cems in power plant

8/2/2019 CEMS in Power Plant

1/37

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

Continuous emission monitoring

in power stations and combined

heat and power plants

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

Thermal Generation Study Committee......................................................................................................

October 1997

Ref : 02003Ren9771


2/37

The Union of the Electricity Industry EURELECTRIC has been formed through a merger of the two associations

and

The Union of the Electricity Industry - EURELECTRIC, formed as a result of a merger inDecember 1999 of the twin Electricity Industry Associations, UNIPEDE

1and EURELECTRIC

2, is the

sole sector association representing the common interests of the European Electricity Industry and itsworldwide affiliates and associates.

Its mission is to contribute to the development and competitiveness of the Electricity Industry and topromote the role of electricity in the advancement of society.

As a centre of strategic expertise, theUnion of the Electricity Industry - EURELECTRIC willidentify and represent the common interests of its members and assist them in formulating commonsolutions to be implemented and incoordinating and carrying out the necessary actions. To that end itwill also act in liaison with other international associations and organisations, respecting the specificmissions and responsibilities of these organisations.

The Union of the Electricity Industry - EURELECTRIC is also the association of the ElectricityIndustry within the European Union representing it in public affairs, in particular in relation to theinstitutions of the EU and other international organisations, in order to promote the interests of its

members at a political level and to create awareness of its policies.

The reports published by EURELECTRIC are the result of the work of its structure of expertise: theyrepresent one of the most direct methods of circulating knowledge and information throughout thesector, on subjects of common interest.

They are intended for wide circulation both within the electr icity supply industry and outside it.

Please do not hesitate to ask for the latest available printedEURELECTRIC publicationscatalogue (with summaries of EURELECTRIC reports) from:

Union of the Electricity Industry EURELECTRICDocumentation

66 Boulevard de l'ImpratriceBE-1000 BrusselsBELGIUM

Tel: +32 2 515 10 00Fax: +32 2 515 10 10Email: [email protected]

You can also use the EURELECTRIC Internet Web site, which provides the followinginformation:

- EURELECTRIC general information

- EURELECTRIC positions and statements

- Events & Conferences

- Publications Catalogue

http://www.eurelectric.org

1

International Union of Producers and Distributors of Electrical Energy2 European Grouping of Electricity Undertakings


3/37

TABLE OF CONTENTS

EXECUTIVE SUMMARY ..................................................................................................1

ACRONYMS........................................................................................................................ 4

ACKNOWLEDGEMENT ...................................................................................................4

1. INTRODUCTION............................................................................................................51.1 BACKGROUND...................................................................................................................51.2 OBJECTIVES ......................................................................................................................51.3 IMPLEMENTATION .............................................................................................................52. REASONS FOR CONTINUOUS EMISSION MONITORING.....................................6

2.1 DEVELOPMENTS IN MONITORING WORK BY UTILITIES ........................................................73. TECHNICAL ASPECTS OF CONTINUOUS EMISSION MONITORING ................ 9

3.1 EXTRACTIVE CEM............................................................................................................93.2 IN-SITU CEM..................................................................................................................103.3 PERFORMANCE CRITERIA ................................................................................................113.4 DETERMINATION OF MASS FLOW MEASUREMENT AND PEM............................................143.5 REPORTING .....................................................................................................................144. EXPERIENCE WITH CEM..........................................................................................17

4.1 EXTRACTIVE CEM..........................................................................................................174.2 IN-SITU CEM..................................................................................................................194.3 ACCURACY .....................................................................................................................214.3.1 Uncertainties of the Measurement Systems ..................................................................214.3.2 Experience with accuracy reported in the different countries....................................... 22

4.4 REPORTED EXPERIENCE OF A GENERAL NATURE..............................................................254.5 INVESTMENT AND OPERATING COSTS ..............................................................................265. GENERAL DISCUSSION .............................................................................................27

LIST OF TABLES

page

TABLE 1 National and regional criteria for CEM ............................................................ 4

TABLE 2 Extractive sampling principles reported for the different flue-gas components in

each country.................................................................................................. 13

TABLE 3 In-situ principles reported for the different flue-gas components in each country14


4/37

02003Ren9771 October 19971

EXECUTIVE SUMMARY

The necessity for Continuous Emission Monitoring (CEM) for process and emission controlof combustion plants is increasing in line with the increased automation of new plants andadditional regulatory requirements. This report presents experience and trend information

from utilities regarding continuous emission monitoring (CEM) at power and Combined Heatand Powr (CHP) production units in Europe. Topics covered include; reasons forinstallations, experience and future expectations of the quality, performance, long-termavailability and costs of CEM installations. The following components are covered; NOx,SO2, CO, dust, "secondary" flue-gas components (mainly NH3) and the continuousmonitoring of flue-gas velocity. Most of the experience comes from those utilitiesrepresented in the UNIPEDE-THERNOX group. Other utilities in these countries maytherefore have differing experiences. It has also been difficult to collect all the relevantexperience for some of the countries.

The most important trend apparent is the increasing requirement for CEM measurements,driven particularly by new requirements of the regulatory authorities. The utilities haveperformed many investigations into CEM and its advantages. CEM is consequently used notonly for control of the environmental performance of a plant but also (particularly in the caseof NOx, CO and dust) for optimisation and control of the operation of the plant. Existing andnew plants continue to be fitted with emission control measures such as low-NO x burners anddeNOx. An increasing number of CEM installations are therefore required, not only formeasurement of NOx for deNOx process control, but also for measurement of NH3 slip whenusing SNCR. CEM is also widely used in Sweden for the determination of fees and taxes forNOx and SO2 emissions respectively.

Continuous emission monitoring can be done with either extractive or in-situ systems. Theuse of extractive CEM to date has greatly exceeded that of in-situ CEM. A trend towardsgreater use of in-situ systems is now becoming apparent on the market. The experiencereviewed in this study shows no definite trends for choice of extractive or in-situ CEM, bothhave been applied successfully in many countries.

The sampling principle of most of the extractive CEM systems is based on extraction of fluegas with condensation of water vapour - all countries in this study have reported experienceof this sampling principle. There are, however, also some countries that have reportedexperience of extractive sampling systems with dilution and/or heated extractive systems.

The reported experience with CEM based on extraction with condensation is better than thatof CEM based on extraction with dilution or heated extractive CEM.

The most common in-situ CEM method is single point measurement with a probe, but manyin-situ CEM systems measure across the duct. Most countries in this study have experiencewith both types of in-situ CEM.

The most common analytic principles for CEM are IR/UV-absorption. Extractive basedCEM mostly uses NDIR or NDUV while the in-situ systems also use differential opticalabsorption spectroscopy (DOAS) or IR/UV-absorption. Electrochemical cell-based methodscould be used for both extractive and in-situ systems. In-situ CEM systems for the

measurement of dust are mostly based on opacity. A common practice in extractive CEM for


5/37

02003Ren9771 October 19972

NOx is the use of converters for the reduction of NO2 followed by analysis for NO todetermine total (NO + NO2).Manufacturers of extractive CEM in Europe reported in this study include; Hartmann &Braun, ADC, Siemens, Rosemount, Bodenseewerk and Maihak. ADC produces CEM basedon extraction with dilution while the remaining manufacturers mainly produce extractive

CEM with condensation. Bodenseewerk also produce heated extractive CEM.

Manufacturers of in-situ CEM in Europe reported in this study include; Lear Siegler, SICK,OPSIS, Codel, Land and Durag. SICK, Codel and OPSIS mostly measure across the ductwhile Lear Siegler and Land use in-situ probes.

The reported experience (accuracy, performance, long term availability, operating costs etc.)with extractive and in-situ CEM systems show no major differences. The main advantageswith in-situ CEM are the removal of the need for sample conditioning and the fact that it ispossible to measure components, such as NH3 and NO2, that otherwise are partly or totallyremoved during sample conditioning. A number of plants are equipped with SNCR-systemsand accurate measurement of the NH3 slip is therefore useful. Measurement of NH3 with in-situ CEM can easily be combined with measurement of NOx and/or SO2. A disadvantage ofmost in-situ systems is the problem of calibration on a regular and routine basis. ExtractiveCEM, on the other hand, is very easy to calibrate.

It takes time and due care to apply CEM in daily operation with adequate accuracy. Ingeneral there have been few problems with extractive CEM. Almost all difficulties reportedrelate to the sampling and calibration systems. Condensate management requires carefulattention if problems are to be avoided. Maintenance requirements can be high unless thesystem is carefully designed with e.g. provision for back purging of probe filters.

Although low levels of maintenance are reported concerning in-situ CEM it is clear that thereis a need for a minimum level of essential maintenance e.g. cleaning of windows and checkingof optical path alignment.

It is not possible from the survey to recognise any clear trend regarding the performance andavailability of CEM systems. It is clear from many responses that some calibration systemspose problems and require further attention. A common difficulty appears to be the qualityand reliability of calibration gas standards. The accuracy of gaseous emission concentrationsin flue gas is expected to be high if calibrations are performed carefully and with certified

calibration gases or calibration cells. One reported experience from an accredited laboratoryis that the deviation between a stationary CEM and a mobile extractive CEM in most cases isbelow 4% of the measured values. The accuracy of dust CEM, especially at lowconcentrations, is low though. The accuracy of yearly emission depends also on uncertaintiesin fuel flow and mass flow calculation. A study on three coal fired power plants reports +/- 5to 11 % uncertainty for NOx-emission and +/- 25 to 30 % for SO2-emission at lowconcentrations after FGD (mainly due to concentration gradients in flue gas ducts and singlepoint measurement).

In most countries measurements are made at a single location (extractive CEM or single pointin-situ probes) which is often determined by grid testing. Most analysers are provided with

systems for automatic calibration. There are also purge air systems to clean the measurementwindows (in-situ CEM). In most of the countries, performance criteria for CEM have been


6/37

02003Ren9771 October 19973

established. Only in Germany (and anticipated in Ireland) does a standard approval systemfor the CEM instruments exist.

The mass flow are mostly calculated from the measured concentrations in conjunction withthe fuel consumption, the excess air level and the fuel analysis and seldom from the measured

concentrations and the measured flue gas velocity.

In the future it is possible that other flue gas components will be measured more widely andother analytic technologies used. Typical examples of such components are NH3 and N2O forprocess optimisation. FTIR and diode laser are example of new measurement techniquescoming into use. It is also very possible that existing analysers will be developed further andtheir availability and performance improved the electrochemical analysis principle is anexample of such an technique.

In some countries Predictive Emission Monitoring (PEM) has been tested for the

determination of NOx emissions. A PEM system can be verified by multivariate data analysisor based on predictions by neural network. The plant must be fired with fuels of constantquality if a PEM system is to be used.


7/37

02003Ren9771 October 19974

ACRONYMS

AFNOR - French Standards Organisation.BEES - Besluit Emissie Eisen Stookinstallaties: Ordinance on Emission Limit Values forCombustion Installations in the Netherlands.BS - British Standard

CEM - Continuous Emission Monitoring.Chemiluminescense - detection based on the luminescense reaction of NO with O3.DOAS - Differential Optical Absorption Spectroscopy.EPA - Environmental Protection Agency.FGD - Flue-gas desulphurisation (deSOx).FTIR - Fourier Transformed Infra-Red.Extractive CEM - The analysis takes place outside the flue gas duct.HMIP - Pollution Inspectorate (United Kingdom)- now part of the UK EnvironmentAgency.In-situ CEM - The analysis takes place inside the flue gas duct.

IR-absorption - Infra-Red absorption. The analysis principle is based on the transmission ofcharacteristic wavelengths in the infra-red spectrum.LCPD - Large Combustion Plant Directive of the EC (88/609/EEC).NDIR - Non Dispersive Infra-Red.NDUV - Non Dispersive Ultra-Violet.PDA - Photo Diode Array.PEM - Predictive Emission Monitoring or (Parameter Monitoring).PEN - National Energy Plan (Spain).SCR - Selective Catalytic Reduction (deNOx).SEPA - Swedish Environmental Protection Agency.SNCR - Selective Non-Catalytic Reduction (deNOx).TA-luft - Technical rules that prescribe the required conditions for CEM in Germany.TV - Technischer berwachungs-Verein, Organisation of Technical Inspection Agencies inGermany. The required conditions for CEM are defined by TA-luft and verified by (TV)test laboratories.USEPA - US Environmental Protection Agency.UV-absorption - Ultra-Violet absorption. The analysis principle is based on thetransmission of characteristic wavelengths in the ultra-violet spectrum.

ACKNOWLEDGEMENT

In addition to the authors, the following persons have made major contribution to the report;

Mr. Brendan BARRY - ESB (Electricity Supply Board)Mr. Roland CSADER - Badenwerk AGMr. Mark ELSWORTH - National Power PLCMr. Daniel MAILLIET - EDF (Electricite de France)Mr. Henk SPOELSTRA - KEMA


8/37

02003Ren9771 October 19975

1. INTRODUCTION

1.1 Background

In 1993 the UNIPEDE THERNOX report on NOx control Technologies was published. Inthis report a short statement was made about NOx measurement technologies.

Widespread use of NOx reduction measures as well as a requirement to meet emission limit

values means that NOx measurement is required under all operating conditions.

Only robust measuring instruments are suitable for power plant operation conditions.

Factors such as high mass flows, large flue gas channel cross sections, particulate

containing flue gas and relatively low NOx and SOx concentrations which are all

characteristic of power station emissions, pose difficulties for representative emissionmeasurements.

In recent times much work has been directed at improving measurement facilities.

In 1994 the Thermal Generating Study Committee charged THERNOX to report in moredetail on CEM in power stations and CHP plants. After that a working group wasestablished to report on the European situation regarding CEM. Decision was made tocollect the information by means of an inventory on experiences of CEM among theTHERNOX members.

1.2 Objectives

The objectives of this study were;

- To document the reasons for CEM installations in Europe; are they installed to meet therequirements of the regulatory authorities?, or for process monitoring and control?, or forboth purposes?

- To follow-up reported studies and to survey current CEM studies in Europe.- To follow-up experience and opinions about the quality, performance, long-term

availability and costs of CEM installations.

1.3 Implementation

To collect (and document) the necessary information a questionnaire was sent to theTHERNOX Group- members. The questionnaires were filled in by all members in theTHERNOX group and were evaluated and processed by the working group.

It should be noted that the results from this study are not fully comprehensive for allcountries. Practice and experience is reported mainly from those utilities represented in theUNIPEDE-THERNOX group.


9/37

02003Ren9771 October 19976

2. REASONS FOR CONTINUOUS EMISSION MONITORING

Continuous monitoring of emissions is required principally to meet the requirements of theregulatory authorities. In the EC the LCPD 88/609/EEC prescribes the minimumrequirements; -continuous monitoring of NOx, SO2, O2 and dust in new plants > 300 MWth,

(with a possibility to exclude SO2 and dust from the CEM demands of the authorities), -determination of annual emissions for SO2 and NOx in new plants and -emission inventory ofSO2 and NOx in existing plants. Appendix A contains the text of article 15 and annex 9 of theLCPD which describe CEM. Also, many countries have had CEM requirements before theadoption of the LCPD. Monitoring of other flue gas components is required in somecountries.

There can be variations at national and regional levels, mostly depending on the thermalrating of the plant. Typical requirements for continuous monitoring are listed in table 1.

TABLE I National and regional criteria for CEM

Country Component Criteria

Austria NOx, SO2 > 30 MWthCO, dust > 10 MWth

Belgium NOx, SO2, dust > 300 MWthFrance NOx, SO2, CO > 150 kg/h

dust > 5 kg/hHCl > 20 kg/h

Germany NOx, SO2, CO, dust yes

Ireland NOx, SO2, dust there will be requirements within 2 yearsItaly NOx, SO2 national law (>300 MWth), local

authoritiesCO national and regional limitsdust > 300 MWth, local authorities

Netherlands

NOx, SO2, dust > 300 MWth

Poland NOx, SO2, dust emission limitsCO authority requirements

Portugal NOx 30 kg/h

SO2 50 kg/hCO 100 kg/hDust -

Spain NOx, SO2, dust large combustion plantsSweden NOx all plants > 50 MW which produce more

than 50 GWh/year, local authorities, feesSO2 all peat and coal fired plants > 50 MW

which produce more than 50 GWh/year,taxes

Hg, NH3, N2O, dust some local limits

Switzerland

NOx, SO2, CO, dust national and cantonal (regional) authorities


10/37

02003Ren9771 October 19977

Country Component Criteria

UnitedKingdom

NOx, SO2 new plants > 50 MWth, existing plants(>50 MWth) within 3 years , local limitsalso exists

CO some new plant >50 MWth

Dust -

Secondary reasons for CEM include process monitoring and control - e.g. monitoring ofNOx, CO and dust for combustion control, of SO2 for control of FGD and of NOx to controldeNOx processes. NH3 may also be measured (as in Sweden and United Kingdom) to controlNH3-emissions from deNOx processes and conditioning of dust control devices respectively.Measurement of SO2 to control deSOx processes is required in some countries - in theNetherlands the efficiency of an FGD must be better than 85 % and be controlled by CEM.Velocity may be measured for the calculation of mass flow, it is possible to direct measurethe velocity but that is normally not preferred. CEM is also used for combustion

modifications or other primary NOx control measures.

In Sweden there is a fee-system for emissions of NOx. Plants with energy production greaterthan 25 GWh/year are included and must quantify their emissions of NO x either by CEM orby estimation using an officially-defined fixed high level. The principle of the fee-system isthat plants which qualify must report their emissions of NOx (as NO2) and their energyproduction to SEPA (the Swedish Environmental Protection Agency) each year. For everykg of NOx emitted the plant must pay SEK 40 (about 4). The total payments made by allplants are refunded. The refund to each plant is based on the ratio of its total energyproduction to the total energy production of all the qualifying plants. Plants with lowemissions and high efficiency therefore make money at the expense of plants with highemissions and low efficiency. In Sweden there are also taxes on sulphur (SEK 30/kg S).This tax is based on the sulphur content in the fuel. A plant emitting less sulphur than thatcontained in the fuel receives a refund. CEM must be used for the determination of thisrefund.

2.1 Developments in Monitoring Work by Utilities

the Netherlands;

(NOx and SO2); Multi-component analysers such as FTIR (Fourier Transform Infra Red) andPDA (Photo Diode Array) are in use. There is a gradual improvement of emission

monitoring through improvements in the quality of the system in all its aspects and throughan increase in the number of CEM-systems. Studies are being made on the precision andaccuracy of the reported data, both for each installation and for the aggregate emissions of allpower stations.

Poland;

In Poland there is a need for a rapid reduction in emissions from utility boilers. An increasingnumber of CEMs are being installed for process control.

Portugal;

(NOx); Measures include research and development to assess the influence of typicalmeasures for NOx abatement and to establish engineering relationships between operatingvariables influencing NOx emissions and global boiler performance in terms of energy


11/37

02003Ren9771 October 19978

efficiency, fouling, slagging, safety and reliability. Several engineering tools are underdevelopment for the prediction of NOx emissions and boiler performance (as a function ofoperating conditions and fuel type) and these are being assessed at present.

Sweden;

(SO2, CO, NOx, NH3); Gradual improvements are planned such as more cost-effective CEM,"stripped" versions of established techniques and CEM based on analysis by electrochemicalcells. Predictive emission monitoring (determination of the NOx-emission from other processparameters; e.g. combustion temperature, oxygen concentration, fuel flow etc.) for oil andgas burners is being studied. Recently, diode lasers have been used for NH3 measurement. Inthe future; testing of continuous measurements by use of FTIR (hydrocarbons) will takeplace.

Switzerland;

(NOx); Installation of CEM to study all parameters of the NOx-emissions and to optimise theemissions through primary measures.

(SO2); Oil with very low sulphur content will be used. CEM is not applied to oil fired boilerbecause the fuel analysis gives a better accuracy.

United Kingdom;

(SO2, NOx); Independent reviews by power producers are planned on the relative accuracyof predictive calculation methods versus direct measurement. Cost effective instrumentationis currently being reviewed.

(NOx); A watching brief will be maintained on developments in predictive emission

monitoring e.g. neural networks etc.

(All components); More cost-effective CEM is being sought this will be done throughreviews of new instruments as and when they appear on the market.


12/37

02003Ren9771 October 19979

3. TECHNICAL ASPECTS OF CONTINUOUS EMISSION MONITORING

Continuous emission monitoring can be done with either extractive or in-situ systems. Todate the use of extractive CEM for gaseous emissions has greatly exceeded that of in-situCEM. A trend towards greater use of in-situ systems is now becoming apparent in the

marketplace.

To reduce uncertainties in the flue gas because of concentration profiles there areinternational standards for location of representative sampling point(s). Standard rules orguidelines for the location of sampling points for CEM have been reported from fourcountries (France, Poland, Sweden and United Kingdom). In most countries measurementsare made at locations where the flue gas inhomogeneity is low at a single location (extractiveCEM or single point in-situ probes) which is often determined by grid testing. In all countriesthe measurement point for emission control is located either at the stack or in the gas ductleading to it. For process control there can also be measurements before and after the de-

NOx/de-SOx process. A number of manufactures in Europe are listed in appeendix B.

3.1 Extractive CEM

The sampling principle of most of the extractive CEM systems is based on extraction of fluegas with condensation of water vapour all countries surveyed in this study have experienceof this sampling principle. Experience of extractive sampling systems with dilution has alsobeen reported by Austria, France, Poland, Spain, Sweden, Switzerland and the UnitedKingdom. Heated extractive systems are used for sampling condensable components such asNH3 and HCl. Experience of such systems has been reported by France, the Netherlands,

Sweden and United Kingdom. Table II contains details of the flue gas constituentsmonitored and the sampling principles reported for each country.

The most common analytic principles used are NDIR or NDUV-absorption.Chemiluminescence is also used for analysis of NOx. Use of converters for the reduction ofNO2 followed by analysis for NO to determine (NO + NO2) is also a common principle.Analysis by electrochemical cells has been done in the United Kingdom and Sweden. Useof UV fluorescence for the determination of the SO2-concentration has been reported fromPoland, Sweden and United Kingdom. Dust has been measured by absorption gauges inFrance and Sweden.

Equipment manufacturers that were mentioned in the answers include Hartmann & Braun,ADC, Emission S.A., Siemens, Rosemount, Bodenseewerk and Maihak. The operatingprinciple of most of those analysers is based on absorption in the IR or UV spectrum ofsamples extracted with condensation. ADC, however, operates by extraction with dilutionand uses Chemiluminescence and UV fluorescence for analysis of NOx and SO2 respectively.Heated extractive analysers, based on 2nd derivative IR-spectroscopy combined with gasfilter correlation, are produced by Bodenseewerk.


13/37

02003Ren9771 October 199710

TABLE II Extractive sampling principles reported for the different flue-gas

components in each country

Country Sampling principles reported Components

Austria extractive with condensation NOx, SO2, COheated extractive NH3extractive with dilution "

Belgium extractive with condensation* NOx, SO2, COFrance extractive with condensation NOx, SO2, N2O

extractive with dilution NOx, SO2heated extractive HCl, dust

Germany extractive with condensation NOx, SO2, COIreland extractive with condensation COItaly extractive with condensation NOx, SO2, CONetherlands extractive with condensation NOx

heated extractive SO2, NOxPoland extractive with condensation NOx, SO2, CO

extractive with dilution "Portugal extractive with condensation NOx, CO

heated extractive SO2Spain extractive with condensation NOx, SO2

extractive with dilution "Sweden extractive with condensation NOx, SO2, CO, N2O

extractive with dilution NOx, SO2, COheated extractive NH3, HCl, dust, N2O,

NOx, COSwitzerland extractive with condensation NOx, CO

extractive with dilution NOx, SO2United Kingdom extractive with condensation, NOx, SO2, CO

extractive with dilution NOx, SO2, COheated extractive NOx, SO2, CO

*Gas turbines only

3.2 In-situ CEM

The principle of most in-situ CEM systems is single point measurement with a probe. Themost common example of an in-situ CEM, the ZrO2-cell which is used for the determinationof Oxygen, is not discussed in this report. There are also in-situ CEM systems whichmeasure across the duct and which are therefore less sensitive for siting. Most countriesstudied have experience with both types of in-situ CEM. All flue gas components discussedin this report except N2O have been measured by in-situ CEM techniques.

The analytic principles for in-situ CEM are IR/UV-absorption or electrochemical cell-basedmethods. In-situ CEM systems based on opacity are used for the measurement of dust.There is also, normally, a purge air supply to clean the sample windows.


14/37

02003Ren9771 October 199711

TABLE III IN-SITU PRINCIPLES REPORTED FOR THE DIFFERENT FLUE-

GAS COMPONENTS IN EACH COUNTRY

Country Monitoring principles reported Components

Austria opacity, light absorption dustDOAS (UV) SO2, NOx, NH3, H2O

Belgium 2nd derivative UV-spectroscopy NO, SO2IR COopacity, light absorption dust

France DOAS (UV) NOx, SO2Germany - -Ireland electrochemical cells NOx

IR COItaly opacity, light absorption dustNetherlands UV, 2nd derivative UV-spectroscopy NO, SO2

opacity, light absorption dust

Poland DOAS, UV, 2nd derivative UV-spectroscopy

NOx, SO2

Portugal UV, electrochemical cells NO, SO2Spain 2nd derivative spectroscopy (IR/UV) NO, SO2Sweden UV NO, SO2, NH3

DOAS (UV) NOx, SO2, NH3, Hg, H2O,HCl

NDIR CO2nd derivative UV-spectroscopy NO, SO2opacity, light absorption dust

diode laser NH3Switzerland IR COUnitedKingdom

UV NO, SO2

NDIR, IR NO, COelectrochemical cells NOx, SO2

The most commonly reported manufacturers of in-situ CEM are Lear Siegler, SICK, OPSIS,Codel, Land and Durag. SICK, Codel and OPSIS mostly measure across the duct while LearSiegler and Land are based on measurement by in-situ probes.

3.3 Performance Criteria

Performance criteria or guidelines for CEM systems have been established in most countriesreported in this survey. Only Germany (anticipated in Ireland) has official approvalprocedures for CEM instruments, only approved types of instruments can be used.Switzerland does not have an official approval procedure as such, but in some cantons(regions), the measurement instruments must be of a type that is officially approved by theGerman TV. In Poland the General Environmental Inspectorate have establishedguidelines for the design of CEM systems.


15/37

02003Ren9771 October 199712

Practice in the USA can be mentioned in this chapter. Germany only approves the specificcomponents in the monitoring system. In contrast, the plant owner in the USA mustimplement a QC/QA-program for the entire measurement system. This approval includesvery detailed tests. Although it may appear that the systems are very different, there aredevelopments on EMAS and QA that tend to go in the same direction in Europe.

Austria;

NORM M9410, 9411. The extraction point is fixed by an authorised institute duringcommissioning ; The availability of data per month must be > 90 %; Zero point and spancalibrations must be done at least once per week; A "technical service" is required everythree months. The equipment and criteria are checked every year by the authorised institute;An independent system calibration is done by this institute every 3 years.

France;

(NOx and SO2); Measurements must be carried out according to AFNOR standards (FrenchStandards organisation) which includes sampling and measurement methods. The licenseissued by the local authority contains general requirements on regular checking and requiresan annual test to be carried out by an accredited independent test laboratory.(Dust); The French authorities require CEM for plant emitting more than 5 kg/h. If theemission exceed 50 kg/h periodic measurements by gravimetric techniques are required also.

Germany;

(SO2, NOx and dust); TA-luft prescribes the required conditions. TV inspection verifiesthis.

Ireland;

(NOx and SO2); The Irish EPA have expressed a preference for CEM which meets TA-luftor which complies with USEPA requirements.

Italy;

(All flue gas components); Precision, rise time, fall time, response time, detection limit,sensitivity, repeatability, linearity, zero drift, span drift, immunity to interference are some ofthe instrument characteristics influencing selection in the purchasing process.

The Netherlands;

(SO2, NOx and dust); Performance criteria are set according to the "Regulations of

Measuring Methods of the BEES". CEM should be used in accordance with generallyaccepted measuring practice.

Poland;

In Poland there is an official issue of guidelines and technical requirements for CEM. Ifpower plants install CEM according to these requirements, results of the monitoring areaccepted by the Regional Environmental Inspectorate. At other plants the RegionalEnvironmental Inspectorate can perform spot emission measurements. If emissions forcertain components exceed upper allowable limits, the power plant must pay a penalty untilmeasurements show that the emissions are below the legal limit.(NOx and SO2); Detection limit = 2 % of measurement range; Zero point drift (between two

calibrations) = 2 % of measurement range; Effect of interferences = 4 % of measurement


16/37

02003Ren9771 October 199713

range or automatic compensation for interfering components; Response time = 240 seconds;The measurement range should be 150 % of average measured value.(NOx and SO2); Analysis should be possible with correction of results for the measuredoxygen concentration and (sometimes) the content of water vapour in the flue gas - sinceoxygen and water vapour determine the dilution level in flue gas.

(CO); Detection limit = 0,2 %; Effect of interferences = 0,5 %; Zero drift = 0,5 %;Temperature affect = 0,2 %/10 oC.(Gas velocity); Error +/- 2 % of range (Zero and test point drift within 2 % of limit).

Portugal;

(NOx, SO2 and CO); The measurement equipment must be submitted to periodic calibrationaccording to the relevant legislation. The span and calibration gases should be certified anddiffer at most by 2 % from the stated value.

Spain;(NOx, SO2); There are standard requirements for the installation of CEM equipment. Allequipment is installed at the same location where the flow is homogeneous.

Sweden;

(NOx, SO2 and O2 /CO2 ); Criteria for CEM for the determination of NOx fees and sulphurtaxes; Detection limit 2 % of measurement range; Zero point drift (between twocalibrations) 2 % of measurement range; Span drift (between two calibrations) 4 % ofmeasurement range; Effect of interferences 4 % of measurement range; Response time(the whole measurement system) 200 seconds.The measurement ranges should be about twice the highest expected concentrations in theflue gas. The span and calibration gases should be certified and differ by a maximum of 2%from the stated value.(Velocity); Criteria for CEM for measurement of velocity where this is used for thecalculation of the mass emission of NOx and sulphur; Detection limit 10 % of measurementrange; Zero point drift (between two calibrations) 2 % of measurement range; Span drift(between two calibrations) 4 % of measurement range; Response time (the wholemeasurement system) 30 seconds.(NOx, SO2 and CO2, O2 or velocity); The whole monitoring system shall be verified by anaccredited laboratory once a year (by parallel measurement). There are then two othercriteria; The standard deviation between the utilities CEM and the accredited laboratorys

CEM result or measured velocity data must be 5 % of measurement range; Systematicerror between the two systems must be 2 % of measurement range.

Switzerland;

(All components); Use of the best measurement technology is required; the Swiss authoritiesset the specification for official measurements; the performance of CEM systems must beaccepted by the Cantonal authorities also; the instruments must generally be approved by theGerman TV.

(NOx); Measurement according to VDI 2456; overall NOx uncertainty +/- 10 %.

(SO2); Measurement according to VDI 2462.(CO); Measurement according to VDI 2459; overall CO uncertainty +/- 10 %.


17/37

02003Ren9771 October 199714

3.4 Determination of Mass Flow Measurement and PEM

The dominant method for determination of mass flow is stoichiometric calculation on thebasis of fuel composition, flue gas O2 content and fuel consumption. This applies in mostcountries. Use of direct measurement of flue gas velocity is normally not used.

Predictive emission monitoring (PEM) for the determination of NOx emissions has beentested in some countries. This has involved prediction by neural network or multivariate dataanalysis.

Italy;

(NOx and SO2); Periodic measurements have been made on thermal units < 300 MWth. Massflow rate is calculated stoichiometrically on the basis of fuel consumption and O 2 content influe gas.

Netherlands;

It is permissible to determine NOx emissions by the continuous measurement of one or morecharacteristic parameters. A relationship is derived between these characteristic parametersand NOx emissions which is verified every 3 years. The statement is only true when a simplerelationship exists between NOx-emissions and load e.g. in natural gas-fired power stationswith constant gas quality but never in coal-fired power stations.

Poland;

In those cases where there is no direct measurement of the flue gas mass flow, the mass flowrate is calculated stoichiometrically on the basis of fuel consumption and the O 2 content ofthe flue gas.

Portugal;

Assessment is being done on several engineering tools which are under development for theprediction of NOx emissions and boiler performance as a function of operating conditions andfuel type.

Sweden;

The principle of predictive emission monitoring (PEM) is accepted by SEPA fordetermination of NOx-emissions. The first applications have been for two biofuel-firedboilers in two separate saw-mills during 1996. The performance criteria for PEM must besimilar to the performance criteria for a CEM (chapter 5.3). Mass flow rate is calculated

stoichiometrically on the basis of fuel composition and flue gas O2 content.

Switzerland;

The principle of predictive emission monitoring (PEM) is generally accepted by theauthorities (for gas and oil combustion). The NOx emissions can be determined by thecontinuous measurement of characteristic parameters. The accuracy must be similar to directmeasurements.

United Kingdom;

A watching brief is being kept on developments in PEM e.g. neural networks.

3.5 Reporting


18/37

02003Ren9771 October 199715

The EC LCPD stipulate reporting of emission from new (large) plants. LCPD describesmonitoring of concentrations of SO2, NOx and dust and the statistical procedures in order toassess compliance with ELV. In addition national or regional authorities may formulateobligations for existing plants, smaller plants and for other pollutants. Also, reports may berequired for other purposes, such as emission fee-systems.

The mass flow are mostly calculated from the measured concentrations in conjunction withfuel consumption, the excess air level and the fuel analysis and seldom from the measuredconcentrations and the measured flue gas velocity.

Some of the additional national and regional limits can be simply concentration limits (ppm ormg/m3n) rather than mass emission limits. In such a case it is necessary to correctmeasurements to a fixed level of O2 or CO2 if emissions are to be made comparable.Otherwise it is not possible to take dilution by excess air or moisture into account.

Some examples of reporting to the authorities are as follows;

Austria;

(NOx, SO2, dust and CO); At least one set of data are measured per minute, mean values arecalculated for every half hour and daily mean values are also calculated. A yearly report ismade to the regulatory authorities; part 1 covers emission mass flows/month and part 2covers compliance with the legal limits: Legal requirements are met if no daily value > limit,if less than 3 % of half hour values > 120 % of the limit and if no half hour value > 200 % oflimit.

Belgium;

(SO2, NOx and dust); Collection of measurement data is done every 15 seconds, the NOxconcentrations are calculated, standardised (for O2) and stored every half-hour. Daily meanvalues are calculated from the half-hour mean values. Daily, monthly and annual reports areavailable. The mass flows and a classification of the half-hour mean values (frequencies > 1,2and > 2 times the limit values) are detailed in these reports. An annual report is sent to theauthorities.

France;

(NOx, SO2, dust and HCl); Collection of measurement data is done every 10 seconds andminute-average figures are calculated as basic data. An average concentration is calculated

every hour from calculated mass flows. Daily and monthly mass flows, exceedances, averageand maximum concentrations are reported monthly to the regional authority. An annualreport is sent to the Environment Ministry.

Germany;

(NOx, SO2 and CO); An annual report is made.

Italy;

(NOx, SO2 and dust); The monthly arithmetic mean, summary statistics of the 48 hourlyaveraged values over the year and the total annual mass emission must be reported on aperiodic report (usually yearly).

The Netherlands;


19/37

02003Ren9771 October 199716

(NOx, SO2 and dust); Reporting is done for both existing and new installations, annually ormore frequent, on an individual basis.

Poland;

(SO2, NOx and CO); Half hourly data must be available to the authorities. If limits are

exceeded a report is made on a special form.

Portugal;

(NOx, SO2 and CO); The measurements of emissions are sent to the authorities every threemonths in the case of CEM, and within 30 days for any spot measurements. RegardingCEM, the statistical parameters are determined with a minimum of 75 % of possible values,measured in normal conditions during a reference year (1 April - 31 March). The sameapproach is applied to the calendar year i.e. the period between 1 January and 31 December.

Spain;

(NOx

and SO2); Half-hourly data is reported to the national authorities in a standard format.

Sweden;

(NOx ); Reporting of NOx for determination of fees; Hourly mass flows are calculated.These mass flows are summated to produce daily mass flows and the daily masss flow areused to generate annual mass flows. The annual mass flows are reported to the SwedishEPA.(SO2); Reporting of SO2 for determination of taxes; Hourly mass flows are calculated.These mass flows (sulphur) are summated to produce daily mass flows and the daily massflows are summated to produce annual mass flows. The annual mass flows are declared tothe taxation authority.

(All components); There are also requirements from the regional authorities, generallyinvolving annual reporting. These requirements refer mostly to specific emission rates(mg/MJ) but in some cases reporting is required of emission concentrations normalised to afixed concentration of O2 or CO2. Continuous monitoring of emissions is not necessary forthese purposes.

Switzerland;

(NOx, SO2 and CO); The cantonal (regional) authorities assume that the emission limits arecomplied with. They verify the power plant emissions (by official measurements) every ~ 2years. A detailed report, based on CEM or PEM of all power plant emissions (NOx, SO2,

CO, dust, volatile organic compounds,..), must be given (~yearly) to the authorities: Thiscovers the type of emission, the emission concentration and its mass flow as a function oftime and further statistical information to fully describe the situation.The SO2 emissions can be calculated from the fuel analysis for oil fired boilers.

United Kingdom;

(SO2, NOx, and CO); Hourly averages are determined every 10 minutes. No averages mustexceed 150 % of the set limit. 95 % of the hourly averages in a month must be below thelimit. Annual mass emissions are reported.


20/37

02003Ren9771 October 199717

4. EXPERIENCE WITH CEM

Experience in different countries of the two measurement systems (extractive and in-situ) isreviewed in this chapter. A common theme in many responses was difficulty in obtaining aconsistent reliable supply of calibration gas.

4.1 Extractive CEM

There is variable experience with extractive CEM but most countries have had goodexperience while the availability in some countries has been poor. In general there have beenfew problems specifically with the analysers and almost all difficulties reported relate to thesampling and calibration systems. Condensate management requires careful attention ifproblems are to be avoided. Maintenance requirements can be high unless the system iscarefully designed with e.g. provision for back purging of probe filters. Only a few reportsabout dilution systems was given and the experience was mixed. A major advantage of

extractive systems is their ease of calibration. Experience in each country is as follows;

Austria;

(All components); Extractive with condensation - generally good experience for allcomponents even at high dust concentrations, some problems with leakage (difficult todetect) and with interferences from moisture in the flue gas (NDUV/NDIR).(All components); Extractive with dilution - generally satisfactory for all components, canonly be used in gases with a low dust content, problems with calibration exist (dilution canvary with temperature), difficulties exist for correction of measurements to standard dryconditions (NDUV/NDIR).

Heated extractive CEM - is used in two cases for measurement of NH3. Experience isacceptable.

Belgium;

(SO2, NOx and CO); Use of NDIR for plant monitoring has been abandoned because of highmaintenance costs. It is still used for comparative measurements but the sampling systemadds many uncertainties. Extractive systems work satisfactorily on gas turbines where theyare the only type suitable for gas monitoring at high temperature (550 C).(CO); Good with clean flue gases; for gas turbines (NDIR).

France;

(SO2 and NOx); No important problems with NDIR instruments which are fitted to mostEDF power plants.(Dust); -absorption method is used. Maintenance is made difficult by the sampling location- generally in the upper portion of stack.

Germany;

(NOx, SO2 and CO); Good experience is reported for one utility (NDUV/NDIR).

Ireland;

(NOx); Highly labour intensive based on a single experience.

(CO); Poor performance. Condensation causes blockage of gas flow, leakage problems atpumps and cooler, probe filter plugging (NDIR).


21/37

02003Ren9771 October 199718

Italy;

(NOx, SO2 and CO); Installed in all ENEL power plants > 300 MWth. Principle isNDIR/NDUV. > 80 installations for NOx and SO2. Problems with automatic calibrationprocedures due to the quality of calibration gases. Problems also with the gas

dehumidification system.

Netherlands;(NOx); Attention must be paid to representative and proportional sampling. Goodagreement with independent measurements. Novel techniques such as FTIR or PDA (photodiode array) are still not sufficiently developed to be used for CEM.(SO2); All parts must be heated to avoid loss of SO2 in condensate, attention must be paid torepresentative and proportional sampling (NDIR/NDUV).(NH3); All parts in the sampling system must be heated to avoid condensation of watervapour and ammonium salts (NDIR).(N

2O); The interferences of CO, CO

2and NO must be checked regularly (NDIR).

(Dust); Dust monitoring is installed after the FGD plant. Sampling systems may clog whenthe plant is restarted after a shutdown. Monitors are adjusted monthly (zeroed) and aremaintained twice yearly.

Poland;

(All components); There are about 40 systems installed, but, within each power plant, thesame CEM type is normally used for measurements on the different boilers.Generally there is good experience of extractive CEM with condensation, some problemsarise relating to leakage (difficult to detect) and moisture. Maintenance problems occur withsample conditioning. It is of limited usefulness for on-line process control because of the

long response time (NDIR/NDUV).

Portugal;

(NOx and SO2); Calibration problems are often encountered. Minor problems related to thecondensate drain pump are sometimes reported (NDIR, NDUV).(SO2); Problems related to H2SO4 formation and associated corrosion (NDIR).(CO); Calibration problems (NDIR).

Spain;

(NOx, SO2); Generally good experience. Maintenance problems with sample conditioning

equipment (NDUV/NDIR/Chemiluminescence).

Sweden;

(NOx, SO2, CO, N2O); Extractive with condensation - generally good experience (availabilityand performance), problems with leakage at connections, coolers or pumps, plugging of theprobe filter or with drain pumps. CO could not be measured in series after the NOx-converter. The maintenance can be much reduced if systems for automatic calibration andback purging of the probe filter are installed (NDIR/NDUV).(N2O); Hydrocarbons can interfere (NDIR).(NH3, H2O, HCl); Heated extractive. Problems with condensation of moisture andammonium salts (2nd derivative IR-spectroscopy combined with gas filter correlation).

Switzerland;


22/37

02003Ren9771 October 199719

Extractive with condensation:

(NOx, CO); Good experience with multi-point probe. Frequent calibration is important. Tocheck for leakage, it is important to be able to inject the different calibration gases near thesample probe and also near the CEM instruments. The fast response time allows the

detection of small combustion changes in optimisation (chemiluminescense/NDIR).

United Kingdom;

(NOx); Generally good availability, some problems with drain pumps (NDIR).(SO2); Problems with high moisture content flue gas analysis (NDIR) after FGD processes(limited experience).(CO); Problems with drain pump and condensation of sample in lines, trace heatingrecommended (NDIR).

4.2 In-situ CEM

There is generally good experience with in-situ CEM. A disadvantage of most in-situsystems is the problem of calibration on a regular and routine basis. The main advantage isthe removal of any need for sample conditioning. Although low levels of maintenance arereported it is clear that there is need for a minimum level of essential maintenance e.g.cleaning of windows and checking of optical path alignment. All dust CEM which uses lightabsorption or opacity must be calibrated against a manual gravimetric dust measurement.The opacity and light scattering from particles from different fuels normally varies.

Some experience in the different countries is as follows;

Austria;(Dust); Problems with water droplets after FGD.(SO2, NOx, NH3, H2O and CO2); Generally good experience is reported with DOAS after alengthy adjustment period in a waste incineration application.

Belgium;

(All components); No severe systematic defaults has been observed. Measurement of theemission parameters requires extremely reliable sensors which provide self calibration,including auto/zero correction, auto/span checks and correction. Highly encouraging resultshave been achieved.

(SO2, NO); Thirty analysers of the 2nd derivative UV type have been in use since 1990.Accurate, free of interferences, calibration intervals are of the order of 6 months. Very lowinstrument drift, inaccuracy is frequently a result of bad calibration, air leakage is a possiblesource of inaccuracy. Equipment intervention has been limited almost completely toreplacement of the UV-lamp.(Dust); ~ 60 dust density monitors in use since 1990. The calibration of optical density tomass concentration varies with type of coal. Although not easily quantifiable, the erroramounts to at least 30 %.

France;(NOx and SO2); A test of DOAS has been in progress for some months in an EDF powerplant.


23/37

02003Ren9771 October 199720

Ireland;

(NOx); Good experience with electrochemical cell, but problems have occurred withvibration and dampness. Cells have a very short life (4 to 6 months). Good auto calibrationsystem.(CO); Good experience with IR cross duct systems, problems occur with cleanliness of

windows, one type of system cannot be calibrated on load, the other is calibrated by filteronly.

Italy;

(SO2 and Dust); Limited experience with in-situ CEM for SO2 monitoring. Extensiveexperience with dust monitoring by optical techniques (100 units). No specific problemsreported.

The Netherlands;

(NOx and SO2); Good experience with (ND)UV, problems with contamination of mirrorsand windows, disturbance in the alignment of the optical path due to vibration of theinstallation, calibration of cross-stack systems is difficult.(Dust); Dust monitors are installed after FGD plant. Generally there are few or no problemswith fouling. Monitors are adjusted (zeroed) each month. Maintenance is done every 6months.

Poland;

(NOx and SO2); About 60 installations. Better availability compared to extractive CEM.

Portugal;

(NOx and SO2); No major problems (UV).

Spain;

(NO and SO2); Very robust and reliable equipment. Low maintenance (2nd derivativeIR/UV).

Sweden;

(NOx, SO2, NH3, H2O); Good experience (availability and performance), general problemsare; soiling of windows, interferences and problems with calibration (DOAS).(CO); Good experience (IR).(NOx, SO2, NH3); Across the duct in-situ analyser (DOAS/UV) can handle a maximum level

of (ppm * m) and it is also difficult to use them in medium to small gas ducts. An in-situanalyser needs a minimum measurement length and is difficult to install in the small plants (25- 40 GWh/year) which will be included in the Swedish NOx-fee system from 1997.(Dust); Good experience with light scattering transmissometer.

Switzerland;

(CO); The accuracy is sufficient for the detection of increased CO due to a reduction inexcess air levels. It thus enables optimum adjustment of combustion conditions to be made(IR).


24/37

02003Ren9771 October 199721

United Kingdom;

(NOx, SO2); Mixed reports, some appear to perform well with long availability periods andfew maintenance requirements. On a number of occasions the degree of confidence in systemaccuracy has been questionable. General problems are fouling of windows, misalignment oftransmitter and receiver, mechanical problem with the chopper wheel and cross sensitivity

with interfering species (UV).(CO); Low maintenance (IR).

4.3 Accuracy

The general emission registration process can be split up into several parts.

4.3.1 Uncertainties of the Measurement Systems

Sampling of flue gases

Representativeness of sampling is one of the major sources of uncertainty. Sampling can beperformed at one point in the flue gas duct (point sampling), at several points (grid sampling)or across the flue gas duct with optical system (line sampling).

There is always a concentration profile in the flue gas duct. A theoretical statistical approachgives the following examples of uncertainties. When sampling at one (random) point in theflue gas duct the deviation of the measured concentration from the average concentration inthe flue gas duct can be 8 % when the ratio between the lowest and highest concentrations inthe flue gas duct is a factor of 1,2 and up to 34 % at a ratio of 2. Performing sampling with a4 x 4 grid the uncertainty (which is 2 times the standard deviation) reduces to 3% at a ratiobetween the highest and lowest concentrations of 1,2 and to 10 % for a ratio of 2. Increasing

the number of sampling points reduces the uncertainty. Deviations from the averageconcentration in the sampling plane with a single random sampling point and with gridsampling are shown infigure 1 and 2.

0

10

20

30

40

50

1 2 3

Ratio between the highest and lowest

concentration in the sampling plane

%d

eviatio

nwithrespecttothe

averageconcentrationinthe

samplingplane

Figure 1. Uncertainty with a single Figure 2. Uncertainty with grid samplingrandom sampling point for different ratios between the

highest and lowest concentration

in the sampling plane

0

24

6

8

10

0 10 20 30 40 50 60 70 80

Number of grid sampling points

Standard

deviation(%)

21.5

1.31.1

2.5


25/37

02003Ren9771 October 199722

Besides concentration gradients velocity gradients can also be present. In principle thecorrect average concentration is determined when the concentrations are velocity weighted.This is however very impractical and costly for continuous emission monitoring systems. Thesystematic error when no velocity weighted sampling is performed is only about 4 % at aratio of 2 between the highest and lowest concentration (and velocity).

Transport and conditioning of the flue gas sample

When using extractive sampling systems losses of SO2 up to 20 % can occur at levels below50 ppm (about 140 mg/m3) when using flue gas coolers (for sampling).

Analysis and calibration

Other important systematic errors occur due to drift and incorrect calibration (adjusting orincorrect calibration gases) of the measuring equipment. These errors are however smallwhen correct calibrations are carried out regularly.

Determination of the volumetric flue gas flow

The determination of the flue gas flow by means of direct measurements is in general lessaccurate than calculations based on the fuel input and the excess air ratio.

Variation in the 24-hour average

For the determination of the uncertainty in the annual emission the variation in the 24-houraverage is also of importance. The smaller this variation is, the smaller will be the uncertaintyin the annual emission..

4.3.2 Experience with accuracy reported in the different countries

The accuracy of the different CEM types for gaseous compounds is generally considered tobe good by users. However, it is important to note that in most cases emissionconcentrations are discussed and in the Netherlands and in Sweden also the question of totalannual mass emissions are considered. This makes direct comparison of experience difficult.

It is not possible to recognise any clear trend regarding the performance and the availabilityof CEM systems from the survey response. The accuracy of gaseous emission concentrationsin flue gas is expected to be high if calibrations are performed carefully and with certifiedcalibration gases or calibration cells. Experience reported from a Swedish accreditedlaboratory is that the deviation between a stationary CEM and a mobile extractive CEM

(owned by the laboratory) is below 4 % of the measured values in most cases. The accuracyof dust CEMs are low, especially at low concentrations. The accuracy of calculations ofyearly emissions depends also on uncertainties in fuel flow and volumetric flow calculation.A study of three coal fired power plants gives +/- 5 to 11 % uncertainty for NOx and +/- 25 -30 % for SO2 in low concentrations after FGD, mainly due to concentration gradients in fluegas ducts.

Belgium;

In-situ:(SO2, NO); During 1988 seven different types of in-situ gas analysers from differentmanufacturers were tested on a power plant. The test extended over six months and anextractive CEM was used as reference. It was found that the mean relative error on NOx-measurement is < 7 % when the calibration is done carefully.


26/37

02003Ren9771 October 199723

(CO); Low accuracy and sensitivity.(Dust); Accuracy is uncertain for dust contents below 100 mg/m3.

Netherlands;

A study has been performed for the estimation of the total uncertainty in the determination of

SO2 and NOx emissions by means of continuous emission monitoring systems at three coal-fired power plants in the Netherlands.The power plants were equipped with primary NOx control and with flue gas desulphurisationunits (FGD). The NOx-concentrations were in the range between 250 and 500 mg/m

3 and theSO2-concentrations in the range between 50 and 200 mg/m

3 . Two plants were equipped witha (single point) in-situ SO2 and NOx monitoring system (UV absorption; Lear Siegler) and thethird power plant was equipped with an extractive emission monitoring system (UV-PDA;Ametek) for SO2, NO and NO2. The sampling occurred at a single point in the flue gas duct.The following major types of uncertainties were determined and quantified:- systematic errors (such as representativeness of sampling, uncertainty in calibration gases,

interferences)- random errors (such as variation in 24-hour average, calculation of the volumetric flue gas

flow)

In table IVthe uncertainties (expressed as standard deviation) are quantified.

Table IV. Uncertainties (expressed as standard deviations) in the emission

registration process for three coal fired power plants (I, II and III) with

FGD.I II III

NO representativeness of 1,3 % 0,7 % 1,5 %

24-hour average 5,2 % 10 % 18,9 %SO2 representativeness of 11,3 % 11,4 % 12 %

24-hour average 27,2 % 24 % 28,1 %O2 representativeness of 0,8 % 2,2 % 2 %

24-hour average 0,62 % - 0,52 %Calibration gases (NO, SO2 and O2) 1 % 1 % 1,5 %Interferences (NO, SO2 and O2) 0,5 % 0,5 % 0,5 %Water vapour content 2,6 % - 3 %Volumetric flow calculationstoichiometric flue gas volume 2,3 % 5,2 % 2,6 %

Coal flow 0,5 % 0,62 % 2,5 %

For the three individual coal fired power plants the uncertainty in the total annual emission ofNOx were around 5 % for two plants and 11 % for the other. This higher value is due to thefact that this power plant uses only one and the same stoichiometric flue gas volume for thecalculation of the volumetric flue gas flow for all fired coal blends during the whole year. Ifthe stoichiometric flue gas volume for each coal blend is used then an uncertainty of about +/-5 % in the annual emission of NOx can also be reached.

For SO2 the uncertainty in the annual emission was in the range between +/- 20 and 25 % for

all three power plants. This uncertainty is mainly due to the single point measurements andthe concentration gradients in the flue gas duct after the flue gas desulphurisation unit.


27/37

02003Ren9771 October 199724

If a sampling grid of about 16 sampling points (4 x 4 grid) is used then an uncertainty ofabout +/- 15 % can be achieved. A further reduction of the uncertainty to a level of +/- 5 %requires at least 100 sampling points.

These uncertainties are solely based on the quantifiable errors in the total emission recording

process. Any bias due to incorrect calibration, leakage, calculation errors (wrong conversionfactors), the inclusion of faulty measuring values etc. are not included in these figures.

(Dust); Comparison measurements against a manual method are made only at installation.Monitors are not calibrated against manual reference methods. Dust levels are very low,generally below 10 mg/m3 this is substantially below the legal limit value which is usually20 or 50 mg/m3 depending on the date of the first license.

Sweden;

(NOx, SO2, CO); The relative error in CEM concentration measurements has been calculated

to be about 4 % in some Vattenfall studies. These studies have been done for mobileextractive CEM but the value is also representative for stationary CEM. The followingparameters have been included; uncertainty in the representativeness of the sampling point(extractive CEM), uncertainty in the calibration gases, uncertainty in the zero point,uncertainty in the linearity, uncertainty because of interferences and finally uncertaintybecause of accidental errors in the recording unit.

The uncertainty in the determination of the mass emissions (kg/h) have been estimated toabout +/- 12 % in another study. In this study the gas flow is calculated from the fuelanalysis and continuous measurement of oxygen.

The experience from the accredited laboratory owned by Vattenfall is that the deviationbetween a stationary CEM and a mobile extractive CEM (owned by the laboratory) is in mostcases below 4 % of the measured values.

Another experience of the same laboratory is that there are almost never concentrationgradients in the flue gas duct or the stack. The laboratory have verified about 30 CEMinstallations in small to average plants (10 - 100 MW). During such a procedure twodifferent CEMs with different probes are used, the first probe is moved across the duct whilethe other probe is fixed at one point.

Concentration gradients have occurred in a few cases and if the concentrations from the twodifferent CEM were standardised to the same level of O2, the concentrations corresponded inall cases except one. The value of possible differences was, with the exception of the oneproblematic case, always less than 10 % (without normalisation to the same O2-concentration). In the problematic case the corresponding concentration levels weremeasured at 7 out of 10 measurement points, in the remaining 3 the difference was up to 30%.

The measurement points were, in all cases, located after the de-dusting and also after anydeSOx-or deNOx-processing. In no cases were the measurement points located close to aflue-gas cleanup process, a rotating air preheater or a fan.

Switzerland;


28/37

02003Ren9771 October 199725

Extractive with condensation:(SO2); Uncertainty +/- 15 % (multi-point grid measurement and NDIR).Extractive with dilution:(NOx, SO2); The accuracy is not sufficient for a heavy oil fired power plant applications(Chemiluminescence, fluorescence).

In-situ:(CO); Uncertainty +/- 20 ppm, reproducibility +/- 10 ppm (IR).

United Kingdom;

(CO); Satisfactory accuracy and reliability (extractive with condensation and NDIR).

4.4 Reported Experience of a General Nature

Experience of CEM is good overall based on extensive experience in many countriesreviewed here.

Experience is better with CEM based on extraction with condensation compared to CEMbased on extraction with dilution or heated extractive CEM. Based on limited data it appearsthat the experience (accuracy, performance, long term availability, operating costs etc.) ofextractive (with condensation) and in-situ CEM are about the same. Belgium has reportedbetter experience of in-situ CEM and Sweden has reported slightly better experience withextractive CEM.

There have been no reports of any difficulties with data collection/processing systems forCEM although major problems with data systems have been widely reported in the US.Further information on data collection/processing, whether of a positive or negative nature,would be of value in completing this review of CEM experience.

Italy;

In Italy the trend is toward the application of extractive with condensation CEM formonitoring of SO2, NOx and CO and of in-situ CEM for monitoring of dust. A unifiedcalibration and maintenance program has now been developed and is being applied for qualitycontrol and insurance. Some problems has been met in automatic calibration due to thedifficulties in finding good calibration gases.

Sweden;

In Sweden there are two trends concerning extractive or in-situ CEM. The new small plants(25 - 40 GWh/year in produced energy) which now are included in the fee-system havechosen extractive CEM or in some cases PEM because it is difficult to measure with an in-situ CEM in small diameter gas ducts. Also, the price of extractive CEM is lower in Swedencompared to in-situ CEM. The larger plants, however, have mostly chosen in-situ CEM. Anincreasing number of plants are gradually being equipped with deNOx and measurement ofNH3 slip is therefore desirable. Measurement of NH3 with in-situ CEM is more accurate andis also easier.Regarding calibration, a recommendation which can be made is to calibrate the analyserweekly in the month following installation. If the accuracy is satisfactory then the analysercan subsequently be calibrated monthly. Most of the new plants that will be included in theNOx-fee system during 1997 are equipped with calibration cells. These CEM systems are


29/37

02003Ren9771 October 199726

therefore only checked with calibration gases 1 - 2 times/year. This strategy have beenaccepted by SEPA.Some CEM suppliers had problems with their data collecting systems when the NO x-feesystem first started (1991). There have not been any important difficulties with the collectionof data in recent years and most established CEM suppliers have developed suitable data

collection programs.

Switzerland;

NOx measurement by Chemiluminescence can suffer drift through use of ambient air for theozoniser this is due to a high sensitivity to the air moisture content.

4.5 Investment and Operating Costs

The costs of CEM-systems depends on several factors; number and types of components,number of sampling points, sampling system, analysis principle and so on. Direct comparison

of reported CEM costs has not been possible. The reported investment span is considerable;from 20.000 - 200.000 and annual operating costs are reported from 3.000 to 9.000.

The lower costs are reported for example from highly standardised CEM used in the SwedishNOx-fee system and only for monitoring of O2 and NO.

The reported experience in this study indicate that costs are lower (both investment costs andmaintenance costs) for in-situ CEM than for extractive CEM. In some countries additionalcost arises each year for independent verification by an accredited laboratory this iscurrently a moderate cost.


30/37

02003Ren9771 October 199727

5. GENERAL DISCUSSION

The most important trend in all the countries reporting is the increasing requirement for CEMmeasurements, driven particularly by new requirements by the regulatory authorities. Theutilities have performed many investigations into CEM and its advantages. CEM is

consequently used not only for control of the environmental performance of a plant but also(particularly in the case of CO and dust) for optimisation and control of the plant operation.Existing and new plants continue to be fitted with primary measures such as low-NO x burnersor with secondary measures deNOx. An increasing number of CEM installations aretherefore required, not only for measurement of NOx for deNOx process control but also, insome countries, for the measurement of NH3 slip when using SNCR techniques.

The experience reviewed in this study shows no definite trends for choice of extractive or in-situ CEM, both have been applied successfully in many countries.

Use of CEM by the extractive condensation principle has, thus far, been more frequent thanin-situ CEM. In most countries the experience (accuracy, performance, long termavailability, operating costs etc.) of extractive and in-situ CEM has been similar. Somecountries e.g. Belgium have reported better experiences of in-situ CEM while Sweden hasreported better experiences of extractive CEM. The reported experience with CEM based onextraction with condensation is, however, better than that of CEM based on extraction withdilution or heated extractive CEM.

In Sweden, despite the slightly better experiences from extractive CEM, there have beenmany new in-situ installations. This is because an increasing number of plants are beingequipped with SNCR-systems and therefore often, mainly for process control, measure theNH3 slip. Measurement of NH3 is more accurate and easier with in-situ CEM which also canoften combine the measurement of NOx and NH3. The trend in Sweden is therefore; plantswith deNOx - in-situ CEM and other plants - extractive with condensation.

In most of the countries, performance criteria for CEM have been established. Only Germany(anticipated in Ireland) has a standard approval system for the CEM instruments. Also somecantons (regions) in Switzerland require German TV approved CEM instruments.

It is not possible to recognise any clear trend regarding the performance and the availabilityof CEM systems from the survey response. A high accuracy can be expected if the CEM is

installed and managed correctly. A much poorer accuracy can be expected if the relevant fluegas concentrations are very low or with a considerable concentration profile e.g.measurement of SO2 after a deSOx unit.

It is clear from many responses that calibration systems pose problems and require furtherattention. A common difficulty appears to be the quality and reliability of calibration gasstandards.

Trends which are apparent include the facts that the more conventional CEM methods arebecoming more cost effective and that the simpler CEM methods are being improved. Newmeasurement techniques such as FTIR and diode laser are also coming into use. Theelectrochemical analysis principle is one example of a technique which may be improvedfurther.


31/37

02003Ren9771 October 199728

In some countries PEM for determination of NOx emissions has been tested (Portugal,Sweden, Switzerland, the Netherlands and United Kingdom). A PEM system can be verifiedby multivariate data analysis or based on predictions by neural network. If a PEM system isto be used the plant must be fired with fuel of constant quality.


32/37

Appendix 1 page 29 of 3

02003Ren9771 October 199729

Article 15

1. In the event of continous measurements, the emission limit values set out in Annexes IIIto VII1 shall be regarded as having been complied with if the evaluation of the resultsindicates, for operating hours within a calender year, that:

a) none of the calendar monthly mean values exceeds the emission limit values; and

b) in the case of:- sulphur dioxide and dust: 97% of all the 48 hourly mean values do not exceed

100% of the emission limit values,

- oxides of nitrogen: 95% of all the 48 hourly mean values do not exceed 110%of the emission limit values.

The periods referred to in Article 8

2

as well as start up and shut down periods shall bedisregarded.

2. In cases where only discontinuous measurements or other appropriate procedures fordetermination are required, the emission limit values set out in Annexes III to VII 3 shallbe regarded as having been complied with if the results of each of the series ofmeasurements or of the other procedures defined and determined according to the ruleslaid down by the competent authorities do not exceed the emission limit values.

3. In the cases referred to in Article 5 (2) and (3)4, the rates of desulphurization shall beregarded as having been complied with if the evaluation of measurements carried out

pursuant to Annex IX, point A.25, indicates that all of the calendar monthly mean valuesor all of the rolling monthly mean values achieve the required desulphurization rates.

The periods referred to in Article 86 as well as start up and shut down periods shall bedisregarded.

1 Council Directive 88/609/EEC



4 Council Directive 88/609/EEC5

Council Directive 88/609/EEC6 Council Directive 88/609/EEC


33/37


02003Ren9771 October 199730

Annex IX

METHODS OF MEASUREMENTS OF EMISSIONS

A. Procedures for measuring and evaluating emissions from new plants

1. Concentrations of SO2, dust, NOx and oxygen shall be measured continously in the caseof new plants with a rated thermal input of more than 300 MW. However, monitoring ofSO2 and dust may be confined to discontinous measurements or other appropriate deter-

mination procedures in cases where such measurements or procedures, which must beverified and approved by the competent authorities, may be used to obtain concentration.

In the case of plants which must comply with the desulphurization rates fixed by Article5 (2) and (3)7, the requirements concerning SO2 emission measurements established

under paragraph 1 shall apply. Moreover, the sulphur content of the fuel which is intro-duced in the combustion plant facilities must be regularly monitored.

3. The competent authorities shall be informed of substantial changes in the type of fuelused or in the mode of operation of the plant. They shall decide whether the monitoringrequirements laid down in paragraph 1 are still adequate or require adaption.

4. Continously-operating measuring systems shall be checked at regular intervals in consul-tation with the competent authorities. The instruments for the measurement ofconcentration of SO2, dust, NOx and oxygen shall undergo basic calibration and an

examination of their operation at appropriate regular intervals. The continously-operating measuring equipment shall be calibrated in accordance with a reference

measuring method approved by the competent authority.

B. Determination of total annual emissions of new plants

The competent authorities shall obtain determination of the total annual emissions ofSO2 and NOx. When continous monitoring is used, the operator of the combustion

plant shall add up separatelys for each pollutant the mass of pollutant emitted each day,on the basis of the volumetric flow rates of waste gases. Where continous monitoring isnot in use, estimates of the total annual emissions shall be determined by the operator onthe basis of paragraph A.1 to the satisfaction of the competent authorities.

Member States shall communicate to the Commission the total annual SO2 and NOxemissions of new combustion plants at the same time as the communication requiredunder paragraph C.3 concerning the total annual emissions of existing plants.



34/37


02003Ren9771 October 199731

C. Determination of the total annual emissions of existing plants

1. Member States shall establish, starting in 1990 and for each subsequent year, a completeemission inventory for existing plants covering SO2 and NOx:

- on a plant by plant basis for plants above 300 MWth and for refineries;

- on an overall basis for other combustion plants to which this Directive applies.

2. The methodology used for these inventories shall be consistent with that used todetermine SO2 and NOx emissions from combustion plants in 1980.

By 1990 Member States shall inform the Commission of full details of methods and basedata used for establishing the emissions of SO2 and NOx from existing combustion

plants, referred to respectively in Annexes I and II8, column 0.

3. The results of this inventory shall be communicated to the Commission in a conveniently

aggregated form within nine months from the end of the year considered.

The methodology used for establishing such emission inventories and the detailed baseinformation shall be made available to the Commission at its request.

4. The Commission shall organize a systematic comparison of such national inventories and,if appropriate, shall present proposals to the Council aiming at harmonizing emissioninventory methodologies, for the needs of an effective implementation of this Directive.



35/37

Appendix 2

02003Ren9771 October 199732

Extractive CEMManufacturer Component Monitoring principle

Hartmann & Braun * NO NDUV or NDIR

(NO+NO2) NDUV or NDIR + converter CO, SO2, N2O NDIR

Bodenseewerk * NO, CO, N2O, NO2 CH4,C2H6 etc

2nd derivative spectroscopy +gas filter correlation (IR)

*** NO, CO, N2O, HCl, NH3,H2O, CH4, C2H6 etc

Siemens * NO, CO, SO2 NDIR (NO + NO2) NDIR + converter

Maihak * NO, CO, SO2 NDIR (NO + NO2) NDIR + converter

Rosemount * NO, CO NDIR NO2, SO2 NDUV

Servomex * NO, NO2, SO2, CO NDIRTecan* NO, NO2, NOx Chemiluminescence + converter

ADC (Monitor Labs)**

NO Chemiluminescence

(NO + NO2) Chemiluminescence + converter CO IR SO2 Fluorescence

* Extractive with condensation

** Extractive with dilution (wet gas)*** Heated extractive (wet gas)

In-situ CEMManufacturer Component Monitoring principleLear Siegler * NO, SO2 2nd derivative

spectroscopy (IR/UV) CO 2nd derivative

spectroscopy (IR)OPSIS */** NO, NO2, H2O, SO2, NH3 DOAS (UV)

Rosemount ** CO IRSICK */** NO, SO2, NH3, dust UV

CO NDIRDurag ** dust Opacity, light absorptionLAND * NOx Electrochemical cell

CO IRCodel NO, CO, H2O IR

Procal ** NO, SO2 IR* In-situ probes** Cross-duct measurement


36/37

Publications Order Form

Name: ...............................................................................................................................................

Position:............................................................................................................................................

Undertaking:.....................................................................................................................................

Address:............................................................................................................................................

Town:.......................................................Country:..........................................................................(with postal code)

Telephone:.......................................................Fax:..........................................................................(with regional code)

E-mail:

EURELECTRIC member: p Yes p No (Tick the appropriate box)

Reference No. Title1

Quantity

To be returned to:

Concetta PALERMO Union of the Electricity Industry - EURELECTRICDocumentation

66, Boulevard de lImpratrice BE-1000 Brussels

Tel.: + 32 2 515 10 00Fax: + 32 2 515 10 10

E:mail: [email protected]: http://www.eurelectric.org

1

Some documents are available in French (FR) and German (DE).Please indicate the language of your choice, when possible.


37/37

cems in power plant

Documents