Minor components trapped in distillation columnscan react to form solids that cause major prob-lems. The conventional responses to this situ-

ation focus upon dealing with the solids afterthey have become formed. A better strategy consists of preventing or minimizing the solids formation in the firstplace, by changing the column operating conditions.

Understanding the situationThe solids-formation problem arises during distillationwhen minor components react with each other under the

process conditions inside the column, forming polymersor other heavy molecules that are not completely solublein the liquid phase in certain sections of the column.The presence of these solids often leads to unpredictable

 vapor-liquid contact, in turn adversely affecting productquality and increasing the plant’s downtime and mainte-nance expenses.

In many cases, engineers approach the problem by aim-ing to modify trays or other column internals to deal with

solids more easily, and/or by injecting additives to eitherinhibit the reactive species or disperse the solids. Not onlyare these responses usually expensive; they also do notaddress the underlying problem. In some cases, taking acloser look at the column operations and exploring waysin which the minor component(s) can be purged out of theprocess will produce better results.

The principles that affect component trapping in distilla-

tion columns has been described in much detail by Kister [1].That reference also summarizes numerous experiences indiagnosing and eliminating such problems at chemical-pro-cess plants. Here, we discuss one such experience in detail.

The processDuring final stages of adiponitrile (ADN) production, a low-boiler stripper (LBS) column removes components having

relatively low boiling points, such as succinonitrile (SN)and acrylonitrile (AN), from the crude ADN stream. Theselow-boilers are taken out in the overhead distillate stream;the bottoms stream, containing mainly product ADN with

traces of low-boiling impurities and other, higher, boilers,goes to further purification. The column operates under

 vacuum to minimize thermal degradation of the product.

The column internals consist of structured-packing beds,above and below the feed point.

Figure 1 describes the basic control scheme for the col-umn. The setpoint for the distillate flowrate is selected bythe operator; the reflux flowrate is set by the liquid level inthe overhead accumulator; steam flow is set by the opera-tor as a function of the column feedrate; the bottoms flow isdetermined by the bottoms level; and the column pressure

control is achieved by vacuum jet ejectors. There is no tem-perature control on this column.

The problemThis column had a history of operational problems, and itsruntime was limited by severe fouling due to polymer for-mation in the packed section above the feed point. Whenclean, the column was able to deliver on-spec product, withminimum product loss in the overhead distillate stream.

But even when the column was essentially clean (namely,soon after its startup following a cleaning), the pressurewould slowly start to build up, and eventually develop intoa process upset that would last several hours. Once thecolumn had “burped” (that is, once the pressure increasehad relieved itself, either by causing the liquid phase torise abnormally or by a cooldown of the column), the op-erating conditions returned to normal and the column ran

well until the next episode occurred.The frequency and severity of these events was predict-

able. Following about six months of operation after columncleaning, these burping episodes would get worse — andbecame accompanied by severe flooding that increased theproduct loss in the distillate stream; finally, the columncould no longer make on-spec products. It had to be shutdown once a year for cleaning and for replacement of the

packing, which resulted in lost production and raised themaintenance costs.

Inspection of the column during shutdowns showed largesections of the structured packing to be completely plugged

CHEMICAL ENGINEERING WWW.CHE.COM MARCH 2006 65

� � �

� � � � �

� � � � � � � � �

� � � � � � � � �

�

�

�

�

�

�

�

�

�

Minimize TrappedComponents inDistillation Columns

Operations & Maintenance

FIGURE 1. Here is the

controlscheme

for thecolumnHere is how one plant remedied a distillation

difficulty by analyzing, then counteracting,

the chemistry that had led to the problemSoundar Ramchandran , Solutia, Inc.

 

with foulant. It was the gradual buildup of the polymer,and the subsequent pluggage of the packing, that causedthe loss in separation efficiency.

 An initial proposed remedyIn an effort to eliminate the problem, a team of experts eval-uated a proposal that would modify the column internals so

as to facilitate easier exit of the solids and reduce the costimposed by the fouling. The team considered various struc-tured-packing configurations, and the design that evolvedrepresented a compromise between (1) achieving the separa-tion required for meeting the product specifications and (2)being able to move the solids out when fouling started.

During this search for a new design, the team had notedthat the version to be replaced had some inherently unde-

sirable features. One in particular was the combination of tall beds with an absence of liquid redistributors; this com-bination posed a strong likelihood of liquid maldistributionand, therefore, the creation of pockets of stagnant liquidthat could form sites for fouling.

Underlying the team’s diagnosis and proposed remedywas a strong belief that the process fouling in the columnwas due to inadequate or less-than-efficient contact betweenthe vapor and liquid phases. So, the rationale for modify-

ing the column internals was that the improvement in thecontact between vapor and liquid would increase the separa-tion efficiency, thereby improve the overall column operationand, hence, lessen the fouling.

 An alternate diagnosisBefore the team’s proposal was acted upon, however, a dif-ferent line of reasoning emerged. As it happened, the in-

ternals of this column had already been modified on morethan one occasion previously; therefore, the expectationthat yet another modification would lead, by some sort of evolutionary optimization, to an optimal solution seemedunrealistic. Furthermore, if liquid maldistribution or

 vapor-bypassing were the real problem, this new line of reasoning went, then the column operation and its inabilityto achieve the desired separation should be affected consis-

tently, not just in the episodic manner observed. Therefore,the outcome from modifying the internals to improve thecontact efficiency and minimize the fouling problem washighly uncertain, and had low probability for success.

FIGURE 2. Shown here are the steady-state vapor-phasecomposition profiles under normal operating conditionsprior to stepping up the distillate flowrate. MS_SN** denotesa proprietary nitrile compound which served in the study asa proxy for succinonitrile

Westfalia SeparatorIndustry

Starch recovery gets really profitable now with the

new Disc-Decanters from Westfalia Separator. Corn

gluten dewatering with Disc-Decanters gives a clear

advantage over systems using rotary vacuum filters:

• drastically reduced upfront costs of up to 60%*

• reduced production costs of up to 60%*.

Westfalia Separator customers also benefit from

higher yields and optimum product quality, reflected

in a fine protein powder, with excellent consistency

and colour. Cut costs, boost profitability – Disc-

Decanters deliver long-term results.

*At a capacity of 1000 t corn/day.

Take the Best – Separate the Rest

Werner-Habig-Straße 159302 Oelde (Germany)

Phone +49 2522 77-0Fax +49 2522 77-2828industry@gea-westfalia.de

www.westfalia-separator.com 3         3         1        0 

  .        1        6 

We’ve cut your costs!

 

Circle 31 on p. 72 or go to adlinks.che.com/5827-31

 

Meanwhile, examination of the structured packing dur-ing one of the maintenance shutdowns showed signaturesof fouling products that were characteristic of nitrile processchemistries used in other process units. Chemical analysisof the foulant confirmed that the fouling in the top sectionwas dominated by cyanide-type polymerization. This typeof polymerization leads to solids resembling coffee grounds;and when exposed to high temperatures, they tend to de-

 velop a black, tarry-looking surface. This polymerizationreaction is autocatalytic — once polymerization is initiated,the rate of reaction rises rapidly. Higher temperatures alsoraise the rates.

Though the column operated under vacuum, the condi-tions within it happened to be quite conducive to chemi-cal reaction. The discovery of the cyanide-type polymerspointed to the presence of a compound that participated,

directly or indirectly, in reactions to form the polymer. Itseemed plausible that the succinonitrile was this com-pound, for reasons given below.

The complications that were assumed to be due to thispresence of SN were consistent with the rationale employedto control the column: the established operating procedureaimed to control the distillate flowrate, in order to minimizethe product loss. Now, SN and ADN happen to be relativelyclose-boiling components, so it was not easy to achieve the

dual goals of separating SN from ADN and minimizing ADNin the distillate stream without also trapping some SN insidethe column. Under the conditions inside the column, SN de-composed into AN and hydrogen cyanide (HCN). The HCN,being the lighter molecule, would tend to become purged viathe overhead product. But in view of HCN’s propensity toreact, operators would find it virtually impossible to stop thepolymerization reaction, once it was initiated,

The only way SN concentration inside the column coulddecrease was during process upset. As the concentration of SN increased and the aforementioned products of SN de-composition (lighter ends) could not leave the column, theoperation became unstable and the pressure rose, eventu-ally leading to a process upset. Once the column had burped,and the operation returned to normal, the process of SN ac-cumulation started again. The situation only became worse

with time, until the polymer growth completely obstructedthe path for vapor and liquid, making it impossible to runthe column and produce on-spec products.

Under this second diagnosis, if the column operating con-

Frankfurt am Main · May 15–19

World Forumof the Process Industries

 www.achema.de

“I trust inChemistryMarkets”

EXHIBITION GROUPSResearch and Innovation ·Engineering, Plant Design andConstruction · Mechanical and

 Thermal Processes · Laboratoryand Analytical Techniques ·Instrumentation, Control andAutomation · Pumps, Compres-sors,ValvesandFittings·Pharma-ceutical Technology · Packagingand Storage Techniques · Mate-rials Technology · Industrial andLabour Safety · Biotechnology ·Environmental ProtectionINTERNATIONAL CON-GRESS  with more than 800lectures and special eventsSPECIAL SHOW InnovativeService Concepts · Chemical Lo-gistics · Supply Chain Manage-ment · Chemical Parks · ContractProduction

Circle 32 on p. 72 or go toadlinks.che.com/5827-32

FIGURE 3. Here is the SN vapor-phase composition profileunder an intermediate distillate flowrate, somwhat greater thanthat corresponding to Figure 2. (See caption to Figure 2 for themeaning of MS_SN**) . . . . .

 

ditions could be modified to provide a path to purge moreSN, a major source of process fouling could be reduced. Sim-ulation studies showed that it was possible to significantlyreduce the SN concentration inside the column by simplyincreasing the overhead distillate flow, thereby reducing therate of fouling and increasing the column run-time

Steady-state simulationsBecause the physical process of a component becomingtrapped and accumulating inside a column is inherentlyan unsteady process, a steady-state analysis of the processwould admittedly seem to be meaningless. In fact, however,a steady-state process simulation can serve as a screeningtool to identify the inherent characteristics and tendenciesof a process system; for example, with respect to the pos-sibility of trapped components.

 Accordingly, such a simulation was made for the processshown in Figure 1. In this simulation, the reboiler heat dutyand overhead distillate flowrate were specified by the user.The column tray efficiencies were adjusted to match thesteady-state process simulation results with the plant oper-ating data and column temperature profile.

Figure 2 shows the resulting vapor-phase compositionprofile for SN and ADN inside the column. The horizontalaxis, corresponding to the tray numbers, represents the

height within the column; the top of the column is at theleft. (The designation MS_SN** identifies a proprietarynitrile compound which served in the study as a proxy forSN.) The simulation results matched well with actual plantoperation, with very small amount of product ADN in thedistillate stream, and small amounts of SN purged via thedistillate and bottoms.

The key point observed was the “bubble” of high SN con-

centration that remained trapped inside the column underthe normal operating conditions, and thus provided a steadysource of foulant species. In Figure 2, the location of thepeak in the SN bubble, slightly above the midpoint of thecolumn, corresponded well with the region of worst foulingin the plant column.

The simulation confirmed that as the distillate flowratewas increased, the SN was purged more thoroughly. The

 vapor-phase composition profile for SN inside the columnshowed that the bubble shifted upwards in the column asthe distillate flow was increased (Figures 3 and 4). Corre-spondingly, the peak SN concentration also dropped from

Games with the Quality of YourCompressed Air or Gas!

Engineers inherit air systems where reliability, integrity

and countless devices determine the fate of productivity.

The wrong moves can result in product rejects or costly

downtime. The right moves keep production, profit and your

career advancing. Since 1946, generations of engineers have

avoided checkmate by insisting upon legendary Pneumatic

Products brand air and gas treatment

systems for their toughest applications

in the nastiest of environments

Pressures to 5,000 psig

Dew Points to -100°F

Specialty Metals

Standard Designs orBuilt to Specifications

Email: pneumatic.products.sales@airtreatment.spx.com Phone: 352-873-5793

©Copyright 2005 SPX Air Treatment. All Rights Reserved.

www.pneumaticproducts-spx.com

 

Pneumatic Products Doesn't Play

because, there's no substitute

Flows to 20,000 scfm•

for the right strategy.

•

•

•

•

Circle 33 on p. 72 or go to adlinks.che.com/5827-33

  CHEMICAL ENGINEERING WWW.CHE.COM MARCH 2006 69

FIGURE 4. . . . . . and here is the SN vapor-phase compositionprofile at a yet higher distillate flowrate. Note that the magni-tude of the peak composition has lessened, and that its loca-tion has moved even closer to the top of the column

 

about 70 wt. % in Figure 2 to about 30% at an intermedi-ate distillate flowrate (Figure 3) and to about 15% at highdistillate flowrate (Figure 4).

Based on these studies made using the steady-state pro-cess model, the engineers recommended that the column beoperated at a higher distillate flow immediately after startup

following a column cleanup. This change required no addi-tional capital expenditure, and it could be implemented im-mediately. Its effectiveness was verified within a short periodof time, and the column did not exhibit the pressure rise andthe burp operation observed in the past.

 Applying the results elsewhere As was the case with this ADN-purification column, symp-

toms of problems due to trapped components often showup in episodes that tend to be more or less repeatable with

 varying frequency. That frequency depends upon a numberof factors, among them the column operating parameters,such as feed flowrates, reflux rates and boilup rates, andthe operating conditions, such as pressure, temperature,and changes to equipment configuration.

In many cases, the real challenge to the engineer lies in

understanding the type and the nature of the componentsthat are actually getting trapped. Unintended and unrecog-nized reaction systems are often silent contributors to suchproblems and can be extremely difficult to diagnose and cor-rect due to the non-stationary process behavior.

One class of reaction systems involves unwanted com-ponents that form because of interaction of the processstreams with additives intended to enhance performance,

such as antifoams, inhibitors and dispersants. Too, changesin the concentration of minor components due to variationsin catalyst performance, reaction selectivity, and other pa-rameters can trigger synthesis reactions.

Symptoms of unwanted reactions become harder to diag-nose if there is added time-delay or transport lag betweenthe reaction section of the plant and the (downstream)purification column where the problem shows up. Goodknowledge of the process chemistry and access to analyti-

cal development tools can provide useful insights into thespecific nature of the problem and may offer ways to reduceits impact or, in some case, completely eliminate it. ■

 Edited by Nicholas P. Chopey

 An earlier version of this article was presented during the DistillationSymposium at AIChE’s spring meeting in Atlanta in 2005.

Operations & Maintenance

References1. Kister, H.Z., Component Trapping in Distillation Towers: Causes, Symp-toms, and Cures, Chem. Eng. Prog., Vol. 100, No. 8, pp. 22-33, August 2004.

 AuthorSoundar Ramchandran is currently the head of pro-cess technology for the Nylon Basic Chemicals Div. of So-lutia, Inc.’s Chocolate Bayou plant in Alvin, Tex. (Solutia,Inc., P.O. Box 711, FM2917, Alvin, TX 77512-9888; Phone:281-228-4248; sbramc@solutia.com). He joined MonsantoCo. at the Chocolate Bayou plant in 1994. He earned aB.S. in chemical engineering from Mangalore Univer-sity, India, in 1987, and an M.S. and a Ph.D. in the samefield from Texas Tech University in 1990 and 1994, re-spectively. During his tenure with Solutia, he has gainedexperience and expertise in process modeling, processdesign, plant startup, process control, reaction engineer-

ing, fluidization, heterogeneous catalysis, computational fluid dynamics, anddistillation operations and other separation processes.

DOUBLE WALLPL

 

 A 

 

S

 

TIC S

 

TOR

 

 AGE TA 

 

NKS

1•800•275•2436 chemtainer.com

 All-in-one Design,Saves Valuable Floor Space.

Elliminates the Need for Containment Pallets.

Sales office open 8AM - 8 PM EST

Shown with optional pump

 

 50 GALLON TANK 

 DESIGNED TO FIT 

THROUGH A 28” 

 DOORWAY/OPENING! 

INDUSTRIES, INC.

Choose From 12Sizes in 50- 6,000Gallon Capacities

FRIATEC-Rheinhütte GmbH & Co. KGP.O.B. 12 05 45, D-65083 Wiesbaden, GermanyPhone / Fax +49 (0)611 604-0 / -328www.rheinhuette.de, info@rheinhuette.de

MORE PUMPS, MORE MATERIALS,MORE CONFIDENCE

100 years of experiencein engineering pumps

over 40 pump designs inmetals, plastics, ceramics

Contact the worldwide pump specialist.FRIATEC-Rheinhütte

THERE IS NO SUBSTITUTE

We look forward to welcoming youACHEMA 2006 · Hall 8.0 · Booth U38-U45

The world of pumps RHEINHÜTTEP U M P S

Circle 35 on p. 72 or go to adlinks.che.com/5827-35

Circle 34 on p. 72 or go to adlinks.che.com/5827-34

70  CHEMICAL ENGINEERING WWW.CHE.COM MARCH 2006