fo

Raniv

EOROptimization

eceR)he2 ped ccadeinstere

refrigerants for CO2 liquefaction are NH3, CO2, C3H8 and R134a. One of the developed liquefaction cyclesthat liquees the CO2 at 50 bar using NH3 refrigerant resulted in 5.1% less power consumption than the

armingin uall CO

sil fuelO2 emian aping.

spheric pressure, CO2 needs to be pressurized to a supercriticalpressure, e.g. 150 bar, before injection into an oil well for EOR. As forstorage or shipping in tankers, the captured CO2 is liqueed ata pressure of about 6 bar so that its volume is reduced. Conven-tionally, CO2 pressurization is done using multi-stage compression

CO2 is liqueed after compression by either seawater or by using anopen cooling cycle that they patented. Moore et al. [14] carried outa similar study. According to Moore et al.s preliminary analysis,a 35% reduction in power is possible when liquefying the CO2 usingabsorption chillers and pumping the CO2 instead of compressing it.Botero et al. [15] compared different compression strategies for CO2compression using HYSYS software [16]. They investigated usingabsorption chillers and cold seawater for CO2 liquefaction at threeliquefaction pressures.

* Corresponding author. Tel.: 1 301 405 5247.

Contents lists available at

Applied Therma

ev

Applied Thermal Engineering 33-34 (2012) 144e156E-mail address: yhhwang@umd.edu (Y. Hwang).CO2 capture and sequestration (CCS) consists of three processes:capture, which is typically done by amine absorption; transport,which is done by either pipeline or ships; and storage, which can bedone in underground geological formations such as oil wells, whichcan be utilized for enhanced oil recovery (EOR). Among the threeprocesses, CO2 capture has attracted the most attention in theliterature since it is more technically challenging. Since CO2captured in post-combustion using amine absorption is at atmo-

[3] used EBSILON software [7] to model a CO2 compression cycle.Sanpasertparnich et al. [8] used an in-house code to model a CO2compression cycle. Moller et al. [2] used IPSEpro software [9] tomodel a CO2 compression cycle. Romeo et al. [10] modeled andoptimized CO2 compression processes using EES software [11].

Few authors investigated the CO2 liquefaction process. Aspelundet al. [12,13] studied CO2 liquefaction for ship transport. Theyconsidered three liquefaction pressures (20, 55 or 95 bars) where1. Introduction

In order to mitigate the global wstationary sources and sequesteredformations. Since about one third offuel energy sources comes from foswhich have the highest density of Cper power output [1], they provideattempt to mitigate the global warm1359-4311/$ e see front matter 2011 Elsevier Ltd.doi:10.1016/j.applthermaleng.2011.09.027, CO2 is captured fromnderground geological2 emissions from fossilburning power plants,ssions in terms of masspropriate target in the

with intercooling. CO2 compression consumes about 100 kWh/tonCO2 [2] while the energy efciency of a power plant can bedecreased by 3e4% points [3]. The other pressurization approach isto liquefy the CO2 and pump it to the target injection pressure, andthen the pressurized CO2 is evaporated so that phase change doesnot occur inside the well. The two approaches are shown in Fig. 1.

Many studies in the open literature developed models for theCO2 compression process. Amrollahi et al. [4] used GT PRO software[5] to model a CO2 compression cycle. Pfaff et al. [6] and Cifre et al.liquefaction cycle. 2011 Elsevier Ltd. All rights reserved.conventional multi-stage compression cycle as well as 27.7% less power consumption than the open CO2Development of CO2 liquefaction cycles

Abdullah Alabdulkarem, Yunho Hwang*, ReinhardCenter for Environmental Energy Engineering, Department of Mechanical Engineering, U

a r t i c l e i n f o

Article history:Received 8 July 2011Accepted 20 September 2011Available online 1 October 2011

Keywords:CO2 liquefactionCO2 compressionCCS

a b s t r a c t

CO2 pressurization is a nenhanced oil recovery (EOconsumes about 4% from tIn this paper, several CO

using an open cycle or clossingle refrigerant and casmodels were validated agaliquefaction parameters w

journal homepage: www.elsAll rights reserved.r CO2 sequestration

dermacherersity of Maryland, College Park, MD 20742, USA

ssary component in any CO2 capture and sequestration (CCS) whereis to be applied. The power demand for the CO2 pressurization processpower plant net power.ressurization methods, such as compression or liquefaction and pumpingycles, were explored and evaluated. New CO2 liquefaction cycles based onrefrigerants were developed and modeled using HYSYS software. Theexperimental data and/or veried against other simulation software. Theoptimized for minimum overall power consumption. The considered

SciVerse ScienceDirect

l Engineering

ier .com/locate/apthermeng

multi-stage CO2 compression, and investigated the design of

The total compression power for compressing the captured gasfrom 1.8 bar to 150 bar is 6.25 MW. The pressureeenthalpy (Peh)diagram for the compression process is shown in Fig. 4.

ssurization approaches.

rmal Engineering 33-34 (2012) 144e156 145several CO2 liquefaction cycles (cascade and single-refrigerant).Further, there has been no research conducted on optimizing theCO2 liquefaction pressure, and none of the authors veried orvalidated their models.

The objectives of this paper are (1) to carry out a comprehensivecomparison between CO2 compression and CO2 liquefaction andpumping processes for EOR application by developing several CO2liquefaction cycles and validating or verifying the developedmodels, (2) to optimize the CO2 liquefaction pressure, and (3) toevaluate the performance of the open CO2 liquefaction cycle.

2. CO2 compression cycle

In order to inject the captured CO2 into an oil well for EOR, itneeds to be pressurized to a supercritical pressure. The injectedCO2 is in supercritical vapor state. The injection pressure used inthis paper is 150 bar, which is typical for an EOR project. Pres-surizing CO2 is done using eight multi-stage centrifugal compres-sors with intercooling, so that it approaches isothermalcompression. HYSYS process simulation software was used tomodel the CO2 compression plant as shown in Fig. 2. The modelingassumptions are listed in Table 1. The seawater temperature wastaken to be 35 C which is a typical value in hot climate regionssuch as the Arabian Gulf and the Red Sea. The equation of stateProprietary CO2 liquefaction processes are available in someapplications such as in the food and beverage industries. However,such applications of CO2 liquefaction differ from power plant interms of quantity and quality of feed gas and product gas. Severalliquefaction cycles exist for natural gas liquefaction but they differsignicantly from CO2 liquefaction in terms of liquefactiontemperature and cooling curves. While absorption chillers area mature technology and many manufacturers exist, they have lowefciency and are considered complex and expensive [17].

There is a lack of literature in detailed comparison between CO2compression and CO2 liquefaction and pumping processes. Noauthor has carried out a comprehensive study on whether lique-fying and pumping CO2 for injection consumes less power than

Fig. 1. Two CO2 pre

A. Alabdulkarem et al. / Applied Theused is the PengeRobinson equation of state. All HYSYS modelconvergence tolerances for the relative residuals were set to be1 104.

The developed model was validated against vendors data fortwo CO2 centrifugal compressors, each having 8 stages withintercooling and different mass ow rate and outlet pressure [18].The model matched the vendors data adequately as shown inTable 2.

A CO2 capturing plant model that captures the CO2 froma natural gas combined cycle (NGCC) was developed (not shown).The captured stream composition mass ow rate is shown inTable 3. CO2 is dehydrated after each compression stage. In the nalstage, the CO2 purity reaches 99.6% by mass. Further CO2 dehy-dration can be achieved using water absorbent or lters.3. CO2 liquefaction cycle

3.1. Introduction

CO2 can be pressurized through the use of pumps after lique-faction. Pumps are less energy intensive than compressors becausethe specic volume of the liquid CO2 is much less than the vaporCO2. However, liquefying the CO2 requires a considerable amount ofenergy.

Since the captured CO2 from the stripper of the CO2 removalplant is at 1.8 bar, the CO2 needs to be compressed to a pressure thatis higher than the triple point pressure (5.17 bar) before liquefac-tion takes place. Otherwise, solid CO2 will be formed with cooling,as shown in the CO2 phase diagram in Fig. 3. Therefore, multi-stageCO2 compressors are introduced upstream liquefaction cycles.

3.2. CO2 liquefaction using cold seawater

Deep seawater is cooler than seawater at the surface. Thetemperature of the seawater can be as cold as 4 C depending onthe depth of the sea. The assumed surface seawater temperatureis 35 C. If seawater was extracted below the surface level, witha temperature of 27 C, it could be used to liquefy the CO2 that isat a pressure higher than critical pressure (73.77 bar) toa temperature that is lower than the critical temperature(30.98 C). A HYSYS model was developed that is similar to the8-compression stage model (Fig. 2) except that it has a pumpinstead of the last compression stage. The P-h diagram for thismodel is shown with the 8-compression stage in Fig. 4. Deepseawater is used to liquefy the CO2 to 30 C with a liquefactionFig. 2. HYSYS model for the CO2 compression plant, 2 stages only shown here out of 8stages (Refer to Fig. 7 for components labeling).

The developed HYSYS model that depicts Dopazo et al.s proto-

Table 2HYSYS CO2 compressor model validation against vendors data.

HYSYSmodel

Vendorsdata

HYSYSmodel

Vendorsdata

No. of stages 8 8Isentropic efciency (%) 80 80Pin (bar) 1.1 1.1Pout (bar) 140 200CO2 ow rate (m3/hr) 27,000 70,000

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156146type is shown in Fig. 6. The model uses same operating parametersas the experiments. The model has a cooling capacity of 9.45 kW ata cooling temperature of 50 C. The validation results, shown inTable 4, matched the experimental data satisfactorily for the topload of 4.1 MW at a 75 bar liquefaction pressure. The resultingtotal power consumption with this conguration is 6.09 MW,which is 0.16 MW or 2.56% power savings as compared to the 8-compression stage case.

3.3. CO2 liquefaction cycles

There is no CO2 liquefaction cycle design available in the openliterature. Thus, new designs that are based on single-refrigerant aswell as cascade conguration for liquefying the CO2 were explored.Four common refrigerants were investigated: NH3, CO2, C3H8 andR134a.

A HYSYSmodel for a cascade refrigeration system using CO2 andNH3 refrigerants was developed and validated against experi-mental data fromDopazo et al. [19]. Dopazo et al.s experiment is ona 9 kW refrigeration capacity system at 50 C evaporationtemperature. The NH3 is the top vapor compression cycle (VCC)refrigerant whereas the CO2 is the bottom VCC refrigerant. Theirexperiment schematic diagram is shown in Fig. 5.

Fig. 3. CO2 phase diagram, adapted with permission from [18].cycle (NH3 VCC) as well as the bottom cycle (CO2 VCC). The inputparameters used for the HYSYS model are the parameters withoutdiscrepancy values in Table 4. The maximum discrepancy is in thecalculated NH3 refrigerant volume ow rate. This could be due tothe accuracy of the equation of state in predicting the specicvolume of the NH3. However, the power consumption and theoverall coefcient of performance (COP) were well predicted.

3.4. Single-refrigerant CO2 liquefaction cycles

Fig. 7 shows a single-refrigerant conguration. The cycle hastwo cooling stages; one is for dehydration and sensible cooling at

Table 1Modeling assumptions used in the work.

Compressor isentropic efciency 80%Pump isentropic efciency 75%Heat exchangers pressure drop 10 kPaHeat exchangers pinch temperature 3 KSeawater temperature 35 C1 C saturation temperature and one is for liquefaction at 57 Csaturation temperature (at liquefaction pressure of 6 bar). The CO2purity reaches 99.93% after the dehydration heat exchanger. Thecycle has two ash tanks at each stage and intercooling to 40 Cbetween compression stages.

The condensing temperature for all cycles is 40 C. Thecondensing pressure is determined from the enthalpy after theexpansion valve, which is an isenthalpic process. The enthalpy afterthe expansion valve is determined to provide the CO2 liquefactionload for a given ow rate at the specied saturation temperature.The Peh diagrams for the NH3 and R134a are shown in Fig. 8.

The number of compression stages is set based on a maximumstage pressure ratio of 5, as well as the outlet temperature ata compressor outlet pressure. To prevent excessive heat, themaximum outlet temperature from a compressor stage was set tobe less than 90 C. The temperature lines in the superheated regionin NH3 refrigerant are closer than the ones in R134a and C3H8refrigerants. Therefore, 7 compression stages were used in the NH3cycle and 2 compression stages were used in the R134a and C3H8cycles.

A new thermodynamic parameter, a, is introduced here thatdescribes the increase in a refrigerant temperature after an isen-tropic compression process. That is, the unit is 1 C temperatureincrease for every 1 bar pressure increase. This parameter can beused as a tool to predict the number of compression stages withdifferent refrigerants. To prevent high compressor outlet temper-ature, high compression stages is required for a certain refrigerantthat has high a.

Temperature increase in isentropic compression process;

a dTdP

s

(1)

Where,dT ToTiTo: Compressor outlet temperature, Ti: Compressor inlet

temperature.

Power (MW) 4.95 5 13.21 14Discrepancy (%) 1.01 5.98dP PoPiPo: Compressor outlet pressure, Pi: Compressor inlet pressure.Table 5 shows values for the introduced parameter, a, for the

three refrigerants at 57 C evaporator temperature. For simplicity,

Table 3Captured steam compositions.

Mass ow rate Value

CO2 mass ow rate (Ton/hr) 72.42H2O mass ow rate (Ton/hr) 8.04O2 mass ow rate (Ton/hr) 0.0053N2 mass ow rate (Ton/hr) 0.013Total mass ow rate (Ton/hr) 80.47

Fig. 4. Peh diagram for liquefying CO2 with seawater option and 8-compression stage option.

Fig. 5. Schematic diagram of the CO2eNH3 cascade refrigeration system, adapted with permission from [19].

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156 147

critical temperature it was expected to consume the least poweraccording to Mclinden [20] observation that a high critical pointrefrigerant yields high COP. Since the required VCC load is 9.01 MW,the COP of the NH3 cycle, Eq. (2), is 1.42.

COP of VCC VCC cooling loadVCC power input

(2)

The Carnot cycle is a reversible VCC that has the least possiblepower consumption. In other words, no actual cycle reaches theefciency of Carnot cycle. Hence, the Carnot cycle is used tocompare how close the actual cycle to a reversible cycle.

Carnot COP for cooling cycle TevapTcond Tevap

(3)

Reversible compression power;Wrev Cooling load;QCarnot COP (4)

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156148Fig. 6. HYSYS model for CO2eNH3 cascade refrigeration system that depicts Dopazoet al.s experiment.dPwas taken to be the difference between the saturation pressuresat the condensing and evaporating temperatures. Ti was taken to bethe evaporation temperature. To was taken to be the temperature atconstant entropy and pressure of the saturation pressure at thecondensing temperature. Values of a show that NH3 has the highesttemperature increase compared to R134a and C3H8. In order toprevent high compressor outlet temperature, liquefaction cyclesthat use NH3 as a refrigerant require a higher number ofcompression stages than the liquefaction cycles that use R134a orC3H8 refrigerant.

The liquefaction cycles power consumption and the CO2compressors power consumption are tabulated in Table 6. The leastpower consumption is for the NH3 cycle. Since NH3 has the highest

Table 4Validation results from the Dopazo et al.s experiment and current HYSYS model for CO2

VCC Bottom (CO2)

Parameter Measurement HYSYS model Discr

Tevap (C) 50 50 eTcond (C) 17.48 17.48 ePevap (kPa) 682 658.6 3.55Pcond (kPa) 2127 2127 eDTsh 15 14.94 0.06DTsc 0.34 0.54 0.2 (KQevap (kW) 9.45 9.45 eQcond (kW) 13.2 12.55 5.2%Wele (kW) 3.93 3.67 7.1%Comp. isen. eff. (%) 57.9 57.9 eRef. ow (kg/h) 124.4 118.1 5.3%COP 2.4 2.57 6.78%Total CO2/NH3DTcascade (K) 3.48 3.48 eCOP CO2/NH3 0.92 0.91 0.83SecondLawefficiency;hIIReversiblecompressionpower;WrevActualcompressionpower;Wvcc

(5)

The Carnot cycle COP is calculated for the CO2 sensible coolingevaporator, 0.962 MW at 2 C, and the CO2 liquefaction evapo-rator, 8.05 MW at 57 C, to be 2.39. Therefore, the Second Lawefciency for the NH3 cycle is 59.34%

The liqueed CO2 is pumped to 150 bar. After pumping, it is ata temperature of 48 C. This liquid CO2 needs to be evaporatedabove the critical temperature so that it achieves supercriticalvapor state before injection. It is proposed for this heating processto be recovered by cooling another stream. The available recover-able heat at 6 bar liquefaction pressure is 3.69 MW.

Total power consumption of NH3 single-refrigerant liquefactioncycle is shown in Table 7. The total power consumption, 8.65MW, is38.4% higher than 8-stage compression with intercooling for theproposed system.

3.5. Cascade CO2 liquefaction cycles

According to Carnot COP, Eq. (3), the COP of a VCC decreaseswhen the ratio between the condensing temperature and evapo-ration temperature increases. In addition, the evaporation pressurein the single-refrigerant cycles was below atmospheric pressure,which can cause air leakage to the evaporator.

/NH3 cascade refrigeration.

Top (NH3)

epancy Measurement HYSYS model Discrepancy

20.96 20.96 e29.72 29.72 e

% 182 179.5 1.39%1158 1158 e

(K) 15 15.02 0.02 (K)) 4.74 0.29 4.5 (K)

13.5 12.55 7.57%17.2 18.03 4.6%6.32 6.68 5.48%58.8 58.8 e45.46 41.18 10.4%2.14 1.87 14.02%%

rmalA. Alabdulkarem et al. / Applied TheCascade conguration employs two vapor compression cycleswhere the top VCC serves as the condensing cycle for the bottomVCC. The bottom VCC is the cooling cycle. The improvements in thecascade conguration include less superheat in the condenser,which results in less entropy generation in the condenser due tosmaller temperature difference between the condenser and theambient temperature. Also, there is improvement in the evaporatorcapacity of the bottom cycle due to a lower condensing tempera-ture [21].

Different CO2 liquefaction cascade cycles were designed andmodeled using HYSYS software. One cascade model for CO2 lique-faction at 8 bar liquefaction pressurewith NH3 refrigerant in the topVCC and CO2 refrigerant in the bottom VCC is shown in Fig. 9. Thetop VCC is designed to have two cooling levels: one for dehydratingand sensible cooling of the CO2 at a 1 C saturation temperature andone for subcooling the CO2 refrigerant in the bottomVCC from40 Cto 15 C. The bottom VCC has one cooling level at a 50 Csaturation temperature where liquefaction occurs. In both cycles,the refrigerants have a pressure higher than the atmosphericpressure. Therefore, the possibility of air leakage into heatexchangers is avoided in the cascade conguration.

The performance of the cascade cycle is a function of the sub-cooling temperature. Optimization studies were done to nd theoptimum subcooling temperature using exergy minimization. Forthe CO2eNH3 conguration, a correlationwas developed for ndingthe optimum subcooling temperature by Lee et al. [22]. Theircorrelation is as follows:

Topt;sc 40:63 0:4Tcond 0:4Tevap DTcascade (6)Where, T is in K.

At 40 C condensing temperature, 50 C evaporation temper-ature and 3 K approach temperature in the subcooling heat

Fig. 7. Single-refrigerant liquefaction cycle at 6 bar liquefactionEngineering 33-34 (2012) 144e156 149exchanger, the optimum subcooling temperature using Lee et al.scorrelation is 15 C. The total cascade VCC power consumption atthese conditions is 7.52 MW.

For further investigating CO2 liquefaction cycle powerconsumption, four different refrigerants for CO2 liquefaction at6 bar liquefaction pressure using cascade conguration were alsomodeled using HYSYS software. The refrigerants are R134a, CO2,NH3 and C3H8. The modeled cycles have 5 C cascade subcoolingtemperature and 55 C CO2 sensible cooling temperature. Theirpower consumption is shown in Table 8, where the lowest powerconsumption is for the C3H8eNH3 cascade cycle.

Since C3H8eNH3 cascade cycle has the lowest powerconsumption, the optimum cascade subcooling temperature for itwas investigated. The effect of the sensible cooling temperature isshown in Table 9. As the cascade subcooling temperature decreases,the power consumption of the top cycle increases while the powerconsumption of the bottom cycle decreases. The table shows thelowest power consumption is at a 5 C cascade subcoolingtemperature.

3.6. Optimization of CO2 liquefaction pressure

CO2 liquefaction is done by VCCs, where the CO2 is liqueedfrom 40 C to the liquefaction temperature. The liquefactiontemperature is the saturation temperature at the liquefactionpressure (e.g. 53.5 C at 6 bar). As the liquefaction pressureincreases, the liquefaction temperature will increase, which willalso increase the Carnot COP of a liquefaction cycle as shown inFig. 10. However, this will have four consequences. First, it will addmore power to the CO2 compressors that compress the CO2 to theliquefaction pressure, which is above the triple point pressure.Second, it will reduce the latent liquefaction load as shown in the T-

pressure with two compression stages and intercooling.

in Fig. 12, calculates the CO2 compressors and a pump power. It alsocalculates the liquefaction latent load and the available cold energy

1

102

103

r]

100C

40C

R134a

Condenser

aTable 6Single-refrigerant liquefaction cycles power consumption at 6 bar liquefactionpressure.

Refrigerant VCC power (MW) CO2 compressors power (MW)

C3H8 7.95 1.98NH3 6.34 1.98R134a 7.78 1.98

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156150R134a

-50 0 50 100 150 200 250 300 35010-1

100

10

h [kJ/kg]

P [b

a

0.2 0.4 0.6 0.8

1C HX

-57C HX

102

103

100C

Ammoniabs diagram Fig. 13. Third, it will reduce the pumping power of theliqueed CO2 from the liquefaction pressure to the injection pres-sure (150 bar). Fourth, it will make the pressurized liquid CO2warmer, which will make it non-usable for cold energy recovery tocool a hot stream, such as captured CO2 gas. As a result, therecoverable cooling in the liqueed and pressurized CO2 decreasesas shown in Fig. 11.

Since there are two conicting objectives (i.e., minimizing CO2compressor power and minimizing CO2 liquefaction cycle power),an optimum liquefaction pressure can be investigated. A methodwas developed to optimize any process simulated in HYSYS usingMatlab optimization tool [23]. Thus, a Matlab code [24] was writtento nd the optimum liquefaction pressure with the lowest powerconsumption. The code is coupled with a generic HYSYSmodel thatwas developed for this purpose. The generic HYSYS model, shown

NH3

-250 150 550 950 1350 175010-1

100

101

h [kJ/kg]

P [b

ar]

40C

0.2 0.4 0.6 0.8

Condenser

1C HX

-57C HX

Fig. 8. Peh diagrams of R134a and NH3 single-refrigerant liquefaction cycles at 6 barliquefaction pressure.

Table 5Temperature increase in isentropic compressionprocess for three refrigerants.

Refrigerant a (C/bar)

C3H8 0.09R134a 0.12NH3 0.21in the liqueed and pressurized CO2 (e.g. 46.3 C at 150 bar thatwas liqueed at 6 bar). This cold energy is recovered in a recoveryheat exchanger. The T-s diagram for the liquefaction process againstcompression process is shown in Fig. 13.

The optimization technique was based on an exhaustive searchtechnique because all designs will be considered in nding theoptimum solution. Further, the design space is not large and thegeneric HYSYS model is computationally simple. The interface codestructure is shown in Fig. 14. The liquefaction pressure was the onlyvariable and it was varied from 5.7 bar to 72 bar with a step value,DP, of 0.1 bar. Thus, 663 designs were considered.

The vapor compression cycle liquefaction power, Wvcc, wascalculated using Carnot COP as following:

Carnot COP for Cooling Cycle TevapTcond Tevap

(7)

Wvcc QCarnot COP hII(8)

Where,Tevap is the liquefaction temperature, Tcond is the condensing

temperature, Q is the latent liquefaction load, hII is the Second Lawefciency.

Since the pressurized liquid CO2 is sensibly heated when it coolsthe natural gas stream (the temperature of the pressurized liquidCO2 is not constant here), the Carnot COP needs to be discretized formore accurate calculation of the vapor compression cycle powerthat will be required to provide an equivalent cooling to therecoverable cold energy. Thus, the Carnot COP calculation for theRecovery heat exchanger (HX), Fig. 12, was discretized for every 1 K.

The results from the exhaustive search are shown in Fig. 15. Thegraph shows that the Carnot COP of the liquefaction cycle variesfrom 2.3 at 5.7 bar to 31 at 72 bar as a result of the variation of theliquefaction temperature. Further, total power for different coolingcycles Second Law efciencies are plotted to show where theoptimum solution is located based on the efciency of the vaporcompression cycle. The Second Law efciency describes how closea real cycle is to an ideal cycle. Thus, by using Carnot cycle andassuming the Second Law efciency, we can nd the optimalsolution.

Table 7

Total power consumption and efciency of liquefying CO2 at 6 bar using NH3 asa refrigerant.

Liquefaction load (MW) to 53 C 9.01Pumping power (MW) to 150 bar 0.33Recoverable Heat (MW) to 47 to 37 C 3.69NH3 liquefaction cycle power (MW) 6.34Minimum reversible liquefaction power (MW) 3.76Liquefaction cycle COP 1.42Liquefaction cycle Second Law efciency (%) 59.34CO2 compressors power (MW) (from 1.8 bar to 6 bar) 1.98Total power (MW) 8.65

ycle

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156 151Fig. 15 shows that for efcient cycles (hII of 100% and 90% cases),the optimal solution is at the lowest liquefaction pressure.However, no real VCC has such efciency. For hII of 70% and 50%cases, the optimum is located at 50 bar liquefaction pressure. Thesmall jump in the total power curves at 51 bar is because thetemperature of the liquid CO2 is greater than 29 C and so was not

Fig. 9. Cascade CO2 liquefaction cycle using NH3 in the top cutilized in cooling another stream.The modeled systems in the graph are complete HYSYS models

for CO2 liquefaction cycle with cold energy recovery (Details inSection 3.7). They are superimposed in the graph to verify the codeand the generic HYSYS model. Furthermore, their location on thegraph indicates that they all have 70% Second Law efciency.

3.7. Using cold CO2 for cooling the CO2 and refrigerant

3.7.1. Single-refrigerant CO2 liquefaction cycles with cold CO2recovery in CO2 liquefaction cycle

The proposed idea of using the cold CO2 to subcool the refrig-erant after the condenser, and to precool the CO2 before liquefac-tion was also investigated and modeled using HYSYS software. Asstated earlier, the available cooling capacity in the liqueed CO2 at6 bar liquefaction pressure is 3.69 MW. Subcooling the refrigerantusing the cold CO2 will make the refrigerant cooler than 40 C after

Table 8Power consumption of different cascade liquefaction cycles at 6 bar liquefactionpressure.

Refrigerants (TopeBottom) NH3eCO2 C3H8eNH3 C3H8eCO2 R134aeNH3

Top VCC PowerConsumption (MW)

2.37 3 2.87 3.15

Bottom VCC PowerConsumption (MW)

4.8 3.31 4.8 3.29

CO2 Compressor Power (MW) 1.98 1.98 1.98 1.98Total VCC Power

Consumption (MW)9.15 8.29 9.65 8.42the refrigerant condenser and before the expansion valve. There-fore, lower refrigerant quality or more cooling capacity is availablefor liquefying the CO2.

The 1 C HX shown in Fig. 7 was replaced by a precooling heatexchanger for the CO2/H2O stream using the cold CO2. Thus, a lowercooling load is required from the VCC.

and CO2 in the bottom cycle at 8 bar liquefaction pressure.The subcooling temperature for the ammonia refrigerant withcold CO2 recovery at 6 bar liquefaction pressure is10 C instead of40 C. Further, the VCC cooling load is 7.23 MW instead of 9.01 MW.The resulting VCC power consumption is 4.39 MW, which is 30.75%less than the cycle without cold CO2 recovery.

Since liquefaction pressure affects the liquefaction load and CO2compressors power, eight liquefaction cycles with cold CO2recovery were modeled using HYSYS software with differentrefrigerants at randomly chosen liquefaction pressure. The resultsin Table 10 show that the lowest total power consumption is at50 bar liquefaction pressure using NH3 refrigerant. Moreover,Table 10 shows that at different liquefaction pressures, the refrig-erants other than NH3 can be used with close power consumption.

Table 9Effect of cascade subcooling temperature on the C3H8eNH3 cycle at 6 bar liquefac-tion pressure.

Cascade subcooling temperature (C) Total C3H8eNH3 cascade power (MW)

33 8.1423 7.5213 6.963 6.472 6.45 6.317 6.3412 6.417 6.3822 6.4727 6.4832 6.64

faction cycle at 6 bar liquefaction pressure without cold CO2 energy

5. Open CO2 liquefaction cycle

5.1. Conventional open CO2 liquefaction cycle

The second way to pressurize CO2 is to liquefy it and then pumpit to the desired pressure. The open CO2 liquefaction cycle for liq-uefying the CO2 from a stationary source is a patented cycle byAspelund et al. [12]. According to Aspelund et al., the liquefaction ofCO2 is best achieved using their open CO2 liquefaction cycle. The

Fig. 12. Generic HYSYS model for nding the optimum liquefaction power.

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156152recovery is 31.44%.

3.7.2. Cascade CO2 liquefaction cycles with cold CO2 recovery in CO2liquefaction cycle

The proposed idea of using the cold CO2 to subcool the refrig-erant after the condenser and to precool the CO2 before liquefactionwas also investigated and modeled for the cascade congurationusing HYSYS software as shown in Fig. 16. The available coolingcapacity in the liqueed CO2 at 8 bar liquefaction pressure is3.38 MW.

HX-1 shown in Fig. 9 was replaced by two precooling heatexchangers for the CO2/H2O stream using the cold CO2 in Fig. 16.These two heat exchangers precool the CO2 to 37 C. Further,a subcooling heat exchanger was introduced to subcool the NH3refrigerant to 0 C in the top VCC using the cold CO2.

As a result, the total cascade power consumption was reducedfrom 7.52 MW to 6.69 MW, which is 0.82 MW or 11.03% powersavings.

4. Absorption chillers for CO2 liquefaction

NH3 absorption chillers can be used to liquefy CO2 if there isIn comparison to CO2 compression only (conventional congu-ration), the savings in power consumption for the NH3 cycle at50 bar is 5.12%. On the other hand, the savings against NH3 lique-

Fig. 10. Liquefaction pressure versus liquefaction temperature and Carnot cooling cycleCOP.enough waste heat. If there is 13.35 MW waste heat at 130 C, NH3absorption chillers can be used to liquefy the CO2 at 12 C or25 bar liquefaction pressure. The liquefaction load is 5.84 MW. Thetotal power consumption (CO2 compressors and pump) for thisoption is 4.37 MW, which is the lowest power consumption forpreparing the captured CO2 for injection (1.88MWor 30.08% powersavings as compared to the multi-stage compression).

Fig. 11. Recoverable cooling in the liqueed and pressurized CO2 variation withliquefaction pressure.purpose here is tomodel the open CO2 liquefaction cycle at ambientconditions equivalent as the previous models.

The working principle of the cycle is that it uses the capturedCO2 to liquefy itself. As shown in the developed HYSYS model forthe open CO2 liquefaction cycle (Fig. 17), the cycle consists of threecompression stages, three intercooler and two multi-stream heatexchangers. A portion of the compressed CO2 is re-circulated andexpanded through an expansion valve. Due to CO2 expansion, itstemperaturewill be reduced and then it can be used to cool anotherincoming CO2 stream. Then, the CO2 that was used to cool theincoming CO2 stream is sent back in a vapor state to an interme-diate compressor. A nal expansion valve is used to expand thecompressed and cooled CO2 to a low pressure at a saturation statewhere the vapor and liquid are separated. The vapor is used incooling the incoming CO2 streamwhile the liquid is sent to storageor pumping.

The open CO2 liquefaction cycle power consumption is functionof the available seawater temperature, the mass recirculation ratioand three pressure levels (i.e., the high side and the two expansionpressures). Due to mass ow rate recirculation, the convergence ofthe model was not simple. In order to understand the performanceFig. 13. CO2 T-s diagrams showing two pressurizing processes. Liquefaction andpumping with cold CO2 recovery (Green). Multistage compression with intercooling(Red). (For interpretation of the references to color in this gure legend, the reader isreferred to the web version of this article.)

inte

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156 153of this cycle, a simplied exhaustive searchwas done by varying theseawater temperature from 28 C to 40 C and the high side pres-sure from 43 bar to 70 bar at 0.35 mass ow rate recirculation ratio.It was observed that the power consumption is not a strong func-tion of the mass ow recirculation ratio because the model willkeep re-circulating the required amount of CO2 until it liquees thedesired amount of CO2 in order to satisfy the mass balance.

24 cycles were resulted from the exhaustive search for liquefy-ing the same feed gas (Table 3). Their power consumption valuesvaried from 14 MW to 34 MW. The power consumption decreased

Fig. 14. Matlab-HYSYSwith lower seawater temperature and higher high side pressure asshown in Fig. 18. The gure also shows that for a given seawatertemperature, the high side pressure needs to be increased in orderto reduce the compression power. The reason for this reduction inpower consumption, even though the pressure is higher, is that ina high pressure case, the re-circulated and expandedmass ow rate

Fig. 15. Total power consumption for CO2 liquefaction and pressurization at differentliquefaction pressures.is reduced and it is in the two-phase region which has lowertemperature and lower enthalpy.

Two open cycles at two different high side pressures are plottedin P-h diagrams, as shown in Fig. 19. Although they are both cooledby an equivalent condensing temperature (40 C), they havedifferent power consumption values for liquefying 72.42 Ton/hr ofCO2 feed gas. In order to simplify plotting the cycles in P-h diagram,the water content was ignored, as it is only a small percentage andmost of it would condense during the rst compression stage. Thehighest pressure side for the low pressure cycle, Fig. 19a, is 50 bar

rface code structure.and its power consumption is 28.7 MW. The highest pressure of thehigh pressure cycle, Fig. 19b, is 93 bar and its power consumption is8.014 MW. As explained, the reason for this difference is that thehigh pressure cycle uses two-phase CO2, which has more coolingcapacity than the low pressure cycle which uses vapor CO2. Thisresulted in three times more re-circulated CO2 mass ow rate forthe low pressure cycle than the high pressure cycle.

5.2. Open CO2 liquefaction cycle model verication

Since there is no available experimental data for the open CO2liquefaction cycle, the developed HYSYSmodel was veried againstEES model for two different high side pressures. The vericationresults show good agreement, as tabulated in Table 11.

5.3. Modied open CO2 liquefaction cycle

The least power consumption using open CO2 liquefaction cycle,Fig. 17, for liquefying the feed gas showing in Table 3 was found to

Table 10Results from HYSYS models for different CO2 liquefaction cycles.

Refrigerant R134a NH3 NH3 NH3 R134a CO2 C3H8 NH3

Liquefaction pressure (bar) 30 30 6 70 70 70 70 50Pump (MW) 0.34 0.34 0.34 0.38 0.38 0.38 0.38 0.33VCC power (MW) 1.43 1.37 4.39 0.15 0.15 0.6 0.16 0.59CO2 Compressor power (MW) 4.36 4.36 1.98 5.44 5.44 5.44 5.44 5.01Total power (MW) 6.13 6.06 6.72 5.97 5.98 6.42 5.99 5.93

mal Engineering 33-34 (2012) 144e156A. Alabdulkarem et al. / Applied Ther154be 8.2 MW. After pumping to 150 bar, the liqueed CO2 is ata temperature of43 C. A simplemodicationwould be to use thisstream in the multi-steam heat exchangers to liquefy the incominggas stream. Thus, less mass ow rate needs to be re-circulated. Thismodication resulted in reducing the power consumption to7.33 MW or 11.87% savings.

6. Comparison

Table 12 lists the least power consumption for all exploredoptions. It shows that the developed VCC that uses NH3 asa refrigerant and recovers the cold energy in the liqueed CO2 atthe optimum liquefaction pressure consumes less energy thanthe conventional multi-stage compression with intercooling by5.12%. Further, the NH3eCO2 cascade cycle consumes less powerthan the single-refrigerant cycle if the cold CO2 energy were notrecovered. Nonetheless, cascade cycles are more complex andcostly than single-refrigerant cycles. The open CO2 liquefaction

Fig. 16. Cascade CO2 liquefaction cycle using ammonia in the top cycle and CO2

Fig. 17. Open CO2 liquefacticycle consumes the highest power due to the large re-circulatedmass ow rate (31.2% more power than conventional

in the bottom cycle at 8 bar liquefaction pressure with cold CO2 recovery.

on cycle HYSYS model.

Fig. 18. Exhaustive search results for varying the condensing temperature and the highside pressure of the open CO2 liquefaction cycle.

Low Pressure Cycle

High Pressure Cycle

-450 -350 -250 -150 -50 50 150100

101

102

103

h [kJ/kg]

P [b

ar]

40C25C

0C

-25C

-55C 0.2 0.4 0.6 0.8

CarbonDioxide

(In)

14

(Out)

MSHX1

MSHX2

-500 -400 -300 -200 -100 0 100 200 300100

101

102

103

h [kJ/kg]

P [b

ar]

40C

-25C

0.2 0.4 0.6 0.8

CarbonDioxide

(In)

(Out)

MSHX1

MSHX2

a

b

Fig. 19. Peh diagrams for two open CO2 liquefaction cycles.

Table 11Open CO2 liquefaction cycle model verication.

Power consumption HYSYSmodel

EESmodel

Discrepancy (%)

Power consumption (HP 93 bar) (MW) 8.014 7.71 3.79Power consumption (HP 50 bar) (MW) 28.7 28.38 1.11

Table 12Summary of the least power consuming options for pressurizing the CO2 to 150 bar.

Option Total PowerConsumption (MW)

Note

Compression 6.25Single-refrigerant 8.65 NH3 VCC at 6 bar

liquefaction pressureSingle-refrigerant

w/energy recovery5.93 NH3 VCC at 50 bar

liquefaction pressureCascade cycle 7.52 NH3eCO2 VCCs at 8 bar

liquefaction pressureCascade cycle

w/energy recovery6.69 NH3eCO2 VCCs at 8 bar

liquefaction pressureOpen cycle 8.2 93 bar high side pressureModied open

cycle w/energy recovery7.33 93 bar high side pressure

NH3 absorption chillers 4.37 Subjected to the availabilityof waste heat

Liquefaction using seawater 6.09 Subjected to the availabilityof seawater at 27 C

A. Alabdulkarem et al. / Applied Thermalcompression at 40 C condensing temperature). The least powerconsuming option is with using absorption chillers. However,absorption chillers require waste heat which needs to be avail-able at certain conditions and quantity. Liquefaction of CO2 usingseawater would be less than 6.09 MW if cooler seawater isavailable.

7. Conclusion

Several options were investigated for pressurizing thecaptured CO2 gas for EOR applications. The investigation wascarried out through development of HYSYS models with modelsvalidated in good agreements. The liquefaction of CO2 wasoptimized using a generic HYSYS model coupled with the Matlaboptimization tool that considers all power consumption and allrecoverable cooling. The optimization results were graphed fordifferent efciency values which can be used as a tool tocompare power consumption from liquefaction of CO2 againstcompression. For a liquefaction cycle that has 70% Second Lawefciency, the results show that the optimum liquefaction pres-sure is 50 bar, where CO2 liquefaction and pumping consumesless power than compression. The developed approach ofcoupling Matlab and HYSYS can be used to optimize any processsimulated in HYSYS software. An open CO2 liquefaction cyclemodel was developed and its performance was investigated. Theopen CO2 liquefaction cycle power consumption was found todecrease as the high side pressure increases at the givencondensing temperature because two-phase uid would be usedto liquefy the incoming CO2 stream. An improvement was madeon the conventional open CO2 liquefaction cycle that resulted ina 10.61% power savings over the conventional open CO2 lique-faction cycle. Last, a new thermodynamic parameter, a, wasintroduced that can determine the temperature increase duringisentropic compression process. This parameter can help indetermining the number of compression stages for a givenrefrigerant so that the compressor outlet temperature is not veryhigh.

Acknowledgements

The authors would like to thank the sponsors of the Centerfor Environmental Energy Engineering (CEEE), University ofMaryland, College Park, MD, USA and the Petroleum Institute ofAbu-Dhabi, UAE, for their nancial support of this researchproject.

Nomenclature

AbbreviationsP Pressure (kPa)_m Mass ow rate (kg/s)h Enthalpy (kg/MJ)s Entropy (kg/MJ-K)_W Power (MW)ex Exergy (MJ/kg)Ref RefrigerantNG Natural gasVCC Vapor compression cycleNGCC Natural gas combined cycleCCS CO2 capture and sequestrationC3H8 PropaneHP High pressure

Engineering 33-34 (2012) 144e156 155EOR Enhanced oil recovery

Greek symbolsh Efciency

Subscriptsc compressori IntermediateH HighEx ExpansionL Lowsc Subcoolingsh Superheating

References

[1] H.J. Herzog, E.M. Drake, Carbon dioxide recovery and disposal from largeenergy systems, Annual Review of Energy and the Environment 21 (1) (1996)145e166.

[2] B. Moller, M. Genrup, M. Assadi, On the off-design of a natural gas-redcombined cycle with CO2 capture, Energy 32 (4) (2007) 353e359.

[3] P.G. Cifre, K. Brechtel, S. Hoch, H. Garca, N. Asprion, H. Hasse, G. Scheffknecht,Integration of a chemical process model in a power plant modelling tool for thesimulation of an amine based CO2 scrubber, Fuel 88 (12) (2009) 2481e2488.

[4] Z. Amrollahi, I.S. Ertesvg, O. Bolland, Thermodynamic analysis on post-combustion CO2 capture of natural-gas-red power plant, InternationalJournal of Greenhouse Gas Control (2010) 422e426.

[5] Thermoow, Inc. GT PRO software, http://www.thermoow.com/CombinedCycle_GTP.htm.

[6] I. Pfaff, J. Oexmann, A. Kather, Optimised integration of post-combustion CO2capture process in greeneld power plants, Energy 35 (10) (2010) 4030e4041.

[7] STEAG Energy Services, EBSILON Professional, http://www.steag-systemtechnologies.com/.

[8] T. Sanpasertparnich, R. Idem, I. Bolea, D. deMontigny, P. Tontiwachwuthikul,Integration of post-combustion capture and storage into a pulverized coal-red power plant, International Journal of Greenhouse Gas Control 4 (3)(2010) 499e510.

[9] Technotrade Inc., IPSEpro software, http://www.technotrade.com.pk.[10] L.M. Romeo, I. Bolea, Y. Lara, J.M. Escosa, Optimization of intercooling

compression in CO2 capture systems, Applied Thermal Engineering 29 (8e9)(2009) 1744e1751.

[11] Engineering Equation Solver, EES, http://www.fchart.com/ees/.[12] A. Aspelund, M.J. Mlnvik, G.D.E. Koeijer, Ship transport of CO2 technical

solutions and analysis of costs, energy utilization, exergy efciency and CO2emissions, Chemical Engineering Research and Design 84 (2006 September)847e855.

[13] A. Aspelund, K. Jordal, Gas conditioning e The interface between CO2 captureand transport, International Journal of Greenhouse Gas Control 1 (3) (2007)343e354.

[14] J. Moore, M. Nored, R. Gernentz, K. Brun, Novel Concepts for the compressionof large volumes of Carbon Dioxide Technical Report. Southwest ResearchInstitute, 2007.

[15] C. Botero, M. Finkenrath, C. Belloni, S. Bertolo, M. DErcole, E. Gor, Thermoe-conomic Evaluation of CO2 Compression Strategies for Post-Combustion CO2Capture Applications, in: Proceedings of ASME Turbo Exp (2009).

[16] Aspen Technology, Inc. Aspen and Hysys software. www.aspentech.com.[17] K. Gluesenkamp, R. Radermacher, Y. Hwang, Trends in Absorption Machines.

ISHPC, Padova Italy, 2011.[18] MAN Diesel & Turbo, www.mandieselturbo.com.[19] J.A. Dopazo, J. Fernndez-Seara, Experimental evaluation of a cascade refrig-

eration system prototype with CO2 and NH3 for freezing process applications,International Journal of Refrigeration (2010) 257e267.

[20] M. Mclinden, Thermodynamic evaluation of refrigerants in the vapourcompression cycle using reduced properties, International Journal of Refrig-eration 11 (3) (1988) 134e143.

[21] E.B. Ratts, J.S. Brown, A generalized analysis for cascading single uidvapor compression refrigeration cycles using an entropy generationminimization method, International Journal of Refrigeration 23 (2000)353e365.

[22] T. Lee, C. Liu, T. Chen, Thermodynamic analysis of optimal condensingtemperature of cascade-condenser in CO2/NH3 cascade refrigeration systems,International Journal of Refrigeration 29 (7) (2006) 1100e1108.

[23] A. Alabdulkarem, A. Mortazavi, Y. Hwang, R. Radermacher, P. Rogers, Opti-mization of propane pre-cooled mixed refrigerant LNG plant, Applied ThermalEngineering 31 (6e7) (2011) 1091e1098.

[24] MathWorks Inc. Matlab software. www.mathworks.com.

A. Alabdulkarem et al. / Applied Thermal Engineering 33-34 (2012) 144e156156

Development of CO2 liquefaction cycles for CO2 sequestration1 Introduction2 CO2 compression cycle3 CO2 liquefaction cycle3.1 Introduction3.2 CO2 liquefaction using cold seawater3.3 CO2 liquefaction cycles3.4 Single-refrigerant CO2 liquefaction cycles3.5 Cascade CO2 liquefaction cycles3.6 Optimization of CO2 liquefaction pressure3.7 Using cold CO2 for cooling the CO2 and refrigerant3.7.1 Single-refrigerant CO2 liquefaction cycles with cold CO2 recovery in CO2 liquefaction cycle3.7.2 Cascade CO2 liquefaction cycles with cold CO2 recovery in CO2 liquefaction cycle

4 Absorption chillers for CO2 liquefaction5 Open CO2 liquefaction cycle5.1 Conventional open CO2 liquefaction cycle5.2 Open CO2 liquefaction cycle model verification5.3 Modified open CO2 liquefaction cycle

6 Comparison7 Conclusion Acknowledgements Nomenclature References