easauarcer ro Che sol has PLANT DESIGN Use glycerol to dehydrate supercritical carbon dioxide This technology increases hydrocarbon recovery M. SWADENER, Denbury Resources Inc, Plai Texas; J. LUNDEEN, K. FISHER and C. BEITLER, Trimeric Corporation, Buda, Texas wenty years after the first cwo glycerol dehydration units were built for superrtcal carbon dioxide (CO,) use, Den- bury Resources found a need to build new units to process the inereasing amount of CO, production from the Jackson Dome area in Mississippi. Denbury has now completed and started up five new glyeerol-based dehydration facilities in the past five years, with other units under construction. Glycerol is used to dehydrate CO; when high-pressures and non-idealities cause excessive vapor-phase glycol (.e., ethylene, diethylene or ‘wetheylene glycol) losses, making normal glycol-based dehydra- tion uneconomical in comparison with glycerol-based dehydra- tion. Denbury uses the supercritical CO; primarily for enhanced cil recovery (ER), ‘This article presents reasons for Denlury’s recent surge ingly col unit construction as well as types of other applications where elycerol dehydration may be well suited. Differences between "using glycerol and glycol for dehydration with superritical, dense- phase CO, gas streams will also be discussed. Since commercial process simulators are not wel suited for designing this type oF system, a model was developed to rigorously predict phase equi- libra in the absorber and simulate other process equipment. The simulator and other data used to build the new glycerol dehydra- tion units are reviewed along with presenting key design features of the dehydration plants. Denbury’s new and existing CO, dehyciration facilities. Denbury is growing independent oil and natural gas company. es the largest oil and navural ges operator in Mississippi: it owns the largest eserves of CO used for tertiary oil recovery east of the Mississippi River, and holds significant operating acreage in Loui- siana, Alabama, Texas, Montana, North Dakota and Wyoming. Denbury’s CO, source field, Jckson Dome, located near Jack son, Mississippi, was discovered during the 1970s while being explored for hydrocarbons (Fig, 1). Ths significant source of COp isthe only one known ofits kind in the US eas of the Mississippi River Mississippi’ fist EOR project began in the mid 1980s in the Livle Creck Feld following the installation of Shell Oil Com- pany’s Choctaw CO> pipeline. The 183-mile Choctaw Pipeline {now referted toas the NEJD Pipeline) transported CO; produced from Jackson Dome tothe Lise Creek Feld, While the CO, flood proved sicesfil in recovering significant ol amounts, commodity prices at that time made the projec unatactve and Shal ater sold itso fields inthis area, swells the CO, source wells and pipelines. (Currently, mote than 1.0 billion 3 per day (Bsefd) of CO; is prodiced from Jackson Dome Area operations. The pipeline ne ‘work has been expanded from the single NEJD Pipeline includ- ing the Free tate Pipeline wo East Mississippi the Delta Pipeline ‘going into northeast Louisiana andthe recent completion ofthe ‘Green Pipeline that delivers CO, across the Gulf Coast Region fiom the end ofthe NEJD Pipeline co south of Houston. Source ‘well operating pressures range from 800 psig to 4,000 psi, Jack- son Dome is generally referred to as the CO, source, but itis also the name ofthe original central dehydration facility that uses « _lycerol process to dehydrate supercritical COprich steams. The Jackson Dome dehydration unit was buil by Shell and based on research performed in the 1980s to understand the dynamics of dehydrating supereritical CO, with plycols and glycerols.™5 The research showed that glycerol can dehydrate supercritical CO, more effectively than glycols. Critical pressure and temperature for pure CO; are 1,070 psia and 87.8°F, respectively. It should be noted that a glycerol unit built 20 years ago and operated by a different company is believed to be no longer operational.6 Tn 2003, an expansion to the Jackson Dome dehydration plant ‘was completed, adding internals tothe inlet separator and conac- tor to almort double the capacity: Since then, Denbucy has bile several new glycezl-based CO, dehydration units to keep pace ‘with is increase in CO, production. Since Denbury continues to find high-pressure source wells its advantageous to dehydrate the CO, with glycerol ae pipeline conditions without compression. ‘The new dehydration units are described briefly in Table 1. The {9s composition is generally 99.4% CO, 0.3% No and 0.3% Cy, HYOROCARBON PROCESSING UY 2041 | 75PLANT DESIGN ‘The post-expansion process performance at Jackson Dome was used to validate modeling results. Improvements were also made ‘0 the new dehydration plants based on Denbury’s Jckson Dome experience and advancements with process technology, Dehydration of supercritical CO, streams with glycerol. ‘The properties of supercritical CO, impact the dehydration sy- tem design in ewo ways: *+ Mutual solubilities of the desiccant and CO, * Saturated water content of CO, with temperature and pressure. Data taken in a laboratory setting, an from field test, «cluded that glycerol had a much lower solubility in supercritical ‘CO, than did echylene glycol (EG), diethylene glycol (DEG) and ‘rlethylene glycol (TEG), Table 2 illustrates that TEG is 10-200 ‘mes more soluble than glycerol in CO, over a pressure range of 1,200 psig-2,000 psig, This means thar TEG would have cacessive vapot losses and makcup requirements, making normal glycol- based dehydration uneconomical. Operational data from Jackson Dome supports the glycerol solubilcy data from these sources. A Foss material balance of the glycerol added co the system indicates 4 ppmv-5 ppmv of glycerol is soluble a 1,300 psig and 15°F. ‘Table 2 also shows that, unlike the glycols, glycerol solubility in CO, did not increase significantly with lower CO, temperatures However, due tothe high glycerol vseosiy and rapidly increasing supercritical CO, density as the temperature decreases, glycerol would probably not be used to dehydrate CO, below 100°F In addition, the CO, solubility in glyeerol was aso determined to be only one-tenth ofthe solubility in glycols. This results in lower ashing losses. 16 | auer2011 Hysecarterrcetng com ‘Water solubility in pure CO is well documented in several sources.” Fig, 2 shows the maximum water amount that can be dissolved into a homogeneous CO} phase—for the purpose of this article, iis called the ‘solubility lim” for water in CO,. As shown in Fig. 2, ara fixed temperature, the solubility limit for ‘water decreases with increasing pressure up to near the critical pressure. As the pressue increases beyond the ctitical pressure (2,070 psa), the solubility limit for water inctesses. However, sinall concentrations of other gases can significant impact the ‘water-holding capacity of CO,.‘ Knowing the correct moisture concent of the supercritical CO ensues tha the absorber and regeneration equipment are not undersized. ‘Other parameters that are important are he density of CO;, which can change greatly in the presence of small amounts of inere gases and hydrocarbons, and the viscosity of glycerol, which is significantly different chan glyeol (1,490 eps and 47.8 eps, respectively, at 20°C) 4 Potential uses of glycerol-based dehydration, Dehy- dration was originally developed for essentially pure supercritical (CO,-streams containing high CO, content that need be dehy- rated ac igh pressure. An example to use this rechnology isthe stream dehydrated using glycerol in Hungary. The Hungarian stream had 81% CO, with the remainder consisting of meth- ane—dehydrated a a preseure near 2,000 psi. Other potential applications ofthe glycerol cechnology may exst where a com bination of high CO; content and a need to dehydrate at high presaure create che operating conditions and compositions where slyceral may have lower solubility losses than glycols and bewer overall economic. Glycerol dehydration simulation development. A simulation was developed specficaly for supercritical dehydra- tion of CO, with glycerol. Tis is because the standard property secs in most process simulators perform poorly when modeling uid properties and phase equilibrium at neae-critial conditions. ‘Careful review of the thermodynamic paramecets inthe process simulator is required, with adjustments made to match reliable data, Key modeling aspects will be discussed shor. TABLE 1. Denbur cerol-based CO, dehydro ion units —_ z ‘Gluck Barksdale Barksdale ‘luck DRI Plant stadt__North South Trace _stadt_Dock Opmtion ste 2006 2007 2006 2010 2040 2011 Flow, Musca 100 300300 150 300300 peraing pressure pig 1550 1800 1800 15001550 180 Operating temperature F120 1011011011518 ‘ater content in Meer 201201 201201201207 Witerconert on, tis 2020 -20”~-20”~20 20 Feboier doy MMBuile 2056 —~=w~S*aTS ‘TABLE 2. Glycol and glycerol solubility in CO, (ppmv)* 1,200 pst 2000 psi Te" Ghcerol EG Glycerol FF Ta00 12 7200 2 35 390 a S 7 Tis ce) 7300 afalues. An important aspect of designing glye- ‘rol-based dehydration systems for supercritical CO, isto havea sg00d estimate of the K-value (i.e, phase equilibrium distribution coefficient) for waterat the conditions in che absorber. Unreliable (CO;-water equilibrium data can produce erroneous results in the sbsorber thac adversely affect predicting operations in downstream vessels, Absorber K-values for water were calculated using the thermodynamic relationship Keyhe= yb" y/DP ® where K » K-alue, phase equilibrium distibution coef y» Mole faction water in the dense CO; phase x» Mole fraction water in the liquid phase *Y = Activity coefficient for water (from vapor-liquid equilib- rium data discussed in the following subsection) P= Vapor pressure of pure water atthe system temperature (Grom steam tables) "= Fugacity coefficient for pure water vapor atthe system emperature and P* (assumed to be unity) ‘1 Poynting correction factor for water (from the component partial molar volumes) © « Fugacity coeicient of water fo the vapor phase = System operating pressure (known). Most ofthe parameters in Bq, 1 are readily known or accesible from literature a indicated previously. However, itis the caleula- tion ofthe vapor-phase fgacity of water in dense-phase CO, near the eitical canitions that is more diffcul. Field data were used to validate the accuracy ofthe simulation in predicting K-values and resulting water contents over a range ‘ofactual operating conditions. Table 3 shows that the treated gas ‘water contents are in very lose agreement with the predicted equilibrium water content for the lean glycerol solution at the absorber conditions (ie., within 546-1796 ofthe measured dat) ‘This indicates thatthe method discussed i reliable for predicting K-values and dehydrator outer water contents. Water-glycerol phase behavior. The vapor liquid equi sium (VLE) of water in glycerol is important because iis used to: + Predict che K-values in the absorber (activity coefficient of | water) * Determine the lean glycerol water content from the reboiler and extra stripping column, VLE data for water and glycerol (in the concentration and. temperature ranges inthe absorber and reboiler) were obtained and used to regres the adjustable parameters inthe Wilson activ- ity coefficient model. CO, solubility in glycerol. itis important to know the CO> solubility in glycerol for several rssons. Fits, i provides a means ‘TABLE 3. Comparison to known field data ‘Weated gas measured, — Predicted equilibrium Inns Tit, tant 7 a = 31 73 32 72, 78 a 38 os nz 132 155 a 2a 25.1 173 PLANT DESIGN to determine how much CO, is picked up in the absorber and will evolve in the flash tank. Second, the CO, solubility in glye- erol is important in prediesing the potential for two-phase flow through the heat exchangers in the process as the temperatures and pressures of the streams vary. Equilibrium data were ised the Wilson activity coefficient model o predict CO, solubility in glycerol Viscosity. The viscosity estimation method in the process simulator that most closely predicted viscosities found in the literature was selected, Viscosity data were used to caleuate equip. ‘ment pressure drop and model absorber hydraulics, ‘Thermal conductivity. A method that accurately predicted the thermal conductivity for both 50-50 glycerol-water mixtures and high-purity glycerol streams was selected. Density. The densities ofthe inlet and outlet CO, streams from the absorber were calculated using the Peng-Robinson eque- tion of state. The calculated densities compared reasonably well to those obtained from other sources. Glycerol solubility in dense-phase COs. The glycrolsolubil- ity in dense-phase CO hasbeen reported in the literature and was used to estimate glycerol losses from the absorber.) Glycerol-based dehydration design and research. Fig. 3 illustrates a typical process flow diagram of che glycerol- based CO, dehydration units for Denburys new facilities. The lycerol process operates much inthe same mannet asa TEG unit 40. detiled process description is not provided.” Operational experience at Jackson Dome and advancements in process technology influenced the new glycerol process designs. Construction material forall process piping and vessels at the ‘original Jackson Dome facility is 31613161. SS. In general, many (of the sime material choices used for glycol are appropriate for slycerol The following materials choices were different in the new units compared to the original Jackson Dome plant: HYDROCARBON ROCESSING JY 201PLANT DESIGN + The contactor was buile with 316L SS in the lower section and a carbon ste! shell above the bottom 2 fi. of packing. This was done since the gas is sub-sarurated ss it moves countercurrent to the glycerol through the packed section. This concept was = cessfully proven elsewhere. * Carbon steel was used for all glycerol piping except for downstream of the glycerol flash tank. In this area, 3161 SS was ‘used instead to provide low temperature integrity if CO; vented through the contro valve if te liquid level contol in the flash absorber was lost. * Carbon stccl was used for therichilean exchanger shells the lyerol air cooler, the reboler shell an filter vessel, Design elements of note from the Jackson Dome process that ‘were incorporated into the new glycerol units will be described Farther, Isolation valves for the hot and cold glycerol exchangers, ‘were included for periodic cleaning/maintenance due to fouling, from upset conditions or high skin temperatures. Due to high glycerol viscosity, glycerol booster pumps were used to ensure Continuous flow fiom the reboiler through the hot and cold glye- trol exchangers. A fr tube liner was included in the eboier to moderate the skin temperature in contact with the glycerol, chus, reducing the degradation rate. The glycerol flash tank and still were constructed of 161 SS dive co the wer CO; in those areas. ‘Other key design aspects include using low-maintenance and low-cost horizontal submersible pumps or gear pumps for glyc- rol circulation and incorporating an extta stripping column for enhanced water removal. A provision fora filter coalescer upstream ofthe glycerol contactors for removal of water with high salt and chloride content asthe well declines was also included inthe design, Jackson Dome data was used to size the absorbers, in the new units. ‘As work with supercritical CO continues, there have been some aeas of research and innovation, A sees of tests were con- ducted to quantify the performance of liquid and solid hydrogen sulfide (HS) scavenger at process conditions when 50 ppm-100 ppm of HS was detected in one of the source wells. The test results indicated a reduced efficiency of these scavenging prod- ucts for HyS removal compared to the typical performance on natural gas! Another atea of study was glycerol solubility CO;, Although the solubility is much less than glycols, the price of glycerol was over $1/Ib in 2008 when oil prices were att peak. Addicives were identified to further reduce glycerol solubil ity in CO, and a full-scale testis planned. Fig. 4 shows a new alycerol unit ‘Summary. ‘The rigorous review of thermodynamic proper- ties associated with the dehydration of supercritical CO» with 18 | mny2011 Hydocteeceingcon lycerol was required for the development ofa reliable process- Simulation tool to design new glycerol processes. Denbury's ‘operational experience at Jackson Dome was important in con- structing high-performing glycerol-based units for dehydration ‘of supercritical CO>. Using glyeerol dehydrator allows Denbury to process more COs for increased hydrocarbon recovery. The information provided can be applied for designing other high- CO,, high-pressure applications, aswell, even ifthe stream is not essentially pure CO, HP LITERATURE CITED Wallace, Cu “Deydrtion of Superciteal COs” Gas Conditioning Confeeac, 1985, 2 Song, and Re Kobayashi, “The Water Content of COyteh ls in rio wi Ligid Water ot Hydate” GPA ResestchRepor 80, May 4, ‘Taha S. and R. Kobayashi “The Water Content and dhe Soll of CO, in Egullbrm with DEGWacer and TEG-Water Solions at Fale [Alsorpion Condens” GPA Technical Publesion 9, December 1982 “Song Ke ¥ and R.Kobayahs, "HO Come Vals af a CO-5.31 Mal % ‘ethane Maur” GPA Reseach Report 120, Janay 1989. Diaz Zand J. He Miley, "Drying Subsnlly Suprerical COp wi {Gyeral” US Peat No, 478,612, Qxober 1984, ‘Ui GL. Ges, ¥. Ouch, F Nagas EA Thoesand CB. Wallace “Prcening Facies for Enhanced Ol Recovery in Hungary Proceedings of {3th Annual GPA Converton, Phoen,Asaona, Mar 12-13, 199, Gas Procesors Supplcs Asosison (GPSA) Engineering Databook, FPS Nolume i welth Editon, Chapter 20, pp. 20-7 and 20-38, 2008. * Melntsh, Is C. Bei M. Sradenr and C. Walle, “Seeing roses for Remoal of HS fiom Enhanced Ol Recoyery CO; Sueams? Laurance Reid Gas Condioning Conference, Norman, Okshom2, 2010, ‘Mark Swadlener io resent engine’ 2 enbur Resources info, ss. He has seen yeas of expences, po- Cuctonloerton nd eserorenginetiog for West Tras na “ee Gul Cost ol els. hes rece, Mr. Swadener 2s the pertos engine eckson Dem, reponable forthe compte tin, production ane martnance cf CO; source we ad acts. He eshed is BE dee mechanical engnesng rom Souther Meth Univers n Daas. ‘oe Lunideen isa vincpal engine a Tere Comp. in Bos, Sona, Hates 2 yeas of opscace proces engineeig.pc> Ces tublesheoing end faclty tall fr oard sas produc ttn and CO, processing chen. Lander vecert epee tas focused on dehy, contaminant ral ae varspon (of tuperticl Coy He ennes 85 and MS does in chemical engneting from ‘he Univer of Mezou—Rola and Teas ABM —Colege Staten respeivey Kevin Fishers prncral engines 2ttimerc Corp. Bus, Tere. He ha over 20 gets of experince in process engine ing reseerch and developmen, ape woubeshestg fr of ard {2 poditin and cl efing cers, as wel a fo priate ard {ovement porte esearch programs Mc Fer species the sreaso c.g rerorlCO, capt, gis roe and ges dehatin He tuned on MS dere inch Enginesring om he Urey of Teast Aus, nd 85 degrees chanical engineering and chery ror Teas A8N—Colege ‘Staton and Sem Houston State Univers respecte Carrie Beitler i sno: cine trimeric Corp. nut, Texas, She has ovr 15 years of experience in process engineer. ing, process nodeing and optimzton of unit operat inthe Mattl ges, petroleum refining 2nd CO processing ars he so specaes nth development cf process eign packages ov the fabcaton of opener technatgy such at cau srubbes, acdeas recon Unis, aha denyerators and ame estes. she eared a BS dare In chemical engrerng eam Pardo Ue 1 ‘