Optimization of Process

0957-5820/01/$10.00+0.00# Institution of Chemical Engineers

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF PROCESS PARAMETERS FORGLYCOL UNIT TO MITIGATE THE EMISSION OF

BTEX=VOCs

A. M. BRAEK1, R. A. ALMEHAIDEB2, N. DARWISH2 and R. HUGHES3

1ADCO, Abu Dhabi, United Arab Emirates

2Chemical and Petroleum Engineering Department, UAE University, United Arab Emirates

3Chemical Engineering Unit, University of Salford, UK

Recent developments in environmental regulations on the emissions of aromaticcompounds including benzene, toluene, ethyl benzene and xylene isomers (BTEX)and other volatile organic compounds (VOCs) have made these emissions a major

concern in the natural gas industry. In particular, one of the sources of emissions is glycoldehydration units. During the dehydration of natural gas, the BTEX and VOC compounds in itare absorbed by the glycol solvent, and subsequently emitted to the atmosphere during thethermal regeneration of glycol. As a result, quanti®cation and means of reduction andultimately elimination of these emissions is gaining importance in the industry.

This problem requires careful attention during the design phase. The environmentalconsiderations mentioned above are increasingly driving the selection and operation of processalternatives to reduce the BTEX=VOCs emissions to the atmosphere.

This paper focuses on the glycol dehydration unit for one of the onshore oil and gasprocessing facilities in Abu Dhabi operated by Abu Dhabi Company for Onshore OilOperations (ADCO). Firstly, the BTEX=VOCs emission levels vented out to atmosphere arequanti®ed, and a description given of the sampling=measurement technique. Secondly, theprocess parameters associated with all major equipments in the glycol dehydration unit, i.e. theabsorber, the ¯ash tank, and the regenerator that may lead to the reduction in BTEX=VOCemissions, are optimized. The optimization study involves building a process simulation modelbased on actual design data=parameters provided by the glycol unit vendor and veri®ed byoperational data.

Keywords: natural gas; dehydration; glycol; process; emissions; aromatics.

INTRODUCTION

Figure 1 shows a simpli®ed process ¯ow diagram of one ofthe onshore oil production facilities located in Abu Dhabi,and operated by ADCO. Reservoir ¯uid from an oil-bearingreservoir is producedvia a numberof oil production wells andis subsequently gathered at the production manifold andseparated into three streams consisting of oil, water andassociated natural gas using a three-phase production separa-tor. The separated oil is then pumped to anothersite for furtherprocessing before it is exported. The water from the separatoris further processed in a water treatment plant before disposal.

The associated natural gas from the production separatoris recompressed and injected back into one of the oilreservoirs for pressure maintenance. This gas is wet andmust be dehydrated before it is recompressed and reinjectedinto the reservoir. Dehydration of this gas is necessary toreduce the potential for corrosion and avoid the possibilityof hydrates formation.

Two identical trains of four-stage compression units areused to recompress the associated natural gas. This study isbased on the glycol dehydration unit located between thesecond and the third stage of compression.

A typical gas dehydration unit is shown in Figure 2. Itconsists of a tri-ethylene glycol (TEG) absorber column,¯ash tank and a regeneration system, which consists of astripping column, reboiler and a TEG surge drum.

The wet gas enters the TEG absorber column for waterremoval by contacting it with lean (containing low concentra-tions of water and BTEX) TEG. The rich (containing a highconcentration of water and BTEX) glycol ¯ows to a ¯ash tankand is then regenerated in a common regeneration unitprovided for both trains to regenerate the glycol solvent.Finally, the dry gas is further compressed in the third andfourth stage of the reciprocating compressor for reinjection.Operating conditions for the above units are listed in Figure 2.

As can be noted from Figure 2, the two direct sources ofBTEX=VOCs emission into the atmosphere are the ¯ash tank

218

Fig

ure

1.

AD

CO

®el

dÐ

centr

alfa

cilities

area

.

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 219

Fig

ure

2.

Set

-up

of

pro

cess

model

for

GD

Uusi

ng

ven

dor

pro

vid

eddat

a.

Trans IChemE, Vol 79, Part B, July 2001

220 BRAEK et al.

and the regenerator identi®ed as streams # 5 and # 9 respec-tively. In the ¯ash tank, BTEX=VOCs gases are released as therich glycol is depressurized. In the regenerator, BTEX=VOCsare released with water vapourexiting during glycol regenera-tion. To quantify these emissions, chemical analyses of theabove streams have been carried out. In addition, the lean andthe rich glycol streams were also analysed to obtain anotherestimate of BTEX=VOCs emission from source # 9.

A process simulation model was set up to investigatethe in¯uence of various process parameters around theglycol dehydration system. Based on this study, severalprocess parameters are suggested to be adjusted to minimizethe BTEX=VOCs emissions.

LITERATURE REVIEW

Background on Aromatic Components

The aromatic series (type formula CnH2n-6), often calledthe benzene series, is chemically active. The aromatics mayform either addition or substitution products, dependingupon the conditions of reaction. Only a few types ofpetroleum contain more than a trace of the low-boilingaromatics such as benzene and toluene1. The aromaticseries of hydrocarbons is chemically and physically verydifferent from the paraf®ns and naphthenes. Also, it containsa benzene ring that is unsaturated but very stable andfrequently behaves as a saturated compound.

The C6-C8 aromatics (benzene, toluene, ethyl benzeneand xylene) are the large volume aromatics used by thepetrochemical industry; greatest demand is for benzene.The product from catalytic reforming contains all of thesearomatics and it is separated into its pure components bya combination of solvent extraction, distillation and crys-tallization. In addition, because of the much greaterdemand for benzene, the excess of toluene and xyleneover market needs can be converted to benzene by hydro-dealkylation2. Present separation methods for recovery ofaromatics from hydrocarbon streams use liquid-liquidsolvent extraction to separate the aromatic fraction fromthe other hydrocarbons; most of the processes used byU.S. re®neries use either polyglycols or sulfolane as theextracting solvent2.

BTEX is the collective term for the volatile organiccompoundsÐbenzene, toluene, ethyl benzene, andxyleneÐand is a constituent of gasoline and other petro-leum products. BTEX compounds are volatile and rela-tively soluble in water. The Gas Research Institute (GRI)lists the health hazards for the BTEX group3. It showstoluene, ethyl benzene and xylene as irritants with nar-cotic effects. Benzene has all these effects, in addition tobeing a human carcinogen. These health effects may comeeither through inhalation or ingestion of contaminatedgroundwater.

The relevant physical properties of the BTEX group areshown in Figure 3. Figure 3(a) shows the range of molecularweights, between 78 and 106, and the range of boilingpoints, between 80 and 145 C. Figure 3(b) shows solubilityof benzene and toluene in various glycols at 25 C. It showsthat TEG has the most af®nity for these compounds,followed by diethylene glycol (DEG), polypropyleneglycol (PG) and then ethylene glycol (EG). Figure 3(c)shows the solubility of BTEX in water. It shows benzene

has the most solubility followed by toluene and then ethylbenzene and the xylenes. Figure 3(d) gives the vapourpressures, showing that benzene is the most volatile of thegroup followed by toluene and ethyl benzene and thexylenes. Figure 3(e) shows the freezing points and Figure3(f ) the density and viscosity of BTEX. These properties areimportant to the understanding of the adsorption of BTEXby the glycols and their subsequent emissions as vent gasesto the atmosphere.

Dehydration Phenomena

Glycol selection for natural gas dehydration applicationsmay be based on a number of factors including dehydrationcapability, glycol losses in the contactor and regenerator,and absorption of VOCs. The most commonly used glycolin the industry is tri-ethylene glycol (TEG). Diethyleneglycol (DEG) and ethylene glycol (EG) may also be usedin dehydration applications; however, DEG and EG areoften not considered due to dry gas water content require-ments. However, using DEG and especially EG instead ofTEG when BTEX is a concern can greatly reduce BTEXemissions, and thus reduce emissions from the glycol stillvent. TEG has a higher degradation temperature and can beregenerated to a higher lean concentration with no modi®ca-tions to the standard regenerator reboiler. However, EG andDEG can meet water content speci®cations when used withenhanced regeneration systems. Enhanced regeneration isany system that improves glycol regeneration to achievea `leaner’ or more concentrated glycol solution. Enhancedregeneration can be achieved either by the injection ofstripping gas into the reboiler, azeotropic regeneration, orother proprietary processes4.

The main objective for dehydration is removal of waterfrom process gas to:

° Meet water dew point requirement of sales gas that isstipulated by users.

° Prevent hydrate formation in downstream units with lowoperating temperatures.

° Prevent pipeline corrosion, since process gas may becontaminated by acid gases (CO2=H2S).

TEG dehydration is a diffusion process; the watermolecules must diffuse through the gas phase, across thegas=liquid interface, and then to the liquid phase. Thedriving force is the difference between partial pressure ofmoisture in the gas phase and concentration of water in theliquid phase.

Henry’s law states: Cl = H Pl (1)

where Cl = concentration of moisture in liquid phase; H =solubility coef®cient; Pl = partial pressure of moisture inequilibrium with liquid of concentration Cl.

Gas phase resistance is inevitable since moisture moleculesalways haveto diffuse through a large amountof insolublegasto reach the interface. Liquid phase resistance is negligible5.

The water vapour in the gas phase is physically absorbedin the glycol solution in an absorption tower referred to as acontactor. In the design of a glycol system, it is generallyassumed that water is the only solute. However, any compo-nent that is present in the inlet gas stream will be present inthe rich glycol stream leaving the bottom of the contactor,and aromatics (BTEX) are particularly soluble in the glycol.

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 221

Fig

ure

3.

Phy

sica

l =ch

emic

alpro

per

ties

of

BT

EX

,w

ater

and

TE

G.(a

)B

oilin

gpoin

tan

dm

ole

cula

rw

eightofB

TE

X.(b

)Solu

bility

ofvar

ious

BT

EX

ingly

col@

25

¯ C.(c

)Solu

bility

of

BT

EX

inw

ater

@25

¯ C.(d

)V

apour

pre

ssure

of

BT

EX

.(e

)B

TE

X,

gly

col

and

wat

erfr

eezi

ng

poin

ts.

(f)

Den

sity

and

vis

cosi

tyof

BT

EX

.

Trans IChemE, Vol 79, Part B, July 2001

222 BRAEK et al.

Therefore, signi®cant amounts of aromatics may beabsorbed during the absorption process in the contactorand ultimately rejected to the atmosphere at high tempera-ture during regeneration.

All glycols may be used in the dehydration process, butTEG is the most common glycol for the commercialdehydration of natural gas. The solubility of hydrocarbonsin TEG is strongly in¯uenced by type of homologous series.Paraf®ns and naphthenes range from insoluble to onlyslightly soluble. However, aromatics are very soluble in TEG.

Historically, the solubility of aromatics in TEG has beenof minor concern in the dehydration of natural gas. This hasbeen the case for several reasons. Firstly, the concentrationof aromatics in naturally produced gas streams is usuallyvery small. Secondly, analytical techniques for accuratelymeasuring the concentration of aromatics are relativelyrecent. Aromatics are typically lumped into the C6 orC6+ fraction and treated as normal paraf®ns for equilibriumcalculation purposes.

Finally, any aromatics that are picked up in the TEG willbe vaporized in the reboiler, and vented to atmosphere. Thepresence of these aromatics in no way interferes with thedehydration of the gas. In fact, the presence of aromatics inthe reboiler can provide stripping gas, which assists the re-concentration of the TEG6.

GDU Process Description

Gas dehydration units (GDUs) are typically representedby a contactor, a ¯ash tank, heat exchangers and a regen-erator, as shown in Figure 4. The lean (i.e. relatively low inwater content) glycol, usually TEG, enters at the top of thecontactor and absorbs water from the wet feed gas comingup as it progresses toward the bottom of the column. A drygas exits at the top of the contactor and may be used forcooling the incoming lean glycol recycle from the regen-erator unit.

The rich glycol stream (rich in water vapour) is used as are¯ux cooling medium for the still vent regenerator gasbefore it ¯ows to the ¯ash tank, where gaseous hydrocar-bons that were absorbed along with some of the water in thecontactor are liberated and may be used as fuel gas. Finally,the glycol is preheated through a heat exchanger by the leanglycol from the reboiler before it enters the still. Water isdistilled or stripped from the glycol in the still and reboiler.In some units, a stripping gas is injected into the reboiler toimprove the water removal ef®ciency and also to help inreducing the reboiler duty. The lean glycol ¯ows to a surgedrum that acts as a heat exchanger on small units. From thesurge drum, the lean glycol is recycled back to the absorber(contactor) by the energy exchange pump, passed through agas=glycol exchanger, and routed to the top of the absorber7.

International Environmental Regulations

In 1990, several new regulatory programmes were estab-lished in the U.S. under the Clean Air Act Amendments(CAAA) that changed the way gas processors viewed theirplants. Under Title III, several aromatic compounds includ-ing benzene, toluene, ethyl benzene and xylene, collectivelyknown as BTEX, were included on a list of 189 hazardousair pollutants (HAP) from major sources and area sources.A major source is any stationary source or group of stationary

sources within a contiguous area and under common controlthat emits more than 10 tons=year (tpy) of any one of theHAPs or 25 tpy of total HAPs. An area source is de®nedsimply as any stationary source of HAPs that is not a majorsource8. These environmental regulations on BTEX emis-sions from glycol units have caused many plants to becategorized as major sources9. For this reason, these facil-ities were subjected to permit requirement under Title V ofCAAA requiring the use of Maximum Achievable ControlTechnology (MACT) to minimize emissions for HAPS.To analyse for BTEX emissions, special procedures werenecessary and the amounts in question were typically inparts per million. The industry has given increased attentionto these compounds, and in mid-1990 the Gas ProcessorsAssociation (GPA) and American Petroleum Institute (API)jointly funded a research project to investigate the solubilityof the BTEX compounds in TEG solutions. The results ofthis work were published in a GPA Research Report.

ADNOC and UAE Governmental Guidelineson Air Pollution

United Arab Emirates (UAE) Federal law number 24 for1999 for Protection and Development of Environment10, onthe issue of environmental monitoring, states: `The (FederalEnvironmental) agency in co-ordination with the competentauthorities and concerned departments shall develop anational monitoring system for the environment includingestablishing, operating and supervising the environmentalmonitoring networks, then the environmental monitoringnetworks shall notify the agency, competent authoritiesand concerned departments of any abusive use of pollutants,and they shall regularly submit reports on the results of theiractivity as per the executive order’.

It is highlighted in the ADNOC guidelines11 that, duringoperation or decommissioning of the project, discharge tothe air, land, surface water, or ground water of theBTEX=VOC components shall not exceed the followinglimits:

Air pollutant Metric ton=yr Kg=24 hr

BTEX 22 90

VOC 227 910

An environmental impact assessment (EIA) is mandatoryfor all new projects and projects involving modi®cations toexisting operations. In the EIA process all relevant signi®-cant environmental issues must be identi®ed and dulyaddressed. It is noteworthy that, because different organiza-tions de®ne VOCs differently, the VOCs are de®ned byADNOC as all volatile hydrocarbon compounds, includingBTEX, except methane12.

Sampling Techniques

As no standard methods have been developed by the U.S.Environmental Protection Agency (EPA), a variety ofsampling points, sampling protocols and analytical tech-niques have been used by industry for measurements ofemissions from glycol dehydration units13. Out of tenmethods identi®ed by an industry group, Rueter et al.9

reported that three of these methods were most practical.These were atmospheric rich=lean (ARL) glycol and pres-surized rich=lean (PRL) glycol sampling, both on the liquid

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 223

Fig

ure

4.

Pro

cess

¯ow

dia

gra

mfo

rG

DU

quan

tify

ing

BT

EX

and

VO

Cs

emis

sions

usi

ng

on-s

ite

com

posi

tions

for

inle

t®el

dgas

.

Trans IChemE, Vol 79, Part B, July 2001

224 BRAEK et al.

side, and total capture condensation (TCC) on the stack gasside. And out of these three, Rueter et al.9 reported that ARLis the most effective method technically and economicallyfor industrial application. The disadvantage of the TCCmethod is the dif®culty in accurately measuring the ef¯uentgas ¯ow rates. The PRL technique did not produce signi®-cantly better results than the ARL technique. They alsohighlighted the advantage of using simulation as a screeningtool to estimate emissions from glycol dehydration units(GDUs). Based on this and other works, a standard ASTMguide14 for collection and sampling of BTEX=VOCs fromglycol dehydration units, using the atmospheric rich=leanglycol technique, has been formulated.

Methods for Minimizing BTEX=VOC Emissions

There are several strategies for reducing the amount ofBTEX=VOCs emissions from glycol still vents4,15. Thesecan be summarized as:

IncinerationIncineration of regenerator vent gases eliminates the

BTEX=VOCs emissions. However, this option may not beviable as it is not consistent with international trends in theindustry and contradicts the ADNOC group goal of eliminat-ing or minimizing all operational ¯aring. The objective of thispaper is to minimize BTEX=VOCs emissions from glycolsystems in order to comply with ADNOC guidelines on airpollution.

CondensationAdding a condensation unit to the regenerator vent stream

to recover BTEX and VOCs is economical and easy tooperate. However, from an environmental standpoint, itcreates the problem of disposal of the water phase contain-ing the BTEX compounds.

Using new process designsThe Drizo process, for example, is a good choice for

many systems requiring steep water dew point depression,as control of BTEX emissions is effectively included withthe package. However, the disposal of the condensed waterstream and non-condensable gases from the Drizo processmust still be considered. Another process is the R-BTEXprocess, which is based on condensing the overhead gasesfrom the regenerator. The R-BTEX process, despite mini-mizing BTEX emissions, still has the problem of disposalof non-condensable gases. The IFPEXOL process, which isrelatively new, uses methanol in a cold process to extractwater vapour from the gas stream.

Process optimizationChanges in the operating conditions of glycol circulation

rates and the ¯ash tank can result in a signi®cant reductionin BTEX emissions. The ¯ashed gas generated from the¯ash tank must be incinerated or used as fuel since itcontains signi®cant VOCs. Also, decreasing absorber pres-sure and increasing absorber temperature tends to decreaseVOCs emissions. Although helpful in reducing emissions,reducing absorber pressure may not be feasible due to thecost of sales gas recompression.

Alternative glycolsSelecting a different glycol such as diethylene glycol

(DEG) or ethylene glycol (EG), which absorbs the leastamount of BTEX, may also play a major role in emissionreduction. DEG and EG have the added bene®t of being lesscostly than TEG and require less energy for regeneration.Additional equipment for enhanced regeneration, if neces-sary, is minimal. However, they have a disadvantage in thatthey are less stable at the high regenerator temperature.

When emissions are a concern, the primary focus shouldbe on meeting emissions requirements. The system must, ofcourse, also meet water dew point requirements. Finding theoptimum system that meets process constraints and require-ments at the least possible cost becomes the objective indesigning a dehydration unit. A process simulator may beused as a tool for determining the optimum design. Thisphase of the work looks into effects of process parameterson BTEX and VOCs emissions; this does not involvesigni®cant capital expenditure and therefore is the easiestto implement.

In¯uence of Process Optimization

System optimization is a pollution prevention techniquethat may be applied to glycol dehydration by adjustingoperating parameters. Since glycol dehydration units maycirculate more glycol than is necessary to meet contractspeci®cations, reducing the glycol circulation rate is onesimple way to lower emissions, i.e. within the range of 2 to 3gallons of glycol per pound of water content to be removedfrom the natural gas stream. Low glycol circulation ratesdecrease the amount of BTEX absorbed from the natural gasstream. Therefore, less BTEX is released from the glycolreboiler vent during regeneration.

In addition, optimizing the ¯ash tank temperature andpressure (i.e. operating at the highest temperature and lowestpressure that are practical) may reduce the VOCs emissionsfrom the regenerator still and allow recovery of these as ¯ashgas to be used in the fuel gas system. However, adjustingthose parameters (pressure and temperature) on the ¯ashtank will not signi®cantly affect BTEX emissions, becausethese mostly remain with the liquid (rich glycol) stream.Other factors that signi®cantly in¯uence the amount ofBTEX absorbed into the glycol, and consequently increasethe rate of emissions, are feed gas temperature and BTEXcontent in the feed gas. Most operators, however, have littlecontrol over these process variables, so adjusting the gastemperature or BTEX content are not feasible emissionscontrol options16.

Thus, the signi®cant parameters that affect the amount ofhydrocarbons emitted in the regenerator vent of gas dehy-dration systems may include17:

° Amount of BTEX hydrocarbons in the plant feed.° Glycol circulation rate through the contactor.° Contactor pressure and temperature.° Flash tank operating conditions.

Of these parameters, only the ®rst cannot be changed dueto the ®xed amount of BTEX content in the wet gas feed tothe contactor. The second and fourth can be varied, and thecontactor pressure and temperature may also be varied tosome degree within the equilibrium needed to achieve thesales gas speci®cations stipulated by the customer.

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 225

RESULTS AND DISCUSSION

Set-up of the Process Model

The case study presented in this work is based on thecomputer simulation of ADCO’s dehydration facility in AbuDhabi. The process ¯ow diagram (PFD) of this unit, assupplied by the vendor, was used in the simulation using aprocess simulator. As the ®rst logical stage is to calibrate thesimulation model, the base case input data supplied by thevendorwere employed.The ¯ow rates, composition and otheroperating conditions of the wet gas assumed in the base caseare presented in Tables 1 and 2. It is noteworthy that the wetgas as in this base case contains no BTEX compounds. Figure3 shows the set-up of the current facility. The predictivecapability of the simulation model for the glycol unit isquite obvious in Table 1 and Figure 3, which means that theactual measured parameters (operating data) and vendordesign data are similar to those predicted by the simulationmodel. In the second step, a recent BTEX lab analysis thatquanti®ed the BTEX components in the wet gas feeding intothe glycolunit was used. This composition is shown in Table 2along with the vendor’s composition data. Then, the simula-tion model was restarted again to measure the BTEX emis-sions from the glycol unit. The summarized results of BTEXmass balance are presented in Table 3 and Figure 4.

As shown in Table 3, based on the simulation results,BTEX emissions from the plant including the ¯ash tank andthe reboiler vent stream are 300 kg=day (110 metric tons=yr).The VOCs emissions rates are 2447 kg=day (903 metrictons=yr). Both ®gures exceed ADNOC’s guidelines. Toverify these results, a laboratory analysis was performedon the composition of the reboiler vent stream. Results areshown in Table 4. Due to the dif®culty in measuring theef¯uent ¯ow rates from the vent stream, the simulated ¯owrates based on the design data are used. Based on these data,the emissions of BTEX and VOCs from the reboiler ventstream are 621 kg=day and 4795kg=day respectively, as

shown in Table 4, again exceeding emissions guidelinesand more than twice the simulated emissions.

In¯uence of Process Operating Parameters

The operating conditions in the glycol system will in¯u-ence the emissions from the glycol regeneration unit.However, suitable selection or control of the operatingconditions will minimize the BTEX=VOCs emissions.

Table 1. Comparison between the vendor design data and the current simulation modelÐbased on vendor’s dataÐand the actual operating plant data.

Description Glycol vendor data

Simulation output Actual plant data

Summer Winter Summer Winter

Gas ¯ow rate (MMSCFD) 17.5 17.54 17.57 9.76 13Contactor operating conditions

Gas side:Temperature, ¯C 56 56 56 56 56Pressure, bara 38 38 38 41 41

Glycol side:Temperature,¯C 61 61 61 60 57Pressure, bara 43 43 43 41 41

Flash drum operating conditionsTemperature, ¯C 67.3 69.65 66.81 70 52Pressure, bara 4 5 5 5.2 5.2

Regenerator operating conditionsTemperature, ¯C 204 204 204 190 190Pressure, bara 1.1 1.1 1.1 atm atm

Glycol circulation rate, m3=hr 2 2 2 2.2 2.12Stripping gas, MMSCFD 0.121 0.121 0.102 0.064 0.034Water dew point, lb=MMSCF 1.886 3.5 3.3 4.5 4.5Reboiler duty, MMBTU=hr 0.59 0.62 0.51 0.932 0.932Still column temperature, ¯C 82.6 87 82 85 85Still column ¯ow rate, MMSCFD 0.27 0.29 0.19 No device No deviceTEG purity, Wt.% 99.8 99.8 99.8 98 99.64

Table 2. Wet feed gas compositions to glycol unit.

parametersMole% Design gas composition On-site gas analysis

Fluid componentsH2S 0.027 0Co2 4.957 4.36N2 0.620 0.76C1 63.314 68.45C2 11.15 10.19C3 9.088 7.32IC4 2.501 1.92Nc4 4.217 3.47IC5 1.536 1.11NC5 1.469 1.07N-C6 0.721 0.65C7 0 0.094C8 0 0.022C9 0 0C10 0 0Water 0.4 0.4Methyl cyclo pentane 0 0.06Cyclo hexane 0 0.07Methyl cyclo octane 0 0.01Benzene 0 0.030Toluene 0 0.014Ethyl benzene 0 0O,M, P-xylene 0 0Total 100% 100%

Trans IChemE, Vol 79, Part B, July 2001

226 BRAEK et al.

In the following discussion, the impact of a number ofoperating parameters simulated by the process model areaddressed:

Glycol circulation rateThe BTEX=VOCs emissions from the ¯ash tank and

regenerator are shown in Figures 5(a)±5(d). The ®gures

show that emissions from both the regenerator vent streamand the ¯ash tank increase directly with higher glycolcirculation rate. Also, as shown in these ®gures, for every0.2 m3=hr increase in rate of TEG, there will be an increaseof about 10 tpy of total BTEX emissions and 37.5 tpy oftotal VOCs emissions. Therefore, if the TEG rate is reducedfrom 2 m3=hr currently to 1.2 m3=hr, a signi®cant reductionof 50 tpy of BTEX emissions and 150 tpy of VOCs wouldbe realized.

Absorber operating conditions (temperature and pressure)The effect of absorber pressure on BTEX and VOCs

emissions is displayed in Figures 6(a)±6(d). For the ¯ashtank, Figures 6(a) and 6(b), the emissions of BTEX=VOCs increase with increasing absorber pressure. For thepractical range of absorber pressure between 500 and600 psig, the graph shows that BTEX emissions increaseby about 0.5 tpy and VOCs emissions increase by about25 tpy, both at the ¯ash tank. For the regenerator, Figures6(c) and 6(d), both BTEX and VOCs emissions decreaseby about 10 tpy if the absorber pressure is increased from500 to 600 psig. Therefore, there would be a net decreasein BTEX emissions and an increase in VOCs emissions ifthe absorber pressure is increased from 500 to 600 psig.The effect of absorber temperature is shown in Figures7(a)±7(d). For the ¯ash tank, Figures 7(a) and 7(b), theBTEX and VOCs emissions increase slightly at higherabsorber temperature. For the regenerator, Figures 7(c)and 7(d), BTEX emissions decrease signi®cantly at highertemperatures while VOCs emissions decrease slightly.Note that the location of the glycol contactor in theplant facilities is between the second and third stages ofthe compression unit; thus a signi®cant change of absor-ber pressure or temperature will not be possible withoutprocess modi®cation, which is not considered in thisstudy.

Re¯ux coil temperatureFigures 8(a) and 8(b) show that BTEX emissions increase

from the ¯ash tank with increased re¯ux coil temperaturewhile VOCs emissions are relatively unaffected. For theregenerator, Figures 8(c) and 8(d) show that neither BTEXnor VOCs emissions are signi®cantly affected by changes inthis parameter.

Flash tank pressureThe uncontrolled BTEX and VOCs emission levels, as

estimated in both the ¯ash tank and reboiler vent streams,are shown in Figures 9(a)±9(d). One can observe thatreductions in emissions at the ¯ash tank at increased ¯ashtank pressure are almost balanced by increases in emissionsat the reboiler vent stream. Therefore, this parameter has aninsigni®cant effect on the total emissions, as whateverBTEX=VOCs not vented at the ¯ash tank will be emittedat the regenerator.

Reboiler temperatureReboiler temperature can only be considered as an

optimization parameter in ranges far from the thermaldegradation point of the TEG, which is around 400 F. Thereboiler temperature is expected to in¯uence the BTEX and

Table 3. Comparison between the current and optimized conditions for thegas dehydration unit.

Current processmodel run

Optimized processmodel run

a) Optimized parameters1) Flash drum operating conditions

Pressure, bara 4 6Temperature, ¯C 67 66

2) Regenerator operating conditionsPressure, bara 1.1 1.1Temperature, ¯C 204 193.3

3) Glycol circulation rate, m3=hr 2 1.24) Stripping gas rate, kgmole=hr 5.43 3

(b) Calculated parameters1) Water dew point, lb=MMSCF 3 32) Still column temperature, ¯C 86 893) Reboiler duty, MMBTU=hr 0.6 0.344) Gas ¯ow rate, MMSCFD 17.5 17.5

5) TEG purity (wt.%)Lean TEG purity 99.8 99.8Rich TEG purity 96 94

(c) BTEX=VOCs emission rate:(1) BTEX emission rate, kg=24 hr 300 157(2) VOCs emission rate, kg=24 hr 2474 1362

Table 4. On-site measurement for the overhead gas from the still-columnregenerator.

Fluid components On-site gas analysis, mole%

H2S 0Co2 7.31N2 0.84C1 53.38C2 10.32C3 9.57IC4 2.70Nc4 5.80IC5 2.22NC5 2.36N-C6 1.92C7 0.41C8 0.14C9 0.07C10 0.08C11 0.01Methyl cyclo pentane 0.25Cyclo hexane 0.29Methyl cyclo hexane 0.14Benzene 0.75Toluene 0.91Ethyl benzene 0.11M, P-xylene 0.27O-xylene 0.15

Total 100%

° Feed wet gas = 17.7 MMSCFD as per design ®gure given by glycolvendor.

° Still overhead gas = 0.2584 MMSCFD as per the simulation output run.° Emission from overhead still column:

BTEX emission, kg=day 621VOC emission, kg=day 4795

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 227

Figure 5. (a) Glycol circulation rate vs. uncontrolledBTEX emission from ¯ash tank. (b) Glycol circulation rate vs. uncontrolledVOC emission from ¯ash tank.(c) Glycol circulation rate vs. uncontrolled BTEX emission from regenerator. (d) Glycol circulation rate vs. uncontrolled VOC emission from regenerator.

Figure 6. (a) Absorber pressure vs. uncontrolled BTEX emission from ¯ash tank. (b) Absorber pressure vs. uncontrolled VOC emission from ¯ash tank.(c) Absorber pressure vs. uncontrolled BTEX emission from the regenerator. (d) Absorber pressure vs. uncontrolled VOC emission from the regenerator.

Trans IChemE, Vol 79, Part B, July 2001

228 BRAEK et al.

VOCs emissions by affecting the water content of the lean glycol. It is evident from Figures 10(a) and 10(b) that the

Figure 7. (a) Absorber temperature vs. uncontrolled BTEX emission from ¯ash tank. (b) Absorber temperature vs. uncontrolled VOC emission from ¯ashtank. (c) Absorber temperature vs. uncontrolled BTEX emission from regenerator. (d) Absorber temperature vs. uncontrolled VOC emission from regenerator.

Figure 8. (a) Re¯ux coil temperature vs. uncontrolled BTEX emission from ¯ash tank. (b) Re¯ux coil temperature vs. uncontrolled VOC emission from ¯ashtank. (c) Re¯ux coil temperature vs. uncontrolledBTEX emission from regenerator. (d) Re¯ux coil temperature vs. uncontrolledVOC emission from regenerator.

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 229

reboiler temperature is not affecting the emission level fromthe regenerator vent stream. This means that the perfor-mance of the regenerator is governed totally by the strippinggas rather than the boiling-off ratio of the reboiler. Thisconclusion is also supported by the results displayed next onthe effect of the stripping gas rate on BTEX and VOCsemission levels.

Stripping gasFigures 11(a) and 11(b) shows an approximate 20%

increase in BTEX emissions if the molar ¯ow rate for the

stripping gas is increased from 0.05 MM SCFD to0.3 MMSCFD. Figure 11(b) shows a marked increase inVOCs emissions for these higher stripping gas ¯ow rates,because the stripping gas injected in the reboiler goes outwith the vent stream as VOCs.

Optimized Run

Based on the parametric study above, the signi®cantparameters chosen for the optimization study were theglycol circulation rate and the stripping gas rate. The ¯ash

Figure 9. (a) Flash tank pressure vs. uncontrolled BTEX emission from ¯ash tank. (b) Flash tank pressure vs. uncontrolled VOC emission from ¯ash tank. (c)Flash tank pressure vs. uncontrolled BTEX emission from regenerator. (d) Flash tank pressure vs. uncontrolled VOC emission from regenerator.

Figure 10. (a) Reboiler temperature vs. uncontrolled BTEX emission from regenerator. (b) Reboiler temperature vs. uncontrolled VOC emission fromregenerator.

Trans IChemE, Vol 79, Part B, July 2001

230 BRAEK et al.

tank pressure and the reboiler temperature were also modi-®ed. An optimized run was made choosing the optimumvalues for these parameters. The results of this run arecompared in Table 3 to the current process model results. Anoverall reduction of 48% in BTEX and 45% in VOCsemissions was achieved from the plant by using theseoptimized parameters.

CONCLUSIONS

Based on the above discussion, the following conclusionscan be drawn:

(1) Based on these preliminary measurements and simula-tions, the plant is shown to exceed ADNOC’s guidelinesfor BTEX=VOCs emissions. Therefore, a method forminimizing BTEX=VOCs emissions needs to be imple-mented.

(2) The process model indicates that the signi®cant para-meters that can be modi®ed to minimize emissionsare the glycol circulation rates and the stripping gasrate. Flow rates higher than 3 gallons TEG=lb waterdo not enhance water dew point depression but willgreatly increase the hydrocarbon and aromaticsabsorption in the absorber, including the BTEXcompounds.

(3) Optimization of glycol unit process parameters resultedin a 48% overall reduction of BTEX and 45% overallreduction in VOCs emissions from both the ¯ash tankand the regenerator still vent.

(4) In order to operate the absorber at a higher tempera-ture and thus take advantage of the effect of absorbertemperature on reduced BTEX emissions, a processmodi®cation is envisioned for further study. Thismodi®cation involves eliminating the air-cooler forthe wet gas stream and reducing the duty of the air-cooler for the lean TEG stream in order to maintainthe required 10¯C temperature difference between theTEG and wet gas streams.

(5) While the reduction in emissions due to processoptimization is signi®cant, in order to meet emissionstandards, the plant probably also needs to apply othermethods such as incineration, condensation or processmodi®cation.

Also, as this is a preliminary study and the aboveconclusions are based on the available data, it isrecommended that the indicated emission levels be veri®edby conducting a more detailed analysis for the BTEX=VOCslevels by sampling the wet feed gas stream, the dry gasstream, the ef¯uent streams from both the ¯ash tank and theregenerator still column, and on the rich and lean glycols,and to obtain a better match between the actual plantdata and the simulation results relevant to BTEX=VOCsemissions.

NOMENCLATURE

Cl concentration of moisture in liquid phaseH solubility coef®cientPl partial pressure

REFERENCES

1. Nelson, W. L., 1985, Petroleum Re®nery Engineering, Fourth Edition(University of Tulsa, McGraw-Hill Book Company).

2. Gary, J. H. and Handwrok, G. E., 1984, Petroleum Re®ning Techno-logy and Economics, Second Edition (Marcel Dekker, Inc).

3. http:==www.gri.org=pub=oldcontent=tech=e + s=mgpsites=chemical.htm.4. Ebeling, H. O., Lyddon, L. G. and Covington K. K., 1998, Reduce

emissions and operating costs with appropriate glycol selection,Proceedings of the Seventy-second GPA Annual Convention, Tulsa.

5. Ghoshal, P. C. and Mukhopadhyay,S., 1994, Improve glycol dehydra-tion unit ef®ciency, Hydrocarbon Processing, March 1994.

6. Fitz, C. W. Jr. and Hubbard, R. A., 1987, Quick, manual calculationestimates amount of benzene absorbed in glycol dehydrator, Oil & GasJournal, November 1987.

7. Hlavinka, M. W., Colllie, J. and Ashworth A., 1998, An analysis ofBTEX emissions from amine sweetening and glycol dehydrationfacilities, Proceedings of the 1998 Laurance Reid Gas ConditioningConference, University of Oklahoma.

8. Rueter, C. O., Murff, M. C., and Beitler, C. M., June 1996, GlycolDehydration Operations, Environmental Regulations, and WasteStream Survey (Gas Research Institute, Environment & SafetyResearch Group).

9. Rueter, C. O., Reif, D. L., Menzies, W. R. and Evans, J. M., 1995,Measurement and enhanced monitoring of BTEX and VOC emissionfrom glycol dehydrators, SPE 29698, Advanced Technology Series4(2).

10. Federal Law No. 24 for 1999 for Protection and Development ofEnvironment, UAE.

11. ADNOC Group Guidelines on Health, Safety & Environmental ImpactAssessment, April 1997.

12. ADNOC Report No. 2.59=197, Methods for Estimating AtmosphericEmissions from E & P Operations, ADNOC ± E&P Forum, September1994.

Figure 11. (a) Stripping gas rate vs. uncontrolled BTEX emission from regenerator. (b) Stripping gas rate vs. uncontrolled VOC emission from regenerator.

Trans IChemE, Vol 79, Part B, July 2001

OPTIMIZATION OF GLYCOL UNIT TO MITIGATE EMISSIONS OF BTEX=VOCs 231

13. Rueter, C. O., Reif, D. L. and Evans, J. M., 1993, Development ofsampling and analytical methods for measuring BTEX and VOC fromglycol dehydration units, SPE 25944 presented in SPE=EPA Explora-tion & Production Environmental Conference, San Antonio, Texas,USA, 7±10 March 1993.

14. American Society for Testing and Materials, 1995, ASTM E 1752-95:Standard Guide for Collection of Multi-Media Field Emission andDischarge Data Associated with Glycol Dehydration Units.

15. Smith, S. and Skiff, B., 1991, Drizo gas dehydration, an aromaticsemissions solution for low dew point projects, Proceedings of the GasProcessors Association Regional Meeting, Midland, Texas, 2 May1991.

16. http:==www.gri.org=pub=oldcontent=tech=e ‡ s=glydehy=emctrl.htm17. Hlavinka, M. W. and Bullin, J. A., 1993, In¯uence of process

operations on VOC and BTEX emissions from glycol dehydrationunits, Proceedings of the Seventy-second GPA Annual Convention,Tulsa.

ACKNOWLEDGEMENT

The authors would like to thank the management of Abu DhabiCompany for Onshore Oil Operations (ADCO) for providing data andtheir permission to publish this paper.

ADDRESS

Correspondence concerning this paper should be addressed to Mr. A.M.Braek, Engineering and Major Projects Division, ADCO, PO Box 270,Abu Dhabi, United Arab Emirates. E-mail: [email protected]

The manuscript was received 6 November 2000 and accepted forpublication after revision 10 May 2001.

Trans IChemE, Vol 79, Part B, July 2001

232 BRAEK et al.