32. Improve design, installation, andmaintenance standards to reducedamper, air-duct, and plenum penetra-tion. Also, improve the design and main-tenance of ducts and duct insulation topromote smooth and efficient air flow.

33. Design and maintain a system thatcan closely control outside air to the actual requirements at all times.

34. Avoid the use of preheat and reheat coils, as well as lockout operationof such coils, when possible.

35. Design and operate systems toraise the mixed-air temperature and to reset hot and cold deck temperatures atpart-load conditions.

36. Size filters properly. Do not “over-filter.” Also, choose the proper filter efficiency for the job. Ensure that filters donot leak around the edges. Install/specifyalarms to measure the differential pressureacross each filter bank with alarms back tothe burner-management system. Imple-ment a filter-maintenance program to ensure that changes are made to maximizeefficiency and minimize filter costs.

37. For VFDs and soft-start fans, usesprocketed belt drives for higher efficien-cies. Inspect all fan belts periodically andreplace them as needed.

38. Relocate or redirect outdoor-airintakes as required to minimize the in-

take of exhaust air.39. Design and maintain air outlets

and returns to make them free from ob-structions.

EXHAUST SYSTEMS40. Shut off exhaust fans during all

unoccupied cycles except economy

purge cycles.41. Check exhaust systems to ensure

that they are exhausting only theamount of air required. If possible, re-duce exhaust-air quantities from toiletrooms, laboratories, etc. to minimumacceptable levels.

42. Control room and process exhaust

W A Y S T O S A V E E N E R G Y

56 March 2001 • HPAC Engineering

fans to operate only when a room is occu-pied or equipment is running.

43. Install an energy-recovery deviceto reclaim energy from building andprocess exhaust-air systems.

44. Develop operations strategiesthat keep laboratory fume-hood doorsclosed and exhaust fans off when they

are not in use.

STEAM AND HYDRONIC SYSTEMS45. Reset chilled-water and heating-

water temperature in accordance withloads.

46. Optimize cooling-tower opera-tion, and use cooling-tower water for hy-

dronic “free cooling” when possible.47. Find and stop steam leaks; insti-

tute a trap-, vent-, and strainer-mainte-nance program; and insulate steam, hot-water, and condensate lines.

48. Replace old chillers and boilerswith new, energy-efficient equipment.

49. Design or convert to variable-flowhydronic systems with VFDs to reducehot- and chilled-water flows at part-loadconditions. Use network-based reset ofdifferential-pressure-set-point or direct-load control of pump speed for lower en-ergy use.

50. Increase pipe diameters to decreasepump head when feasible.

51. Determine whether the boilerplant can be shut down and replacedwith smaller boilers and water heatersduring the summer.

52. Ensure that all condensate is re-turned to the boiler plant.

53. Consider replacing PRVs withsteam turbines for pump and fan drives.A turbine can operate as a PRV valve tomeet low-pressure-steam needs.

54. Use proper water treatment to re-duce the fouling of transfer surfaces inboilers, chillers, cooling towers, and heatexchangers. Ensure that only the neces-sary amount of water-treatment chemi-cals is used. To cut down on water usage,use improved chemical treatment.

55. Maintain cooling-tower and boilerbleed and blowdown cycles to ensurethat water and chemicals are not wasted.Consider sidestream filtration for coolingtowers to reduce water and chemical use.

56. Install water-efficient plumbing toreduce hot-water consumption andpumping.

57. Maintain cooling towers, evapora-tive coolers, and air-cooled condensersfor peak efficiency.

58. Use low-flush (1.6-gpm) toilets,occupancy sensors at flushometers, etc.

59. Add a water-side economizer cyclefor free cooling instead of operating thechiller.

MISCELLANEOUS60. Periodically retest, balance, and

adjust air and hydronic systems.61. Commission building systems af-

ter retrofits and renovations.62. Improve access to filters, coils, and

other system components to make main-tenance easier and more frequent.

W A Y S T O S A V E E N E R G Y

57HPAC Engineering • March 2001

63. Optimize multiple-chiller oper-ation to stage chillers for maximum ef-ficiency and the lowest electrical de-mand.

64. Design and operate systems forcoordinated control to minimize si-multaneous heating and cooling in pe-rimeter and interior systems or withinair handlers and unitary equipment.

65. Reduce heating and raise coolingsetpoints in unoccupied areas.

66. Notice when personal heatersand fans are being used. Find out whyand then solve problems with thebuilding HVAC system.

ALTERNATIVE ENERGY67. Design or retrofit solar collectors

to heat domestic and process water.68. Consider photovoltaic roofing

materials and curtain wall panels.69. Consider infrared heating in

place of forced-air heaters for areas sub-ject to much outside-air infiltration.

COMBUSTION EQUIPMENT70. Design and periodically check

combustion equipment for proper airsupply and adequate flue conditions.

71. Periodically check and adjust fuel-air ratios on combustion equipment.

72. Consider installing automatic air-gas combustion controls.

73. Do not overfire equipment. Care-fully plan and facilitate operation to min-imize energy usage while meeting processor comfort-load requirements.

74. Replace indirect-fired makeup-airunits with direct-fired equipment whenfeasible.

75. Preheat combustion air with wasteheat.

76. Check boilers, furnaces, ovens, etc.for cracks or damage and repair as re-quired. Keep combustion surfaces cleanand well-maintained.

INDUSTRIAL PLANTS77. Update plant heating and air-

conditioning systems as comfort andprocess requirements change.

78. Consider operations schedules tominimize peak electric-demand periods.

79. Plan work so that the whole plantcan be shut down on given weekends.

80. Shut off machinery when not inuse, such as during lunchtime.

81. Keep covers on tanks and vatsclosed to reduce evaporation losses.

82. Use cold-water detergent inwashers whenever possible.

83. Combine operations when possi-ble to reduce the number of washers.

84. Shroud openings of furnaces,ovens, paint booths, and washers to min-imize the amount of exhaust air required.

85. Eliminate stratification of air inhigh-bay plants during the winter byheating at the floor level.

86. Use spot heating or cooling forpeople when they are located far apart.

87. Use evaporative cooling for hu-man cooling when practical.

W A Y S T O S A V E E N E R G Y

Continued on page 75

75HPAC Engineering • March 2001

88. Determine whether pressureblowers could replace some compressed-air usage. Do not use compressed air athigher pressures than required.

89. Design for reduced exhaust airwith the use of local, rather than general,exhaust systems, such as ventilated weld-ing guns, hoods for portable grindingequipment, and local traveling hoods formolten-metal pouring.

90. Consider recycling industrial-waste streams.

91. Consider heat recovery or filtra-tion for recycling industrial exhaust air.

92. Determine the feasibility of utiliz-ing waste heat in production operationsbefore exhausting it.

93. Analyze interplant truck runs.Consolidate loads and eliminate trips.

94. Replace worn-out machinery withmodern, efficient equipment.

95. Keep heat- and smoke-relief ventsclosed during the winter.

LIGHTING AND POWER96. Design and retrofit high-effi-

ciency lighting systems.97. Reduce excessive illumination.98. Consider a group relamping

program for bulbs and ballasts.99. Install daylighting utilizing dim-

ming ballasts and photo sensors.100. Utilize multiple switching for

selective lighting levels in offices, con-ference rooms, etc. Utilize dimming-ballast technology to have the lightlevel match the task.

101. Reduce lighting in areas thatdo not require high levels of it, such asstockrooms and corridors.

102. When redecorating, use lightcolors on ceilings and walls to achievegood illumination levels with less light-ing.

103. Reduce decorative and adver-tising lighting.

104. Use timers or photocells tocontrol outdoor lighting.

105. Implement bi-level lightingtriggered by motion sensors to lowerlight levels when no one is around.

106. Reduce parking-lot lighting tothe minimum levels required for safety.

107. Use proper-sized high-effi-ciency motors operated by VFDs whencapacity requirements fluctuate.

108. Apply power-factor correctionwhen possible.

109. Stage the startup of loads tokeep demand down, and install de-mand-limiting equipment.

110. Connect lighting and comfort-conditioning systems to the buildingautomation system to minimize energyuse in unoccupied areas.

111. Replace low-efficiency fixtureswith high-efficiency fixtures (HID andHPS).

112. Install motion sensors anddimmable fluorescent fixtures in ware-houses for maximum energy savings.

W A Y S T O S A V E E N E R G Y

Continued from page 57

75HPAC Engineering • March 2001

88. Determine whether pressureblowers could replace some compressed-air usage. Do not use compressed air athigher pressures than required.

89. Design for reduced exhaust airwith the use of local, rather than general,exhaust systems, such as ventilated weld-ing guns, hoods for portable grindingequipment, and local traveling hoods formolten-metal pouring.

90. Consider recycling industrial-waste streams.

91. Consider heat recovery or filtra-tion for recycling industrial exhaust air.

92. Determine the feasibility of utiliz-ing waste heat in production operationsbefore exhausting it.

93. Analyze interplant truck runs.Consolidate loads and eliminate trips.

94. Replace worn-out machinery withmodern, efficient equipment.

95. Keep heat- and smoke-relief ventsclosed during the winter.

LIGHTING AND POWER96. Design and retrofit high-effi-

ciency lighting systems.97. Reduce excessive illumination.98. Consider a group relamping

program for bulbs and ballasts.99. Install daylighting utilizing dim-

ming ballasts and photo sensors.100. Utilize multiple switching for

selective lighting levels in offices, con-ference rooms, etc. Utilize dimming-ballast technology to have the lightlevel match the task.

101. Reduce lighting in areas thatdo not require high levels of it, such asstockrooms and corridors.

102. When redecorating, use lightcolors on ceilings and walls to achievegood illumination levels with less light-ing.

103. Reduce decorative and adver-tising lighting.

104. Use timers or photocells tocontrol outdoor lighting.

105. Implement bi-level lightingtriggered by motion sensors to lowerlight levels when no one is around.

106. Reduce parking-lot lighting tothe minimum levels required for safety.

107. Use proper-sized high-effi-ciency motors operated by VFDs whencapacity requirements fluctuate.

108. Apply power-factor correctionwhen possible.

109. Stage the startup of loads tokeep demand down, and install de-mand-limiting equipment.

110. Connect lighting and comfort-conditioning systems to the buildingautomation system to minimize energyuse in unoccupied areas.

111. Replace low-efficiency fixtureswith high-efficiency fixtures (HID andHPS).

112. Install motion sensors anddimmable fluorescent fixtures in ware-houses for maximum energy savings.

W A Y S T O S A V E E N E R G Y

Continued from page 57

NETWORKED CONTROLS

10October 2005

Despite numerous technological ad-vances, digital building controls arewoefully underutilized. Why? Becausethe fundamental control methodolo-

gies they most often are configured to apply are little changed from the early 1900s, the timeautomatic building controls were introduced. As a result, the control applied in many buildingcomfort systems consists primarily of a large number of independent, stand-alone controlloops. As building systems become more inte-grated and complex, the continued use of this control threatens to severely undermine perform-ance, precision, stability, efficiency, and reliability.

Rising energy costs and a widening gap betweenelectricity-generation capacity and electricity demand are driving interest in the networking capabilities ofmodern control systems. Takingadvantage of these capabilitiesand creating tremendous syner-gies between equipment and op-erational requirements is a seriesof new control methodologiestermed “relational”1 because they optimize the operation ofHVAC-system components in“relation” to one another.

This article will examine the broad concept of relational control, focusing on its potential benefits and long-range prospects.

CURRENT CONTROL METHODOLOGIESFor about as long as building controls have

been around, the control model shown in Figure 1has been employed in one form or another inHVAC-system applications to adjust capacity orflow by modulating valves, dampers, vanes, motorspeed, and other devices and variables. Once a very simple analog device, the “controller” modulehas grown in sophistication over the last severaldecades. Today, this PID (proportional, integral,derivative) controller is extremely flexible, incorporating variable-gain and even self-tuning

By THOMAS HARTMAN, PE

The Hartman Co.

Georgetown, Texas

RelationalNetwork-enabled approaches to improving the efficiency and effectiveness of HVAC operation

Feedback signal

Referencesignal

(set point)

Controller

Sensor

Controlleddevice

Controlsignal

Device output(controlled variable)

FIGURE 1. “Closed-loop” control.

capabilities, which ensure deviceoutput remains at or near the reference signal (set point) underwidely varying conditions. Unfortunately, for modernHVAC applications, the controlarchitecture is fundamentallyflawed, most significantly interms of:

• Applications-level control.Figure 1 depicts “closed-loop”control. “Closed loop” meansthe feedback signal links thecontroller and controlled vari-able so that the output signal tothe controlled device is adjustedautomatically to maintain thedesired reference signal (setpoint) as conditions change. Inthe very simple comfort-condi-tioning systems employed a century ago, closed-loop controlcould be provided at the applica-tions level easily. For example, by modulating a steam valve to a simple heating device, with aspace-temperature sensor pro-viding feedback to the con-

troller, a space-temperature set point could be maintainedunder a wide range of load conditions.

In today’s typical HVAC system, closed-loop control at the applications level is not soeasily achieved. For example, in a variable-air-volume (VAV)system, if spaces are not beingsatisfied, there normally is nofeedback to drive down supply-air temperature. This open-loopcharacteristic significantly limitsthe performance of today’s morecomplex HVAC systems.

• Control coordination. Today’sVAV-system controls employtwo major control loops: one for supply-air temperature andanother for duct static pressure.For optimum efficiency andcomfort, changes in airflow and temperature in response to changes in a VAV system’scooling requirements need to becoordinated. Most VAV systems,however, lack the control mech- NETWORKED CONTROLS

11October 2005

anism necessary to coordinatethe temperature and flow of supply air in response to changesin load. Likewise, VAV-box dataseldom are employed effectivelyenough for the air system to beadjusted optimally as coolingload changes.

RELATIONAL CONTROLFOR SYSTEMS SERVINGMULTIPLE PROCESSES

Consider a chilled-water distribution pumping system. If the system were to be controlledtraditionally, a pressure sensorwould be incorporated acrossthe supply and return headers atthe end of the distribution loop,the idea being that as long as differential pressure between theheaders is adequate, every load in the circuit will have sufficientpressure to deliver design flowand, therefore, meet design loadconditions. For simple distribu-tion configurations, this strategyworks. In many configurations,

l Control

NETWORKED CONTROLS

12October 2005

however, branches from mainshave adjacent loads that can experience high concurrentloading, meaning local areasmay experience low differentialpressure and flow starvation under certain conditions—evenwhen the design pressure ismaintained at the sensor loca-tion. At least as problematic isthe fact this method of controlresults in poor pumping effi-ciency at part-load conditions,which constitute nearly all of atypical pumping system’s operat-ing time.

Relational-control techniquesreplace this pressure-based control with more-direct flow-based control. The network collects data on the condition of each load served, which can be condensed easily for much-more-efficient pumping-systemoperation at all load conditions.Because each load is checkedcontinuously, proper servicingcan be ensured. Such a methodof control is shown in Figure 2.Here, pump speed is calculatedat timed intervals based on valve-position (and load) data retrieved from the local loadcontrollers (ASC [application-specific controller] 1 throughASC 5). This multi-inputmethod of control can be muchmore stable than conventionalPID control because it does notinvolve continual readjustmentbased on a set point and errorsignals. Methods of self-correct-ing control, which make adjust-ments at loads, rather than atplants, have been developed.2,3,4

These result in smoother, morestable control.

At the applications level, thetype of relational control de-picted in Figure 2 is closed loopbecause of the direct link be-tween multiple processes (cool-ing loads) and a single controlledresource (a chilled-water distri-bution pump). To a limited extent, this link can be madewith conventional PID con-trollers by using valve-positionfeedback from all of the loads

to “reset” the distribution-pres-sure set point. Because the controlled variable (pressure) isnon-linear with respect to thecontrolled output, however,such an approach often becomes

unstable under certain circum-stances. Not only must gainschange as set point and load conditions change, the accuracyof pressure readings at low-flowconditions often is a problem.

Chilled-watersupply

Pump-speedcontroller

Chilled-waterdistribution

pump

Load

1

ASC1 Lo

ad 2

ASC2 Lo

ad 3

ASC3 Lo

ad 4

ASC4 Lo

ad 5

ASC5

Chilled-waterreturn

FIGURE 2. Closed-loop relational control.

Chillercontroller

Cooling-tower-fancontroller

Condenser-pumpcontroller

Coolingtower

Chilled-watersupply

Pump-speedcontroller

Chilled-waterdistribution

pumpASC

1ASC

2ASC

3ASC

4ASC

5

Chilled-waterreturn

Chiller

Load

1

Load

2

Load

3

Load

4

Load

5

FIGURE 3. Closed-loop relational control.

RELATIONAL CONTROL FOR SYSTEMS WITH MULTIPLE COMPONENTS

Many HVAC systems today employmultiple components to provide desiredoutput. Consider the system in Figure 2.While the chilled-water distributionpump appears to be the only element required to provide chilled water, weknow that is not the case. Figure 3 is amore realistic diagram of the system.

According to the Equal Marginal Per-formance Principle,5 to optimize operat-ing efficiency, every component in acooling system must be adjusted simulta-neously as load conditions change. Although such integrated, multicompo-

nent control is beyond the capacity ofconventional control with independentPID controllers, it can be achieved with a flexible, network-based control system.

To get an idea of how much more effective a cooling system will be with relational control based on the EqualMarginal Performance Principle, con-sider that chilled-water systems generallyincorporate some degree of decouplingto accommodate stand-alone control of their various components and that low delta-T almost always is an issue.When relational control is applied, how-ever, such decoupling is unnecessary, andchiller-capacity-loss problems from lowdelta-T vanish.

In addition to solving problems that have plagued HVAC systems fordecades, relational-control strategiesgreatly enhance the operational effi-ciency of systems with multiple compo-nents and, when coupled with relational-control approaches for systems servingmultiple processes or loads, ensure thateach load in a system is met effectively.

CONCLUSIONThe use of simple stand-alone, single-

process feedback control limits both theeffectiveness and efficiency of modernHVAC systems. The networking capac-ity of digital control systems, on the other

hand, opens our industry to new meth-ods of relational control that offer com-plete closed-loop and fully coordinatedcontrol. These new approaches will significantly improve both the efficiencyand effectiveness of HVAC operationwhile allowing simpler and more stablesystems. In this era of rising costs andlimited electric-energy resources, rela-tional-control methods have an impor-tant role to play in the development ofenergy-efficient HVAC systems.

REFERENCES1) Anderson, R., & Hartman, T.

(2005). Internal technical communica-tions.

2) Hartman, T. (2003, September).Presenting intelligent iterative control:PID replacement for setpoint control (pt. 1). HPAC Engineering, pp. 13-14.

3) Hartman, T. (2003, October). Pre-senting intelligent iterative control: PIDreplacement for setpoint control (pt. 2).HPAC Engineering, pp. 9-10.

4) Hartman, T. (2003, November).Presenting intelligent iterative control:PID replacement for setpoint control (pt. 3). HPAC Engineering, pp. 9-10.

5) Hartman, T. (2005). Designing efficient systems with the equal marginalperformance principle. ASHRAEJournal, 47, 64-70.

ABOUT THE AUTHORPrincipal of The Hartman Co.

(www.hartmanco.com) and a member ofHPAC Engineering’s Editorial AdvisoryBoard, Thomas Hartman, PE, is an internationally recognized expert in thefield of advanced high-performancebuilding-operation strategies. His accomplishments include developmentof one of the first hourly building energysimulation programs and refinement ofan integrated approach to chiller-plantcontrol that reduces commercial-build-ing annual cooling-energy requirementsby more than 50 percent. He can be contacted at tomh@hartmanco.com. NC

Relational control will significantly improve both the efficiency and effectiveness of HVAC operation while allowing simpler and more stable systems.

Circle 00