Acid Gas Water Content and Physical Properties
Content
Acid Gas Water Content and Physical Properties: Previously Unavailable Experimental Data for the Design of Cost Effective Acid Gas Disposal Facilities, an Emission Free Alternative to Sulfur Recovery Plants
By M.A. Clark, W.Y. Svrcek. W.O. Monnery: University of Calgary A.K.M. Jamaluddin: Hycal Energy Research Laboratories
E. Wichert: SogaproEngineering Ltd.
Abstract
The economicsof recoveringsulfur from sour natural gas have become unfavorablefor small fields. Hydrocarbonproducing companies require a cost effective yet environmentallysound alternativeto deal with acid gas. Compressed acid gas re-injection into producing,depleted or non-producing formationshas emergedas a viable alternativeto traditionalsulfur recovery. Most injection schemes include dehydration facilities to remove the saturated water from the gas,
preventing corrosion and hydrate formation. An alternative, less
costlyapproach is proposed:
keep the water in the acid gas phase throughout the injection circuit, eliminating the need to dehydrate.
To design an optimized injection strategy, determination of thermodynamic and physical propertiessuch as water content, dewpoint, bubble point, hydrateconditions and density of the acid gas is necessary. A comprehensivejoint industry/research project is underway to obtain data for several different acid gas compositionsover a range of operating temperatures and pressures. An online, repeatable and precise method of water content analysis using gas chromatographyhas been designed and tested. Pure CO2 data has been generated from the equipmentand matchesacceptedvalues with an averagestandarddeviationof 5 %. To date the phase behaviorof an acid gas mixturecontaining89.5%CO2, 9.9% H2Sand 0.6% CH. has been studied and comparedto equation of state predictions. Other mixture data is being generated and should be completedin six months. Until this or other experimental data becomesavailable, operating companieswill be forced to rely on equation of state predictions and over-design accordingly .
1/16
Introduction
The unfavorable economics of constructing small Claus sulfur recovery plants, due to their inherent operating difficulties when the H2S feed concentration is low and the decreased demand and oversupply of elemental sulfur, necessitate an altemative technology. Air emission standards and regulations are increasingly stringent, compounding the need for an environmentally-friendly, cost-effective method to deal with acid gas. Acid gas compression and re-injection into depleted reservoirs or disposal zones, similar to produced water disposal, is a viable alternative to traditional sulfur recovery processes with the added advantages of reducing greenhouse gas emissions and providing pressure support for producing reservoirs. 1.2,3
Acid gas mixtures, separated from hydrocarbons in a sweetening plant, are compressed through several stages, dehydrated and pumped into a reservoir. Phase behavior, water content and physical properties of the acid gas are required for facility design. Phase behavior data is used to avoid the two-phase region during multi-stage compression to ensure equipment is not damaged by the appearance of a liquid phase. The injection pressure required is dependent on several factors including the formation pressure and the density of the acid gas mixture.1.4 The hydrostatic head of the acid gas column in the well bore aids injection, particularly if the fluid is injected above its critical point as a dense phase. Water content data determines the need for and aids in the design of dehydration facilities.
Although experimental data is available for pure hydrogen sulfide and pure carbon dioxide, little work has been done with acid gas mixtures5,6,7,8,9. Equation of state predictions can be used for mixtures, but their applicability for these polar/non-polar mixtures has not been proven and liquid density calculations are known to be unreliable. An acid gas mixture study is underway to fill the information gap, providing the required thermodynamic and physical properties over a range of temperatures and pressures of a series of acid gas mixtures. This pertinent, practical project is a joint effort between industry and research institutions.
Background
Hydrogen sulphide and
carbon dioxide are removed from sour gas by absorption with a
regenerative solvent in an amine plant. The acid gas mixture of H2S, CO2 and a small amount of light hydrocarbons leaves the sweetening unit saturated with water at the amine still conditions of low pressure and high temperature, represented by point A in Figure 1. The gas mixture is then compressed from points A to F in 3 or 4 stages. After each stage, the gas mixture is cooled,
2/16
without entering the two-phase region. Condensed water is removed after each stage. After the last stage, the mixture travels down a pipeline into the disposal well. Ideally, at the final
compressor discharge pressure the mixture will be supercritical. Further cooling in the pipeline will increase the density without a phase change, increasing the hydrostatic head of fluid in the well and reducing the required injection pressure. The operator must ensure that the mixture does not cool below its water saturation temperature, to avoid corrosion and hydrate plugging of the pipeline and wellbore.
Corrosion
and hydrates may occurwhenthe gas is saturated with water. Due to the safety
hazard associated with acid gas equipment failure, most injection schemes currently include dehydration facilities to ensure the acid gas is undersaturated throughout the system. Unfortunately.dehydrationfacilities and stainless steel comprise a major portion of the capital cost of re-injectionfacilities. Methanol injection is an option to combat corrosion and hydrate formation,but can significantlyincreaseoperatingexpenses. When the gas is not in contact with a water phase and is undersaturated,hydrates cannot form until the temperature drops sufficientlythat the gas can no longerhold all the water in solutionand "free"water is availablefor the formationof hydrates. If the mixtureis below its saturatedhydrateformationcurve, hydrates form preferentially a liquidwater phaseonce free water is available. to Although there is little experimental data on acid gas mixtures, the solubility of water in pure
hydrogen sulfide and pure carbon dioxide lead to some interesting hypotheses. The ability of the pure compounds to hold water in the vapor phase decreases as the pressure increases up to about 3000 kPa (400 psi) for H~ and 6000 kPa (900 psi) for CO2. At higher pressures the water holding capacity of the gases increases, corresponding to a higher water absorption capacity in the liquid phase or dense phase compared to the vapor phase. In both cases, increasing the temperature allows more water to be absorbed in the gas phase. Small amounts of methane substantially reduce the water absorption ability of both components.
If it is assumed that solubility of water in the gas mixtures mimics the trend of the individual components, then a minimumwater holdingcapacityexists at some pressure. Figure 2 1 demonstrateshow this informationcould eliminate the need for dehydrationfacilities. Points A through G in Figures 1 and 2 correspondto the same stages in a re-injectionfacility. Over each compressionstage, the pressureand temperatureincreaseand after each compressionstage the gas is cooled. Initiallythe water holdingcapabilityof the gas decreasesfrom stage to stage, until the minimumwater holdingcapacityis reached. If the condensedwater is removedat this point, the gas will be undersaturated with water throughoutthe rest of compression. Stainlesssteel will not be requiredin the compressorsor coolersafter point E. If the temperatureof the compressed 3/16
gas does not drop to the new water saturation temperature, point G, in the pipeline or wellbore, dehydration can be eliminated and stainless steel materials and methanol injection will not be necessary.
! ~
II) II) G) ... no
Temperature
,-
c 0 c
>-'; :'511)Q) ,.Q~ =a. OE U)o 'aU CC) IV C -.C ... Q) = -0 C II) 0 IV
U<.? ~
-Q)'C.IV U ~ct
Pressure
If it is concluded that dehydrationcannot be completely eliminated due to a particular set of conditions,for example in an extremelycold climate, the experimentaldata will still be beneficial. The operator will know the inlet water content of point A and the conditionsof the lowest water solubilityof the system. The glycol contactortower, regeneratorand circulationsystem can then be designed appropriately.
Currently, estimates of the density, water content and phase behavior of acid gas mixtures are being predicted with equation of state models, as experimental data is unavailable. The equations can produce considerable error and result in over or underdesigned facilities. Experimental data results in proper decisions on the equipment and materials required for each particular set of conditions, which can lead to considerable cost-savings.
Apparatusand Experimental Procedure
The apparatus consists of a temperature controlled air bath, a high-pressure cell with a sight glass, a positive displacement pump. a Hewlett Packard 6890 gas chromatograph (GC), and an Anton Parr density meter. The sight cell has an internal volume of approximately 80 cm3 (5 in3) and a maximum working pressure of 70 MPa at 150°C (10000 psi at 300°F). The 2.5 cm (1 in) thick sight glass located on the front and back of the cell allows visual observation of the cell contents throughout experimentation. As shown in Figure 3. the high-pressure sight cell is
mounted in the center of the oven as are the cylinders containing gas mixtures and distilled water. The density measuring cell is mounted at the bottom of the oven and wired to the density meter outside the oven.
A resistancedetectionthermometer locatedinside the oven next to the sight cell and is wired to is a digital temperaturecontroller. The temperaturecontrollerswitchesthe three oven heaterson or off depending on the difference betweenthe set and detected temperatures. Two fans in the oven circulatethe air, preventingtemperaturegradientswithin the oven. The oven can be heated up to 150°C (300°F) and the temperatureis measuredand controlledto within :t 0.1°C (O.2°F). The temperatureof the sight cell contentsis measuredto within :t O.1°Cwith a second resistance detectionthermometerinstalledin a port on the side of the sight cell. The sight cell volume and pressure are controlled by the addition and withdrawal of mercury througha port at the bottomof the sight cell. Samplefluids are pumpedin and out of a port at the top of the sight cell. The pump measuresvolume displacementwith a precisionof :f: 0.02 cm3 (0.001 in1. A dead-weightcalibrateddigital pressure gauge is connectedto the pump outlet. 5/16
The oven is connectedto a motor via a steel arm. The motor rotates the arm, rocking the oven and its contents in a 1800arc. The mercury in the sight cell agitates the fluids and enhances mixingwhen the oven and cell are rocked.
A heated, insulated line runs between the cell and the GC. A metering valve is used to control the flow and delivery pressure to the GC. A thermocouple downstream of the metering valve verifies that the temperature of the depressurized gas is sufficiently high to prevent water condensation.
Figure 3
-
Experimental Apparatus for Acid Gas S~dy
6/16
The apparatusis contained in a sour gas laboratory. The lab is continuouslyflushed with fresh air pumped in from the ceiling and drawn out of vents located in the bottom four comers of the room to an incineratorstack. A permanentH~ monitor is located close to the floor, below the apparatus. If the monitordetects 10 ppm HzS,an alarm soundsoutside the sour gas lab and the incineratorfires up. At 20 ppm the main lights in the lab shut off and emergency lights flash. PersonnelworKing in the lab with supplied air breathing apparatus (SABA) may not hear the alarm and the flashing lights ensure their evacuation. The lab is equippedwith a video camera for continuousremotemonitoringof personnelperformingdangerousworK. An -HzSPanic Button-which summonsemergencyrescue and medicalservices is located immediatelyoutside
the sour gas lab.
A gas mixture is synthesizedfrom pure componentsin the laboratory using partial pressures. Concentrationsare verified using the GC. The system is purged and gas is transferredto the sight cell. The gas in the cell is saturatedwith water at the desiredtemperatureand pressureand allowed to equilibratewhile rocking. At equilibrium,usually reachedwithin a few hours, a stable free water phase should be visible in the cell, ensuringthe gas is fully saturated. Equilibriumis verified by consistent GC measurements. Gas samples are then transferred to the density measuring cell and the GC for analysis, while maintaining constant cell pressure and temperature. The density measuringcell containsa U-shapedtube, electromagnetically excited by the density meter and vibrating at its natural frequency. The addition of sample fluid to the tube causes a frequencychange. The density is calculatedfrom the change in frequency using two calibration constantsand is read directly from the meter in g/cm3. The density meter has been calibratedat differenttemperatures and pressureswith methaneand pentane. Tests with carbon dioxide have revealedan accuracyof:t 0.5 % of the densityreadingcomparedto acceptedliteraturevalues. Water Content Analysis The accurate measurementof the small amountof water (as little as 100 ppm) that is soluble in the acid gas phases is crucial. Literature on the subject of water content of acid gases and analytical methods used to determinemoisturecontent was researchedand reviewed. In 1984 and 1987. Kobayashiand SongS.1O measuredthe water content of carbon dioxide and a carbon dioxide/methanemixture using a "tedious modified chromatographictechnique" developed by Block and Lifland11 The techniqueuses a glycerolpacked absorptioncolumn to withdrawwater
,
from a measuredquantity of gas. The amountof water withdrawn is determinedin a calibrated
7/16
split column gas chromatograph system. Kobayashi and Song reported an accuracy of 1: 5-6% based on calibration results.
In 1985, Huang et al.9 analyzed two mixtures of H~,
CO2, CH4 and water by direct
chromatography. The GC was equipped with a therrno-conductivity detector (TCD) and a 3.2 mm OD by 1 m long Porapak as column and the experiments were run between 30°C and 60°C. Calibrations were performed with pure carbon dioxide. The reported repeatability of
.t;
0.2 mole
percent is reasonable at high temperatures, where the water content is greater than 2 mole percent, ten times the repeatability. At other conditions, the water content measured less than 0.2 mole %, the repeatability of the analyses. Carroll et. al.7 reported in 1989 that water measurements in H2S performed by iodometric titration were accurate to within 0.1 mole percentage for amounts above 1 mole % water, but were inaccurate for measurements below 1 mole % water. None of these procedures determine water content with the precision, accuracy. repeatability and simplicity necessary for our project. Consultation and experimentation with Hewlett Packard A Plot-Q of resulted in the use of direct chromatography, as in Huang's method, with a TCD.
capillary column, 30 m long by 0.53 mm diameter with a 40Jlm film thickness
Divinylbenzene/styrene porous polymer is being used. Hydrogen is used as the carrier gas for improved resolution. Water content analysisis performedby withdrawinga sampleof gas througha meteringvalve to
the GC sampling loop. The line, valve and loop are heatedto preventcondensation. Once the line and the sampling loop are adequatelypurged with sample,the GC run is started and the switchingvalve sendsthe carriergas throughthe samplingloop. The componentsare separated in the column and travel to the TCD. The TCD containsa filament,which is electricallyheated to a constanttemperature. The carTier and sample gases flow intermittently over the filament The differencein power requiredto maintainthe filament temperature with the sample gas compared to the carTiergas is measured. The resultant peaks, with a height measuredin microvolts(~V), are integratedover time. seconds.to give area counts measuredin ~V's. To convert area counts to mole fractions, a response factor is required for each component analyzed. The responsefactor is determinedby runningexperiments with known concentrations of each of the components. The percent area count of a componentdivided by its response factor equates to the mole percent of that component present in the sample. Determiningthe responsefactors for the componentsto be tested is commonlyreferredto as calibratingthe GC.
8/16
Between runs the line, valve and loop are purged with heated nitrogen until no water is detected in the nitrogen stream.
Many tests were run without success until modifications to the gas analysis were perfected. The GC column was changed from a methyl silicone coated column to the Plot-Q, thus allowing the runs to proceed at higher temperatures and preventing water freeze out in the column. The heat tracing and insulation of the line between the sight cell and the GC was increased. Calculations indicate that for the most severe conditions studied, the gas must be heated to above 200°C to ensure that water does not condense over the metering valve. The metering valve replaced a pressure regulator due to carryover between runs, despite lengthy nitrogen purges, attributed to a large, unswept dead volume in the regulator.
WaterContentExperimental Results
While testing the apparatus, the water content of a gas mixture of carbon dioxide and 5.06% methane with traces of ethane, nitrogen, propane and butane was determined at 6200 kPa (900 psia) and 27°C (80.6°F). Water area counts of 125.0, 119.6, and 126.6 ~V's were obtained over three successive runs. The measured data has a standard deviation between runs of less than 3% and average water content of 0.135 mole %, proving the repeatability of the equipment at low water concentrations. The data is plotted vs. published data for a carbon dioxide and 5.31% methane gas mixture in figures 4 and 5. Considering the compositional difference between the two gases, the measured water contents are very dose.
9/16
A series of experiments were run with pure carbon dioxide to verify the accuracy of results over a range of temperatures and pressures. There is ample experimental data available for pure
10/16
carbon dioxide in the literature. As shown in figures 4 and 5, the applicationof responsefactors to the raw data resultedin a close matchto publisheddata. Figure4 shows the data plotted in 51 units while figure 5 shows the data plotted on a logarithmicscale with field units for ease of comparisonwith plotsfound in the GP5A EngineeringData Book.
The water content of carbon dioxide at 31°C (87.8°F) was measured at three different pressures. At 2500 kPa (370 psia) and 7200 kPa (1040 psia) the match was within 3% of published data and the standard deviation between runs at the same conditions was less than 4%. At 5100 kPa (735 psia) some difficulty was encountered obtaining repeatable data and only the last data point of five experiments is plotted. The match is within 10%. At 50°C, water content was measured at two different pressures. The match at 10100 kPa (1470 psia) was exact and the repeatability was within 1%. Difficulties were again encountered at 7600 kPa (1100 psia), the last three of fIVe runs is plotted, and the match is within 9%.
The poorer matches of 10% and 9% are
suspected be the result of insufficient pressure control to correctlyany fluctuations in analyzed water
t
during the experiments, affecting the repeatability and accuracy of the results. Since very good repeatability is obtained when the experiment is run content of greater than 5% alert the operator to inaccurately obtained samples and the experiments should be repeated. With proper pressure control results with less than 5% error can be obtained.
Phase Behavior Experimental Results
Severalisothemls were obtained to establish the phase envelope of a mixture of 89.5% CO2,
9.9% H2S and 0.6% CH.. The cell temperature was set and allowed several hours to equilibrate. The cell mercury volume was increased incrementally, resulting in 15-30 psi pressure steps. Transient and stabilized phase behavior was observed and recorded. The change in mercury volume was recorded as a function of pressure. At pressures close to the dew and bubble points, the volume/pressure increment was reduced. bubble points were established.
By taking a series of data points immediately
above and below the appearance and disappearance of the two-phase region, the dew and
As recorded in Table 1, dew pointswere observed at 7°C/4100 kPa, 9°C/4342 kPa, 15°C/5045
kPa, 26°C/6355 kPa, 34°C/7479 kPa and 37°C/7955 kPa. Bubble points were observed at 7°C/4528 kPa, 9°C/4755 kPa, 16°C/5510 kPa, 27°C/6900 kPa and 34°C/7844 kPa. Volume of mercury in the cell is plotted in Figure 6 vs. pressure for each isotherm. As seen in figure 6, two
11/16
distinct slope changes occurred each isotherm. The slope changes correspond to the dew for
and bubble points, verifying the visual observations.
Figure 6 Pressure vs Volume Data 9.9% H2S,89.5%CO2, 0.6% C1
-
1400
1200
.CI 'i Do
-
f) f) e Q.
e ~
1000
800
800
400
Cell Volume (cc)
12/16
Above 37.5°C(99.5°F),a stabletwo-phaseregionwas not observed. Some droplets,and elongatedbubblesappearedduring a volume/pressure changeand while the system was stabilizing, uponreaching but equilibrium, system singlephase all pressures. the was at
At 37.5°C (99.5°F) and 8253 kPa (1197 psia) the critical point was observed. At all other
temperatures the contents of the cell were clear and colorless in the vapor, liquid and two-phase regions. In the critical region, a small change in pressure (3-5 psi) resulted in the entire cell contents becoming a murky, grey cloud and then stabilizing out into a variety of shades of yellow. Above 8303 kPa (1205 psia) the contents were single phase, clear and colorless. At about 8274 kPa (1200 psia), the see-through single phase took on a slightly yellow tint. At the critical point of 8253 kPa (1197 psia), two phases appeared with an indistinct thick yellow interface, a darker yellow color at the bottom of the cell and a lighter yellow color on top. At 8212 kPa (1191 psia) the bottom half of the cell was a distinct dark orange liquid and the top half a colorless vapor. At 8198 kPa (1189 psia) the liquid phase faded to yellow and below 7957 kPa (1154 psia) the cell contents were again a colorless single phase. In Figure 7, the calculated equation of state phase envelope is plotted along with the experimental data. The widths of the two phase envelopes are similar, but the calculated
envelope falls below and to the left of the experimental data. The calculated critical point occurs at 34.9°C and 7633 kPa (94.8°F, 1107 psia), 2.6°C and 620 kPa below the experimentally determined critical point of 37.5°C and 8253 kPa (99.5°F, 1197 psia). The deviations between actual and calculated phase behavior emphasize the importance of obtaining an experimental data set for acid gas mixtures. The equation of state was regressed to fit the phase behaviour data obtained experimentally. The regressed curves and critical point match the measured data within the experimental error. The modified equation of state allows some extrapolation to different conditions, but experimental verification will be necessary until more data becomes available and a general regression is completed.
13/16
Figure 7 Pressure vs TemperaturePhaseEnvelope Experimentalvs Equation of State
-
{ !
I
~
..
5.0
10.0
15.0
200 Temperatu,. (C)
ao
30.0
35.0
40.0
The two-phaseregion can be avoided during compressionby cooling the gas to a minimum of 37.5°C betweencompressionstages. The fluid is in the supercritical,dense phase above 8253 kPa and above37.5°C. Conclusions & Future Plans Acid gas re-injectionmay be the optimum solution for producingsmall sour gas fields. As the issue of greenhousegases heats up, re-injectionwill becomeviable for even moderateto large operations. The operating company must avoid the two-phase region during compression. Water condensation and hydrate formation in the post-compressionequipment must be prevented to ensure safe, cost-effectiveoperation. Experimentaldata on the water content, density.and phasebehaviorof acid gas mixturesis thereforenecessary.
A comprehensive joint interest project is underway to obtain data for several different acid gas compositions over a range of operating temperatures and pressures. An online, repeatable and precise method of water content analysis using gas chromatography has been designed and tested. The reproduction of carbon dioxide data has proven the reliability of the method.
14/16
Experimentalphase behavior of a single mixture has shown that actual behavior does deviate
from equation of state predictions. Additional mixture data is being generated and should be completed in six months. Until this or other experimental data becomes available, operating
companies will be forced to rely on equation of state predictions and over-design accordingly.
Acknowledgements
The authors would like to acknowledge the support of staff at both Hycal Energy Research Laboratories Ltd. and the University of Calgary.
15/16
References
WIChert,E. and T. Royan, .Sulphur Disposal By Acid Gas Injection-, SPE 35585, Gas
Technology Conference, Calgary, May 1996. Keushnig, H., -Hydrogen Sulphide - If You Don't Like It, Put It Back", Journal of Canadian Petroleum Technology, June 1995, 34,6.18-20.
2.
3. Longworth, H.L., G.C. Dunn and M. Semchuck, .Underground Disposal of Acid Gas in
Alberta, Canada: Regulatory Concerns and Case Histories", SPE 35584, Gas Technology Conference, Calgary, May 1996.
4.
Carroll,J. and D. Lui, "Phase Equilibrium and Physical Property Considerations for Acid Gas
Injection Systems", Canadian Gas Processors Association, Calgary, November 1996.
5.
Song, K. and R. Kobayashi, "Water Content of CO2 in Equilibrium with Liquid Water and/or
Hydrates", SPE Formation Evaluation, December 1987, 500-508.
6.
Selleck, F.T., L.T. Carmichael and B.H. Sage, .Phase Behavior in the Hydrogen SulfideWater System.,Ind. Eng. Chern.,September,1952,44,9,2219-2226 Carroll, J.J. and A.E. Mather, .Phase Equilibriumin the System Water-HydrogenSulfide: Experimental Determination the LLV Locus.,CanadianJournalof Chern. Eng.,June, 1989, of 67,468-470.
7.
8.
Ng, Ho, D. Robinson and A. Leu, .Critical Phenomena in a Mixture of Methane, Carbon
Dioxide and Hydrogen Sulfide-, Fluid Phase Equilibria, 1985, 19,273-286.
9.
Huang, S., A.-D. Leu, H.-J. Ng and D.B. Robinson, "The Phase Behaviorof Two Mixturesof
Methane,Carbon Dioxide, HydrogenSulfide, and Water", Fluid Phase Equilibria, 1985, 19, 21-32. 10. Song, K.Y., and R. Kobayashi, -The Water Content of CO2-richFluids in Equilibriumwith LiquidWateror Hydrate,"ResearchReportRR-80,GPA,Tulsa, 1984. 11. Bloch, M.G. and P.P.Lifland, "Catalytic ReformingImprovedby Moisture Metering",Chern. Eng. Prog.69(9), 1973,49-52
By M.A. Clark, W.Y. Svrcek. W.O. Monnery: University of Calgary A.K.M. Jamaluddin: Hycal Energy Research Laboratories
E. Wichert: SogaproEngineering Ltd.
Abstract
The economicsof recoveringsulfur from sour natural gas have become unfavorablefor small fields. Hydrocarbonproducing companies require a cost effective yet environmentallysound alternativeto deal with acid gas. Compressed acid gas re-injection into producing,depleted or non-producing formationshas emergedas a viable alternativeto traditionalsulfur recovery. Most injection schemes include dehydration facilities to remove the saturated water from the gas,
preventing corrosion and hydrate formation. An alternative, less
costlyapproach is proposed:
keep the water in the acid gas phase throughout the injection circuit, eliminating the need to dehydrate.
To design an optimized injection strategy, determination of thermodynamic and physical propertiessuch as water content, dewpoint, bubble point, hydrateconditions and density of the acid gas is necessary. A comprehensivejoint industry/research project is underway to obtain data for several different acid gas compositionsover a range of operating temperatures and pressures. An online, repeatable and precise method of water content analysis using gas chromatographyhas been designed and tested. Pure CO2 data has been generated from the equipmentand matchesacceptedvalues with an averagestandarddeviationof 5 %. To date the phase behaviorof an acid gas mixturecontaining89.5%CO2, 9.9% H2Sand 0.6% CH. has been studied and comparedto equation of state predictions. Other mixture data is being generated and should be completedin six months. Until this or other experimental data becomesavailable, operating companieswill be forced to rely on equation of state predictions and over-design accordingly .
1/16
Introduction
The unfavorable economics of constructing small Claus sulfur recovery plants, due to their inherent operating difficulties when the H2S feed concentration is low and the decreased demand and oversupply of elemental sulfur, necessitate an altemative technology. Air emission standards and regulations are increasingly stringent, compounding the need for an environmentally-friendly, cost-effective method to deal with acid gas. Acid gas compression and re-injection into depleted reservoirs or disposal zones, similar to produced water disposal, is a viable alternative to traditional sulfur recovery processes with the added advantages of reducing greenhouse gas emissions and providing pressure support for producing reservoirs. 1.2,3
Acid gas mixtures, separated from hydrocarbons in a sweetening plant, are compressed through several stages, dehydrated and pumped into a reservoir. Phase behavior, water content and physical properties of the acid gas are required for facility design. Phase behavior data is used to avoid the two-phase region during multi-stage compression to ensure equipment is not damaged by the appearance of a liquid phase. The injection pressure required is dependent on several factors including the formation pressure and the density of the acid gas mixture.1.4 The hydrostatic head of the acid gas column in the well bore aids injection, particularly if the fluid is injected above its critical point as a dense phase. Water content data determines the need for and aids in the design of dehydration facilities.
Although experimental data is available for pure hydrogen sulfide and pure carbon dioxide, little work has been done with acid gas mixtures5,6,7,8,9. Equation of state predictions can be used for mixtures, but their applicability for these polar/non-polar mixtures has not been proven and liquid density calculations are known to be unreliable. An acid gas mixture study is underway to fill the information gap, providing the required thermodynamic and physical properties over a range of temperatures and pressures of a series of acid gas mixtures. This pertinent, practical project is a joint effort between industry and research institutions.
Background
Hydrogen sulphide and
carbon dioxide are removed from sour gas by absorption with a
regenerative solvent in an amine plant. The acid gas mixture of H2S, CO2 and a small amount of light hydrocarbons leaves the sweetening unit saturated with water at the amine still conditions of low pressure and high temperature, represented by point A in Figure 1. The gas mixture is then compressed from points A to F in 3 or 4 stages. After each stage, the gas mixture is cooled,
2/16
without entering the two-phase region. Condensed water is removed after each stage. After the last stage, the mixture travels down a pipeline into the disposal well. Ideally, at the final
compressor discharge pressure the mixture will be supercritical. Further cooling in the pipeline will increase the density without a phase change, increasing the hydrostatic head of fluid in the well and reducing the required injection pressure. The operator must ensure that the mixture does not cool below its water saturation temperature, to avoid corrosion and hydrate plugging of the pipeline and wellbore.
Corrosion
and hydrates may occurwhenthe gas is saturated with water. Due to the safety
hazard associated with acid gas equipment failure, most injection schemes currently include dehydration facilities to ensure the acid gas is undersaturated throughout the system. Unfortunately.dehydrationfacilities and stainless steel comprise a major portion of the capital cost of re-injectionfacilities. Methanol injection is an option to combat corrosion and hydrate formation,but can significantlyincreaseoperatingexpenses. When the gas is not in contact with a water phase and is undersaturated,hydrates cannot form until the temperature drops sufficientlythat the gas can no longerhold all the water in solutionand "free"water is availablefor the formationof hydrates. If the mixtureis below its saturatedhydrateformationcurve, hydrates form preferentially a liquidwater phaseonce free water is available. to Although there is little experimental data on acid gas mixtures, the solubility of water in pure
hydrogen sulfide and pure carbon dioxide lead to some interesting hypotheses. The ability of the pure compounds to hold water in the vapor phase decreases as the pressure increases up to about 3000 kPa (400 psi) for H~ and 6000 kPa (900 psi) for CO2. At higher pressures the water holding capacity of the gases increases, corresponding to a higher water absorption capacity in the liquid phase or dense phase compared to the vapor phase. In both cases, increasing the temperature allows more water to be absorbed in the gas phase. Small amounts of methane substantially reduce the water absorption ability of both components.
If it is assumed that solubility of water in the gas mixtures mimics the trend of the individual components, then a minimumwater holdingcapacityexists at some pressure. Figure 2 1 demonstrateshow this informationcould eliminate the need for dehydrationfacilities. Points A through G in Figures 1 and 2 correspondto the same stages in a re-injectionfacility. Over each compressionstage, the pressureand temperatureincreaseand after each compressionstage the gas is cooled. Initiallythe water holdingcapabilityof the gas decreasesfrom stage to stage, until the minimumwater holdingcapacityis reached. If the condensedwater is removedat this point, the gas will be undersaturated with water throughoutthe rest of compression. Stainlesssteel will not be requiredin the compressorsor coolersafter point E. If the temperatureof the compressed 3/16
gas does not drop to the new water saturation temperature, point G, in the pipeline or wellbore, dehydration can be eliminated and stainless steel materials and methanol injection will not be necessary.
! ~
II) II) G) ... no
Temperature
,-
c 0 c
>-'; :'511)Q) ,.Q~ =a. OE U)o 'aU CC) IV C -.C ... Q) = -0 C II) 0 IV
U<.? ~
-Q)'C.IV U ~ct
Pressure
If it is concluded that dehydrationcannot be completely eliminated due to a particular set of conditions,for example in an extremelycold climate, the experimentaldata will still be beneficial. The operator will know the inlet water content of point A and the conditionsof the lowest water solubilityof the system. The glycol contactortower, regeneratorand circulationsystem can then be designed appropriately.
Currently, estimates of the density, water content and phase behavior of acid gas mixtures are being predicted with equation of state models, as experimental data is unavailable. The equations can produce considerable error and result in over or underdesigned facilities. Experimental data results in proper decisions on the equipment and materials required for each particular set of conditions, which can lead to considerable cost-savings.
Apparatusand Experimental Procedure
The apparatus consists of a temperature controlled air bath, a high-pressure cell with a sight glass, a positive displacement pump. a Hewlett Packard 6890 gas chromatograph (GC), and an Anton Parr density meter. The sight cell has an internal volume of approximately 80 cm3 (5 in3) and a maximum working pressure of 70 MPa at 150°C (10000 psi at 300°F). The 2.5 cm (1 in) thick sight glass located on the front and back of the cell allows visual observation of the cell contents throughout experimentation. As shown in Figure 3. the high-pressure sight cell is
mounted in the center of the oven as are the cylinders containing gas mixtures and distilled water. The density measuring cell is mounted at the bottom of the oven and wired to the density meter outside the oven.
A resistancedetectionthermometer locatedinside the oven next to the sight cell and is wired to is a digital temperaturecontroller. The temperaturecontrollerswitchesthe three oven heaterson or off depending on the difference betweenthe set and detected temperatures. Two fans in the oven circulatethe air, preventingtemperaturegradientswithin the oven. The oven can be heated up to 150°C (300°F) and the temperatureis measuredand controlledto within :t 0.1°C (O.2°F). The temperatureof the sight cell contentsis measuredto within :t O.1°Cwith a second resistance detectionthermometerinstalledin a port on the side of the sight cell. The sight cell volume and pressure are controlled by the addition and withdrawal of mercury througha port at the bottomof the sight cell. Samplefluids are pumpedin and out of a port at the top of the sight cell. The pump measuresvolume displacementwith a precisionof :f: 0.02 cm3 (0.001 in1. A dead-weightcalibrateddigital pressure gauge is connectedto the pump outlet. 5/16
The oven is connectedto a motor via a steel arm. The motor rotates the arm, rocking the oven and its contents in a 1800arc. The mercury in the sight cell agitates the fluids and enhances mixingwhen the oven and cell are rocked.
A heated, insulated line runs between the cell and the GC. A metering valve is used to control the flow and delivery pressure to the GC. A thermocouple downstream of the metering valve verifies that the temperature of the depressurized gas is sufficiently high to prevent water condensation.
Figure 3
-
Experimental Apparatus for Acid Gas S~dy
6/16
The apparatusis contained in a sour gas laboratory. The lab is continuouslyflushed with fresh air pumped in from the ceiling and drawn out of vents located in the bottom four comers of the room to an incineratorstack. A permanentH~ monitor is located close to the floor, below the apparatus. If the monitordetects 10 ppm HzS,an alarm soundsoutside the sour gas lab and the incineratorfires up. At 20 ppm the main lights in the lab shut off and emergency lights flash. PersonnelworKing in the lab with supplied air breathing apparatus (SABA) may not hear the alarm and the flashing lights ensure their evacuation. The lab is equippedwith a video camera for continuousremotemonitoringof personnelperformingdangerousworK. An -HzSPanic Button-which summonsemergencyrescue and medicalservices is located immediatelyoutside
the sour gas lab.
A gas mixture is synthesizedfrom pure componentsin the laboratory using partial pressures. Concentrationsare verified using the GC. The system is purged and gas is transferredto the sight cell. The gas in the cell is saturatedwith water at the desiredtemperatureand pressureand allowed to equilibratewhile rocking. At equilibrium,usually reachedwithin a few hours, a stable free water phase should be visible in the cell, ensuringthe gas is fully saturated. Equilibriumis verified by consistent GC measurements. Gas samples are then transferred to the density measuring cell and the GC for analysis, while maintaining constant cell pressure and temperature. The density measuringcell containsa U-shapedtube, electromagnetically excited by the density meter and vibrating at its natural frequency. The addition of sample fluid to the tube causes a frequencychange. The density is calculatedfrom the change in frequency using two calibration constantsand is read directly from the meter in g/cm3. The density meter has been calibratedat differenttemperatures and pressureswith methaneand pentane. Tests with carbon dioxide have revealedan accuracyof:t 0.5 % of the densityreadingcomparedto acceptedliteraturevalues. Water Content Analysis The accurate measurementof the small amountof water (as little as 100 ppm) that is soluble in the acid gas phases is crucial. Literature on the subject of water content of acid gases and analytical methods used to determinemoisturecontent was researchedand reviewed. In 1984 and 1987. Kobayashiand SongS.1O measuredthe water content of carbon dioxide and a carbon dioxide/methanemixture using a "tedious modified chromatographictechnique" developed by Block and Lifland11 The techniqueuses a glycerolpacked absorptioncolumn to withdrawwater
,
from a measuredquantity of gas. The amountof water withdrawn is determinedin a calibrated
7/16
split column gas chromatograph system. Kobayashi and Song reported an accuracy of 1: 5-6% based on calibration results.
In 1985, Huang et al.9 analyzed two mixtures of H~,
CO2, CH4 and water by direct
chromatography. The GC was equipped with a therrno-conductivity detector (TCD) and a 3.2 mm OD by 1 m long Porapak as column and the experiments were run between 30°C and 60°C. Calibrations were performed with pure carbon dioxide. The reported repeatability of
.t;
0.2 mole
percent is reasonable at high temperatures, where the water content is greater than 2 mole percent, ten times the repeatability. At other conditions, the water content measured less than 0.2 mole %, the repeatability of the analyses. Carroll et. al.7 reported in 1989 that water measurements in H2S performed by iodometric titration were accurate to within 0.1 mole percentage for amounts above 1 mole % water, but were inaccurate for measurements below 1 mole % water. None of these procedures determine water content with the precision, accuracy. repeatability and simplicity necessary for our project. Consultation and experimentation with Hewlett Packard A Plot-Q of resulted in the use of direct chromatography, as in Huang's method, with a TCD.
capillary column, 30 m long by 0.53 mm diameter with a 40Jlm film thickness
Divinylbenzene/styrene porous polymer is being used. Hydrogen is used as the carrier gas for improved resolution. Water content analysisis performedby withdrawinga sampleof gas througha meteringvalve to
the GC sampling loop. The line, valve and loop are heatedto preventcondensation. Once the line and the sampling loop are adequatelypurged with sample,the GC run is started and the switchingvalve sendsthe carriergas throughthe samplingloop. The componentsare separated in the column and travel to the TCD. The TCD containsa filament,which is electricallyheated to a constanttemperature. The carTier and sample gases flow intermittently over the filament The differencein power requiredto maintainthe filament temperature with the sample gas compared to the carTiergas is measured. The resultant peaks, with a height measuredin microvolts(~V), are integratedover time. seconds.to give area counts measuredin ~V's. To convert area counts to mole fractions, a response factor is required for each component analyzed. The responsefactor is determinedby runningexperiments with known concentrations of each of the components. The percent area count of a componentdivided by its response factor equates to the mole percent of that component present in the sample. Determiningthe responsefactors for the componentsto be tested is commonlyreferredto as calibratingthe GC.
8/16
Between runs the line, valve and loop are purged with heated nitrogen until no water is detected in the nitrogen stream.
Many tests were run without success until modifications to the gas analysis were perfected. The GC column was changed from a methyl silicone coated column to the Plot-Q, thus allowing the runs to proceed at higher temperatures and preventing water freeze out in the column. The heat tracing and insulation of the line between the sight cell and the GC was increased. Calculations indicate that for the most severe conditions studied, the gas must be heated to above 200°C to ensure that water does not condense over the metering valve. The metering valve replaced a pressure regulator due to carryover between runs, despite lengthy nitrogen purges, attributed to a large, unswept dead volume in the regulator.
WaterContentExperimental Results
While testing the apparatus, the water content of a gas mixture of carbon dioxide and 5.06% methane with traces of ethane, nitrogen, propane and butane was determined at 6200 kPa (900 psia) and 27°C (80.6°F). Water area counts of 125.0, 119.6, and 126.6 ~V's were obtained over three successive runs. The measured data has a standard deviation between runs of less than 3% and average water content of 0.135 mole %, proving the repeatability of the equipment at low water concentrations. The data is plotted vs. published data for a carbon dioxide and 5.31% methane gas mixture in figures 4 and 5. Considering the compositional difference between the two gases, the measured water contents are very dose.
9/16
A series of experiments were run with pure carbon dioxide to verify the accuracy of results over a range of temperatures and pressures. There is ample experimental data available for pure
10/16
carbon dioxide in the literature. As shown in figures 4 and 5, the applicationof responsefactors to the raw data resultedin a close matchto publisheddata. Figure4 shows the data plotted in 51 units while figure 5 shows the data plotted on a logarithmicscale with field units for ease of comparisonwith plotsfound in the GP5A EngineeringData Book.
The water content of carbon dioxide at 31°C (87.8°F) was measured at three different pressures. At 2500 kPa (370 psia) and 7200 kPa (1040 psia) the match was within 3% of published data and the standard deviation between runs at the same conditions was less than 4%. At 5100 kPa (735 psia) some difficulty was encountered obtaining repeatable data and only the last data point of five experiments is plotted. The match is within 10%. At 50°C, water content was measured at two different pressures. The match at 10100 kPa (1470 psia) was exact and the repeatability was within 1%. Difficulties were again encountered at 7600 kPa (1100 psia), the last three of fIVe runs is plotted, and the match is within 9%.
The poorer matches of 10% and 9% are
suspected be the result of insufficient pressure control to correctlyany fluctuations in analyzed water
t
during the experiments, affecting the repeatability and accuracy of the results. Since very good repeatability is obtained when the experiment is run content of greater than 5% alert the operator to inaccurately obtained samples and the experiments should be repeated. With proper pressure control results with less than 5% error can be obtained.
Phase Behavior Experimental Results
Severalisothemls were obtained to establish the phase envelope of a mixture of 89.5% CO2,
9.9% H2S and 0.6% CH.. The cell temperature was set and allowed several hours to equilibrate. The cell mercury volume was increased incrementally, resulting in 15-30 psi pressure steps. Transient and stabilized phase behavior was observed and recorded. The change in mercury volume was recorded as a function of pressure. At pressures close to the dew and bubble points, the volume/pressure increment was reduced. bubble points were established.
By taking a series of data points immediately
above and below the appearance and disappearance of the two-phase region, the dew and
As recorded in Table 1, dew pointswere observed at 7°C/4100 kPa, 9°C/4342 kPa, 15°C/5045
kPa, 26°C/6355 kPa, 34°C/7479 kPa and 37°C/7955 kPa. Bubble points were observed at 7°C/4528 kPa, 9°C/4755 kPa, 16°C/5510 kPa, 27°C/6900 kPa and 34°C/7844 kPa. Volume of mercury in the cell is plotted in Figure 6 vs. pressure for each isotherm. As seen in figure 6, two
11/16
distinct slope changes occurred each isotherm. The slope changes correspond to the dew for
and bubble points, verifying the visual observations.
Figure 6 Pressure vs Volume Data 9.9% H2S,89.5%CO2, 0.6% C1
-
1400
1200
.CI 'i Do
-
f) f) e Q.
e ~
1000
800
800
400
Cell Volume (cc)
12/16
Above 37.5°C(99.5°F),a stabletwo-phaseregionwas not observed. Some droplets,and elongatedbubblesappearedduring a volume/pressure changeand while the system was stabilizing, uponreaching but equilibrium, system singlephase all pressures. the was at
At 37.5°C (99.5°F) and 8253 kPa (1197 psia) the critical point was observed. At all other
temperatures the contents of the cell were clear and colorless in the vapor, liquid and two-phase regions. In the critical region, a small change in pressure (3-5 psi) resulted in the entire cell contents becoming a murky, grey cloud and then stabilizing out into a variety of shades of yellow. Above 8303 kPa (1205 psia) the contents were single phase, clear and colorless. At about 8274 kPa (1200 psia), the see-through single phase took on a slightly yellow tint. At the critical point of 8253 kPa (1197 psia), two phases appeared with an indistinct thick yellow interface, a darker yellow color at the bottom of the cell and a lighter yellow color on top. At 8212 kPa (1191 psia) the bottom half of the cell was a distinct dark orange liquid and the top half a colorless vapor. At 8198 kPa (1189 psia) the liquid phase faded to yellow and below 7957 kPa (1154 psia) the cell contents were again a colorless single phase. In Figure 7, the calculated equation of state phase envelope is plotted along with the experimental data. The widths of the two phase envelopes are similar, but the calculated
envelope falls below and to the left of the experimental data. The calculated critical point occurs at 34.9°C and 7633 kPa (94.8°F, 1107 psia), 2.6°C and 620 kPa below the experimentally determined critical point of 37.5°C and 8253 kPa (99.5°F, 1197 psia). The deviations between actual and calculated phase behavior emphasize the importance of obtaining an experimental data set for acid gas mixtures. The equation of state was regressed to fit the phase behaviour data obtained experimentally. The regressed curves and critical point match the measured data within the experimental error. The modified equation of state allows some extrapolation to different conditions, but experimental verification will be necessary until more data becomes available and a general regression is completed.
13/16
Figure 7 Pressure vs TemperaturePhaseEnvelope Experimentalvs Equation of State
-
{ !
I
~
..
5.0
10.0
15.0
200 Temperatu,. (C)
ao
30.0
35.0
40.0
The two-phaseregion can be avoided during compressionby cooling the gas to a minimum of 37.5°C betweencompressionstages. The fluid is in the supercritical,dense phase above 8253 kPa and above37.5°C. Conclusions & Future Plans Acid gas re-injectionmay be the optimum solution for producingsmall sour gas fields. As the issue of greenhousegases heats up, re-injectionwill becomeviable for even moderateto large operations. The operating company must avoid the two-phase region during compression. Water condensation and hydrate formation in the post-compressionequipment must be prevented to ensure safe, cost-effectiveoperation. Experimentaldata on the water content, density.and phasebehaviorof acid gas mixturesis thereforenecessary.
A comprehensive joint interest project is underway to obtain data for several different acid gas compositions over a range of operating temperatures and pressures. An online, repeatable and precise method of water content analysis using gas chromatography has been designed and tested. The reproduction of carbon dioxide data has proven the reliability of the method.
14/16
Experimentalphase behavior of a single mixture has shown that actual behavior does deviate
from equation of state predictions. Additional mixture data is being generated and should be completed in six months. Until this or other experimental data becomes available, operating
companies will be forced to rely on equation of state predictions and over-design accordingly.
Acknowledgements
The authors would like to acknowledge the support of staff at both Hycal Energy Research Laboratories Ltd. and the University of Calgary.
15/16
References
WIChert,E. and T. Royan, .Sulphur Disposal By Acid Gas Injection-, SPE 35585, Gas
Technology Conference, Calgary, May 1996. Keushnig, H., -Hydrogen Sulphide - If You Don't Like It, Put It Back", Journal of Canadian Petroleum Technology, June 1995, 34,6.18-20.
2.
3. Longworth, H.L., G.C. Dunn and M. Semchuck, .Underground Disposal of Acid Gas in
Alberta, Canada: Regulatory Concerns and Case Histories", SPE 35584, Gas Technology Conference, Calgary, May 1996.
4.
Carroll,J. and D. Lui, "Phase Equilibrium and Physical Property Considerations for Acid Gas
Injection Systems", Canadian Gas Processors Association, Calgary, November 1996.
5.
Song, K. and R. Kobayashi, "Water Content of CO2 in Equilibrium with Liquid Water and/or
Hydrates", SPE Formation Evaluation, December 1987, 500-508.
6.
Selleck, F.T., L.T. Carmichael and B.H. Sage, .Phase Behavior in the Hydrogen SulfideWater System.,Ind. Eng. Chern.,September,1952,44,9,2219-2226 Carroll, J.J. and A.E. Mather, .Phase Equilibriumin the System Water-HydrogenSulfide: Experimental Determination the LLV Locus.,CanadianJournalof Chern. Eng.,June, 1989, of 67,468-470.
7.
8.
Ng, Ho, D. Robinson and A. Leu, .Critical Phenomena in a Mixture of Methane, Carbon
Dioxide and Hydrogen Sulfide-, Fluid Phase Equilibria, 1985, 19,273-286.
9.
Huang, S., A.-D. Leu, H.-J. Ng and D.B. Robinson, "The Phase Behaviorof Two Mixturesof
Methane,Carbon Dioxide, HydrogenSulfide, and Water", Fluid Phase Equilibria, 1985, 19, 21-32. 10. Song, K.Y., and R. Kobayashi, -The Water Content of CO2-richFluids in Equilibriumwith LiquidWateror Hydrate,"ResearchReportRR-80,GPA,Tulsa, 1984. 11. Bloch, M.G. and P.P.Lifland, "Catalytic ReformingImprovedby Moisture Metering",Chern. Eng. Prog.69(9), 1973,49-52
