Refinery air emissions Good Practice Series management 2012 · Good Practice Series 2012 ......

Refinery air emissionsmanagement Guidance document for the oil and gas industry

OperationsGood PracticeSeries2012

www.ipieca.org

The global oil and gas industry association for environmental and social issues

5th Floor, 209–215 Blackfriars Road, London SE1 8NL, United KingdomTelephone: +44 (0)20 7633 2388 Facsimile: +44 (0)20 7633 2389E-mail: [email protected] Internet: www.ipieca.org

© IPIECA 2012 All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in anyform or by any means, electronic, mechanical, photocopying, recording or otherwise, without theprior consent of IPIECA.

This publication is printed on paper manufactured from fibre obtained from sustainably grownsoftwood forests and bleached without any damage to the environment.

Refinery air emissionsmanagementGuidance document for the oil and gas industry

Revised edition, July 2012

This document was produced in collaboration with Jeffrey H. Siegell and ICF International.

Photographs on the cover and pages 2, 26 and 38 reproduced courtesy of ©Shutterstock.com.

IPIECA

ii

REFINERY AIR EMISSIONS MANAGEMENT

Contents

Executive summary 2

Introduction 3

Air emissions overview 3Emission types 3

Potential emissions impacts 3

Control scenarios 4Source pollutant emission limits 5

Source pollutant concentration emission limit 5

Ambient concentration limit 5

Specified control equipment 5

Specified control performance 6

Specified control practice 6

Developing emission inventories 7

Sources 7Hydrocarbons 7

Combustion products 7

Estimating methods 7Average factors 7

Correlations 8

Computer models 8

Measurements 8

Quality assurance 9Good practices for emissions inventory development 9

Auditing an emissions inventory 9Review procedures 10

Checklist 10

Reporting results 10

Sources and control of hydrocarbon emissions 11

Fugitives and piping systems 11How to quantify emissions 12

Open-ended lines 12

Pump, compressor and valve stem sealing 12

Enhanced sealing techniques 14

Valve quality: materials and finishes 15

‘Leakless’ components 15

Leak detection and repair 16

Good practices for control of fugitive emissions 18

Storage tanks 18How to quantify emissions 21

Tank types: fixed and floating 21

Floating roof rim seals 21

Roof fittings: gasketing and slotted guidepoles 23

Roof landings 24

Cleaning operations 25

Good practices for control of storage tank emissions 26

Product loading 26How to quantify emissions 27

Splash, bottom and submerged loading 27

Vapour balancing 27

Vapour recovery: adsorption, absorption 28and refrigeration

Vapour destruction: flares, thermal oxidizers 30and catalytic oxidizers

Good practices for control of loading emissions 31

Wastewater collection and treatment 32How to quantify emissions 33

Source reduction 33

Sewers, drains, junction boxes and lift stations 33

Primary separators, IAF/DAF, biological treatment 34and treatment tanks

Good practices for control of air emissions from 35wastewater collection and treatment

Process vents 36Good practices for controlling process 36vent emissions

Flares 36Source reduction 36

Gas recovery 37

Sources and control of combustion emissions 38

Boilers, heaters and furnaces 38How to quantify emissions 39

PM (particulate matter) control 39

SOx control 40

NOx control 42

Cogeneration 43

Good practices for control of boiler, heater 43and furnace emissions

1


Catalytic cracking 43How to quantify emissions 44

PM (particulate matter) control 45

SOx control 45

NOx control 45

Good practices for control of catalytic 46cracker emissions

Sulphur plants 46How to quantify emissions 46

Sulphur recovery 46

Amine absorption 46

Sulphur recovery units 46

Good practices for control of sulphur plant emissions 47

Gas turbine NOx 47

Flares 47Source reduction 47

Gas recovery 47

Odour control and management 49

Problem assessment 49

Source identification 50

Impact assessment and verification 50

Problem resolution 52Good practices for addressing odour problems 53

References 54

List of Tables and Figures

Table 1: Examples of air emissions control scenarios 4

Table 2: Relative emission contribution for hydrocarbons 11

Table 3: Controls for reducing fugitive emissions 12

Table 4: Controls to reduce storage tank emissions 20

Table 5: Seal system impact on emissions from 22

external floating roof tanks

Table 6: Seal system impact on emissions from 23

internal floating roof tanks

Table 7: Controls to reduce product loading emissions 27

Table 8: Characteristics of vapour recovery technologies 28

Table 9: Advantages and limitations of vapour 29

recovery technologies

Table 10: Characteristics of vapour destruction technologies 30

Table 11: Advantages and limitations of vapour 31

destruction technologies

Table 12: Controls to reduce wastewater collection 32

and treatment emissions

Table 13: Controls to reduce PM emissions 40

Table 14: Controls to reduce SOx emissions 41

Table 15: Controls to reduce NOx emissions 43

Table 16: Control option applicability for catalytic 44

cracking units

Table 17: Example odour detection thresholds, 51

exposure limits and descriptions

Table 18: Exponents for Steven’s Law equation 52

Figure 1: Leak detection: US EPA ‘Method 21’ 17

Figure 2: Leak detection: optical imaging 17

Figure 3: A leaking valve, viewed using optical 18

gas imaging equipment

Figure 4: Air flow across a slotted guidepole 24

promotes evaporation

Figure 5: A sleeve placed around a slotted guidepole 24

eliminates air flow through the slots

This document describes ‘good practices’ andstrategies that can be used in petroleum refineriesto manage emissions of air pollutants, and includesa special section on how to identify odour sources.Many of the techniques may also be applicable tothose chemical plants and petroleum distributionfacilities having similar equipment and operations.

Since individual refineries are uniquely configured,the techniques, which comprise a collection ofoperational, equipment and procedural actions,may not be applicable to every site. Applicabilitywill depend on the types of processes used, thecurrently installed control equipment and the localrequirements for air pollution control.

This document will assist plant personnel to identifythose techniques which may be used to optimizethe management of air emissions and to selectappropriate techniques for further site evaluation.

The document is organized as follows:� Introduction� Developing emission inventories� Sources and control of hydrocarbon emissions � Sources and control of combustion emissions� Odour control and management

IPIECA

2

Executive summary

Air emissions overview

Petroleum refineries are complex systems ofmultiple linked operations that convert the refinerycrude and other intake into useful products. Thespecific operations used at a refinery depend onthe type of crude refined and the range ofrefinery products. For this reason, no tworefineries are exactly alike. Depending on therefinery age, location, size, variability of crudeand product slates and complexity of operations,a facility can have different operatingconfigurations and significantly different airemission point counts. This will result in relativedifferences in the quantities of air pollutantsemitted and the selection of appropriate emissionmanagement approaches.

For example: refineries that are highly complex witha wide variety of hydrocarbon products are likely tohave more product movements and hence apotential for relatively higher fugitive, tank andloading emissions; refineries that process heavier orhigh sulphur crude and which have higherconversion are likely to have relatively highercombustion emissions because of their higherenergy demand. Each refinery will have site-specificair pollution management priorities and uniqueemissions management needs as a consequence ofall these factors. National or regional variations infuel quality specifications can also affect refineryemissions as stricter fuel quality requirements willoften require additional processing efforts.

Emission types

Refinery air emissions can generally be classifiedas either hydrocarbons, such as fugitive andvolatile organic compounds, or combustionproducts such as NOx, SOx, H2S, CO, CO2, PMand others. When handling hydrocarbon materials,there is always a potential for emissions throughseal leakage or by evaporation from any contact ofthe material with the outside environment. Thus, theprimary hydrocarbon emissions come from piping-

system fugitive leaks, product loading, atmosphericstorage tanks and wastewater collection andtreatment.

A refinery uses large quantities of energy to heatprocess streams, promote chemical reactions, andprovide steam and generate power. This is usuallyaccomplished by combustion of fuels in boilers,furnaces, heaters gas turbines, generators and thecatalytic cracker. This results in the emission ofproducts of combustion.

In addition to hydrocarbon losses and corecombustion emissions, refineries emit small quantitiesof a range of specific compounds that may need tobe reported if threshold limits are exceeded. Controlson core emissions may also be effective for these(e.g dust controls are effective for reducing emissionsof heavy metals, VOC controls are effective forspecific hydrocarbons such as benzene).

Potential emissions impacts

Management of refinery emissions is focused onmeeting local and national standards. Air qualitystandards are expressed as concentration limitvalues for specific averaging periods or as thenumber of times a limit value is exceeded. Theactual concentrations generated depend on thecharacteristics of specific site emission points andalso on the local meteorological conditions.Emission limit standards may also apply wherelong range or regional pollution is of concern.Here, the details of the site emission areunimportant but the total site emission of certainpollutants may be subject to a national or regionalemission reduction plan.

The purpose of air quality standards is to protectthe human population from adverse impacts ofpollution from all sources. The rationale behindspecific standard values can be found in, forexample, the technical documentation for theWorld Health Organization Air Quality Standards.Not all pollutant concentrations can be directly

3


Introduction

linked to simple source emissions. NOx andvolatile organic compounds (VOCs) can react inthe lower atmosphere under suitable conditions tocreate higher than natural environmentalconcentrations of ozone. A regional or nationalemission control plan is needed to regulate suchepisodic ozone events.

Understanding potential impacts of emissionsTo better understand impacts, both ambient airquality monitoring and modelling is used.Dispersion modelling is sometimes conducted onspecific emission sources to evaluate off-sitepotential concentrations. Using local meteorology(e.g. wind speed and direction) and details of theemission release (e.g. stack height, temperatureand quantity), the location and magnitude ofmaximum concentrations can be predicted.Ambient air quality monitoring may be used toverify these predictions, especially if limit values arepredicted to be approached, or to provideassurance that no breaches occur.

Regional air quality modelling can be used toevaluate the impact of multiple sources onbackground air quality.

Control scenarios

Regulatory agencies can specify air pollutionemission limits and control requirements in avariety of ways. These include limits on the quantityof a pollutant that may be emitted, the allowableconcentration of the emission, the resultant localambient concentration, a target emission reductionand specific monitoring and repair procedures, etc.Sometimes, more than one of these emission limitsand control requirements are applied to the samesource. Guidance on emission control techniquesmay also be provided, for example information oneffectiveness, cost and applicability.

Table 1 provides examples of the ways thatregulatory agencies may control air emissions. In

IPIECA

4

Table 1 Examples of air emissions control scenarios

Scenario Example control requirement Example application

• Maximum quantity of SOx, NOx, PM from stack or site (site ‘bubble’ limit).

• Maximum hydrocarbon or toxics from vent.

• Maximum ppm of SOx or NOx in flue gas.• Maximum mg/m3 of PM on flue gas.• Maximum ppm of hydrocarbon from vent.

• Maximum concentration of SOx, NOx or PM in ambient air.

• Use of specific control equipment (e.g. SCR, wet gas scrubber(WGS), electrostatic precipitator (ESP), etc.).

• Application of specific rim seals on atmospheric storage tanks.• Multi-seal pumps.• Use of natural gas to replace liquid fuel firing

• Percent removal of PM and SOx from catalytic crackerregenerator stack.

• Destruction efficiency for oxidation unit on a product loadingsystem.

• Piping system component monitoring and leak repair.• Monitoring of tank rim seals and floating roof gaskets.

Maximum tonnes/annum

Maximum mg/m3 in flue gas

Maximum micrograms/m3 inambient air

Agreed technology step or operational measure

Pollutant removal efficiency

Inspections and repair

Pollutant emissionquantity limit

Pollutant emissionconcentration limit

Ambient concentrationlimit

Selected control

Specified controlperformance

Specified controlpractice

most cases, the control scenarios are not unique.They are often copied from other countries thathave well established national air pollutionreduction programmes. It is also common that themore stringent control requirements tend to bepropagated.

In many locations, facilities must apply what isoften called ‘best available technology’ (BAT) and‘best environmental practice’ (BEP). The definitionof BAT and BEP can vary from agency to agency,but it generally refers to well-establishedcommercially available control equipment, designs,principles or practices that are technically andeconomically applicable. The cost-effectiveness ofimplementing a specific control should be assessed,particularly where a retrofit to an existing unit isconcerned.

Source pollutant emission limits

Regulating emissions by setting a limit on the totalquantity (e.g. kilograms) of a pollutant emitted in agiven time can obscure environmental performancebecause comparison of different facilities ofdifferent sizes or function is not easily made. It ispreferable to set a concentration limit where theconcentration is expressed at some standardcondition. The limit can be set for an individualsource, a group of similar sources or for the entirefacility (i.e. a bubble limit). Typical applications ofthis type of limit are for SOx, NOx and particulatematter (PM) from combustion sources and forhydrocarbons from process vents or from productloading operations.

Source pollutant concentrationemission limit

A concentration limit on the pollutant beingreleased is typically defined as an averageconcentration over a given time period. Timeperiods may be hourly, daily, annual, dependingon the pollutant in the stream being released. Theconcentration should be referenced to a given

dilution, for example, for flue gas stackconcentrations this is usually 3% oxygen at 1 atmand 0 °C of dry flue gas vapour. It is important touse consistent units. In Europe, for stack gases(except CO) and dust, the concentration limit isexpressed in units of mg/m3.

Ambient concentration limit

Care has to be taken over units for ambient airconcentration limits because notation can beconfusing, particularly if measurements are cited involume units and the standards in mass units. Massunits are necessarily expressed at one atmosphereand 0 °C, and a µg/m3 scale is used. Anaveraging time has to be specified, and somestandards have more than one period specified.Common periods are hourly, daily, annual. As acompanion to the limit, and recognizing thatconcentrations in the atmosphere are highlyvariable, a certain number of limit exceedancesmay be allowed. The limit may be equivalentlyexpressed as a percentile of suitably averagedconcentrations rather than an overall maximum.

As discussed above, dispersion modelling can beused to perform an ambient air quality impactassessment to predict how the maximum expectedconcentrations from a source will compare to theambient concentration standards. Ambient airquality monitoring can be used to inform on actualconcentrations, especially where sources apart froma refinery, for example traffic, are present anddominant.

Specified control equipment

It is preferable that the refinery has flexibility inselecting from alternative methods of emissionreduction where this is needed and feasible, ratherthan the regulatory agency requiring the use ofspecific emissions control equipment. In most cases,an alternate control that provides equivalentemissions reduction is allowed to be substituted forthe specified equipment.

5


Specified control performance

In cases where the regulatory agency sets a specificcontrol performance, it is usually expressed as therequired removal efficiency of a specific pollutantfrom the discharged stream under normaloperating conditions. Examples include PM andSOx from catalytic cracker regenerator vents, andresidual hydrocarbons from product loadingemission control systems. Alternate controlequipment or procedures are usually allowed aslong as the percent reduction in emissions isachieved.

Specified control practice

In cases where the regulatory agency requires aspecified practice to be applied, it is important thatstandard procedures are used and that thefrequency of inspection is appropriate to the levelof control required and reflects any demonstratedcontinuous improvement. Examples of these aremonitoring and repair of piping systems (e.g.valves, flanges, pumps, etc.) for leaks andinspection and repair of atmospheric storage tankrim seals with excessive gaps.

IPIECA

6

An essential part of any emission managementprogramme is a representative assessment of currentand projected emissions. The emissions inventoryallows comparison of potential sources for controland provides a mechanism to quantify potentialreductions. Emphasis should be placed on makingthe inventory complete and of high quality so that itis as representative of plant emissions as possible.

In this report, each of the sections on emissionscontrols is preceded by a brief discussion of themethods available for estimating emissions for thattype of source. Detailed methods for estimatingemissions are available in the references.

Sources

There are two general types of refinery emissions:hydrocarbons and combustion products such asSOx, NOx and CO2. Most of the major pieces ofprocess equipment handling hydrocarbons atrefineries do not emit any combustion products.However, the combustion sources such as heatersand boilers will typically emit air pollutants andgreenhouse gases as well as small amounts ofhydrocarbons (VOC) due to incompletecombustion.

Hydrocarbons

When handling hydrocarbons, there is always apotential for leakage through seals and byevaporation from any contact with the outsideenvironment. Examples of leaking though seals includeleaks from piping connectors, valves, compressorsand pumps. Examples of sources of evaporationinclude atmospheric storage tanks, productloading, and wastewater collection and treatment.

Combustion products

A refinery uses large quantities of energy to heatprocess streams, promote chemical reactions,

provide steam, isolate and recover excess sulphurand generate power. This is usually accomplishedby combustion of fuels, typically those generated onsite such as refinery fuel gas and the coke depositedon cracking catalysts. Examples of combustionsources include furnaces, boilers, heaters, turbinesand the catalytic cracker regenerator.

Some sources of combustion products are unitsoperated to safely control hydrocarbon emissionsand which do not normally supply useful energy forplant operations. Examples of these are flares andincinerators/thermal oxidizers.

Estimating methods

For most emission sources, there are several ways toestimate emissions. These have mostly beendeveloped by regulatory agencies, e.g. the USEnvironmental Protection Agency (US EPA) andindustry groups such as CONCAWE and theAmerican Petroleum Institute (API). Methodsrequiring more detailed design and processoperating data provide more representative emissionestimates and usually require more effort to applythe more detailed input data. The choice of emissionestimating method may be prescribed or may be anoperator’s choice but should be recorded. The choiceof methods should be consistent with the objective ofthe emission inventory, the intended use, informationavailability, time allowed, and resource needs.

In order of increasing data requirements andcalculation efforts, estimating methodologiesinclude average emission factors, correlations,computer models and direct measurement. This isalso the general order of obtaining morerepresentative emission estimates.

Average factors

Industry average emission factors have beenpublished for a wide range of source types (see

7


Developing emission inventories

References) and are often used for initialinventories and until more representative andsource-specific input data are available. Typically,these factors are used by multiplying the factor byan operating parameter, such as throughput or fuelcombusted, to obtain the estimated emissions.

An example of industry average emission factorsare those for NOx emissions. In this case the factorsrepresent the quantity of NOx emitted for aquantity of fuel burned (tonne NOx/GJ fuel fired).In the case of a single factor for NOx, there is noconsideration of specific equipment design ordifferences in specific operating conditions.

Improved NOx emissions quantification can beobtained through direct measurement of the specificsource. In some cases, equipment vendors provideequipment-specific estimates. Models based onlimited source measurements have proved veryreliable. For example, measuring NOx emissions ina furnace under known operating rates may resultin an emission factor that may reasonably beapplied to other similar operating and similarlydesigned heaters.

Correlations

In some cases, many of the major design andoperating parameters can be input to equationsthat attempt to provide more representativeemission estimates. Theoretically, the more complexthe correlation and the more operating variables itincorporates, the more representative the emissionsestimate. This assumes that actual operating dataare used and not the model defaults.

Correlations can also be developed semi-empirically using discrete monitoring campaigns(e.g effect of load or fuel changes on NOxemissions from a heater). More simply, fuel sulphurcontent can be used to calculate SO2 emissions.

Correlations are widely used for estimating tankand wastewater treating emissions. As these

equations can be complex, they are typically usedas part of a computer model.

Another set of correlations are those for estimatingfugitive losses from piping components. In this case,measurements of local hydrocarbon concentrationsat each component are converted to an emissionrate. They are then aggregated to quantify the totalplant emissions.

Computer models

A wide range of computer software is availablewhich can be used to calculate almost all plantemissions as a labour-saving device. As withmanual approaches, the accuracy of the emissionestimate will improve as more source-specific inputdata is used.

The two most widely used emissions estimatingcomputer programs are those for atmosphericstorage tanks and wastewater treating. Versions ofthese are available from the US EPA (seeReferences). The manual calculation methods forestimating emissions from these two sources arevery tedious, and the use of computer models isrecommended. Although significant equipment-specific and operating input data are required, theemission estimating results are widely accepted byregulatory agencies.

Measurements

The most representative way to estimate emissions isby continuous monitoring of important parameters.This can be a combination of stack measurementusing in-situ continuous emission monitors (CEMs) ordiscrete sampling campaigns and monitoring of fuelconsumption from which flue gas volume flow atstandard dilution can be assessed. Continuousmonitoring of oxygen concentration is needed bothfor this step and for efficient combustion control.

CEM devices are useful for determining NOx, SO2,CO concentrations and for monitoring changes in

IPIECA

8

dust. Manual sampling is still needed forcalibration purposes, especially for dust where aCEM device cannot measure concentrationsdirectly. CEMS are best applied to the largestsources (e.g. combustion systems > 100 MWth) .

As described above, measurement can be combinedwith correlation techniques to parameterize theperformance of furnaces (e.g NOx emissions)where there are defined changes in, for example,load or fuel mix in the case of dual-fired systems.

It is important to recognize that continuousmonitoring is not synonymous with continuousmeasurement as not all inputs need to bedetermined with the same frequency in order tocalculate emissions.

Quality assurance

The inventory of emissions to air is a key componentof a refinery environmental management system(EMS). The support and active involvement of seniormanagement is needed to provide the resourcesfor the inventory activity and to ensure properevaluation and review of the results.

The principal quality assurance steps are to ensurethat the methodology used to quantify emissionsfrom each source is adequately documented andthat results are reviewed on a regular basis.Transparency is very important especially whereinventory results are used interactively in refinerymanagement, for example in verifying compliancewith refinery bubble limits or for demonstratingcontinuous improvement in reducing emissionswhich can assist decisions on the frequency of leakdetection and repair programmes.

Where specific inventory results are required forregulatory reporting purposes the EMS shouldensure that the internal methodologies areconsistent with reporting requirements.

In many refineries necessary data for the inventoryis gathered and held in the refinery data collectionsystem. Automated links to the data collectionsystem for such key data can usefully support theinventory effort.

Guidelines on auditing an inventory are givenbelow.

Good practices for emissionsinventory development

� Check that all emissions sources are included ininventory.

� Use the most appropriate estimating methodsand follow the application guidance.

� Collect representative equipment design andplant operating input data.

� Emphasize the need for inventory results thatare representative of operations.

� Ensure continuity of personnel skills, experienceand knowledge.

� Conduct an independent review of the inventorydevelopment and results.

� Address deficiencies found in review andconsider recommended improvements.

� Document all assumptions and methodologiesused.

Auditing an emissions inventory

The complexity of collecting operating data andusing various methods to obtain emissionsestimates introduces many opportunities forimprovements over time. Conducting a systematicaudit of the emissions inventory developmentprocess can identify potential improvement areas,check calculation methods, minimize errors andprovide recommendations for results that are morerepresentative of actual plant emissions.

Whenever possible, audits should be conductedby specialists with extensive experience in

9


applying and developing emissions estimatingtechniques. The more knowledgeable andexperienced the auditors, the more likely theresults will be meaningful. Audit teams shouldalso include plant personnel for training purposesas well as for their knowledge of the facility andcurrent practices.

Review procedures

The primary focus of an independent review is toconfirm the quality of the inventory and to identifyany errors or omissions in inventory development.Evaluating estimating methods and the input dataare essential parts of the review process. Duringthe review, all input data are checked forreasonableness.

The first step in reviewing the emissions inventory isidentifying how the inventory will be used. Often,there are several uses for the inventory includingregulatory reporting and corporate emissionstracking. Knowing the reasons that the inventorywas developed will help guide the reviewers inidentifying appropriate recommendations forimprovement.

Initially, a check of all potential emission sourcesconsistent with the emission inventory purpose ismade. All calculation models and factors used toestimate emissions are checked to confirm that theyare appropriate for representing the sources andare being used correctly.

All assumptions and input data should bethoroughly reviewed. The quality of the inventorywill depend on the quality of the specific plantoperating data. Checks should be made to makesure that all assumptions are reasonable and arefully documented. Improvements to improveaccuracy should be recommended.

Checklist

To ensure that all emission estimating proceduresare reviewed, a preliminary list of emissioninventory pollutants, sources and items to check isdeveloped. The source lists are the most criticalitems to develop correctly and sufficient time shouldbe allocated to making sure that all appropriatesources are included in the inventory.

Input data for calculating emissions from eachsource is checked with emphasis on themethodology used and the input data quality. Thevalidity of the detailed input data is checked andconfirmed to be representative of actual. Thisincludes a review of all the details of how the dataare used in obtaining an estimate of the emissions.

Documentation for all assumptions made tocomplete the inventory is confirmed. Improvementsto improve accuracy should be recommended.

Reporting results

Documentation of the results and recommendedimprovements is as important as doing a thoroughreview of the estimating procedures. The audit is oflimited value if the issues raised are not clear andthe plant is not able to implement therecommendations.

Audit findings will fall into two general areas: itemswhere there are errors that need to be corrected,and items where improvements may be made tomake the estimate more representative. Where thecurrent estimating procedure is adequate, qualityand accuracy may be improved and therecommended improvement(s) may be consideredfor use at the next emission inventory update.

Documentation should include the emission source,the issue that needs to be addressed and specificrecommendations on how to proceed with follow-up. The recommendations should have sufficientdetail so that plant personnel can implement them.

IPIECA

10

11


Sources and control of hydrocarbon emissions

The primary sources of hydrocarbon emissions areleaks from piping system components, evaporationfrom product loading, losses from atmosphericstorage tanks and evaporation from wastewatercollection and treatment. The relative emissionquantities from these sources might appear asprovided in Table 2.

This represents a refinery with good tankmanagement (appropriate storage of volatilematerial in floating roof tanks, appropriatelyequipped tanks) and avoiding unnecessarydischarges of hydrocarbons to the wastewatertreatment system. Adding vapour balancing andvapour recovery systems for product loading cansignificantly reduce this contribution. Fugitiveemissions from equipment leaks present acontinual challenge.

Fugitives and piping systems

Refineries typically contain hundreds of thousandsof piping components such as valves, connectors,flanges, pumps and compressors. Each of thesehas the potential for the process fluid to escapearound the seal into the environment. While thequantity of emissions from each individualcomponent is usually very small, the large numberof components in a refinery may make fugitiveemissions the largest aggregate source ofhydrocarbon emissions.

Studies have found that while almost everycomponent has a very small leak rate, more than80% of emissions typically come from a smallpopulation of the components that are considered‘high’ leakers. Finding and fixing these larger leaksshould be a priority and is the driver for a leakdetection and repair programme.

Leaks are not usually visible. They have typicallybeen found through the use of sensitive gassampling devices to ‘sniff’ for ppm concentrations

on the piping component. As the ‘sniffer’ has to bevery close to the leak site this is labour-intensiveprocess. New optical gas imaging equipment canvisualize leaks and make detection simpler andmuch more cost-effective. These techniques arediscussed later.

Because fugitive piping system emissions are apotential large contributor to refinery hydrocarbonemissions, a number of controls have beendeveloped and successfully applied. These fall intothree general areas: improved seals; improvedmaterials and metallurgy; and finding and repairingthe large leakers. Some trade-offs can be madebetween these. For instance, using better designsand equipment can reduce maintenance costs.However, all successful fugitive control programmeswill include some monitoring and repair.

Table 3 lists the most common controls for fugitiveemissions and their relative costs.

These controls are discussed in more detail in thefollowing sections. The most effective results areobtained when several control methods are applied.For example, if improved valve packing and pumpseals are installed, the monitoring and repairprogramme can be conducted more cost-effectively.If low emission control valves with dual packingsets are installed, then leak monitoring of thesecomponents can be done much less frequently.

Table 2 Relative emission contribution for hydrocarbons

Source Relative %

40–50

30–40

10–15

10–15

Fugitive equipment leaks

Product loading*

Storage tanks

Wastewater collection and treatment

*Without vapour control

IPIECA

12

How to quantify emissions

The quantity of fugitive emissions is obtained bydetermining the emission from each piping systemcomponent in the refinery and summing theseemissions to obtain the refinery total. There aremany ways to determine the individual componentemission rates. The simplest, and potentially leastrepresentative or least accurate, is to use industryaverage emission factors for each component type.If a periodic monitoring and component repairprogramme is conducted, a reduction of 75% forcontrol efficiency can be applied to this number. If amore representative and accurate estimate offugitive emissions is desired, the ppm readings fromthe monitoring programme gas detection instrumentcan be used in correlation equations to calculate themass emission rate for each component. There arefinite leak rates generally applied even when thedetection instrument reads zero for the backgroundconcentration. There are numerous publications thatprovide guidance for estimating fugitive emissions,including the ‘1995 EPA Protocol’ (US EPA, 1995a)and a calculation manual from the AmericanPetroleum Institute (API, 1998b).

Open-ended lines

Open-ended lines—pipelines with a single valvepreventing loss of fluid to the environment—shouldbe avoided.

The recommended control for open-ended lines isto use a second valve, a plug or a cap at the endof the line. Valves on small bore sampling linesshould be maintained.

Pump, compressor and valve stemsealing

In pumps, compressors and rising stem valves,there are shafts that pass through the device,between areas containing pressurized process fluidand the surrounding environment. These provide apotential path for process fluid to leak from thepump, compressor or valve. Various seals are usedto minimize the quantity of leakage. A properchoice of sealing system can significantly reducepotential emissions. Numerous vendors can providedesigns with excellent sealing performance. Use ofsuperior sealing systems will often reduce fieldemissions control maintenance costs.

Pumps using mechanical seals may be of a single-seal or multi-seal design. The choice of design willdepend on the specific gravity of the process fluidand on the desired level of emissions control.Design selection may sometimes be balancedagainst the cost of an emissions monitoringprogramme. The seals incorporate both rigid andflexible elements that maintain firm contact at thesealing interface, allowing the rotating shaft to passthrough a sealed case while minimizing leakage of

Table 3 Controls for reducing fugitive emissions

Emission control Relative cost

Low/medium

Low

Low

Low

Medium

Medium

Medium

High

Initiate a component leak detection and repair (LDAR) programme

Install improved packing in block valves

Optimize valve stuffing box and stem finishes

Install second valve, cap or plug on open-ended lines

Use low emission type control valves

Upgrade pump seals

Use low emission quarter-turn valves

Use leakless technology (bellows valves; canned and magnetic drive pumps)

13


the process fluid. The elements can be bothhydraulically and mechanically loaded with aspring or other device to maintain firm contact withthe rotating shaft.

A single mechanical-seal pump is the mosteconomical choice and can often provideadequate emissions control provided that the sealface design and materials are appropriatelychosen. Seal face materials should have a highmodulus of elasticity, superior heat transferproperties and a low coefficient of friction. Sinceseals use the process fluid to lubricate the sealfaces, there is potential for emissions of theprocess fluid. A single mechanical seal can alsoinclude a closed vent system that captures anyleaking process fluid and returns it to the processor to a control device.

Dual mechanical seals provide excellent controlperformance with near zero emissions. There aretwo basic types of dual-seal systems: double-sealand tandem-seal systems. In a double-sealarrangement, a non-regulated barrier fluidbetween the seals is at a higher pressure than theprocess pressure. Leaks of process fluid into thebarrier fluid are, therefore, prevented. In atandem-seal arrangement, a non-pressured barrierfluid is used and, although process fluid can leakinto the seal fluid, a collection system can beincorporated to remove and capture any processfluid that leaks.

Emission controls for centrifugal compressorsrequire the use of mechanical seals equipped witha barrier fluid and controlled degassing vents orenclosure of the compressor seal and venting ofleakage emissions to a control device. Seal designscan be labyrinth, carbon ring, bushing,circumferential or face seals. Combinations of sealtypes in a single compressor are typical. Sealsystems can use liquid buffer fluids (wet seals) orgas buffer fluids (dry seals). With oil wet seals,there is usually a need for systems to remove thebarrier oil from the process gas.

A labyrinth seal design incorporates a complexpath for the process fluid, making it difficult for thefluid to pass through and thus creating a barrier tohelp prevent leakage. Such a design typicallyincludes multiple paths or grooves spaced tightly sothat there is high resistance against escape of thefluid. To be effective, very small clearances arerequired between the labyrinth and the runningsurface. Labyrinth seals on rotating shafts provide anon-contact sealing action by controlling thepassage of fluid through a variety of chambers bycentrifugal motion. At higher speeds, centrifugalmotion forces the liquid towards the outside andtherefore away from the passages. Process gas istrapped in the labyrinth chamber preventing itsescape. When leakage of process gas must beprevented, a buffer fluid is injected between thelabyrinths. Labyrinth seals are often utilized as endseals with other mechanical seal designs. Overtime, the emissions control effectiveness of alabyrinth seal may decrease due to wear andchanges in spacing alignment.

Other seal designs are generally applicable tohigher pressure applications than labyrinth designs.A buffer fluid is injected between the ring sets toprevent leakage. Leakage is dependent on sealsize, compressor speed and process pressure.These seals use a fluid buffer which may leak intothe process gas and also into the environment.Systems may include automatic shutdown if thebuffer fluid pressure is lost.

Controlling emissions from reciprocatingcompressors requires minimization of gas leakagealong the cylinder rod. This may be accomplishedusing appropriate packing systems on the rod andpressurizing the packing box.

Pump and compressor seal designs should bespecified by the plant rotating equipment specialistafter consultation with the plant environmental staff.Vendor reliability and experience with low emissionrequirements is critical.

IPIECA

14

There is a wide variety of packing designs andmaterials available to control leakage along avalve stem. Packing is installed in a stuffing boxsurrounding the valve stem and maintained undermechanical pressure to prevent the escape ofprocess fluid along the stem or through the stuffingbox. The mechanical pressure is provided by ascrew or nut forcing a flange to compress thepacking. Newer packing materials are typicallygraphite or polymeric. The polymeric materialsoften provide better emissions control performancebut may not pass fire safety testing requirements.

Some valve packing is appropriate for factoryinstallation in new equipment, and some is moreappropriate for field packing replacement.Typically, preformed solid ring packing is forfactory installation and continuous spool packing,cut in the field, is typical for repairs. Somepreformed ring packing is provided pre-cut or canbe field cut for repair applications. Somemanufacturers may provide unique shapes to apacking in an attempt to improved emissionscontrol performance.

For rising stem block valves, a basic packing set,consisting of three die-formed graphite sealingrings with two braided end rings to preventpacking extrusion, has been shown to providegood emissions control performance. Somemanufacturers have incorporated the performanceof both sealing rings and end rings into a spool-type packing for field repairs.

Use of more than five rings does not typicallyimprove emissions control performance and may,in fact, reduce the pressure on some of the sealingrings allowing higher emission rates through thestuffing box. Some old valves may have very deepstuffing boxes allowing many extra packing rings.Spacers should be used in these to reduce thenumber of packing rings required to no more thanfive to seven.

In applications where valves are cycled frequently,such as control valves, dual packing sets with leakdetection between the packing sets will providebetter emissions control. In addition, ‘live loading’using springs may be utilized to maintain constantpressure on the stuffing box.

Valve leakage can often be eliminated bytightening the screws or nuts on the flange toincrease pressure on the packing in the stuffingbox. Care should be taken so that the screws arenot tightened to the point that the valve becomesinoperable. When tightening screws or bolts nolonger reduces emissions, it is usually a sign thatthe packing or valve needs to be replaced.

Enhanced sealing techniques

In some situations, the leak may be repaired byinjecting a sealing liquid directly into the stuffingbox. This technique may be useful for emissionscontrol if the leak is large and the valve cannot beremoved from service for repacking or repair. Useof this technique should be done after technicalevaluation as the technique may cause damage tothe stuffing box and an additional path foremissions, and is not appropriate for all valves,valve types or service (e.g. valves that are likely tosee more than occasional usage).

Quarter-turn valves typically provide loweremissions and maintenance compared to risingstem valves. These types of valves have beenapplied more in chemical plants than refineries.Prior to using this type of design, the plantmechanical equipment specialist should be involvedin discussions with the vendor.

Most valve and packing suppliers will be able toprovide results from testing their products for lowemissions. There are several tests available andcomparison between vendors may be difficult.Many vendors offer guarantees for various leaklevels. What they are really offering is a lowerprobability that, over time, the valve will leak. It is

15


sometimes advantageous to purchase a betterperforming valve and packing system to reduce theneed for costly field maintenance later.

Valve packing should be specified by the plantmechanical equipment specialist after consultation withthe plant environmental staff. Vendor reliability andexperience with low emission requirements is critical.

Valve quality: materials and finishes

In rising stem block valves, as the stem risesthrough the packing, there is potential for the stemto cause damage to the packing and hence createa path for increased emissions. The stem must bemaintained in a clean and good condition tominimize this damage. The stuffing box finish mustalso be addressed as the packing can be damagedby a rough surface as it is lowered into the box,possibly creating a path for process fluid leakage.

To reduce the likelihood of packing damage as thevalve stem is raised and lowered, it is important tokeep the stem clean, straight and corrosion free.Choosing stem materials appropriate for theprocess application will help reduce corrosion. It istypical to find leaks from valves with corroded ordamaged stems.

Stem and stuffing box finish is also important asthere is a balance between packing damage as thestem is moved or the packing is installed and theability of the packing to seal against the walls ofthe stuffing box and the stem. Too smooth a finishmay not necessarily be beneficial. Material andfinish should be selected after discussion with theplant mechanical equipment specialist and thevalve and packing supplier.

Valve stems should be kept clean to avoid damageto the packing as the valve is operated. Cleaningwith a dry soft cloth is recommended before thevalve is turned. Use of grease on valve stems is notrecommended since it may attract debris and resultin packing damage.

‘Leakless’ components

In general, use of good seals and componentdesigns in combination with a periodic leakdetection and repair programme can provideemissions control almost equivalent to that of‘leakless’ designs. The significant increase in coststo apply ‘leakless’ equipment is normally notwarranted. In addition, the failure modes of‘leakless’ designs can result in significant releasesof process fluid, making them somewhat lesseffective in overall emissions control.

Leakless components are those that do notincorporate any leak paths between the processfluid and the environment. Seal-less pumps aredesigned without a shaft penetrating the pumphousing. These may be diaphragm, canned ormagnetic drive designs. Bellows seal valves have awelded sealed bellows between the process fluidand the environment to prevent emissions.

Even ‘leakless’ components can fail, and a meansof monitoring is usually provided to detect suchfailure. In diaphragm pumps, holes may develop inthe diaphragm. In canned or magnetic drivepumps, the casing may develop leaks. In bellowsseal valves, the bellows may crack or the edge mayseparate allowing leakage of fluid. On bellows sealvalves, a back-up packing system is usuallyinstalled to address this failure. Although in manylocations emissions from components with ‘leakless’design are assumed to be zero, in some locations afinite leak rate, usually equal to that from anuncontrolled flange, is applied.

Leakless technology should be considered inapplications dealing with highly toxic process fluidsor if there is a potential for release of highlyodorous materials. The need for mitigationmeasures in the event of seal failure should beconsidered in these cases.

IPIECA

16

Leak detection and repair

The most effective fugitive emission control methodis to conduct periodic surveys to find and repairleaking components. These surveys are commonlyreferred to as ‘leak detection and repair’ (LDAR),‘monitoring and maintenance’ (M&M) or‘inspection and maintenance’ (I&M) programmes.Each of these has two parts. The first part is to findthe leaking components. The second part is torepair or replace the leaking components so thatthey are no longer hydrocarbon emission sources.

Even with the use of excellent sealing equipment,there will be some, but perhaps fewer, leakingcomponents, and a monitoring programme willidentify these for repair. Emission reductions of50–90% have been demonstrated by LDARprogrammes and, in some cases, the cost of theprogramme is more than compensated for by thevalue of the material no longer emitted from theleaking components.

Fugitive leaks occur randomly, and it is essentiallyimpossible to predict which specific componentswill leak. Therefore, all components selected forinclusion in an inspection programme need to bemonitored. The critical parameters in conducting anLDAR programme are the choice of components toinclude, the frequency of monitoring and the leaklevel above which component repair is required.There is also an option to apply optical gasimaging which is a more cost-effective monitoringmethodology than the traditional ‘sniffing’procedure (see below).

It is not necessary to include all component types inthe monitoring programme. Emissions fromcomponents in heavy liquid service (kerosene andheavier) have been found to leak much less thancomponents in gas or light liquid service and are,therefore, usually excluded from LDARprogrammes. It is not economically justifiable tomonitor these heavy liquid components because ofthe very small emission reduction that can be

achieved. Also, many LDAR programmes do notinclude flanges since their low relative leak rateand high number make them uneconomic tomonitor. However, once LDAR has been applied toother components such as valves, open-ended-lines,pumps and compressors, leaks from flangesbecome a much larger fraction of the remainingfugitive emissions, and including them in the LDARprogramme, at longer time intervals, may becomejustified if further emission reductions are required.

The sooner a leak is found and repaired, the lessprocess fluid will enter the environment. There is abalance, however, between the cost of morefrequent monitoring and the value of the materiallost or its impact on the environment. Many LDARprogrammes are conducted annually. In somelocations, however, there is a requirement tomonitor more frequently, especially when there arehigh percentages of leaking components.Sometimes, quarterly monitoring is required if morethan 2% of components are leaking. However, thereis also the opportunity to monitor less frequently ifthe percentage of leaking components is lower.Therefore, there is an incentive to use componentswhich are of high quality or improved design toachieve lower leak percentages, and hence beallowed to monitor less frequently.

The most widely used monitoring method is theUS EPA Reference Method 21. This is known as‘sniffing’ and uses a sensitive gas-samplinginstrument to measure the concentration ofhydrocarbon adjacent to a potentially leakingcomponent. Each component is monitoredindividually, as shown in Figure 1.

Guidelines for conducting Method 21 monitoringhave been developed by the American PetroleumInstitute (API, 1998a).

If the measured gas concentration is above acertain threshold, the component is considered a‘leaker’. This concentration was originally set at10,000 ppm. Since the major contribution to

17


fugitive emissions is from the high leakers, setting alower leak level for repair is not as good anemissions reduction approach as is finding andrepairing the large leakers sooner.

If starting a new Method 21-based programme,annual monitoring of valves, pumps, compressorsand open-ended lines in gas and light liquid serviceis recommended with a leak definition for repair of10,000 ppmv. Including more components,conducting more frequent monitoring and loweringleak definitions for repair can be incorporated ifadditional fugitive emissions reduction is required.

With Method 21, each component must bemonitored individually, so it is a very manpower-intensive activity. The process involves placing theprobe of a hydrocarbon detection instrument at thepotential leak surface of the component. Air andany leaked hydrocarbon are drawn into the probeand passed through a detector (flame ionization isthe most widely used type of detector).

The instrument measurement in ppmv is correlatedto the mass emission rate from the component, but

this is a relatively poor correlation. In practice,some large leaks may give lower relative readingsand some small leaks may give higher relativereadings depending on the nature of the leak.These are termed false negatives and falsepositives when they have an impact on repairdecisions, and can result in the misapplication ofrepair activities.

The majority of fugitive emissions—typically morethan 80%—come from a very small fraction ofcomponents with relatively high leak rates. Sincemost components do not leak at concentrationshigh enough to require a repair, most of the effortassociated with Method 21 ‘sniffing’ is spentmonitoring the non-leaking components.

A new method of component monitoring whichuses optical gas imaging to detect leaks has beensuccessfully applied at refineries and chemicalplants around the world. Use of this technique isshown in Figure 2.

Optical gas imaging allows an instrument operatorto easily view all components and detect leaking

Figure 2 Leak detection: optical gas imaging Figure 1 Leak detection: US EPA Reference Method 21 The most widelyused monitoringmethod is theUS EPA ReferenceMethod 21, alsoknown as ‘sniffing’(Figure 1), whichuses a gas-sensitiveinstrument tomeasure theconcentration ofhydrocarbonadjacent to apotentially leakingcomponent.

Optical gasimaging (Figure 2)enables the operatorto visually detectleaking hydrocarbongas, and allowsleaks to be identifiedmore quickly and atlower cost than the‘sniffing’ method.

IPIECA

18

Figure 3 A leaking valve, viewed using optical gas imaging equipment

hydrocarbon gas in a real-time video image. Usingthis equipment, components may be viewed asshown in Figure 3, and leaks identified morequickly and at lower cost compared to using the‘sniffing’ method.

The remote sensing and instantaneous detectioncapability of optical gas imaging allows an operatorto monitor larger areas of a process unit much moreefficiently, eliminating the need to measure thehydrocarbon concentration at each individualcomponent. When using optical gas imaging to findleaks, all components showing evidence ofhydrocarbon leakage are scheduled for repair.

The initial repair for valves found to be leaking isto tighten the packing gland to further compress thepacking and seal the leak path. At locations thatare just starting an LDAR programme, thistechnique has a very high success rate. If the glandtightening is not successful, then the next time thevalve is out of service, the packing should bereplaced with a new low-emission packing chosenafter consultation with the plant mechanicalequipment specialist and the packing vendor.

Flange repairs involve retightening of the bolts andreplacement of the gasket when next removed fromservice. Pump and compressor repair should becoordinated with the plant machinery specialist.Equipment should be monitored after repair toensure that the repair was effective in stopping thehydrocarbon leak.

Good practices for control of fugitiveemissions

� Use low-leak multi-seal arrangements for pumpsand compressors.

� Use low-leak dual-seal designed control valves.� Use low-leak block valve packing and keep stem

clean.� Consider use of quarter-turn valves where

appropriate.� Install a second valve, a plug or a cap on all

open-ended lines.� Using available techniques such as the optical

gas imaging camera in combination with‘sniffing’ according to Method 21, performannual leak detection and repair on gas andlight liquid valves, pumps, compressors andopen-ended lines.

� Repair or replace leaking components.

Storage tanks

Atmospheric storage tanks are utilized in a refineryfor a variety of hydrocarbon liquids including crudeoils prior to processing, products waiting forshipment and intermediate streams. There are twogeneral types of atmospheric storage tanks: fixedroof tanks and floating roof tanks. There are threetypes of floating roof tanks: external floating roof,internal floating roof and covered (or domed)floating roof. Typically, lower vapour pressureliquids such as heating oils and kerosene arestored in fixed roof tanks. Crude oils and lighterproducts such as gasoline are stored in floatingroof tanks.

19


A fixed roof tank consists of a shell and a fixedroof with a gas space above the liquid surface,which is vented to the atmosphere through apressure relief device. Some of the hydrocarbonliquid in the tank evaporates into the gas spaceand, when the tank is filled and the gas isexpelled through the pressure relief device, thisvaporized hydrocarbon is emitted. This is called‘filling loss’. A small amount of gas is alsoreleased due to daily changes in atmosphericpressure and temperature. This is called‘breathing loss’ or ‘standing loss’. Typically, fillinglosses constitute 80–90% of the total losses forfixed roof tanks.

Floating roof tanks consist of a shell and a roof thatfloats on the hydrocarbon liquid. In the case of anexternal floating roof, the top of the floating roof isopen to the environment. In the case of an internalor covered floating roof, there is a gas spacebetween the floating roof and the roof on the top ofthe tank. The internal floating roof and coveredfloating roof tanks resemble a fixed roof tank witha floating roof placed internally on top of thehydrocarbon liquid.

In floating roof tanks there is a rim seal thatreduces the quantity of hydrocarbon vapourspassing through the space between the floatingroof and the shell. There are also a number of roof‘fittings’, which are openings in the floating roof,that provide for inspection and maintenance aswell as sampling of the liquid.

With floating roof tanks, the hydrocarbon liquidevaporates and vapours can pass around thefloating roof rim seal and also around openings forfittings in the floating roof. This is called ‘standingloss’. In addition, a small amount of material cancoat the shell and any vertical poles when the tankroof is lowered. This material evaporates and iscalled ‘withdrawal loss’. The quantity of loss forfloating roof tanks depends on the rim seal designand emission controls on the roof fittings.

Emissions from internal and covered floating roofsare much lower than for external floating roofs dueto the elimination of wind driven pressuredifferences across the roof. Most of the emissionsfrom floating roof tanks are due to standing losses.

Table 4 describes the most common controls forreducing tank emissions and their relative costs. Forfixed roof tanks, the primary focus is on thecollection of hydrocarbon vapours that are expelledwhen the tank is being filled. A standard approachis known as ‘vapour balancing’, where the vapourexiting the tank is sent to the space created wherethe liquid is coming from. This works well if theliquid is being offloaded from a nearby vessel,truck or another fixed roof tank. There are vapourtransporting and safety issues that need to beaddressed with this control option. However,vapour balancing can work well if the receivingvessel is situated close enough that costs for thenecessary ducting and blowers are reasonable.

Vapours expelled from a fixed roof tank can alsobe collected for recovery or destroyed. Recovery isgenerally only used for very high value productsand its application has typically not been foremissions control purposes. Recovery anddestruction are the most costly controls and arediscussed in more detail in the section on Productloading (page 26).

If the emissions from a fixed roof tank aresignificant, the material might be better stored in afloating roof tank. If a floating roof tank alreadyexists, costs may be moderate depending onavailable piping and current use of the floatingroof tank. Alternatively, the fixed roof tank can beconverted into an internal floating roof tank, butcosts to do this are relatively high.

In floating roof tanks, emissions are mostly due tostanding losses which come from vapour passingthe rims and roof fittings. A first step in emissionreduction is to ensure that the controls on these arein good condition. Roof fitting gaskets and wipers

IPIECA

20

should be checked to ensure that they are in goodcondition and are providing a proper vapour seal.The rim seals should be inspected for excessivegaps. If none exist, a secondary rim seal can beinstalled to reduce the vapour losses across theprimary seal. If a vapour mounted primary seal isbeing used, this can be changed to a mechanicalshoe primary seal with a secondary seal. Thiscombination will provide excellent vapour controlperformance for the rim emissions.

If additional emissions reduction is needed,external floating roof tanks can be converted tocovered floating roof tanks, which will eliminate thewind driven emissions. This option is relativelyexpensive but is sometimes justified by productcontamination issues (e.g. eliminating rainwater) inaddition to emissions reduction needs.

In extreme circumstances, usually for very odorousor toxic liquids, an internal floating roof tank mayrequire collection of the vapours and use of vapourrecovery or destruction. However, in these cases,use of a closed pressurized vessel may be moreappropriate than an atmospheric storage tank.

The controls mentioned above are discussed inmore detail in the follow sections. Options shouldbe reviewed with the site tank specialist andvendors should be contacted to discuss locallyavailable options and equipment. The mosteffective results for floating roof tanks are obtainedwhen several of the controls are applied. Forexample, when both improved rim seals are usedalong with gaskets and bolts on roof fittings.

Table 4 Controls to reduce storage tank emissions

Emission controlTank type Relative cost

Medium

Site specific

High

Very high

Very high

Low

Low

Medium

High

High

Low

Low

Medium

High

Very high

Very high

Fixed roof

External floating roof

Internal floating roof

Install vapour balance system

Use existing floating roof tank

Install internal floating roof

Apply vapour destruction

Apply vapour recovery

Check and repair roof fitting gaskets

Check and repair existing rim seals

Install secondary rim seal

Change rim seal to mechanical shoe seal

Convert to covered floating roof tank

Check and repair roof fitting gaskets

Check and repair existing rim seals

Install secondary rim seal

Change rim seal to mechanical shoe seal



21



The methodology for estimating tank emissions iscomplex. A set of semi-empirical equations basedon laboratory tests on different seals and fittingshas been developed by the American PetroleumInstitute (API, 2002/03) and has been adopted bythe US Environmental Protection Agency (US EPA,1995b). Use of these equations for estimating tankemissions requires many inputs including the tanktype, details of design, construction and operationand properties of the stored hydrocarbon liquid.

Typically, a spreadsheet is developed or a standardcomputer program such as the EPA’s Tanks(US EPA, 2010) is used for the calculation.Hydrocarbon emissions from atypical operationssuch as floating roof landings and openings fortank cleaning also need to be included.

Tank types: fixed and floating

The design and emissions mechanism differences offixed and floating roof tanks were discussed above.The floating roof can be an emission control for thefixed roof tank design. It reduces contact of thehydrocarbon liquid with the gas which is thenexpelled. The gas has a lower concentration ofhydrocarbon vapour since it is not in constantcontact with the liquid. In many locations, highervolatility liquids such as crude oil and gasoline mustbe stored in floating roof tanks to reduce emissions.

There are generally two types of floating rooftanks: internal floating roof and external floatingroof. An internal floating roof tank is similar to afixed roof tank with the placement of a floating roofinside. The external floating roof tank has the roofsubject to the environment; to wind and rain.Hydrocarbon emissions from an internal floatingroof tank are usually much lower because the wind-driven evaporation is limited by the fixed roof.

Sometimes, internal floating roof tanks aredistinguished between internal floating roof and

covered floating roof. The internal floating roofthen refers to tanks that were originally designedas internal floating roof tanks, often with lessconcern for losses from rim seals and roof fittingsdue to the expected presence of the fixed roof onthe original design. They typically have riveteddeck seams, no secondary rim seal and lesscontrol on the deck fittings.

A covered floating roof tank often refers to a tankthat was originally designed as an external floatingroof tank that then had a fix roof installed. Thefloating roof construction is often quite different asthe deck seams are usually welded rather thanbolted and better seals are placed on the rim androof fittings.

Floating roof rim seals

Floating roofs are designed to have an annularspace between the perimeter of the floating roof andthe tank shell to allow easy vertical movement of theroof as liquid is added or removed. As a fully openspace would allow significant evaporation of liquid,the annular space is closed using a rim seal system.

There are many types of rim seal combinations andsome unique vendor designs. Effective rim sealsystems provide good closure of the annular space,accommodate irregularities in the tank shell and helpthe floating roof stay centered in the tank whileallowing easy vertical movement of the floating roof.

Rim seal systems can consist of a primary rim sealand a secondary rim seal. For most internal floatingroof tanks, a secondary rim seal is usually notnecessary because the fixed or domed roof limitsevaporation caused by the wind. For externalfloating roof tanks, secondary rim seals are usuallyrecommended, depending on the volatility of theliquid stored.

There are three general types of primary rim seals:vapour-mounted, liquid-mounted, and mechanicalshoe. Vapour-mounted and liquid-mounted primary

IPIECA

22

seals are typically made of non-metallic materialsand are often foam filled. They resemble asausage-shaped tube or envelope that it fastenedaround the outside circumference of the floatingroof. Vapour-mounted primary seals have a vapourspace between the liquid and the bottom of theseal. In liquid-mounted primary seals, the bottom ofthe seal touches the liquid. Both vapour-mountedand liquid-mounted non-metallic seals arevulnerable to damage from rivet heads and weldburs on the tank shell as the roof moves up anddown, which can tear the fabric.

Liquid-mounted primary seals provide a muchbetter emission control compared to vapour-mounted primary seals because the vapour spacebetween the seal and the liquid surface isminimized. However, when torn, they easilybecome contaminated with liquid seeping into theinterior of the seal. Therefore, it may be advisableto avoid the use of liquid-mounted primary seals soas not to have to deal with the contaminated sealwhen replacement is required.

A mechanical shoe primary seal uses light gaugemetallic sheets that are formed together as a ringcontacting the tank shell. These sheets are most oftenheld against the shell by weights or springs attachedto the floating roof. A seal fabric is connectedbetween the top of the metal band and the floatingroof to prevent emission of the evaporated liquidvapours contained above the surface of the storedliquid and below the fabric seal.

Mechanical shoe seals generally have a longservice life and are not subject to the materialintegrity issues associated with non-metallic liquid-and vapour-mounted fabric seals. In addition,when paired with a secondary rim seal, mechanicalshoe seals provide excellent emissions controlperformance. API has evaluated the relativeemissions control of different rim seal combinationsand provides detailed descriptions of their designcharacteristics (API, 2002/03).

Tables 5 and 6 provide comparisons of controlefficiencies for different rim seal configurations. Forexternal floating roof tanks, Table 5 shows thepercent reduction in emissions from a single vapourmounted seal as a secondary seal is added or theseal is replaced with a mechanical shoe typeprimary seal and then a secondary seal is added.The table shows the superior performance of themechanical shoe seal in reducing rim losses.

The mechanical shoe primary seal with asecondary seal is considered best technology forstoring typical volatile hydrocarbons in externalfloating roof tanks.

For internal floating roof tanks, Table 6 on thefollowing page shows the percent reduction from asingle vapour mounted seal as a secondary seal isadded or the seal is replaced with a mechanicalshoe type primary seal and then a secondary seal isadded. Similar to external floating roof tanks, use ofa secondary seal or changing to a mechanical shoe

Table 5 Seal system impact on emissions from external floating roof tanks

Seal system configuration Approximate control efficiency* (%)

BASE

60 – 70

80 – 90

90 – 95

95 – 99

Vapour mounted resilient primary rim seal

Vapour mounted primary rim seal with a secondary seal

Mechanical shoe primary rim seal

Mechanical shoe primary rim seal with a wiper seal

Mechanical shoe primary rim seal with a secondary seal

* Control efficiency is dependent on the size of the tank, the properties of store material, meteorological conditions and throughput.

23


primary seal will result in lower emissions but thereduction will be relatively less because the fixedroof already provides significant emissions control.

Emissions for internal floating roof tanks arealready lowered significantly by the fixed roof,hence rim seal improvements may not provide cost-effective reductions of overall tank emissions. Inmany cases, a vapour mounted primary sealprovides adequate emissions control for an internalor covered floating roof tank.

To ensure good emissions control, it is importantthat, whichever rim seal system is used, it providesan effective closure of the annular space betweenthe floating roof and the tank shell. Many locationsrequire periodic inspection of these seals. Due toaccess constraints, inspections of internal floatingroof tank seals are usually done visually rather thanwith hands-on physical inspection.

For internal floating roof tanks, the seals may beinspected through a hatch opening in the fixedroof. For external floating roof tanks, inspectionmay include measurement of gaps between the sealand the tank shell. Excessive gaps will result inhigher emissions and will need to be repaired.

Roof fittings: gasketing and slottedguidepoles

There are numerous fittings that are attached to orpass through the floating roof. These allow forsampling, inspection and maintenance hatches and

for support and positioning columns. When fittingsrequire an opening in the floating roof, they becomea potential source for evaporative emissions.

There are two general types of fittings. Hatchesallow access to the liquid below the deck forsampling of the liquid and for measuring level.Larger hatches allow access for maintenancepersonnel. Columns and guidepoles providesupport for a fixed roof on internal floating rooftanks and prevent rotation of the floating roof as itmoves up and down. In some cases, the columnsmay also be used for gauging and sampling.

To minimize evaporative losses past hatches, agasket can be placed around the hatch rim toprovide a seal, and the hatch cover can be latchedor bolted shut when not in use. For columns andpoles, the annular opening between the pole andthe floating roof needs to be sealed to preventevaporative emissions. This can be done with afabric and rubber wiper arrangement that restrictsvapour passage and wipes liquid hydrocarbon offthe pole as the roof is lowered. These seals andwiper systems are available from many tank vendors.

Guidepoles come in two types: slotted and un-slotted. Unslotted guidepoles have openingsallowing fluid to pass only near the bottom of thepole. There is concern that liquid samples takenthrough these poles are not representative of theentire tank contents. For this reason, APIrecommends the use of a ‘slotted’ guidepole forproper sampling and gauging.

Table 6 Seal system impact on emissions from internal floating roof tanks

Seal system configuration Approximate control efficiency* (%)

BASE

50 – 60

60 – 70

70 – 80

Vapour mounted resilient primary rim seal

Mechanical shoe primary rim seal

Vapour mounted primary rim seal with a secondary seal

Mechanical Shoe primary rim seal with a secondary seal

* Control efficiency is dependent on the size of the tank, the properties of store material, meteorological conditions and throughput.

IPIECA

24

While the slottedguidepole designhas advantages forsampling andgauging, it providesadditional pathwaysfor evaporativeemissions.

Figure 5 providesan example of asleeve placedaround the slottedguidepole (thedashed linesindicate analternative locationfor attachment ofthe pole sleeve).

In a slotted guidepole, there are holes or ‘slots’along the entire pipe which allows liquid to freelyflow in and out. While the slotted guidepoledesign has advantages for sampling andgauging, it provides additional pathways forevaporative emissions; air from above the roofcan enter and leave the region below the deckthrough the openings (see Figure 4).

In external floating roof tanks, an uncontrolledslotted guidepole can be a significant source ofemissions for lighter hydrocarbons. For this reason,it is recommended that consideration be given toplacing a sleeve around the slotted guidepole in theregion where it passes through the floating roof(see Figure 5). The sleeve should cover all the holesin the guidepole from just above the deck to belowthe liquid surface.

Gasketing and wipers should be installed to closethe annular opening to prevent evaporation andminimize liquid on the pole as the floating roof islowered. In some cases, the cost of installing thesleeve can be completely offset by the value of thereduced product emissions.

Roof landings

Standard operation of floating roof tanks assumesthat there is continuous contact of the floating roofwith the liquid below the floating roof.

As material is removed from the tank and the floatingroof lowered, the floating roof reaches a level whereit becomes supported on roof or deck legs whichprevent it from moving any lower. This preventsdamage to equipment inside the lower part of thetank, or to deck fittings penetrating below the floatingroof. Once the floating roof reaches this level, furtherwithdrawal of liquid causes atmospheric vents toopen automatically to avoid excessive vacuum insidethe space below the floating roof. At this point, thevapour space under the floating roof is freely ventedto the environment above the floating roof, allowinga significant increase in hydrocarbon emissions.

While the floating roof is on its legs and thevacuum breaker vents are open, any liquid thatremains in the tank can evaporate, as can anymaterial clinging to the tank walls and poles. Inaddition, emissions will occur as the tank is refilledcausing the vapour below the floating roof to beexpelled through the open vents until the floatingroof is refloated by the rising liquid.

Figure 4 Air flow across a slotted guidepolepromotes evaporation

Figure 5 A sleeve placed around a slottedguidepole eliminates air flow through the slots

25


The quantity of hydrocarbon emissions due to aroof landing depends primarily on the elapsed timeof each operation, the quantity of material thatremains in the tank while the roof is landed on itsdeck legs and the vapour pressure of the liquid. Inaddition, if the tank is drained, the degree ofsaturation of the remaining gas under the roof hasa significant impact. The degree of saturationdepends on the design of the tank bottom and howcompletely the remaining liquid is drained. ‘Draindry’ tanks will have lower emissions than tanks witha liquid heel because, in addition to the liquid onthe walls and poles that evaporates, the materialremaining in the heel will evaporate and be emittedas long as the roof remains landed on its legs.

The primary control to reduce these emissions is toavoid all unnecessary roof landings. If roof landingsare necessary to prepare the tank for repair or tochange the liquid that is stored, the liquid should bedrained as quickly as possible and as completely aspossible. Minimizing the elapsed time that the roofremains landed on the deck legs with hydrocarbonliquid present below it will reduce the standinglosses. In all cases, vapours will be expelled as thetank is re-filled; collection of these vapours isdifficult as there are multiple vents, and access ontothe floating roof is not always possible.

Details of the potential loss mechanisms wereexplored, and methodology for estimatingemissions from landing roofs developed, by theAmerican Petroleum Institute (API, 2005).

Cleaning operations

Cleaning and maintenance operations on storagetanks are typically unique to the site, tank andspecific event. Many steps are usually involved andnot all may occur during a specific cleaning ormaintenance event. The steps in preparing a tankfor cleaning or maintenance most often includeemptying of the hydrocarbon liquid from the tank,removing any of the remaining liquid as best as ispossible, purging the tank of hydrocarbon vapours,

removing the sludge from the tank floor and tankwall, cleaning the floor and walls and then, finally,refilling the tank with hydrocarbon. There arealternative procedures available for each step, andthe ability to reduce emissions during cleaning andmaintenance will be site- and tank-specific. Detailsof the hydrocarbon loss mechanisms have beenexplored, and estimating methodology for tankcleaning operations developed, by the AmericanPetroleum Institute (API, 2007).

Initially, liquid is removed from the tank asthoroughly as possible, first through the normalwithdrawal procedures, after which any remainingliquid may be collected using vacuum hoses. Asliquid is being removed, there are essentially no airemissions from the tank because, for all tank types,the flow of air will be into the tank. It is importantto remove as much liquid as possible, because anyliquid remaining after this step will likely evaporatewhen the tank is opened.

After all of the liquid is removed, the remainingvapours in the tank are purged. Several purges arenormally required to ensure that all hydrocarbonvapours are removed from the tank. In somelocations, the first tank volume (sometimes severaltank volumes) of this vapour must be collected andtreated because of the potentially high hydrocarboncontent. Suggested vapour recovery anddestruction processes for treating these vapours arediscussed in the section on Product loading(overleaf).

Removal and collection of sludge may releasehydrocarbon vapours. Depending on the specificoperation, it may not be possible to collect vapoursfor treatment during this operation. Operations forcleaning of the tank walls and removal of sludgefrom the tank floor are usually site-specific anddepend on the contractor and methods used.Additional hydrocarbons may be releaseddepending on the procedures and chemicals used.When the tank is returned to service, the normalfilling losses occur.

IPIECA

26

Good practices for control of storagetank emissions

� Inspect roof fitting gaskets and seals and rimseals.

� For external floating roof tanks, replace avapour mounted primary rim seal with amechanical shoe seal.

� Install a secondary rim seal on external floatingroof tanks.

� Gasket and latch or bolt all roof hatches onexternal floating roof tanks.

� Install a sleeve around the slotted guidepole inan external floating roof tank.

� Avoid causing a floating roof to land on its legswhen withdrawing liquid.

� Drain-dry a tank prior to opening to theenvironment for cleaning.

Product loading

When hydrocarbons are loaded into rail cars, tanktrucks, barges or vessels some of the materialloaded evaporates into the vapour space in thecompartment. The vapours are then expelled fromthe compartment as they are displaced by theadded liquid. This is similar to the emissionsmechanism for fixed roof tank filling losses.

Hydrocarbon emissions during loading are usuallyfrom two sources. Initial emissions arepredominantly due to vapours from the previouscargo transported (unless the compartment wascleaned). Once these existing vapours aredisplaced, emissions become predominantlyvapours evaporated from the new liquid beingloaded.

Loading emissions can be a large source of sitehydrocarbon emissions depending on the amountof material loaded, the vapour pressure of thematerial and the application of any vapouremissions control. Because of the magnitude ofloading emissions, some type of vapour control isusually recommended for the higher vapourpressure products such as gasoline. Vapourcontrols are also typically required on benzene andother toxic liquid loading operations.

The choice of control technology will depend onthe quantity and volatility of the material beingloaded, the value of any recovered andcondensed vapours, the desired emissionreduction, local support for the technology, andcosts. The costs include both capital and operatingcosts, and can be significant. As vapour controls onloading are rarely cost-effective based on recoveryof the hydrocarbon liquid, they are most oftenapplied due to a regulatory directive. In thesecases, the choice of technology must meet theregulatory requirement.

Most vapour control technologies are supplied aspackage units by vendors who specialize in thesetypes of units. In most cases a complete system ispurchased from a vendor who will guarantee thelevel of performance and provide ongoingoperations support.

The typical methods used to control loadingemissions are listed in Table 7. A significantreduction in vapour generation is possible bydecreasing the turbulence created when liquid isintroduced to the compartment. This can be done

27


by using bottom or submerged loading rather thansplash loading. Vapours can be collected usingvapour balancing where the displaced vapour isreturned to the container from where the liquid isbeing emptied. End-of-pipe controls involvecollecting the vapours and sending them to arecovery or destruction process. These end-of-pipecontrols are usually quite expensive and aretypically applied only if required by regulation or ifjustified for very high value products.

The major types of controls are discussed in moredetail in the following sections. In most cases,where vapour recovery or destruction is applied, itprovides high levels of control efficiency, but withhigh capital and operating costs.

How to quantify missions

The quantity and composition of vapours emittedfrom loading hydrocarbons will depend on thematerial previously contained in the compartment,any compartment cleaning before loading, thevapour pressure of the material being loaded, themethod of loading and the use of any vapourcontrols.

There are several ways to estimate emissions fromproduct loading operations. The simplest, andpotentially least representative, is to use industryaverage emission factors for each type of liquid.Improved estimates are possible if moreinformation is known about the vessel, its conditionand the type of loading. If specific vapour pressureinformation on the material being loaded isavailable, even more representative emission

estimates are possible. These emission estimatingoptions are discussed in US EPA, 1995b.

Splash, bottom and submergedloading

How material is placed into the receiving containermakes a significant difference in hydrocarbonvapour generation. Turbulence that tends toincrease vapour concentrations should beminimized. If new product is added above theliquid, significant splashing can occur which willincrease vapour concentrations and create dropletsthat can be entrained with the escaping vapour asit is displaced by the liquid. For this reason, splashloading from above the liquid surface is notrecommended.

Alternatives to splash loading include submergedand bottom loading. In submerged loading thepipe through which new liquid is loading is placedvery near the bottom of the container. Thisminimizes splashing as the liquid entry point willquickly become submerged in the compartment. Inbottom loading, the pipe enters from below thefloor of the compartment and is always maintainedbelow the liquid surface.

Vapour balancing

In vapour balancing, hydrocarbon vapours arecollected from the compartment where the liquid isbeing loaded and returned to the tank from whichthe liquid is being sent. This works because thevolume of displaced vapours is almost identical tothe volume of liquid removed from the tank. The

Table 7 Controls to reduce product loading emissions


Low

High

Very high

Very high

Use submerged or bottom loading

Install a vapour balance system



IPIECA

28

technique is used mostly when loading tank trucksfrom fixed roof tanks. It cannot be applied whenloading from floating roof tanks since there is noclosed vapour space in the tank to which vapourscan be returned.

Vapour balancing sends the displaced hydrocarbonvapours to the tank in place of external air thatwould normally enter the tank as the level islowered. This reduces the volume of air thatbecomes contaminated with hydrocarbon vapours.When refilling the tank, these vapours will beexpelled and controls need to be considered if theycannot be sent to another compartment forbalancing.

Vapour recovery: adsorption,absorption and refrigeration

Vapour recovery provides control of emissions bycollecting the vented material for recycle and reuse.The three types of technologies most often appliedare adsorption, absorption and refrigeration. Theoperating characteristics of these technologies areprovided in Table 8. In some cases, combinationsof technologies such as absorption followed byadsorption have been applied. These combinationssometimes provide both higher capacity and lowervapour concentrations more cost-effectively.

Adsorption involves capture of vapour by a highlyporous solid. Carbon is the most commonly usedadsorbent but many proprietary adsorbents arecurrently offered by vendors, including silica gel,alumina or zeolite based products. Adsorbents withhigh surface area to volume ratio are desired.

There are two types of adsorption unit, each ofwhich uses a different method to regenerate theadsorbent. One uses vacuum regeneration and oneuses thermal regeneration, often with steam. Somesmall and infrequently operated vapour controlunits may, more economically, use carbon canisterswhich are replaced rather than regenerated on site.

A typical adsorption unit that is in continuousoperation will consist of two or more adsorbentbeds. This allows at least one to be treatingvapours while the other is being regenerated.Vapours removed during regeneration are oftencondensed and recycled to the product tanks.

Control efficiencies for adsorption depend on thevapour flow rate, temperature, operating pressureand adsorbent. In general, adsorption increaseswith increasing molecular weight of thehydrocarbon and operating pressure, anddecreases with increasing temperature. Adsorptionunits can often provide high control efficiency butat high operating costs. They are mostly applicableto lower flow rates and lower hydrocarbon vapourconcentrations.

Table 8 Characteristics of vapour recovery technologies

Adsorption

Absorption

Refrigeration

Solid waste

Solid wasteWastewater

Solid wasteWastewater

CoolingDehumidificationParticulate removal

Particulate removal

Dehumidification

20 – 2,000

1,000 – 20,000

5,000 +

90 – 99

50 – 95

50 – 80

Technology Secondaryimpact

Pretreatment requirements

Feed concentration range (ppm)

Approximate control efficiency (%)

29


Absorption recovery units involve the mass transferof hydrocarbon from a gas stream into a liquidsolvent. In many cases this is just a liquidhydrocarbon of higher molecular weight. The rateof absorption and maximum capacity is determinedby the equilibrium concentrations of the liquid andgas phases. Removal efficiency is mostly determinedby the solubility of the hydrocarbon in the absorbentand the intimacy of vapour-liquid contact.

Absorption tends to be less efficient than some othertechniques and, as previously mentioned, is oftenused in combination with another technology such asadsorption. Absorption can handle a wide range ofvapours and can accommodate larger changes invapour flow and concentration. However, absorptionunits cannot usually achieve the same level ofremoval efficiency as carbon adsorption beds.

In refrigeration, also known as condensation,hydrocarbons are removed from the gas bylowering the gas temperature. Sometimes increasedpressure is used in place of lower temperature.There are two types of condensers: surface anddirect contact. In surface condensers, coolant flowson one side of a heat exchanger, condensing thehydrocarbons as the gas flows on the other side ofthe heat exchanger. In shell and tube heatexchangers, the coolant usually flows through thetubes and the gas stream is on the shell side.Contact condensers operate by spraying a coolliquid directly onto the gas stream. A majoradvantage of vapour recovery by refrigeration isthe ability to recover relatively pure hydrocarbon.However, these systems tend to have relatively lowerrecovery efficiencies and higher operating costs.

Table 9 Advantages and limitations of vapour recovery technologies

AdvantagesTechnology Limitations

• Limited capacity with maximumthroughput and concentration allowances

• Requires low humidity• Requires low temperatures• Potential to plug or poison adsorbent

packed bed• May require prefiltering of gas• Less efficient for light molecular weight

species• Swing operation to allow regeneration in

continuous operation

• Not efficient for low concentrations• Not efficient for low gas volumes• Recovered product may need to be

separated from lean oil• Lower capture efficiencies

• Potential high energy costs• Lower collection efficiencies• Not efficient for low flash temperature

streams

Adsorption

Absorption

Refrigeration

• Very high efficiency• Potential recovery of very high purity

stream• No open flame (but exothermic reaction

needs monitoring)• Applicable to very low stream

concentrations• Applicable to batch operations

• Typical process-like operation• Potential recovery of high purity stream• Low pressure drop• Lower energy consumption• No open flame

• Direct recovery of pure product possible• No open flame• Applicable to high concentration

streams

IPIECA

30

Condensers are often used in combination withother control techniques. They can be locatedupstream of absorbers, carbon beds or oxidizers torecover some product and to reduce thehydrocarbon being treated by these other units.

The advantages and limitations of these vapourrecovery technologies are summarized in Table 9.

Vapour destruction: flares, thermaloxidizers and catalytic oxidizers

Vapour destruction provides control of emissions bycombustion of the hydrocarbon to form carbondioxide and water vapour. The three types oftechnologies most often applied are flares, thermaloxidizers and catalytic oxidizers. The operatingcharacteristics of these are provided in Table 10.While vapour destruction reduces emissions ofhydrocarbon, because it is a combustion process, itcan increase emissions of NOx and, if sulphur ispresent, SOx. Destruction efficiency is controlled byresidence time, combustion temperature andoxygen availability.

In some cases, vapours can also be controlled byrouting the gas to a boiler, heater or furnace. Thisallows recovery of the heating value of thematerial. Applicability will depend on locationwithin the plant and on compatibility of the vapoursas a supplemental fuel.

The least expensive destruction technology is oftenflaring. Flares can be elevated and open orenclosed at ground level. Open, elevated flares

have a flame located at the top of a stack and usespecially designed burners. They are most oftenused for controlling emissions from plant upsets.Steam assist is often used to provide increasedturbulence and mixing with air which increasesdestruction efficiency.

Ground level flares are enclosed and contained inan insulated shell. The shell provides reduced noiseand visibility. Enclosed ground level flares are usedfor continuous streams and provide more stableoperating conditions than do elevated flares.

Thermal oxidizers or incinerators are controlledcombustors. They include a combustion chamberthat is designed to completely contain the flame atmaximum firing rate to ensure sufficient residencetime at flame temperatures to maximize destruction.Operating parameters that affect emissions controlinclude the supplemental fuel firing rate, residencetime, gas mixing and exit temperature. Good gasmixing with supplemental combustion air isessential for high destruction efficiencies.

There are three different designs for thermal oxidizersdepending on the type of heat recovery employed:direct flame, recuperative and regenerative. Directflame units do not include any heat recovery fromthe combustion process. Recuperative designsinclude heat exchangers to preheat the feed gaswith the flue gases from combustion. Regenerativedesigns transfer the heat first to an intermediatematerial (usually ceramic beds) and then to thefeed gas. The regenerative designs are usuallymore thermally efficient due to better heat transfer.

Table 10 Characteristics of vapour destruction technologies

Flares

Thermal oxidizer

Catalytic oxidizer

CO2, SOx, NOx

CO2, SOx, NOx

CO2, SOx, solid waste

Liquid removal

Preheating

PreheatingParticulate removal

20 – 20,000

50 – 10,000

98 +

95 – 99

90 – 98

Technology Secondaryimpact

Pretreatment requirements

Feed concentration range (ppm)

Approximate control efficiency (%)

31


In catalytic oxidizers, the combustion occurs at amuch lower temperature in the presence of acatalyst that promotes complete combustion.Destruction efficiency depends on the hydrocarbonconcentration, operating temperature, residencetime, catalyst activity and oxygen availability. Asresidence time and temperature increases,destruction efficiency increases. Since these unitsoperate at lower temperatures, there is much lessformation of thermal NOx. However, catalyticoxidizers are susceptible to plugging and poisoningby metals in the hydrocarbon vapour. As thecatalyst activity decreases, it must be replaced orregenerated.

The advantages and limitations of vapourdestruction technologies are summarized inTable 11.

The selection of a specific vapour controltechnology will depend on the regulatoryrequirements and the capital and operating costs. If

several technologies will meet the regulatoryrequirements and have similar costs, factors suchas the range of applicable flow rates, vapourconcentrations and stream consistency becomeimportant in selection. Options should be discussedwith local vendors supplying package units as theywill typically provide ongoing operations supportand a performance guarantee.

Good practices for control of loadingemissions

� Apply bottom or submerged loading of materialto reduce vapour generation.

� Consider vapour balancing opportunities.� Consider vapour recovery opportunities.� Work with local vendors supplying package

vapour control units.� Have vendor guarantee recovery efficiency.

Table 11 Advantages and limitations of vapour destruction technologies

AdvantagesTechnology Limitations

• No product recovery

• No energy recovery

• Routine flaring to be minimized

• Emissions of products of combustion(CO2, SOx, NOx)

• Limits on placement due to noiseand safety

• Emissions of products of combustion(CO2, SOx, NOx)

• No product recovery

• Catalyst plugging and poisoning

• High pressure drop

Flares

Thermaloxidation

Catalyticoxidation

• Low cost

• Treats wide range of flows andconcentrations

• High control efficiency

• Treats a wide range of streamconcentrations

• Possible energy recovery

• High control efficiency

• Possible energy recovery

• Lower energy consumption thanthermal oxidation

• Lower NOx formation

IPIECA

32

Wastewater collection and treatment

Hydrocarbon emissions from the wastewatercollection and treatment system occur as a result ofevaporation of hydrocarbon from the water wherethere is water-air contact and release of the vapourto the environment where the system is open to theoutside. In most plants, many parts of the collectionand treatment system are typically open to theenvironment.

The wastewater collection system includescomponents such as drains, manholes, junctionboxes and sewers. Where each is open to theenvironment there may be emissions ofhydrocarbon. The treatment system in each plant isunique but will typically include a primaryoil/water separator, holding and equalization tanksor basins and possibly flotation and biologicaltreatment units. Where each of these treatmentunits is open to the environment, and there is air-water contact, emissions of hydrocarbon can occur.

The emission control strategies for wastewater systemsare to minimize wastewater generation, to reducehydrocarbon concentrations entering the systemand to reduce the area of air/water interface.

Wastewater typically contains a variety ofhydrocarbon compounds with widely varyingconcentrations. The compounds present andconcentrations are usually site specific. The bestway to minimize hydrocarbon emissions from thewastewater collection and treatment system is toreduce the quantity of hydrocarbon that entersthe system.

Sometimes it is possible to better managewastewater by segregation or recycle. Drains andsumps should not be used for routine disposal ofprocess fluids or for discharge from sampling lines.Drains and pads can be raised to reduce stormwater entry. Having separate process and stormwater systems will reduce contamination.Optimizing strippers to lower hydrocarbonconcentration will reduce the quantity entering thecollection and treatment system. Once wastewaterenters the treatment units, control usually involvescovering them, and in some cases, controls on thecollected and vented vapours.

Water can become contaminated withhydrocarbons intentionally, through direct contact,or accidentally, through indirect contact. Intentionalsources include units that use water for washing

Table 12 Controls to reduce wastewater collection and treatment emissions


Low

Low

Low

Medium

Medium

High

High

High

Very high

Very high

Decrease wastewater volume

Decrease wastewater hydrocarbon concentration

Leak detection programme for heat exchangers

Optimize stripper operation

Install sewer system emissions controls

Segregated process and storm water systems

Reduce air/water interface

Cover separation and treatment units



33


such as desalters and sour water strippers. Watermay also be formed as a byproduct of reaction ormay be mixed with process fluids in tank draw-off,storm water run-off and steam eductor condensate.

Accidental or indirect sources are those that do notnormally come into contact with hydrocarbon aspart of the process. These sources include processfluid leaks from cooling heat exchangers,condensers and pumps. In addition, pump sealwater may be collected in local drains that flowinto the wastewater sewer system. A heatexchanger leak detection programme is the mosteffective approach to minimizing hydrocarboncontamination of cooling water. Individualexchangers with leaks can be identified and planscan be made for tube repair.

Once contaminated water enters the treatmentunits, emissions control opportunities are limited toreducing the air-water contact area and collectingvapours for treatment. Although not justified bycost, some separator and treatment units can becovered to reduce wind driven evaporation.Options to reduce hydrocarbon emissions fromwastewater collection and treatment and theirrelative costs are provided in Table 12.


There are two general ways to estimatehydrocarbon emissions from wastewater collectionand treatment. One involves the use of industryaverage emission factors. While this method issimple, and all that is usually needed is the unitflow rate, it is not likely to produce representativeestimates and the predicted quantities will tend tosignificantly exceed the actual unit emission rates ina well-run plant.

A more representative method for estimatingemissions involves the use of computer models;however, this is complex and requires a significantamount of design and operating data. Also, thesemodels calculate individual chemical compound

emissions so knowledge of all the hydrocarboncompounds in the wastewater is required toaccurately estimate total hydrocarbon emissions.However, as stated, these models will provide muchmore representative estimates of emissions. The USEPA has developed a computer model forestimating hydrocarbon emissions from thewastewater system (US EPA, 2004).

Source reduction

As previously mentioned, the best way to minimizeemissions from the wastewater collection andtreatment system is to reduce the quantity ofhydrocarbon that enters the system. Wasteminimization includes flow and/or concentrationreduction as well as recycling. Sometimes, sourcereduction can be achieved through process orequipment modification, stream segregation, orimproved work practices.

Guidelines for reducing the quantity andhydrocarbon content of wastewater can be foundin IPIECA, 2010.

Sewers, drains, junction boxes andlift stations

Most wastewater from process areas enters thecollection system through drains which areconnected to the sewer system. There can besignificant evaporation from the system if the drainis not sealed and vapours are allowed to escape.A water seal, often called a p-trap, can be used toeffectively eliminate these emissions. This reducesthe vaporization of hydrocarbon in the wastewateras it enters the drain and also preventshydrocarbon vapours already in the sewer systemfrom escaping. In open sewers, air passingthrough can be a promoter of increased emissionsas wind passing over the drain can create avacuum-like effect.

Inspection and cleaning of sewer lines isaccomplished by the presence of manholes which

IPIECA

34

allow access for maintenance personnel andequipment. A sewer line may have many of theseplaced along its length. If the manhole is notproperly sealed, hydrocarbon vapours from thesewer system can be released. As with opendrains, air passing through due to an ineffectiveseal or holes in the manhole can promoteincreased emissions as wind passes over themanhole and creates a vacuum-like effect. Aneffective control can be obtained by completelysealing all openings in the manhole so that vapourrelease and air passage is prevented.

Junction boxes are used to collect the flow fromseveral sewer lines. Where these are open to theenvironment, hydrocarbon vapour emission canoccur. Turbulence in the junction box cansignificantly increase the potential for hydrocarbonemissions. The junction box should be designed sothat wastewater streams enter below the liquidsurface to minimize splashing and turbulence.Also, to reduce overall evaporation, residencetime of the wastewater in the junction box shouldbe minimized.

The final part of the wastewater collection system isusually a lift station. This uses a pump to physicallyraise the water so that it has sufficient pressurehead to flow to the treatment system. In somecases, there is an open top on the lift station.Typically, there is a periodic operation of the pump.As water enters the lift station sump and the waterlevel increases, the pump will begin to operatewhen the water reaches a specified level and willcease operation when the water level has beenreduced to a lower, specified level.

In lift stations, there are two mechanisms thatpromote emissions. At lower water levels,hydrocarbon emissions are enhanced bysplashing and turbulence as new wastewaterenters the lift station. As the water rises in the liftstation, it pushes existing hydrocarbon-contaminated vapours out of the top of the liftstation. As with the other collection system units,

emissions can be essentially eliminated by closingthe lift station so that vapour cannot flow into theenvironment.

Controlling the hydrocarbon emissions from thewastewater collection system by enclosing it willnormally result in the wastewater having a higherconcentration of hydrocarbons when it enters thetreatment units. Emissions prevention duringcollection can result in almost the same quantity ofhydrocarbon emissions being released duringtreatment. Therefore, for the system to beconsidered effective, at some point thehydrocarbons need to be recovered or destroyed.

If some control of hydrocarbon vapours occurredin the collection system there could be a reductionin the emissions from the treatment system. Thiscould be achieved using carbon canisters oncollection system vents, which would need to beperiodically replaced.

Primary separators, IAF/DAF,biological treatment and treatmenttanks

Wastewater treatment systems generally consist ofprimary, secondary and occasionally, tertiarytreatment units. In the treatment system,hydrocarbon species are removed from thewastewater via three pathways. They canbiodegrade into other species and, eventually, intocarbon dioxide and water; they can adsorb intothe sludge and be removed; or they can bereleased as vapours into the air. The specific designand operation of each unit will determine therelative amounts for each pathway.

Primary treatment separates free oil and mostsolids from the wastewater. In the primaryoil/water separator, oil and solids are separatedfrom the wastewater by gravity. As thecontaminated wastewater flows across theseparator, oil floats to the top of the water phasewhere it can be removed for reprocessing. Heavier

35


solids sink to the bottom of the separator and areremoved for disposal.

Hydrocarbon emissions from oil/water separatorsdepend on a number of factors including theconcentration of the specific hydrocarbon, thehydrocarbon volatility, wastewater and airtemperatures, and wind speed. Though not cost-effective, covering the unit with either a fixed orfloating roof results in the greatest emissionreduction since the ambient air-water interface iseliminated. Fixed roofs may require inert gasblanketing to avoid an explosion hazard. Theemissions control effectiveness of the roof is largelydependent on the sealing between the walls of theseparator and the roof.

After the primary separator, a dissolved airflotation (DAF) or induced air flotation (IAF) unit isused for further oil/water separation. In these units,the water mixture is subjected to physical andchemical procedures that promote the aggregationof suspended solids into particles large enough tobe removed. In the DAF unit, a recycle stream ispressurized with air which, when returning to theDAF unit, creates small air bubbles that promotethe flotation of oil and some solids. In an IAF unit,air is induced into the fluid by a rotatingmechanism and then rises, collecting oil and solids.Any oil collected on the surface of the flotation unithas the potential to volatilize. Covering these unitswith a fixed roof can allow any volatilized vapoursto be collected and prevented from being releasedto the atmosphere. Proper precautions to mitigateor prevent explosion hazards on these coveredunits are important aspects of unit design.

Equalization basins are used to reduce fluctuationsin wastewater flow rate and concentration tosubsequent treatment units. They may be locatedbefore or after the primary treatment units. Aswastewater flows through the basin, hydrocarbonsdiffuse through the water to the liquid surface andevaporate. In some cases, aerators are used inequalization basins to add oxygen for subsequent

treatment. This aeration has the potential toenhance evaporation due to the additionalturbulence and increased surface areas of theliquid coming into direct air contact.

Secondary treatment units remove dissolvedhydrocarbons remaining in the wastewater.Biological treatment is normally conducted inaerated basins. These units use diffused ormechanical aeration to provide oxygen for thebiological processes. In these units organic materialis converted into cell tissue, water and carbondioxide.

Some treating operations can be conducted in fixedor floating roof tanks. This will usually providesignificant emission reduction compared to open-top systems. By covering the surface of thewastewater, evaporation of the hydrocarbons isreduced. If there is sufficient hydrocarbon to coverthe liquid surface, emissions from fixed roof tankscan be significant and a floating roof tank mayneed to be considered. The section on Storagetanks (page 18) discusses emission controls.

As with other treatment units, tank emission controlsoften result in the hydrocarbon continuing toremain in the water until a later opportunity arisesfor it to evaporate. Actual emission reduction isonly achieved if the hydrocarbon is recovered ordestroyed.

Good practices for control of airemissions from wastewatercollection and treatment

� Reduce concentration of hydrocarbons inwastewater.

� Reduce volume of wastewater that needs to betreated.

� Consider whether seals can be applied todrains, manholes and junction boxes.

� Consider whether the air/water interfaces onwastewater treatment units can be covered.

IPIECA

36

Process vents

Atmospheric process vents are generally pipesconnected to vessels that emit process gases directlyinto the environment. They may be located onreactor vessels or other process fluid receivingvessels that require removal of material afterreaction or separation. In many cases the vents willbe open directly to the atmosphere. Alternatively,there may be a pressure relief valve that allowsonly occasional release of the process fluid to theenvironment. The best control for these process ventemissions is to eliminate the need for discharge byaltering the process operation or recycling thematerial. If this cannot be done, vapour controlssuch as recovery or destruction can be consideredfor application to the vent stream.

The first step in controlling process vent emissions isto determine whether there are economicalternatives to venting such as direct recycle ortemporary storage and reuse or recovery.Possibilities will be specific to the process unit.Recovery options such as refrigeration, absorptionand adsorption are discussed in the section onProduct loading.

In some cases, vapour destruction as a control forprocess vent emissions may be the best option. Typesof vapour destruction controls may include flares,incinerators or other oxidation type units discussedin the section on Product loading. Generation ofcombustion emissions from all of these destruction-based controls should be accounted for andconsidered in the technical evaluation.

Good practices for controllingprocess vent emissions

� Evaluate options for recycle and reuse ofdischarged material.

� Consider vapour recovery options such ascondensation, adsorption and adsorption.

� Consider vapour destruction options such asflares, incineration and catalytic oxidation.

Flares

The refinery flare system exists to prevent a majoraccident should a process upset or other conditionrequire the shut-down of a process unit releasingflammable gases. As such the flare system is anessential safety device.

Intermittent releases from vents and pressurisedrelief valves may be directed to flare as a preferredalternative to venting to atmosphere.

Routine flaring of waste gases is deprecated forboth overall efficiency and environmental reasons.A refinery flare minimisation plan should bedeveloped as part of the overall environmentalmanagement system.

Although flares are external combustion sourceswith limited control of the combustion process, awell operated flare will generally have acombustion efficiency of 99.5%.

There are many designs of flare used in refineries.The flare tip design is important to assure flamestability in different meteorological conditions, lownoise, good combustion performance, etc., whichmay be aided by steam injection for minimizationof visible smoke.

Source reduction

The most effective way to reduce flare emissions isto minimize the amount of material sent to theflare.

The sources are:� necessary fuel for the flare pilot flame to ensure

ignition in the event of emergency; � emergency venting;� vented gases from process upsets, start-up and

shut-down of operations;� surplus gases from production processes; and� vented gases from normal operations, pressure

relief valve operation, etc.

37


Improved operation of processes can lead to areduction in the number of emergency flaringepisodes, thus reducing the quantity of gas sent tothe flaring system. In addition, the installation of aflare gas recovery system to recycle thehydrocarbons back into the process system and/orrefinery fuel gas system is an option that is widelyapplicable in refineries where there is a surplus ofgas from processes.

Gas recovery

A flare gas recovery system comprises pipeheaders to collect the gases, condensers to removeany liquids, and compressor(s) to recycle the gasback to the process or into the refinery fuel gassystem.

The cost-effectiveness of flare gas recovery mayvary widely between sites depending upon thenature and distribution of the sources of materialgoing to flare. It will be less favourable fornumerous distributed sources and for gasesneeding pre-treatment for further use.

Combustion processes can be divided into twogroups. The first group involves the burning of fuelsto use the heat of combustion to raise thetemperature of a fluid other than the fuel (e.g. invarious pieces of process equipment includingfurnaces, boilers and heaters). The second groupinvolves combustion of part of the material beingprocessed to provide heat to conduct the reaction(e.g. catalytic crackers, catalytic reformers,steam/methane reformers and sulphur plants). Bothof these groups can result in the generation ofsulphur oxides, nitrogen oxides and particulatematter depending on the fuel and fuel components,unit design and operation.

Boilers, heaters and furnaces

Boilers, furnaces and heaters are utilized to raisethe temperature of water, other heat transfer fluid ora process stream. This is accomplished by burning afuel and using the heat of combustion to raise thetemperature. The heat is transferred to the water,heat transfer fluid or process fluid by radiation fromthe flame and convection and conduction from thecombustion gases. Reduction in fuel use typicallyresults in a proportional reduction in mostemissions. Therefore, energy conservation efforts

can often have direct impacts on reducingemissions. An exception to this is combustion airpreheating which raises the flame temperature andcan sometimes result in increased NOx formation.

Particulate matter (PM) can take more than oneform. ‘Primary particulates’ usually refers to solidmaterials such as soot or catalyst. The soot is oftenformed from incomplete combustion of the fuel orfrom intermediate combustion products. Typically,gaseous fuels have lower amounts of primaryparticulate emissions than liquid fuels which, inturn, have lower primary particulate emissionsthan solid fuels. ‘Secondary particulates’ usuallyrefers to precursors of sulphates and nitrates, andalso to aerosols that can be condensed from theflue gas at specific temperatures. The quantity ofaerosol emitted is defined by the condensingtemperature of the flue gas sampling system. Aswith primary particulates, gaseous fuels typicallyhave lower amounts of secondary particulateemissions than liquid fuels which, typically, havelower secondary particulate emissions than solidfuels. Carbon deposited on (and subsequentlyburned off from) a catalyst, such as in FCC units,is considered a solid fuel.

Sulphur oxides are created in combustion processeswhen sulphur in the fuel is burned. The amount ofSOx emitted is directly proportional to theconcentration of sulphur in the fuel. Typically, gaseousfuels have lower amounts of sulphur than liquid fuelswhich have lower sulphur than solid fuels. Somegaseous fuels that are generated within a refineryprocess (e.g. refinery fuel gas) can have highersulphur contents and may require pre-treatment tolower the amount of sulphur prior to use.

There are two combustion mechanisms that producenitrogen oxides. Nitrogen oxides are created whennitrogen and oxygen in the air combine at hightemperatures. This formation mechanism produces‘thermal NOx’. For fuels with significant elementalnitrogen content (e.g. liquid fuels like oil, and solidfuels like coal or coke) this fuel-bound nitrogen can

IPIECA

38

Sources and control of combustion emissions

39


be a large contributor to NOx emissions. Thisformation mechanism produces ‘fuel NOx’.


Emissions of NOx and PM are typically estimatedusing industry average emission factors from AP-42(US EPA,1995b) and CONCAWE (CONCAWE,2009). There are different factors for different unittypes, fuels and sometimes operations. Whenestimating NOx emissions, adjustments can be madefor a myriad of operating parameters such as fuelhydrogen content and ambient humidity. PM factorsare provided for both primary PM and secondaryPM for most units. Although average emissionfactors are fully accepted, stack testing provides amore representative method to assess PM and NOxemissions, but at much higher cost.

Emissions of SOx are usually based on assuming allthe sulphur in the fuel is combusted and forms theoxide. Thus, a direct material balance assessmenton the quantity of sulphur in the fuel burned is thestandard procedure for calculating SOx emissions.

PM (particulate matter) control

PM, or dust, generally refers to any solid, aerosolsand liquid droplets. The technologies available forPM control include mechanical collectors, baghouses and filters, electrostatic precipitators and wetscrubbers. The choice of control will depend on thesize and weight distributions of the particles to beremoved and the emission limit. In some locations,‘total’ PM is no longer regulated, as the focus ofcontrol is now on particulate matter less than10 microns in diameter. Most of these controltechnologies require significant plot space, whichmay be a limitation in retrofit applications.

PM control efficiency is based on the mass ofparticulates collected rather than the number ofparticles collected. Most PM control devices havehigh efficiencies over specific particle size ranges.The control device should be selected based on the

particle size distribution containing the greatestmass percentage of PM. In that way, the greatestPM mass reduction can be achieved.

Mechanical collectors control particulates by usinggravity or inertial forces to separate the particlefrom the gas stream. These include cyclones, otherinertial collectors and settlers. In cyclones,particulates follow the gas stream along thecircumference of the cyclone and move toward theexterior wall due to inertial forces. They aredisengaged from the gas at the bottom of thecyclone where the gas quickly changes direction toexit the cyclone. Multi-stage cyclones can oftenachieve high particulate removal efficiencies.Smaller-sized cyclones can also provide greaterremoval efficiencies, but pressure drops are higher.

In inertial collectors, separation is based on thedifference between the densities of the particlesand the gas. The gas is redirected through quickturns which, although easy for the gas, is moredifficult for the particles because of their higherinertia. Units may include baffles and otherobstruction-like systems where the particulates areintercepted as the gas flows by; the particulatesthen settle and are removed. Sometimes theseunits are called ‘impingement separators’. Thesedevices are generally not very efficient except forthe removal of larger-sized particulates.Sometimes, an inertial separator can be used as apreliminary ‘cleaning’ unit for larger particlesprior to removal of smaller particle ‘fines’ in adownstream unit.

The cyclone is a special type of inertial separatorsince it combines promoting particulate movementdown and to the sides of the cyclone in addition toa quick change in gas direction. As describedabove, cyclones are often used in multiple stages toachieve higher PM removal efficiencies.

In settling chambers, the gas velocity is loweredand the particulates fall out of the gas due togravity. The advantages of settling chambers

include low maintenance and low pressure drop.Although these units are not efficient for small-sizedparticulates, they can be used for preliminary‘cleaning’ prior to removal of ‘fines’. The chamberis designed such that the flue gas is at a sufficientlylow velocity for a large fraction of the particulatesto settle to the bottom region of the chamber wherethey enter hoppers and are removed. Gas velocitymust be set to minimize re-entrainment of theparticulates once they are collected. In some cases,improved performance can be obtained byincluding a number of flat plates in the chamber tocreate shorter settling distances.

Bag houses contain fabric filters that collectparticulates as the gas passes through the fabric.They usually provide very high collection efficiencybut are unsuitable for high temperature combustionstreams without additional cooling. Therefore, baghouses usually require cooling of the combustionflue gas. Also, the pressure drop can increase overtime as particulates are collected; therefore, thebags are usually cleaned by agitation and/orreverse flow using clean gas. Collection efficiencymay be lower after the bags are cleaned sincesome build-up of particulate on the fabric isrequired to enhance the removal of PM.

In wet gas scrubbers (WGSs), water is sprayedinto the gas stream and the water dropletsintercept and collect the particulates. Scrubberscan handle high temperature combustion streamsbut they produce liquid waste containingparticulates that may need to be treated.

Scrubbers can also be used to reduce SOx byselecting an appropriate liquid absorbent. In caseswhere emissions of both PM and SOx need to bereduced, use of a WGS should be considered.Scrubber technology is generally sold by vendorswho often offer ‘turnkey’ services.

In electrostatic precipitators (ESPs), the particles arecharged and then pass through a chamber withelectrodes that attract the particulates due to thedifference in charge. The collecting electrodes areusually plates that are occasionally cleaned by‘rapping’ which causes the particulates to fall to thebottom of the chamber and into a hopper forremoval. ESPs are generally sold by vendors aspackage units specifically designed for individualsite application.

Substitution of liquid fuel for imported natural gasis an option for emissions control for all three mainpollutants.

Note that the pressure drop across combustionsystems is a limiting factor for the use of secondarymeasures.

SOx control

The emissions of SOx from furnaces, boilers,heaters and other external combustion units can becontrolled in three ways. Two involve reducing thequantity of sulphur in the fuel to the unit. This maybe done by treating the fuel to reduce sulphur orby substituting a lower sulphur fuel such as a gas

Table 13 Controls to reduce PM emissions


Medium

Medium

High

High

Substitution liquid fuel for imported natural gas

Cyclones, settling chambers and inertial collectors

Fabric filters

Wet gas scrubbers

Electrostatic precipitators

IPIECA

40

41


for a liquid. The third way is to apply end-of-pipecontrols such as scrubbers.

Since any sulphur in the fuel will, when the fuel isburned, result in formation of SO2, reduction ofsulphur in the fuel will result in a proportionalreduction in SO2 emissions. If currently using fueloil, switching to gas (which usually has much lowersulphur content) will typically reduce SO2 emissions.In some cases, sulphur may need to be removedfrom refinery fuel gas if it is high in sulphur content.

Treating liquid fuels (usually with hydrogen in ahydrodesulphurization process) to remove sulphurcompounds is another way to decrease theformation of SO2. However, costs can be highdepending on the required level of sulphur removaland this is not usually seen as a technique which isused specifically to prepare internal refinery fuelsbut to control the refinery product slate. Refineryfuels may benefit as a consequence.

Scrubbers provide for gas contact with anabsorbing liquid (usually an alkali water mix)which removes the SO2. There are severalalternative scrubbing technologies. These aregenerally characterized as either once-through orregenerative systems, and either wet or dryprocesses. The once-through systems use anabsorbent that is not recyclable. Regenerativesystems recycle the spent absorbent after sulphurcompound removal to produce sulphur or sulphuricacid. As mentioned above, wet scrubber systemscan also be used to simultaneously reduce PMemissions.

The most commonly applied wet systems use limeor limestone. Although these systems are lower incost and have the largest experience base, theyrequire treatment and disposal of waste sludge. Inmany wet systems, the SO2 is converted into asulphate. The liquid is collected and treated beforerecycle. After time, fresh liquid is added and aportion of the contaminated liquid is removed as awaste stream. Usually, a significant amount of

equipment is required for treating the spent slurryfrom the wet systems, including clarifiers, filters,centrifuges and settling ponds.

Scrubbers require significant plot space, which maybe a limitation in retrofit applications. Anotherconcern in wet scrubbers is scale formation andplugging at higher pH. Slightly basic conditions areusually optimal for SO2 removal. However, theseare the same conditions that will enhance scaleformation and plugging.

Some newer WGS applications have usedseawater. This is applicable only to sites near thesea. Water treatment can sometimes result in therelease of SO2 during aeration, decreasing thecontrol efficiency of the system.

There are ‘dual-alkali’ processes where a sulphateis used to absorb the SO2 in place of the limestoneslurry. The spent slurry is then regenerated usinglime. These processes eliminate the scalingproblems of the lime and limestone scrubbers byeliminating calcium in the scrubber. However, thesetwo-step systems are more complex and costly tooperate since the lime needed for regeneration isgenerally more expensive than limestone.

In dry scrubbing, a solid sorbent is injected into theflue gas stream. The sorbent particles, which collectsulphur compounds, are collected downstreamusing conventional PM control methods. Thesesystems reduce the need for much of the waterhandling, and the waste is a more easily handledsolid rather than sludge.

There is much less experience with the dry systemsthan with wet systems, and they often use lime

Table 14 Controls to reduce SOx emissions


Medium

High

Fuel substitution

Scrubbers

rather than limestone. It is believed that much of theSO2 removal occurs in the particle collection deviceas the gas contacts the sorbent. For this reason,filters are the preferred PM collection device.


NOx control

As discussed earlier, there are two types of NOx:thermal NOx created in combustion processeswhen nitrogen and oxygen in the air combine athigh temperatures; and fuel NOx which involvesconversion of elemental nitrogen chemically boundto carbon in the fuel. Controls for thermal NOxinvolve reducing NOx formation by altering thecombustion process or by end-of-pipe treatment.Controls for fuel NOx include reducing nitrogen inthe fuel used and end-of-pipe treatment.

Thermal NOx formation is a function of the flametemperature, residence time and the air-to-fuelratio. Combustion modifications can beimplemented that reduce the flame temperature andthe time that the flue gases are at the maximumtemperature. Usually, operating in the most efficientmanner for heat recovery, with high temperaturesand longer residence times, tends to promotethermal NOx formation. Reducing excess air (i.e.excess oxygen) and operating closer to thestoichiometrically required amount of air forcomplete combustion will reduce NOx formation.Unit design and control to promote thorough air-fuel mixing for complete combustion becomescritical at near stoichiometric conditions.

There are several methods to reduce flametemperature including staged combustion, flue gasrecycle and steam injection. In staged burners, theinitial stage operates fuel rich (i.e. oxygen deficient)at high temperature while the second stage, whichcompletes the combustion, operates at a lowertemperature resulting in lower overall NOx formation.

These ‘low-NOx’ burners have varying performancedepending on the size of the burner and the fuel.

Newer burner designs alter flame geometry,turbulence and combustion sequencing to reducethermal NOx formation. These can include localizedgas recirculation and delayed mixing of combustionair. The latter reduces flame temperature since thereis lower localized excess air, resulting in reducedNOx formation. Vendors offer advanced designsthat can provide for very low NOx formation whenusing gaseous fuels. Several vendors offer ‘ultra-LowNOx’ burners for new units. However, these newerburner types are not always applicable for retrofitsin existing units due to space limitations.

Small amounts of flue gas can be returned to thecombustion chamber to reduce the flametemperature. In this configuration, oxygen isdiluted, resulting in slower combustion and lessoxidation of the nitrogen.

For fuel oil fired units, the majority of NOxformation in low NOx burners is due to nitrogenbound to the carbon in the fuel. This ‘fuel NOx’ canbe addressed by processing the fuel oil to reducenitrogen content. An alternative is to switch to agaseous fuel which has minimal nitrogen-bound-to-carbon content. The same end-of-pipe controls usedfor gas fired units are generally applicable to oilfired units provided PM loading is not an issue.

End-of-pipe NOx controls include selective catalyticreduction (SCR) and selective non-catalyticreduction (SNCR). In SCR, flue gas is contactedwith gaseous ammonia in a catalyst bed convertingthe NOx to N2 and water vapour. Althoughincreasing the amount of ammonia will improve theconversion and removal efficiency of NOx, it canresult in ammonia slip (i.e., residual, unreactedammonia in the flue gas stream), which can beharmful to any downstream treatment units due todeposits and can also create visible plumes. Thereare also operational issues with maintainingcatalyst activity.

IPIECA

42

43


Ammonia itself is an important pollutant and theamount of ammonia slip will have to be seriouslyconsidered and monitored.

In the SNCR NOx reduction process, the conversiontakes place completely in the gas phase at highertemperature without a catalyst. Good mixing ofammonia or urea and gas is essential forreasonable conversion efficiency. The SNCR processoperates well within a relatively narrow temperaturerange set by the chemical used. SNCR is usually lesseffective than SCR in reducing NOx in the flue gas.


Cogeneration

Where power is generated on site, an option toimprove efficiency that can be considered, whichalso reduces overall combustion emissions, is togenerate both heat and electric power in onesystem. This is known as cogeneration. Whengenerating power, some heat must be rejected tocooling. With cogeneration, this heat is used togenerate steam or raise the temperature of aprocess fluid. The thermal efficiency of the system ismuch higher overall. Less fuel is consumed thanwould be the case in separate heating and powergeneration units, resulting in lower overallcombustion and CO2 emissions.

Good practices for control of boiler,heater and furnace emissions

� Reduce fuel consumption by applying energyconservation measures.

� Use lower sulphur containing fuels.� Substitute gas for liquid fuel.� Reduce sulphur in refinery fuel gas used in

combustion units.� Use ‘low-NOx’ burner designs.� Consider cogeneration opportunities when

refurbishing on-site power generation or steamraising facilities.

Catalytic cracking

During the cracking reaction, some of the carbon inthe gas oil feed is deposited as coke on thecatalyst. This reduces catalyst activity, thus requiringthat the catalyst be regenerated. The catalyst iscirculated from the cracking reactor into aregenerator vessel where the coke is burned off torestore catalyst activity and reheat the catalyst. Theheated catalyst is then returned to the reactorwhere it provides heating for the endothermiccracking reaction.

The catalytic cracker regenerator can be asignificant source of combustion emissions. Thecombustion of coke during catalyst regenerationcan result in SOx, NOx and PM emissions. The PMemissions can include catalyst dust as well as thePM from the coke combustion.

Sulphur oxides are created in the catalystregeneration process when sulphur in the coke isburned. This sulphur originated in the feed to thecracker. The amount of SOx emitted is directlyproportional to the concentration of sulphur that iscontained in the coke.

In the regenerator, nitrogen oxides are createdwhen nitrogen and oxygen in the air combine athigh temperatures (resulting in ‘thermal NOx’) and

Table 15 Controls to reduce NOx emissions


Low

Medium

Medium

Medium

High

Very high

Install lower NOx burners

Steam injection

Flue gas recirculation

Substitute gas for liquid fuel

SNCR

SCR

when elemental nitrogen in the coke is burned (resulting in fuel NOx’). In typical catalytic crackingoperations, the fuel NOx fraction of NOx emissionsis much greater than thermal NOx due to the highquantity of fuel-bound nitrogen in the gas oil feedto the catalytic cracker.

PM emissions from the catalytic cracker can beboth primary PM and secondary PM. There aretwo sources of primary particulate: crackingcatalyst and soot from incomplete combustion.Secondary particulate (e.g. aerosols of NH3 andSOx) can be the larger fraction of particulateemissions after the primary cyclones collect andreturn most of the catalyst fraction. The quantity ofsecondary PM identified as being emitted asaerosols is defined by the condensing temperatureof the flue gas sampling system.


The most appropriate way to quantify emissionsfrom catalytic cracker regenerators is to conductstack test measurements at different operatingconditions.

Sometimes it is possible to develop a parametricemissions model based on the stack testing for aspecific unit. These models use test data from theunit to correlate emissions to changes in operatingconditions. The model can then be used to helpevaluate how changes in operations may impactemissions.

Choosing the best combination of emission controlsfor a catalytic cracker can be a complex process.Many of the end-of-pipe controls discussed abovefor furnaces, heaters and boilers are applicable tocracking regenerators. While these controls can beused to address only one of the pollutants, it isdesirable, if possible, to use controls that canaddress at least two of the pollutants to reduce bothcapital and operating costs.

One combination control option that has been usedis a wet gas scrubber for both PM and SOxreduction in combination with additives used forNOx reduction. Another control option is to use anelectrostatic precipitator (ESP) for PM reduction withadditives used for NOx control. The potential toreduce SOx emissions with additives should beexplored together with the reduction of PMemissions using multi-stage cyclones. Where lowerNOx emission levels are required, SCR and SNCRunits have been used but applicability and the cost-effectiveness for these techniques are poor. Dustmust be removed upstream of the reactor. As forcombustion, it may be noted that the pre-treatmentof liquid fuel by hydroprocessing can reduce feedsulphur content and remove some metals, but thismeasure is not an abatement step per se.Investments in feed pre-treatment will be driveninstead by the need to tailor the FCC product slate.

Table 16 provides an overview of the availableemission controls for catalytic cracking units. Itshows the available control options indicating which

Table 16 Control option applicability for catalytic cracking units

Control options

Cyclones Fabric Electrostatic Wet gas Feed Additives Selective non-catalytic Selective catalyticfilters precipitator (ESP) scrubber (WGS) pretreatment reduction (SNCR) reduction (SCR)

Pollutant

PM � � � � – – – –

SOx – – – � � � – –

NOx – – – � � � � �

IPIECA

44

45


pollutants can be addressed. Several of the optionscan be used for controlling more than one pollutant.

PM (particulate matter) control

The first step in PM emissions reduction is to usecyclones to disengage the larger catalyst particles.Most existing units already do this since there is acost saving associated with reducing excessivecatalyst loss. Multi-stage cyclones can sometimesachieve sufficient PM removal efficiencies so that noother controls are required. Tertiary cyclones havebeen used to achieve greater recovery of primaryPM but with increased system pressure drop.

Wet gas scrubbers are an option when cyclonesalone will not meet the PM control limit. Ifnecessary, the WGS can be designed to alsocontrol SOx emissions using an appropriatesorbent. New developments indicate that it ispossible to also reduce NOx emissions in a wet gasscrubber but the technology has not yet beenapplied commercially.

Wet gas scrubbers necessarily emit water vapourthat can appear as a significant white plume that isclearly visible.

Electrostatic precipitators (ESPs) can be used toreduce dust emissions. The efficiency depends uponthe particle size distribution, the resistivity of the dustand the inlet dust loading. For high efficiency largeESP volumes and up to four fields may be needed.Space requirements for retrofit are likely to belimiting. There are explosion hazards associated withthe use of ESPs during startup and shutdown periodsso that operating permits must address operation ofthe FCC without PM control at such times.

SOx control

The emissions of SOx from cracking units can becontrolled in three general ways. One is to reducethe quantity of sulphur in the feed entering the unit.

Although this may be achieved by hydrotreating thefeed, this is very expensive. Hydrotreating the feedis usually considered as a process modificationwhere there is a need to change the FCC productslate. Another, and preferred, SOx control is to useadditives to reduce the feed sulphur that deposits onthe catalyst. In this case, the sulphur stays with thegas or liquid product streams from the cracker andis more readily handled in downstream processunits. The third way is to apply end-of-pipe controlssuch as scrubbers.

With existing units, the first option to consider is theuse of SOx reduction additives. This is often theleast expensive control option and, in some cases,can provide the necessary emissions reduction.Additives can also play a role even if feed treatingor end-of-pipe controls are necessary since theycan often reduce the size or operating cost of theseunits. It is very difficult, however, to predictperformance of these additives. Several vendorsoffer a range of different additives. Therefore, atrial period is essential to determine how well theadditives will work in a specific application underdifferent operating conditions.

NOx control

The emissions of NOx from cracking units can becontrolled in three ways. One is to treat the feed toremove some of the chemically bound nitrogen.Again, this is not a preferred option. Another is touse NOx reduction additives. Finally, there are theend-of-pipe controls such as SNCR and SCR, whichare generally not cost-effective options due to thelarge investments needed compared to the NOxremoval achieved. With existing units, the firstoption to consider is the use of NOx reductionadditives. This is often the least expensive controloption and produces useful reductions in emissions.

Additives can play a role even if end-of-pipecontrols are applied since they can often reduce thesize or operating cost of these units. However, it is

very difficult to predict performance of additives.Several vendors offer different additives. A trialperiod is essential to determine how well they willwork in a specific application.

Good practices for control ofcatalytic cracker emissions

� Use cyclones to control emissions of largerparticles and to return catalyst to theregenerator.

� Consider the use of additives for NOx and SOxemission reduction.

� Consider end-of-pipe equipment that can controlmore than a single pollutant.

Sulphur plants

Refinery gas streams contain hydrogen sulphide. Itcan enter as a by-product of processes that removesulphur from unit feeds and products. It can alsoenter from sour water stripping. The potential SO2emission can be reduced if the H2S is removed fromsuch gas streams by, for example, amine treatingprior to their addition to the refinery fuel gas system.Not all gas streams need to be treated in this wayand the final refinery fuel gas sulphur content will bethe result of blending. The H2S removed from therefinery fuel gas supply streams needs to be routedto the sulphur recovery unit where it is destroyedand sulphur recovered as a solid product.


Any sulphur in gases that are combusted isassumed to form SOx. Material balances that tracktotal sulphur content of the sour gases, amount ofsulphur recovered and amount in combusted gaseswill provide sufficient information to calculateemissions with and without controls.

Periodic sampling of fuel gas for total sulphur isrecommended. Stack sampling of sulphur plant tail

gas emissions will also provide improvedquantification of SOx emissions.

Sulphur recovery

Hydrogen sulphide can be removed from gasstreams using chemical solvents, physical solventsor dry adsorbents.

Amine absorptionMany gas treating processes in refineries useamines to absorb the H2S. The choice of amineand system operating pressure will determine theenergy consumption and the removal efficiency.

In an amine treating unit, H2S is absorbed from thefuel gas stream and cleaned fuel gas is sent to therefinery fuel gas system. The contaminated amine issent to a regenerator for removal of H2S. Thecleaned amine solution is then returned to theabsorber to repeat this cycle. The concentrated H2Sgas next goes to the sulphur recovery unit.

Sulphur recovery unitsMost refineries use the Claus process to convert theconcentrated H2S stream from the amine treatingunit into elemental sulphur. In the Claus process, thegas stream containing the H2S is burned,producing SO2. The combustion gas then goesthrough a reactor where it is reduced to elementalsulphur. The sulphur is condensed and separated asa liquid. There are usually two or three stages ofthis reducing reaction depending on the desiredlevel of conversion to elemental sulphur.

The ‘tail gas’ from the Claus plant will still containa small amount of unconverted SO2, H2S andsome organic sulphides. For additional sulphurreduction, some refineries use a tail gas clean-upunit. These may provide further absorption usingamines or combustion and catalytic conversion. Theconversion of the third stage of a three-stage Claussystem to a tail-gas unit, using ‘SuperClaus’technology, is a common option, and can achieveup to 99.3% sulphur recovery. Other technologies

IPIECA

46

47


are needed to achieve higher overall efficiencies,but at more significant cost. Many locations send thegas from the Claus unit directly to an incinerator ifresidual sulphur concentrations are low enough toallow this option.

Other, smaller, potential H2S emission sourcesassociated with the sulphur recovery processinclude sulphur collection pits and sulphurstorage tanks. Emissions from these units mayalso need to be controlled. In some cases, it maybe more economical to collect and incineratethese H2S-containing gases and, if necessary, touse end-of-pipe controls to reduce the resultantSOx emissions.

Good practices for control of sulphurplant emissions

� Periodically sample sulphur content of refineryfuel gas.

� Avoid using high sulphur containing gases asfuels.

� Concentrate H2S streams using recovery unitssuch as amine scrubbers.

� Recover sulphur using Claus or other processes.� Collect and treat off-gases from sulphur pits and

sulphur storage tanks.� Monitor sulphur plant stack emissions.

Gas turbine NOx

As with other combustion sources, NOx controls forgas turbines either reduce NOx generation byreducing combustion temperatures or by removingNOx from the flue gas using end-of-pipe controls.The most widely used method to reduce NOxgeneration is steam injection. This reduces thecombustion temperature. The most widely usedmethod to remove NOx from flue gases is selectivecatalytic reduction (SCR).

Flares

Flares are a widely used control device forreducing the emissions of hydrocarbon vapours. Asplants have become more environmentally efficient,the amount of routine flaring has been significantlydecreased and most of the emissions are fromepisodic (upset) events. When the gases arecombusted they produce NOx and PM and, if anysulphur compounds are present, SO2.

There is an emission factor in AP-42 (US EPA,1995b) for estimating NOx from flaring. There areseveral factors in AP-42 for primary PM (soot)depending on the amount of smoking observed.These emission factors have high uncertaintybecause of differences in flare design and gascomposition. Emissions of SOx can be determinedby ‘material balance’ assuming all of the sulphurspecies are completely oxidized.

Source reduction

The most effective way to control flare emissions isto minimize the material sent to the flare. Improvedoperation of processes to reduce potentially ventedmaterial that is flared, as well as recycling ventedmaterial back into the process system are options.Some plants have initiated a ‘flare minimizationplan’ to focus efforts.

Flare minimization plans generally include alisting of measures to address flaring as a resultof planned major maintenance. This includesminimizing gas sent to the flare during startupand shutdown operations. It also includesidentifying methods to reduce flaring caused byfailure of air pollution control equipment, processequipment, or processes.

Gas recovery

Flare gas recovery is a system designed to collectthe hydrocarbon vapours that are sent to the flareand process them so that they can be returned to

the process or utilized as fuel. The system includespipe headers to collect the gases and condensers toremove any liquids. When the gas is used as afuel, the system includes a compressor to introducethe gas into the fuel gas system.

Although there is some economic benefit when therecovered gas is used as fuel, these systems aregenerally not cost-effective. They can be quiteexpensive if there is need to transport waste gasesfrom widely separated areas within a plant or ifthere are significant impurities in the gas that mustbe removed prior to use as a fuel.

IPIECA

48

49


Odour control and management

This section focuses on the management of odour. Itdoes not address the public communicationnecessary in the event of a major accident resultingin the bulk release of hazardous chemicals and theconsequent liaison with emergency services.

As an air emission odour presents specialchallenges. The source may not be known orreadily identified. If the odour is detectable outsidethe refinery fenceline and there is a negative publicreaction, the technical investigation needsappropriate public affairs support. Where odourevents are intermittent rather than persistent, thesuccessful location of the problem may dependcritically on analysis of the public response.

There is usually a wide range in individualresponses when people are exposed to an odorouscompound. Each person may detect odours at adifferent concentration and there may not be aconsensus on the degree to which an odour isconsidered objectionable. Some odours may evenbe considered pleasant. Habituation may reducesensitivity to nuisance associated with an odour.Perceptions of a change in odour intensity may beparticularly strong.

A process to address odour problems can bedivided into four steps. The first is to assess theproblem and establish the need for the facility totake action. This is followed by identifying thepotential sources of the odorous material bothwithin and external to the plant (some sources havebeen found to be external to the plant that receiveda complaint). Analysis of the potential sources andreported impact is then used to confirm the locationof the release and that it is the cause of the odourproblem. Finally, emission reduction actions areconsidered to bring the ambient concentration tolevels below the odour detection limit.

Problem assessment

To a substantial degree, the information needed toassess an odour issue will comprise reports fromthe public, and an appropriate infrastructure isneeded to handle and record, in a factual manner,information that is very likely communicated bytelephone. The same infrastructure can also workto handle reports from on-site personnel, andoperational issues that create temporary odoursources should be reported and recorded as soonas possible. It is also important to providefeedback to the public, and a means of recordinginformation provided, including details of whenand to whom the information was given, shouldbe included.

The first information needed to assess an odourproblem includes: the time and date of the call andthe identity and contact details of the reporter; thetime and date that the odour was noticed; thelocation at which the odour was noticed; thedistinguishing characteristics of the odour; and anyknowledge of the chemical and its source.

This time and location information will be needed tocross-reference with information on wind-speed anddirection, general weather conditions, etc., and withinformation on plant operations during the relevanttime, as an attempt to locate the source.

Where there is enough information to confirm andexplain the source of odour, and if the cause hasbeen recognized and quickly remedied, theproblem may be considered solved andappropriate feedback given.

When progressing to source identification, thepossibility that the source arises from outside thefacility should be considered. Where the potentialis high, for example in a manufacturing complexwith several operators, a coordinated approachmay be appropriate.

Source identification

Once the need for action has been established,steps are taken to identify the emission sourcesthat are causing the odours. Based on the odour,it is often possible to identify specific chemicalsand potential sources within the plant. Plantpersonnel may be able to determine the probableprocess units where the odour is originating sincethey are likely familiar with odours associatedwith specific process units. Many classes ofchemicals have distinctive odours that are easilyrecognizable (CONCAWE, 1975). In some cases,analysis of air samples may be necessary toidentify or confirm the chemicals involved. At thispoint, the likelihood of any potential hazardsassociated with exposure of plant personnel or thecommunity to the suspected chemical should bedetermined. Adequate safeguards should also bein place to detect releases of hazardous materials(e.g. H2S), together with appropriate emergencyresponse planning.

When the chemicals causing the odour have beenidentified, a list of locations in the plant wherethese chemicals may exist should be created. Theselocations are then surveyed to determine whetherthey are the source of emissions contributing to theodours reported. The location, wind direction andtiming of the odour reports should be examined todetermine whether specific plant areas may be theemission source.

There may be several plant sources that arecontributing to an odour. Emissions from all sourcesmay not need to be reduced to solve the odourproblem. While it is necessary to reduce theambient concentration of the chemical to below theodour detection threshold, a cost-effective solutionmay be to focus on the most appropriate source toremedy. This may be guided by carrying out animpact assessment.

Impact assessment and verification

In this step, the ambient concentrations of the odour-causing chemicals are quantified to help focusemission reduction efforts on the most appropriatesources. This can involve ambient air monitoring inthe community surrounding the plant, or emissionsestimating combined with dispersion modelling topredict concentrations. Actual or predicted ambientconcentrations are then compared to odourdetection thresholds and the locations of odourcomplaints; this provides a better understanding ofthe source contributions and allows odourreduction efforts to be focused appropriately.

Table 17 provides odour detection thresholds,published exposure limits and odour descriptionsfor some common industrial chemicals. Plantsshould check with their regulatory agencies for thelatest local limits on ambient concentrations.Information on a wider list of chemicals is availablein US EPA, 1992. The plant industrial hygienistshould be consulted to confirm acceptable exposurelimits for both plant personnel and the community.

The lack of a relationship between odourdetection thresholds and recommended exposurelimits can be demonstrated using the informationin Table 17. Hydrogen sulphide has a much lowerodour detection threshold than its exposure limits.This is a chemical that will be detected at muchlower concentrations than will cause any health orsafety issues, but this might not be understood bythe public.

An opposite example is ammonia, where the odourdetection threshold and the exposure limit arerelatively close. In this case, odour detection is anindication of a potential health hazard. Table 17shows that there are even some chemicals wherethe odour detection threshold is well above therecommended exposure limit.

IPIECA

50

51


Dispersion modelling can usefully guide monitoringprogrammes provided that there is sufficientinformation available to reasonably quantify theemissions and account for the local meteorology.The model should be able to account forconcentration fluctuations in order to reflect theshort-time peak concentrations typical of someodour problems. The guidance that is most useful isfor the location of monitors and the likelyconcentration ranges to be encountered.

Conducting an ambient air monitoring programmewill generally cost much more than emissionsestimating and dispersion modelling. In addition,monitoring has some limitations and care must beused in its application. When monitoring, it ispossible to miss the odour plume by having too fewmonitors or by placing them in non-optimal

locations for the local meteorology. In addition toplacing monitors in areas where there have beenpast odour complaints, typically expected winddirections should be considered when placingmonitors, so as to increase the likelihood that anyplant emissions will be detected.

In some cases, the odour detection threshold maybe lower than the analytical detection method limitfor the chemical. Monitoring may not always be ofbenefit in these cases. However, if the chemical isdetected above the odour threshold, this providesconfirmation of the problem and subsequentmonitoring can be used to show successfulelimination of the odour problem.

If emissions estimating and dispersion modelling ofworst-case conditions show that certain sources

Table 17 Example odour detection thresholds, exposure limits and descriptions

Chemical Odour threshold (ppm)

Exposure limits (ppm) OdourdescriptionTWA* STEL†

Acrylonitrile

Ammonia

Carbon disulphide

Chlorine

Dimethyl amine

Ethylene dichloride

Hydrogen sulphide

Methanol

Methyl amine

Methyl ethyl ketone

Naphthalene

Styrene

Sulphur dioxide

Toluene

Trichloroethylene

Garlic/onion

Irritating/pungent

Sulphur

Bleach

Fishy

Chloroform

Rotten eggs

Sweet and sour

Fishy

Sweet

Tar

Sweet

Irritation/pungent

Sour

Ether

1.6

17

0.0081

0.01

0.34

26

0.0005

100

4.7

10

0.027

0.047

2.7

2.8

82

2

25

1

0.5

5

1

10

200

5

200

10

50

2

100

50

10

35

10

1

15

2

15

250

15

300

15

100

5

150

100

*TWA: time weighted average—the allowable exposure level for a normal 8-hour day, 40-hour work week. †STEL: short-term exposure limit—the allowable exposure level for short periods of about 15 minutes.

would not result in ambient concentration near theodour detection threshold, it may be possible toeliminate these sources from further considerationas primary sources. However, they may still need tobe considered as secondary sources if they couldcause perceptible odours when combined withother sources.

There is a need to consider combinations of severalsources of the chemical which, in combination, mayresult in concentrations approaching or exceedingthe odour detection threshold. Analysis will showwhich combinations of sources can be reduced toeliminate the odour problem. If there are alternativesolutions, costs should be considered.

Problem resolution

Once the need for an odour control plan has beenestablished, specific emission sources can beidentified and applicable emission reductiontechniques can be evaluated. Many of the emissionreduction technologies discussed in this manualcan be applied. Emission reduction opportunitiesshould be evaluated based on the contribution ofeach source to the odour problem and the cost ofthe control.

The odour intensity of a specific chemical is relatedto the ambient concentration by Stevens’ Law:

I = kCn

where:I = odour intensityC = ambient concentrationK, n = constants

The value of the exponent in Stevens’ Law rangesfrom about 0.2 to 0.6, with most sulphur compoundsnear 0.3. Thus, for example, a 50% reduction inthe ambient concentration of a chemical with aStevens’ Law exponent of 0.3 would result in aperceived odour reduction of less than 20%.

Therefore, the emissions reduction required toachieve ambient concentrations below the odourdetection limit will be greater than if the perceivedodour intensity was proportional to the ambientconcentration. A consequence is that emissionreductions might require high cost investments.

Steven’s Law exponents are only available for alimited number of chemicals. The reportedexponents have some uncertainty since they weredeveloped using different protocols and odourpanels. Some Steven’s Law exponents are providedin Table 18. If no exponent is available for achemical of concern, the use of 0.3 isrecommended in initial evaluations.

Where there is no clear option to eliminate odoursources on a short time scale, alternative emissionreduction approaches should be discussed inconsultation with the public focal points.Consideration should also be given to balancingreductions of odour intensity with reductions ofodour impact frequency. In some cases, it may beadvantageous to decrease the frequency of theodour rather than the strength of the odour. Thismay be acceptable to the surrounding communityand may be beneficial for the plant since thefrequency of odour problems can often bereduced at lower cost through improvedoperations and housekeeping.

Table 18 Exponents for Steven’s Law equation

Chemical Exponent

0.26

0.29

0.58

0.25

0.34

0.53

Acetaldehyde

Butyl mercaptan

Chloroform

Ethyl mercaptan

Methyl mercaptan

Toluene

IPIECA

52

Good practices for addressing odourproblems

� Operations personnel should monitor for odoursand report them promptly.

� Keep surrounding community fully engaged.� Identify potential odour sources outside the

plant.� Compare actual impact areas with predictions

from emission and dispersion modelling.� Consider maximum short-time concentrations

when evaluating impacts.� Remember that the technical solution may not

always satisfy the surrounding community.

53


API (1998a). Fugitive Emissions from EquipmentLeaks I: Monitoring Manual. Publication No. 342,American Petroleum Institute, Washington, DC.May 1998.

API (1998b). Fugitive Emissions from EquipmentLeaks II: Calculation Procedures for PetroleumIndustry Facilities. Publication No. 343, AmericanPetroleum Institute, Washington, DC. May 1998.

API (2002/03). Manual of Petroleum MeasurementStandards, Chapter 19, Evaporative LossMeasurement. American Petroleum Institute,Washington, DC.

API (2005). Evaporative Loss from Storage TankRoof Landings. Technical Report 2567. AmericanPetroleum Institute, Washington, DC. April 2005.

API (2007). Evaporative Loss from the Cleaning ofStorage Tanks. Technical Report 2568. AmericanPetroleum Institute, Washington, DC. November2007.

CONCAWE (1975). The Identification andMeasurement of Refinery Odours. CONCAWEReport No. 8/75. The Hague. December 1975.

CONCAWE (2009). Air Pollutant EmissionEstimating Methods for E-PRTR Reporting byRefineries. CONCAWE Report No. 1/09. Brussels.January 2009.

IPIECA (2010). Petroleum RefiningWater/Wastewater Use and Management. IPIECAOperations, Fuels and Product Issues Committee‘Operations Best Practice’ Series, London, UK.October 2010.

US EPA (1992). Reference Guide to OdorThresholds for Hazardous Air Pollutants Listed inthe Clean Air Act Amendments of 1990.EPA-600/R-92-047. Environmental ProtectionAgency, Office of Research and Development,Washington, DC. March 1992.

US EPA (1995a). Protocol for Equipment LeakEmission Estimates. EPA-453/R-95-017.Environmental Protection Agency, Research TrianglePark, NC. November 1995.

US EPA (1995b). AP-42, Compilation of AirPollutant Emission Factors. Environmental ProtectionAgency, Research Triangle Park, NC. January1995.

US EPA (2004). Water9 (emissions estimationsoftware, Version 2). Environmental ProtectionAgency, Research Triangle Park, NC. July 2004.

US EPA (2010). Tanks (emissions estimationsoftware, Version 4.09d). Environmental ProtectionAgency, Research Triangle Park, NC. February2010.

IPIECA

54

References

www.ipieca.org

IPIECA5th Floor, 209–215 Blackfriars Road, London SE1 8NL, United KingdomTelephone: +44 (0)20 7633 2388 Facsimile: +44 (0)20 7633 2389E-mail: [email protected] Internet: www.ipieca.org

© IPIECA 2012 All rights reserved.

Association members

African Refiners Association (ARA)

American Petroleum Institute (API)

Australian Institute of Petroleum (AIP)

Brazilian Petroleum, Gas and Biofuels Institute (IBP)

Canadian Association of Petroleum Producers (CAPP)

Canadian Petroleum Products Institute (CPPI)

European Petroleum Industry Association (EUROPIA)

International Association of Oil & Gas Producers (OGP)

Japan Petroleum Energy Centre (JPEC)

Petroleum Association of Japan (PAJ)

Regional Association of Oil and Natural Gas Companies inLatin America and the Caribbean (ARPEL)

South African Petroleum Industry Association (SAPIA)

The Oil Companies’ European Association for Environment,Health and Safety in Refining and Distribution (CONCAWE)

United Kingdom Petroleum Industry Association (UKPIA)

World Petroleum Council (WPC)

Occidental

OMV

Petrobras

Petronas

Petrotrin

PTT EP

Qatargas

RasGas

Repsol

Saudi Aramco

Shell

SNH

Statoil

Talisman

Total

Woodside Energy

Addax Petroleum

BG Group

BP

Chevron

CNOOC

ConocoPhillips

eni

ExxonMobil

Hess

Hunt Oil

Husky Energy

KPC

Madagascar Oil

Mærsk

Marathon

Nexen

NOC Libya

Company members

IPIECA is the global oil and gas industry association for environmental and social issues. It develops,shares and promotes good practices and knowledge to help the industry improve its environmental andsocial performance, and is the industry’s principal channel of communication with the United Nations.

Through its member-led working groups and executive leadership, IPIECA brings together the collectiveexpertise of oil and gas companies and associations. Its unique position within the industry enables itsmembers to respond effectively to key environmental and social issues.

Date post:	04-Apr-2020
Category:	Documents
Upload:	others
View:	6 times
Download:	0 times

Refinery air emissions Good Practice Series management 2012 · Good Practice Series 2012 ......

Documents