Ethane Recovery in Natural Gas Processing Plant

10
CHEMICAL ENGINCERING RESEARCH AND DESIGN 8 8 (2 0 10)779-787 ELSEVIER Contents lists available at ScienceDirect Chemical Engineering Research and Design journal homepage: www,elsevier.com/locate/cherd Optimum ethane recovery in conventional turboexpander process R. Chebbi'^-% N.S. AI-Amoodi^ N.M. Abdel )ahhar°, G.A. Husseíni^ K.A. Al Mazroui" ^ Department 0/Chemical Engineering, American uniuersity o/Sharjah, P.O. Box 26666, Sharjah, United Arab Emirates ^ Department 0/Chemical Engineering, The Petroleum Institute, P.O. ßox 2533, Abu Dhabi, United Arab Emirates ABSTRACT Ethane recovery in a conventional turboexpander process is optimized considering different demethanizer pressures and different feeds: a lean gas and a rich one. The design variables are varied, while meeting process constraints, in order to find the optimum conditions achieving the maximum profit. The analysis covers the whole process including the refrigeration part, and the entire typical demethanizer pressure range. The optimum ethane recovery is compared with the nnaximum possible recovery for each value ofthe demethanizerpressure. Recommendations are given regarding the selection of the level of ethane recovery, along with the demethanizer pressure, and refrigeration recovery system. I I © 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Natural gas liquids (NGL); Ethane: Recovery; Optimization; Tbrboexpander; Simulation 1. Introduction Natural gas liquids are valuable components in natural gas. Several extraction processes were proposed. Reviews about processes for NGL recovery can be found in Manning and Thonnpson (1991), Arnold and Stewart (1999), GPSA (2004) and Kidnay and Parrish (2006). A recent overview is given by Chebbi et al. (2008), and briefly reviewed here. Differ- ent options include JT (Joule Thompson) valve expansion, external refrigeration (Rüssel, 1977) using propane as a refrig- erant, and turboexpansion. Typically, a turboexpander is used in combination with JT expansion and propane refrigera- tion. Other methods include cascade refrigeration which is complex and requires high compression cost (Manning and Thompson, 1991). Mixed refrigerant (MacKenzie and Donnelly, 1985; Manning and Thompson, 1991) is commonly used in LNG processes, but much less in NGL recovery. Refrigerated lean oil absorption is expensive in terms of equipment and energy requirements and is hard to operate (Arnold and Stewart, 1999; GPSA, 2004), The turboexpander process is dominating ethane recovery processes. Different processes have been proposed. The evo- lution in design for the "old" generation was summarized by McKee (1977). The simple plant consists of turboexpansion. The other options include the use of side reboiler, refrigera- tion, and two stages of expansion (McKee, 1977). The next generation of ethane recovery processes includes the residue recycle (RR) process, the gas subcooled process (GSP) (Pitman et aL, 1998; GPSA, 2004). In the GSP, the gas stream leaving the cold separator is split into two streams, one of them feeding the turboexpander, and the second one is subcooled by the demethanizer overhead stream, flashed in a valve, and then sent to the demethanizer as a reflux. The next generation also includes the CRR process. It has one addition to the GSP and is reported to achieve very high ethane recovery {Pitman et al., 1998); but with a cryogenic compressor, cost may be prohibitive (Lee et al., 1999). The IPSI is another modification of the GSP (Lee et al., 1999; GPSA, 2004). Combining GSP with liquid subcooled process (LSP) (Jibril et al., 2006) was found to yield higher recoveries than GSP and LSP used individually using process simulation for eight different feeds. RSV (recycle split-vapor) process {Pitman et al., 1998) is another modifica- tion ofthe GSP, and RSVE (recycle split-vapor with enrichment) process is itself a modification of the RSV (Pitman et al., 1998) found to lower capital cost compared to RSV. Studies have been carried out in order to optimize ethane recovery. Wang considered different feeds with low CO2 content and found that a combination of turboexpansion • Corresponding author. E-mail address: [email protected] (R. Chebbi). Received 25 October 2008; Received in revised form 6 October 2009; Accepted 3 November 2009 0263-8762/$ - see front matter © 2009 The Institution of Chemical Engineers, Published by Elsevier B.V. All rights reserved. doi:10.1016/j.cherd.2009.11.003

description

Natural gas processing

Transcript of Ethane Recovery in Natural Gas Processing Plant

Page 1: Ethane Recovery in Natural Gas Processing Plant

CHEMICAL ENGINCERING RESEARCH AND DESIGN 8 8 (2 0 10 )779 -787

ELSEVIER

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www,elsevier.com/locate/cherd

Optimum ethane recovery in conventionalturboexpander process

R. Chebbi' -% N.S. AI-Amoodi^ N.M. Abdel )ahhar°, G.A. Husseíni^ K.A. Al Mazroui"

^ Department 0/Chemical Engineering, American uniuersity o/Sharjah, P.O. Box 26666, Sharjah, United Arab Emirates^ Department 0/Chemical Engineering, The Petroleum Institute, P.O. ßox 2533, Abu Dhabi, United Arab Emirates

A B S T R A C T

Ethane recovery in a conventional turboexpander process is optimized considering different demethanizer pressuresand different feeds: a lean gas and a rich one. The design variables are varied, while meeting process constraints,in order to find the optimum conditions achieving the maximum profit. The analysis covers the whole processincluding the refrigeration part, and the entire typical demethanizer pressure range. The optimum ethane recoveryis compared with the nnaximum possible recovery for each value ofthe demethanizerpressure. Recommendations aregiven regarding the selection of the level of ethane recovery, along with the demethanizer pressure, and refrigerationrecovery system.

I I © 2009 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Natural gas liquids (NGL); Ethane: Recovery; Optimization; Tbrboexpander; Simulation

1. Introduction

Natural gas liquids are valuable components in natural gas.Several extraction processes were proposed. Reviews aboutprocesses for NGL recovery can be found in Manning andThonnpson (1991), Arnold and Stewart (1999), GPSA (2004)and Kidnay and Parrish (2006). A recent overview is givenby Chebbi et al. (2008), and briefly reviewed here. Differ-ent options include JT (Joule Thompson) valve expansion,external refrigeration (Rüssel, 1977) using propane as a refrig-erant, and turboexpansion. Typically, a turboexpander is usedin combination with JT expansion and propane refrigera-tion. Other methods include cascade refrigeration which iscomplex and requires high compression cost (Manning andThompson, 1991). Mixed refrigerant (MacKenzie and Donnelly,1985; Manning and Thompson, 1991) is commonly used in LNGprocesses, but much less in NGL recovery. Refrigerated leanoil absorption is expensive in terms of equipment and energyrequirements and is hard to operate (Arnold and Stewart, 1999;GPSA, 2004),

The turboexpander process is dominating ethane recoveryprocesses. Different processes have been proposed. The evo-lution in design for the "old" generation was summarized byMcKee (1977). The simple plant consists of turboexpansion.

The other options include the use of side reboiler, refrigera-tion, and two stages of expansion (McKee, 1977).

The next generation of ethane recovery processes includesthe residue recycle (RR) process, the gas subcooled process(GSP) (Pitman et aL, 1998; GPSA, 2004). In the GSP, the gasstream leaving the cold separator is split into two streams,one of them feeding the turboexpander, and the second one issubcooled by the demethanizer overhead stream, flashed in avalve, and then sent to the demethanizer as a reflux. The nextgeneration also includes the CRR process. It has one additionto the GSP and is reported to achieve very high ethane recovery{Pitman et al., 1998); but with a cryogenic compressor, cost maybe prohibitive (Lee et al., 1999). The IPSI is another modificationof the GSP (Lee et al., 1999; GPSA, 2004). Combining GSP withliquid subcooled process (LSP) (Jibril et al., 2006) was foundto yield higher recoveries than GSP and LSP used individuallyusing process simulation for eight different feeds. RSV (recyclesplit-vapor) process {Pitman et al., 1998) is another modifica-tion ofthe GSP, and RSVE (recycle split-vapor with enrichment)process is itself a modification of the RSV (Pitman et al., 1998)found to lower capital cost compared to RSV.

Studies have been carried out in order to optimize ethanerecovery. Wang considered different feeds with low CO2content and found that a combination of turboexpansion

• Corresponding author.E-mail address: [email protected] (R. Chebbi).Received 25 October 2008; Received in revised form 6 October 2009; Accepted 3 November 2009

0263-8762/$ - see front matter © 2009 The Institution of Chemical Engineers, Published by Elsevier B.V. All rights reserved.doi:10.1016/j.cherd.2009.11.003

Page 2: Ethane Recovery in Natural Gas Processing Plant

780 CHEMICAL ENGINEERING RESEARCH A N D DESIGN 8 8 [ 2 0 I o ) 1 7 9 - l S ?

and external refrigeration achieves minimum energy require-ments (Bandoni et al, 1989}. Bandoni et al. (1989) divide theprocess into two sectors. The first one includes compression,recompression and atmospheric heat exchange while the sec-ond sector includes separation, expansion refrigeration, andheat exchange below ambient temperature. A general super-structure embedding different features was considered. Themain optimization variables they used are: cold tank tem-perature and pressure in addition to demethanizer columnpressure. The energy balance over sector 2 was found toprovide guidelines for the optimum ethane recovery processselection, and the economic objective function was foundto increase with demethanizer pressure due to less energyrequirement for gas recompression. Fernandez et al. (1991)extended the analysis to CO2-containing mixtures. One con-straint was added: prevention of carbon dioxide precipitationin the demethanizer. The effect of increasing the demeth-anizer pressure was also found to enhance the net annualbenefit (calculated as sales minus operating cost minus annualinvestment cost).

A cost-revenue comparison was performed for GSP, SFR{split-flow reflux), ISS (typical industry-standard stage piant)a typical turboexpander process, and CRR processes byWilkinson and Hudson (1992) for one gas feed containing6.76% C2+P a flow rate of lOOMMscfd, and 1050 psig nominalpipeline pressure. The SFR process is based on GSP, with thesubcooled stream used to reflux the demethanizer indirectly.The GSP was found to be the best choice if minimum capitalcost is needed, and CRR process was found to be the processof choice if ethane market is expected to be favorable. SFR pro-cess was found to be the best option if ethane price projectionsare low.

LSP, BTEP (basic turboexpander process), GSP, and 2-sDP(two-stage demethanization process) were compared by Diazet al. (1997) for different feed gas mixtures with % C2+ rang-ing from 6% to 25%. The GSP was found to give the maximumprofit for most mixtures, especially the lighter ones (lower C2+content). Optimal design and operating conditions were foundusing a superstructure embedding different expansion alter-natives together with mixed-integer nonlinear programming(MINLP). The impact of CO2 content was found to decreaseethane recovery.

Konukman and Akman (2005) analyzed flexibility and oper-ability of a H EN-integrated natural gas expander plant, usingHYSYS process simulation. They showed that HEN synthe-sis design and flexibility analysis, based on nominal values,would provide incorrect results due to significant changesof physical properties. Konukman and Akman concluded thedifficulty or impossibility of automated HEN synthesis for ahighly energy-integrated plant like the turboexpander processusing process flowsheet simulators.

Mehrpooya et al. (2006) simulated an existing NGL recov-ery unit using HYSYS. Two modifications were consideredsuitable for optimization: turboexpander, and turboexpander-exchanger configurations to find the best revamping alterna-tive. A genetic algorithm that used for optimization to ensurethe maximum profit achieved is global and not only local.

Five different turboexpander ethane recovery processeswere compared by Chebbi et al. (2004), and it was shown thatmore complex processing scheme may yield the same or lessethane recovery. TWo very common processes were consid-ered in Chebbi et al. (2008): the conventional turboexpanderprocess and the GSP which adds reflux to the conventionalprocess. Several other processes include modifications of GSP

as we have mentioned above. Considering four different gasfeeds and four different demethanizer pressures, and varyingall available design variables, Chebbi et al. (2008) showed thatthe conventional turboexpander process yields larger maxi-mum ethane recovery in most cases except when both thefeed is lean and the demethanizer pressure is low.

In the present work, we consider the effect of demethanizerpressure on optimal ethane recovery level selection to achievethe maximum profit in a conventional turboexpander process.Optimization is performed by changing all the design variablesfor each value of the demethanizer pressure, while satisfyingprocess constraints. In contrast to the works in Bandoni et al.(1989) and Diaz et al. (1997), the process simulation does notstart at the cold tank, and does include the whole NGL recov-ery unit. Recommendations are given based on the presentresults.

2. Simulation and optimization

As described in Chebbi et al. (2008), in the conventional processconsidered (Fig. 1 in Chebbi et al. (2008)), the feed is first cooledwhile providing the reboiler duty, then cooled further by heatexchange with the residue gas, followed by heat exchangewith propane in a chiller to reach a temperature of -31 FAfter separation from the liquid, the gas leaving the sepa-rator is cooled by heat exchange with the overhead streamleaving the demethanizer, then separated from the liquid inthe cold separator, expanded in a turboexpander and sent tothe demethanizer. The liquids leaving the two separators areexpanded through JT expansion and sent to the demethanizerat lower levels. The turboexpander provides part of the powerneeded to recompress the residue gas. The other portion isprovided by a recompressor to raise the residue gas pressureto882psia.

The temperatures after cooling by heat exchange with theresidue gas leaving the demethanizer are design variablesthat are changed in the optimization process to maximizeethane recovery for different demethanizer pressures. Theconstraints are conditions preventing temperature cross inthese two heat exchangers.

Fig. 1 also shows the refrigeration loop. The chiller is com-mon to both the turboexpander process and the refrigerationunit. A propane economizer cycle is selected (Manning andThompson, 1991). The compression cost is smaller comparedto propane refrigeration simple cycle since part of the vapor isobtained at an intermediate pressure, called economizer pres-sure after the first valve expansion (Manning and Thompson,1991); therefore the economizer cycle requires less power ofrecompression.

The optimum economizer pressure achieving minimumtotal compression power selected in our process is given byFraser (Manning and Thompson, 1991):

"cond.out

Pch(1)

in which Pcord,out ar d Pd, are the pressures of propane aftercondensation (achieved by air cooler in our case) and insidethe chiller.

We consider two different feeds: A and D (Bandoni et al.,1989; Diaz et al, 1997). Their compositions in terms of molefractions are given in T^ble 1. Feed A is a lean gas with 6%C2+ content, and feed D is a rich gas with 30% C2+ content.The four demethanizer pressures considered are 100,215,335,

Page 3: Ethane Recovery in Natural Gas Processing Plant

CHEMICAL E N G I N E E R I N G RESEARCH A N D DESIGN 8 8 ( 3 0 I o ) 761

Feed

NGL

REFRIGERATIONLOOP

Fig. 1 - Conventional ethane recovery process with the propane refrigeration loop at low demethanizer pressure.

and 450 psia as in Chebbi et al. (2008). In all cases, the feed gasis at 100 F and 882 psia. the residue gas is recompressed to882psia. and in the NGL stream, the molar ratio of Ci to C2 isset at 0.02 (Manning and Thompson, 1991). Also in all casesthe feed gas rate is 10.980 Ihmol/h as in Chebhi et al. (2008).

2.1. Lou; demethonizer pressure

The demethanizer pressure considered here is 100 psia. Theoutlet temperature from the first gas-to-gas heat exchangeris a design variable that affects the cooling partition betweenthe ftrst gas-to-gas heat exchange and refrigeration.

As explained in the next section, the simulation and opti-mization study for this case shows that the profit is enhancedwhen the first gas-to-gas heat exchanger is discarded, withrefrigeration providing the full cooling duty required to lowerthe feed gas temperature to -31 F. The process flowsheetshown in Fig. 1 reflects the abovementioned process change.

Table 1 - Feed gas composition.Component

NitrogenMethaneEthanePropaneButanesPentanesHexanes

0.010,930.030.0150.0090.0030,0036

0.010.690.150.0750.0450,0150.015

30

2.1. Intermediate demethanizer pressure I

The pressure considered as intermediate is 215 psia. As will bementioned in the next section, it is more economical in thiscase to keep the first gas-to-gas heat exchanger to cool thefeed gas before refrigeration to recover refrigeration from theresidue gas leaving the demethanizer, after it exits the secondgas-to-gas heat exchanger operating at low temperature. Theprocess at intermediate demethanizer pressure correspondsto the typical flowsheet in which two gas-to-gas heat exchang-ers are used (Fig, 1 in Chebbi et al. (2008)).

2.3. High demethanizer pressures

The pressures considered here are 335 and 450 psia. As for theintermediate pressure case, the results show that a first gas-to-gas heat exchanger is also economically justified and yieldsa higher profit. Another modification, dictated by thermody-namics, is required at high demethanizer pressures. As thedemethanizer pressure increases, the temperature profile inthe column is shifted up. The simulation results show that, at335 and 450 psia, the feed gas temperature is not high enoughto provide heat in the reboiler. Thus, an external heat sourcehas to be used in this case. The process flowsheet shown inFig. 1 was modified to reflect the two abovementioned processchanges.

3. Results and discussion

HYSYS simulation was performed with the Peng-Robinsonequation selected as the thermodynamic model. The design

Page 4: Ethane Recovery in Natural Gas Processing Plant

782 C H E M I C A L E N G I N E E R I N G R E S E A R C H A N D D E S I G N 8 8 ( 2 0 I 0 )

Table 2 -

P (psia)

100215335450

Results from optimization for feed D (high NGL to gas price ratio).

CUT

Mainprocess unit

4.623.353,93,41

(MM$)

Refrigerationcycle

3.612,113,433,39

FCl (MM$)

Main Refrigerationprocess unit cycle

18,210,79,177,08

13.49.62

1312.9

Total FCI{MM$)

31.620.422.120

Objectivefunction (MM$)

657.16S3.9637.3619.1

Ethanerecovery (%)

9185.376.264,3

variables are varied to maximize the profit P calculated in$/year as

P = SG + SNGL - CRM - (2)

in which SG and SNGL represent residue gas and NGL sales,respectively, CRM is the cost of feed gas (raw material), andCOMd is the cost of processing without depreciation. This costis given for a typical chemical process as (TUrton et al., 2003)

= 0.180FCI + 2,73CoL + 1-23 {CUT (3)

where FCI is the fixed capital investment, COL is the cost ofoperating labor. CUT is the cost of utilities, CWT is the cost ofwater treatment, and CRM is the cost of raw materials.

We define our objective function as

B - SG + SNGL - 0,180FCI - 1,23CUT (4)

The operating labor and feed gas costs. COL. and CRM, are con-stant; therefore P-B is constant, and in order to maximize theprofit, we can simply maximize B,

The fixed capital cost is obtained using the Lang factor tech-nique CRirton et al., 2003), and the purchase cost of the majorequipment is obtained using the cost correlations in Turton etal, (2003),

3.1. High NGL to gas price ratio

To determine residue gas and NGL sales, the output fromHYSYS is used along with the following assumed prices:8$/MMBtuand 1,53 $/US gallon. The ratio in this case is high:0.191 ($/US gallon)/($/MMBtu). The utility is assumed to be pro-vided by the residue gas. If more ethane is extracted from thefeed gas, it is clear that this will result in less residue gas sales,and more NGL sales. The basis used for updating the costs isthe chemical engineering plant cost index (GEPCl) for January2008.

For the low demethanizer pressure case (100 psia), opti-mization shows that the first gas-to-gas heat exchangerbetween the feed gas and the overhead stream from the

demethanizer should be discarded in order to maximize profit,which means that the refrigeration duty needs be higher toachieve the same final temperature, -31 F, after chilling.

3.1.1, First gas-to-gas heat exchange uersus adciítíonaíre/rigeration loadThe optimization results show that for the 100 psiademethanizer pressure case, the first gas-to-gas heatexchanger needs to be discarded in order to maximizethe profit.

Although this seems to be contradictory because of lessrefrigeration recovery and more load on the refrigeration unitin this case, the overall effect can be determined by con-sidering the impact on the main process side as well. Inorder to explain the apparent contradiction, we consider asan example two cases in which the overall duty (needed tolower the temperature of the feed gas to the same tempera-ture of -31 F) is shared equally between the first gas-to-gasheat exchanger and the chiller (refrigeration unit) in the firstcase, and shared as 40% for the refrigeration part in thesecond case, with more refrigeration recovery (60% of theload) provided by the residue gas in the first gas-to-gas heatexchanger. Tables A1-A3 in the appendix show the effectof changing the partition from 50-50% (case 1) to 60^0%(case 2). Also included in the appendix are the equations thatillustrate how most CBM values for case 2 can be obtainedsimply from those calculated for case 1. Although the par-tition change results in a reduction of the refrigeration costas expected, the process side is affected adversely: moregas-to-gas heat exchange yields a higher residue gas tem-perature at the inlet to the booster, and therefore a highercost of compression (in the booster) and air cooling. At inter-mediate and high demethanizer pressures, the compressioncost {process side) is less (compared to the 100 psia pres-sure case) and the simulation study shows that, for a givenintermediate (215 psia) or high demethanizer pressure (335 or450psia), higher profit is achieved by having a gas-to-gas heatexchanger for refrigeration recovery before the chiller in addi-tion to the low temperature gas-to-gas heat exchanger shownin Fig. 1.

Table 3 - Results from optimization for feed A (high NGL to gas price ratio).

P (psia)

100215335450

CUT

Mainprocess unit

4,913.773.032.03

(MM$)

Refrigerationcycle

1,630,71,241,23

FCI

Mainprocess unit

24,114,410.88.2

(MM$)

Refrigerationcycle

7.254,186.16,05

Total FCI(MM$)

31,318.616,914,3

Objectivefunction {MM$)

353,23S4,9349,9344.7

Ethanerecovery (%)

92.682.465.445,9

Page 5: Ethane Recovery in Natural Gas Processing Plant

C H E M I C A L E N G I N E E R I N G R E S E A R C H A N D D E S I G N 8 8 ( 2 0 I 0 )

3.1.2. Maximum profitleíble 2 clearly shows that selecting lower demethanizer pres-sures yields higher profit for feed D, which is rich. The higherNGL recovery offsets the additional cost of compression atlow pressure. For feed A, which is lean, there is an optimumdemethanizer pressure at which the ohjective function, andtherefore the profit, is at a maximum (see Table 3). Comparedto feed A, the profits are substantially higher in the case of feedD (due to the substantially higher NGL content) with compa-rable capital costs at low and intermediate (100 and 215psia)demethanizer pressures.

3.1.3. Ouerall capital costFor feed A. as the demethanizer pressure decreases, the overallcapital cost increases, which is expected due to higher recom-pression cost needed to raise the residue gas pressure for thepurpose of transportation. For feed D, the overall capital costincreases as pressure increases from 215 to 335 psia. Althoughthe recompression cost is lower at higher demethanizer pres-sure, the refrigeration load is higher due to a lower potential ofcooling for the residue gas which leaves the demethanizer ata higher temperature. Another reason for the increase of thecapital cost at higher demethanizer pressures is the necessityof an external heat source.

3.1.4. Rc/rigeration costsThe results in lïihle 3 clearly show that refrigeration is lesscostly (both in terms of capital cost and utility) in the 215 psiacase compared to both 335 and 450 psia cases, which isexpected since the residue gas reaches the first gas-to-gas heatexchanger at a lower temperature, and has a larger potentialto cool the feed gas, and as a result, the load on the refriger-ation unit is reduced. We would expect the refrigeration costto be lower in the 335 psia case compared to the 450 psia onefor the same reason; nevertheless another factor should betaken into account. At these high demethanizer pressures,the temperatures of the residue gas leaving the demethanizerbecome closer, making the difference in potential for coolingless; however, the heat capacity of the residue gas. is larger forthe 450 psia case due to the fact that more ethane and heaviercomponents are left in the residue gas, and this tends to givemore potential for cooling. Simulations show that the secondeffect (higher heat capacity) slightly surpasses the first effect(lower residue gas temperature) at these high demethanizerpressures (335 and 450 psia).

The refrigeration cost (both in terms of capital cost and util-ity) is the highest in the 100 psia case, due to the fact that thefirst gas-to-gas heat exchanger is eliminated, which adds thecorresponding cooling load to the refrigeration unit throughthe chiller. .

3.1.5. Main process capitoX and utility costsFor both feeds A and D. the main process capital cost decreasesas the demethanizer pressure increases due to lower recom-pression cost (see Tables 2 and 3). The same trend is noticedfor the utilities cost for all demethanizer pressures in the caseof feed A and for low and intermediate demethanizer pres-sures (100 and 215 psia) in the case of feed D. At high pressures(335 and 450psia). the utility cost needs to include the addi-tional external heating utility cost, which is by far larger inthe case of the rich feed D. For this reason the utility cost risesto a higher value instead of declining as for feed A when thedemethanizer pressure increases from 215 to 335 psia.

Table 4 - Results for maximum ethane recovery in thecase of feed D (high NGL to gas price ratio).

Maximum ethane recovery (%) Objectivefunction (MM$)

91.085.376.364.3

657.1653.9637.3619.1

T^ble 5 - Results for maximum ethane recovery in thecase of feed A (high NGL to gas price ratio).

Maximum ethane recovery (%) Objectivefunction (MM$)

94.182.465.445.9

353.1354.9349.9344.7

3.1.6. Comparison between optimum and maximumrecouery casesUsing the design variables values obtained to get the max-imum ethane recovery for each demethanizer pressure inChebbi et al. (2008) yields the profit at maximum recovery.Results are shown in Tables 4 and 5 for feeds A and D. respec-tively.

Comparing results from Tables 2 and 4 clearly shows thatthe optimum occurs at the maximum ethane recovery for feedD. which is rich. For feed A, which is lean, the same con-clusion is reached at intermediate and high demethanizerpressures (see Tobies 3 and 5); however, even at low demeth-anizer pressure. 100psia. the difference is not significant; andoperating at the highest ethane recovery level would give aprofit nearly equal to the optimum, along with a higher ethanerecovery (1.5% difference at the lowest demethanizer pressure100 psia).

The trends are better illustrated In Figs. 2-5, in which thedifference in profit and ethane recovery are plotted versusdemethanizer pressure.

720

680-

2 640-

ou 600-

520-

480

FeedD

o * - - - - - _ _ ,

50 100 150 200 250 300 350 400 450 500

Demethanizer Pressure, psia

Fig. 2 - Objective function as a function of demethanizerpressure for feed D (circle: optimum value, triangle: valuecorresponding to maximum ethane recovery, continuouscurve: high NGL to gas price ratio, dashed curve: low NGL togas price ratio).

Page 6: Ethane Recovery in Natural Gas Processing Plant

784 C H E M I C A L E N G I N E E R I N G RESEAKCH A N D D E S I G N 8 8 ( 2 0 I O ) 1 1 S - 1 Z 7

390

34050 100 150 200 250 300 350 400 450 500

Demethanizer Pressure, psia

Fig. 3 - Objective function as a function of demethanizerpressure for feed A (circle: optimum value, triangle: valuecorresponding to maximum ethane recovery, continuouscurve: high NGL to gas price ratio, dashed curve: low NGL togas price ratio).

4050 100 150 200 250 300 350 400

Demethanizer Pressure, psia450

Fig. 5 - Percent C2 recovery as a function of demethanizerpressure for feed A (circle: optimum value, triangle:maximum ethane recovery, continuous curve: high NGL togas price ratio, dashed curve: low NGL to gas price ratio).

100

50 100 150 200 250 300 350 400 450 500Demethanizer Pressure, psia

Fig. 4 - Percent C2 recovery as a function of demethanizerpressure for feed D {circle: optimum value, triangle:maximum ethane recovery, continuous curve: high NGL togas price ratio, dashed curve: low NGL to gas price ratio.The dashed curve coincides with the continuous one).

3.2. Loiu NGL to gas price ratio

The simulations were repeated with the following prices:9.60$/MMBtu and 1.00$/US gallon. The ratio in this case islow: 0.104 ($/US gallon)/($/MMBtu), Only differences or impor-tant similarities with respect to the previous case (high NGLto gas price ratio) will be discussed in this part. The comparedresults can be seen in Figs, 2-5 introduced previously.

3.2.1. Pro/itThe results obtained are shown in Tables 6 and 7, In all cases,the objective function is seen to show a peak, with an opti-mum pressure at which the profit is maximum. This was notthe case in part 1 (high NGL to gas price ratio) for rich feed D, forwhich producing the maximum NGL, by operating at the low-est pressure 100 psia, yields the highest profit, a likely resultat high NGL prices for a rich gas. At low NGL to gas price ratio,the values of the objective function, and therefore the profit,are smaller in the case of feed D, which is expected since thegas is rich, and NGL price is low. On the contrary, the profitsare higher in the case of feed A, which is also anticipated sincethe feed is lean (with a significantly smaller NGL productionrate), and gas prices are high. Considering feeds A and D sep-arately, the objective function values are relatively close forthe different demethanizer pressures at low NGL to gas priceratio.

3.2.2. Main process and refrigeration utilities costsCompared to the high NGL to sales gas price ratio case, the util-ities costs are higher in all cases due to the higher gas priceexcept for feed A at low demethanizer pressure (100 psia), forwhich the refrigeration utility cost increases but the main pro-cess utility cost declines. Optimization yields an increase inthe outlet temperature of the feed gas exiting the cold gas-to-gas heat exchanger shown in Fig. 1 (only one such heatexchanger is included at 100psia, for the reasons mentionedabove), which enhances the power of the turboexpander, andreduces the load on the booster, and therefore the utility coston the process side.

Table 6 - Results from optimization for feed D (low NGL to gas price ratio).

P (psia)

100215335450

CUT

Mainprocess unit

5,554.024.694.1

(MM$)

Refrigerationcycle

4,322.544.114,07

FCI

Mainprocess unit

18.210.79.177.08

(MM$)

Refrigerationcycle

13.49.62

1312.9

Total FCI(MM$)

31.620.422.120

Objectivefunction (MM$)

535.7538,7530,6525

Ethanerecovery (%)

9185.376.264.3

Page 7: Ethane Recovery in Natural Gas Processing Plant

C H E M I C A L E N G I N E E R I N G R E S E A R C H A N D D E S I G N 8 8 ( 2 0 1 0 I

Table 7 - Results from optimization for feed A (low NGL to gas price ratio).

P (psia) CUT ( M M $ ) FC! (MM$) Total FCI(MM$)

Objectivefunction (MM$)

Ethanerecovery {%)

Main Refrigeration Main Refrigerationprocess unit cycle process unit cycle

100215335450

4.164.523.642.44

2.438.41.491.47

23.314.410.88.2

8.314.186.16.05

31.618.616.914.3

374.2378376.6376.3

83.382.465.445.9

Table 8 - Results for maximum ethane recovery in thecase of feed D (low NGL to gas price ratio).

4. Conclusions

Maximum ethane recovery (%) Objectivefunction (MM$)

91.085.376.264.3

535.7S38.7530.6525.0

Table 9 - Results for maximum ethane recovery in thecase of feed A (low NGL to gas price ratio).

Maximum ethane recovery (%) Objectivefunction (MM$)

94.182.465.445.9

374.0378.0376.6376.3

3.2.3. Capital costsThe main process and refrigeration unit capital costs are thesame as for the high NGL to sales gas price ratio case, except forfeed A at demethanizer pressure equal to 100 psia. This specialcase is the only case in which the design variable (temperatureof the feed gas exiting the cold gas-to-gas heat exchanger) atoptimum conditions does not correspond to the limiting casewhere the design variable is set just to avoid temperature crossin the gas-to-gas heat exchanger. In all other cases, equipmentsizing is the same, and equipment prices are considered notaffected by the relative costs of NGL and sales gas. As indicatedin the previous part, there is less load on the booster in thisspecial case, and this yields not only a decrease of the utility onthe process side, but also a drop in the capital cost comparedto the high NGL to sales gas price ratio case.

3.2.4. Ethane recoueryThe results are shown in Tables 6-9. As for the high NGLto sales gas price case, for a given demethanizer pressure,the optimum is found to occur at maximum ethane recov-ery except for the lean feed A at tow demethanizer pressure,100 psia. The deviation from maximum ethane recovery is sig-nificantly higher for the low NGL to sales gas price case (seeTables 7 and 9). This result is likely since it yields more ethaneleft in the residue gas, at optimum condition, in the case ofhigher sales gas price. However for both prices ratios, the max-imum profit (optimum) is only slightly higher than the profitat maximum recovery.

The present work shows that in all studied cases except one,there is an optimum demethanizar pressure at which theprofit is maximum. The exception corresponds to rich gasD and high NGL to gas price ratio. The optimum is reachedin this special case at the lowest demethanizer pressure,100 psia. In contrast to the works in Bandoni et al. (1989) andDiaz et al. (1997) the process simulation does not start at thecold tank, and does include the whole NGL recovery unit inour analysis. In addition, the demethanizer pressure rangeis wider in our work and does cover the typical range indi-cated in Manning and Thompson (1991). Recovery is foundto be adversely affected at higher demethanizer pressures asexpected. For a given demethanizer pressure, the recovery atoptimum conditions is found to be equal to the maximumrecovery at the specified pressure, except for feed A at lowdemethanizer pressure (100 psia). In addition, the maximumprofit (optimum) is found equal to the profit at maximumethane recovery except for the special case of feed A at lowdemethanizer pressure for which the optimum profit andvalue at maximum ethane recovery are found not equal hutvery close.

Optimization shows that the use of a gas-to-gas heatexchanger before the chiller to recover refrigeration yieldslower profit at low demethanizer pressure (100psia), and thusshould be discarded at low pressure. At high demethanizerpressures (335 and 450 psia), the feed gas cannot be used toprovide the reboiler duty. Therefore, less cooling (refrigera-tion recovery) can be achieved, and an externa! heat sourceis required. At intermediate pressure (215 psia). the main pro-cess is a conventional turboexpander unit with two gas-to-gasheat exchangers, and where the demethanizer reboiler dutyis provided by the feed gas, which results in further coolingbefore refrigeration. The costing structure is relatively com-plex and accounts for the overall cost including the interactionbetween the main process part and the refrigeration unitthrough the chiller. Details are given to explain the costingstructure. Due to fluctuations in the prices of NGL and salesgas, it is essential, for the purpose of optimization, to havea control system capable of adapting to changes needed inthe design variables (including the demethanizer pressure) foroptimization, along with a flexible process that encompassespossible required changes in terms of refrigeration recoveryas the demethanizer pressure changes as discussed in thepresent paper.

Acknowledgement

This work was supported by the American University of Shar-jah under grant no. FRG07-040,

Page 8: Ethane Recovery in Natural Gas Processing Plant

786 C H E M I C A L E N G I N E E R I N G R E S E A R C H A N D D E S I G N 8 8 ( 2 0 ! 0 ) 7 7 9 - 7 8 7

Appendix A. Appendix

The relevant bare module equipment costs (CBM) (TUrton etal., 2003), utility costs, and total annual costs are comparedin Tables A1-A3. The CBM values for case 2 can be obtaineddirectly as for case 1; however most of those values can beobtained by simply using the equations given below.

The utilities in Table Al are proportional to the compressorduties, therefore we have

The mass flow rate of the refrigerant, propane, is obtained as

(A2)0.4 .-m

The intensive properties inside the propane loop remain thesame. The utilities cost on the refrigeration side can be simplyobtained as

0404Utilityj - fTT Jtilityi (refrigeration loop) (A3)

Duty2

m

The sizes of the compressors and their drives in the refrigera-tion loop can be found as

(A4)

l^ble Al - Compared annual utility cost estimates for two different cooling partitions between gas-to-gas heat exchangeand refrigeration.

Equipment Refrigeration part: 50%

Duty (Btu/h) Duty ($/year)

Refrigeration part: 40%

Duty (Btu/h) Duty ($/year)

RefrigerationFirst compressorSecond compressorTbtal {refrigeration)

7,635,0007,298,000

14,933,000

Main Process (equipments affected by the partition change)Compressor (booster) {2) 70,437,000Total 85,370,000

535,100511,400

1,046,500

4,936,2005,982,700

6,108,0005,838,000

11,946,000

75,224,00087,170,000

428,100409,100837,200

5,272,0006,109,000

Table A2 - CBM estimates compared for two different cooling partitions between gas-to-gas heat exchange andrefrigeration (the only equipment included in the total cost are those affected by the partition change).

Equipment (number ofunits if more than 1)

Refrigeration part: 50%

Size per unit CBM

Refrigeration part: 40%

Size per unit CBM ($)

250.6 ft820 ft=1,606,000 Btu/h2,294,000 Btu/h

1,535,000 Btu/h2,192,000 Btu/h

3189 ft

RefrigerationFirst separatorSecond separatorChiller (2)First compressorDrive for the ñrst

compressorSecond compressorDrive for the second

compressorAir coolerTotal CBM {refrigeration)

Main process {equipment affected by the partition change)First gas-to gas heat 786 ft

exchangerCompressor (booster) (2) 7,936,000 Btu/hDrive for the booster (2) 10,581,000 Btu/hAir cooler 3278 ft'Total CBM (process

equipment affected by thepartition change)

Total

13,50095,500

1,461,4001,454,400

343,700

680,900331,400

241,3004,622,100

183,200

4,610,9002,319,100

261,3007,374,500

11,996,600

14.3 ft179.4ft'750 ft1,285,000 Btu/h1,835,000 Btu/h

1,228,000 Btu/h1,754,000 Btu/h

2552 ft

1054 ft=

8,475,000 Btu/h11,300,000 Btu/h3525 ft=

11,80077,800

1,342,5001,210,800

287,300

566,400277,000

217,1003,990,700

212,800

4,817,4002,442,600

270,7007,743.500

11,734,200

Table A3 - Compared annual cost for two different cooling partitions between gas-to-gas heat exchange and refrigeration(the only equipment included in the total cost are those affected by the partition change).

Cost Refrigeration part: 50% Refrigeration part: 40%

.23Cm-($/year) 9,906,800 10,006,400

Page 9: Ethane Recovery in Natural Gas Processing Plant

C H E M I C A L E N G I N E E R I N G R E S E A R C H AND DESIGN 8 8 ( 3 0 I O ) 787

The volumes of the two separators in the refrigeration loopcan be obtained using

A2 p2AiDi J VQi/Ua

3/2 3/2

3/2

V 0 5 /(A5)

where V, A, D and L represent the volume, the cross-sectionalarea, diameter and length of the vessel, respectively, and Uathe maximum allowable vapor velocity.

The sizes of the booster and the first gas-to-gas heatexchanger cannot be calculated directly from the valuesobtained for case 1 since there are changes in relevant inten-sive properties. The same is also valid for the chiller since ithas one stream on the process side.

The size of the equipment being found, and the bare mod-ule factors being the same for cases 1 and 2, the bare moduleequipment cost, CBM2, can be obtained using

CBM2 = (A6}

where the K values, for the corresponding equipment, aregiven in "nirton et al. (2003).

References

Arnold, K. and Stewart, M.. (1999). (second edition). SurfaceProduction Operations (Gulf Publishing Company),

Bandoni, J,A,. Eliceche, A.M.. Mabe, G,D,B. and Brignole, E,A., 1989,Synthesis and optimization of ethane recovery process.Computers & Chemical Engineering, 13: 587-594,

Chebbi, R,, Al-Qaydi, A,S., Al-Amery, A.O,, Al-Zaabi, N.S. andAl-Mansouri, H.A., 2004, Simulation study compares ethanerecovery in turboexpander processes. Oil & Gas Journal,102(4): 64-67,

Chebbi. R., Al Mazroui, K,A, and Abdel Jabbar, N,M,, 2008, Studycompares Cj-recovery for conventional turboexpander, GSP,Oil & Gas Journal, 106{46): 50-54,

Diaz, M.S,, Serrani, A., Bandoni, J,A, and Brignole, E.A,, 1997. 1Automatic design and optimization of natural gas plants.Industrial & Engineering Chemistry Research. 36: 2715-2724,

Fernandez, L.. Bandoni, J.A., Eliceche. A.M, and Brignole, E.A.,1991, Optimization of ethane extraction plants from naturalgas containing carbon dioxide. Gas Separation & Purification.5: 229-234,

GPSA Engineering Data Book,, (2004). Sec, 16 (twelfth edition). ,(Gas Processors Suppliers Association), '

Jibril, K,L,, Al-Humaizi. A.!,, Idriss, A,A, and Ibrahim, A,A,, 2006.Simulation study determines optimum turboexpanderprocess for NGL recovery. Oil & Gas Journal, 104(9): 58-62,

Kidnay. A,J. and Parrish, W,R,, (2006), Fundamentals of Natural GasProcessing {first edition). (Taylor and Francis),

Konukman. A.E.S. and Akman, U,, 2005. Flexibility and operabilityanalysis of a HEN-integrated natural gas expander plant.Chemical Engineering Science. 60: 7057-7074,

Lee, R.J,. Yao, J, and Elliot, D,G., 1999, Flexibility, efficiency tocharacterize gas-processing technologies. Oil & Gas Journal.97(50): 90-94.

MacKenzie. D.H and Donnelly. S.T, 1985, Mixed refrigerantsproven efficient in natural-gas-liquids recovery process. Oil &Gas Journal, 83(9): 116-120,

Manning. F.S. and Thompson. R,E,. {1991), {first edition), OiljieldProcessing of Petroleum (PennWell Publishing Company),

Mc.Kee. R.L., 1977, Evolution in design, In Proceedings of the 1Fifty-sixth Annual GPA Conuention Dallas,

Mehrpooya. M., Gharagheizi, F, and Vatani, A.. 2006, Anoptimization of capital and operating alternatives in a NGLrecovery unit. Chemical Engineering & Technology, 29(12):1469-1480.

Pitman, R,N.. Hudson, H.M., Wilkinson. J.D, and Cuellar, K,T..1998. Next generation processes for NGL/LPG recovery. InProceedings of the Seuenty-seuenth Annual GPA Conuention Dallas,

Rüssel. T,H,, 1977. Straight refrigeration still offers processingflexibility. Oil & Gas Journal, 75(4): 66-72,

TVirton, R,. Bailie, R.C.. Whiting. W,B. and Shaeiwitz. J,A,, (2003).Analysis, Synthesis and Design 0/Chemical Processes (secondedition). {Prentice Hall),

Wilkinson, J,D. and Hudson. H.M,, 1992. Improved NGL recoverydesigns maximize operating flexibility and productrecoveries. In Proceedings o/the Seventy-Jirst Annual GPAConuention Anaheim,

Page 10: Ethane Recovery in Natural Gas Processing Plant

Copyright of Chemical Engineering Research & Design: Transactions of the Institution of Chemical Engineers

Part A is the property of Elsevier Science and its content may not be copied or emailed to multiple sites or

posted to a listserv without the copyright holder's express written permission. However, users may print,

download, or email articles for individual use.