Applied Thermal Engineering 128 (2018) 51–63

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier .com/locate /apthermeng

Research Paper

Feasibility study of environmental relative humidity through thethermodynamic effects on the performance of natural gas liquefactionprocess

http://dx.doi.org/10.1016/j.applthermaleng.2017.08.0901359-4311/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: mynlee@yu.ac.kr (M. Lee).

Muhammad Abdul Qyyuma, Le Quang Minh a, Wahid Ali a, Arif Hussain a, Alireza Bahadori b,Moonyong Lee a,⇑a School of Chemical Engineering, Yeungnam University, Gyeongsan 712-749, Republic of Koreab Southern Cross University, Lismore, NSW, Australia

h i g h l i g h t s

� Effects of relative humidity on theperformance of SMR was investigatedsuccessfully.

� Compression energy for SMR processwas reduced significantly.

� Compression power has a linearrelation with the relative humidity.

� The UA value of LNG cryogenicexchanger increases as 4th-orderpolynomial function.

g r a p h i c a l a b s t r a c t

SMR PROCESS

a r t i c l e i n f o

Article history:Received 11 March 2017Revised 13 May 2017Accepted 18 August 2017Available online 24 August 2017

Keywords:LNGNatural gas liquefactionMCD optimizationSMRThermodynamic effectsRelative humidity

a b s t r a c t

This study examined the thermodynamic effects of relative humidity (RH) on the performance of the nat-ural gas liquefaction process. A single mixed refrigerant (SMR) liquefaction process was chosen for thisstudy because of its simplicity and compactness. In addition, it is considered the most promising processfor the liquefied natural gas (LNG) floating production, storage and offloading (FPSO) unit. The SMR pro-cess was optimized using a modified coordinate descent methodology, which resulted in 13.6% energysavings. Subsequently, an interface between commercial software Aspen Hysys� and MS-Excel VBAwas carried out to study the effects of RH. The results showed that RH has pronounced effects on the per-formance of the LNG cycle by affecting the enthalpy balance around the air coolers, which ultimatelyaffects the overall compression power, LNG exchanger performance, and other design and operationalparameters. Furthermore, when the RH was increased from 0% to 95%, the UA value (product of overallheat transfer coefficient and heat transfer area) of the air coolers and the overall compression powerdecreased and increased linearly, respectively. Moreover, the heat transfer coefficient of the LNG cryo-genic exchanger increased as a 4th order polynomial function in terms of the log-mean enthalpy differ-ence. The results can provide insight into the selection of the appropriate design and operationalparameters for the LNG plants associated with the regions of low or high relative humidity.

� 2017 Elsevier Ltd. All rights reserved.

Nomenclature

SMR single mixed refrigerantDMR dual mixed refrigerantC3MR propane precooled mixed refrigerantS molar entropyH molar enthalpyRH relative humidityMCD modified coordinate descentLNG liquefied natural gas

MR mixed refrigerantMITA minimum internal temperature approachNG natural gasBOG boil-off gasTDCC temperature difference between composite curvesTHCC temperature-heat flow composite curvesTD dry-bulb temperatureTW wet-bulb temperature

52 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

1. Introduction

The location of natural gas reserves has key importance in theeconomics of natural gas transportation either in gaseous form(through pipelines) or in liquid form (LNG). A reduction of physicalvolume (1/600th) of liquefied natural gas (LNG) compared to gas-eous natural gas can reduce the transportation costs by allowingdelivery using cargo ships or transport trucks instead of pipelines[1]. Ideally, LNG plants are located near the seashore to minimizethe delivery distance between the liquefaction plant and the LNGcargo terminal [2]. In addition to the transportation, other factors,such as high energy density and high purity compared to the gas-eous form, also motivates the liquefaction of natural gas. Therefore,the worldwide demand for natural gas is increasing rapidly. To sat-isfy this requirement, the LNG industry is growing and expandinginto new natural gas sources in different climatic zones rangingfrom the North Pole to the South Pole [3]. The regions associatedwith climate (with respect to natural gas sources and LNG plants)can be classified into three categories, i.e. Desert (e.g. Qatar), Trop-ical (e.g. Island of Borneo), and Arctic (e.g. Yamal Peninsula). Fig. 1presents the statistical result of a NG reservoir [4], where countriesnorth and south of the equator have different seasonal air proper-ties, such as temperature levels, wind velocity, precipitation, andrelative humidity.

In fact, these different seasonal air properties have an impact onthe energy performance of air-cooled LNG processes. In this study,the thermodynamic effect (due to a change in enthalpy) was inves-tigated at LNG production, overall compressor power, overall air-coolers, and LNG exchanger duty when varying the RH. Further-more, this study analyzed the impacts of relative humidity onthe overall transfer coefficient (U) of a LNG cryogenic exchangerfor air-cooled LNG plants. A commercial simulator, Aspen HYSYS�

V9 [5], was used to study the announced effects on the singlemixed refrigerant liquefaction process by varying the relativehumidity.

Fig. 1. Natural Gas proven reserves in the world in 2015 [4].

1.1. Air-cooled LNG plants and the impact of the relative humidity

Natural gas liquefaction is an energy intensive process, whichrequires 1188 kJ of energy to liquefy one-kilogram of natural gas[6]. This liquefaction energy depends on the selected liquefactioncycle (such as SMR, DMR, Cascade and C3MR etc.) and the climaticsite conditions. The operational and design parameters can differfrom site to site for the same liquefaction cycle. For example, afteran evaluation of more than 20 locations based on the environmen-tal and climatic conditions, ExxonMobil, BP, ConocoPhillips, andTransCanada choose a site in the Nikiski area on the Kenai Penin-sula for the proposed Alaska (North of Equator) LNG project [7].Fig. 2 presents the graphical data [8] of the average relative humid-ity according to the months of the Kenai Peninsula.

In addition, Western Australia Inc. currently operates three LNGexport projects: the NorthWest Shelf (Karratha), Pluto, and Gorgon[9]. Table 1 lists the annual average temperature, average morningrelative humidity percentage, and average evening relative humid-ity percentage for the corresponding sites of the North West shelfLNG plant (Karratha) and the Gorgon LNG plant (Barrow Island),which are located in Western Australia (South of the Equator)[10]. Similarly, Equatorial Guinea (EG) LNG (known as the PuntaEuropa LNG) plant is located at Malabo [11], which is also an air-cooler LNG plant. Fig. 3 shows the climatic conditions at the Mal-abo LNG site [12].

For LNG plants, seawater or air is mainly used as the coolingmedium to cool the compressed mixed refrigerant and is some-times used to reduce the temperature of the natural gas beforeentering the liquefaction section. The selection of cooling mediumdepends strongly on the site climatic conditions. Table 2 lists thecountries (here classified in the northern and southern of the equa-tor) with air-cooled LNG plants [13].

1.2. Thermodynamics of the relative humidity

Recently, Park et al. [14] examined the effects of varying ambi-ent air temperatures on the performance of a single mixed refrig-erant (SMR) liquefaction cycle and presented the optimal designand operational solutions for the SMR under various ambient tem-peratures. Xu et al. [15] investigated the correlation between themixed refrigerant composition and ambient conditions in theSMR process. Many factors can influence the ambient air tempera-ture [16]. This variation in the ambient air temperature (also referas Dry-bulb temperature [17]) is a significant factor that affects therelative humidity of the air. This does not mean that the ambientair temperature and the relative humidity have the same effecton the LNG cycle at the same site or at the different sites. Accord-ingly, the general and common definition of the relative humidityis shown as follows

RH ¼ Pw

PS¼ Actual water vapor pressure

Saturation water vapor pressureð1Þ

Fig. 2. Average relative humidity by month at Kenai, Alaska, USA [8].

Table 1Meteorological data of the two different LNG plant sites (Western Australia) [10].

Month Karratha LNG plant Gorgon LNG plant

Temperature (�C) RH% Temperature (�C) RH%

Min Max 9 AM 3 PM Min Max 9 AM 3 PM

Jan 26.7 36.0 59 51 25.8 33.2 70 58Feb 26.7 35.8 64 55 26.3 33.7 72 59Mar 25.9 36.1 54 46 26.5 33.2 69 59Apr 22.7 34.3 47 40 24.6 31.1 62 53May 18.3 30.0 45 42 21.3 27.4 60 55Jun 15.1 26.5 47 44 18.8 24.3 61 56Jul 13.8 26.2 45 40 17.6 23.7 65 58Aug 14.3 28.3 39 35 17.8 25.2 61 52Sep 16.9 30.8 40 36 19.5 27.0 62 51Oct 20.8 34.1 40 38 21.8 29.6 64 51Nov 23.1 35.0 43 41 23.0 30.5 65 52Dec 25.5 35.8 51 47 24.4 32.1 67 54

Fig. 3. Climatic conditions at the site of Equatorial Guinea (EG) LNG plant [12].

Table 2Air-cooled LNG plants [13].

Globe equator poles Countries

North MalaysiaNigeriaEquatorial GuineaAlaskaRussiaEgypt

South AustraliaTrinidad

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 53

Or,

RH¼ Actual amount of water vapor in air at specified TemperatureSaturated amount of water vapor at same specified Temperature

ð2Þ

where PS is the saturated vapor pressure of water, which is a func-tion of the dry bulb temperature. In other words, there is an inverserelationship between the RH and temperature. An increase ordecrease in RH has a significant effect on the enthalpy change at a

constant temperature. In this study, the Aspen Hysys� V9 [5]built-in stream saturator (SAT) was used to investigate the thermo-dynamic properties, such as the heat flow, enthalpy, and entropy ofair under various relative humidity. The mathematical equations forthe thermodynamic properties of humid air are given below.

From the ideal gas law,

PV ¼ nRT ð3ÞIf P = Pw, n = actual moles of water, Nactual, and V = volume of

humid air, the above definition of the relative humidity at the con-stant temperature is,

Pw:V ¼ Nactual:RT ð4Þ

Pw:V ¼ ðNSÞðRH%Þ:RT ð5Þ

54 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

where NS is the saturated amount of water vapor at a specifiedtemperature.

The above equation states that at the constant saturatedamount of water vapor there is a linear relationship between therelative humidity of air and the product of the partial pressure,Pw and the volume of humid air. Fig. 4 presents a graphical presen-tation of this relationship.

The wet-bulb temperature is considered as an importantparameter where direct-contact cooling/heating through air isinvolved such as cooling tower operations. The wet-bulb tempera-ture correlation with dry-bulb temperature and RH can beexpressed as in [18]:

Relative Hum idity %0 20 40 60 80 100

(PwV)

kPa

.m3 /h

0

10

20

30

40

50

60

Fig. 4. Relative humidity and PV of air.

Relative Humidity %

Enth

alpy

(kJ/

Km

ole)

-8000

-6000

-4000

-2000

0

Relative Humidity %0 20 40 60 80 100

Mol

es o

f hum

id a

ir (k

mol

e/h)

0.000

0.005

0.010

0.015

0.020

0.025

(a)

(c)

0 20 40 60 80 100

Fig. 5. Effect of relative humidity on th

TW ¼ TDatan 0:151977ðRH%þ 8:313659Þ0:5h i

þ atanðTD þ RH%Þ � atanðRH%� 1:676331Þþ 0:00391838ðRH%Þ1:5atanð0:023101RH%Þ � 4:686035 ð6Þ

where, TW and TD denote the wet-bulb and dry-bulb temperature,respectively.

Furthermore, the enthalpy of humid air, H, kJ/kg mole, is givenas

H ¼ Ha þ ðwÞðHvÞ ð7Þwhere ‘‘w” is the molar mixing ratio, kgmole-H2O/kg mole-Dry air.

The molar mixing ratio in terms of the relative humidity isexpressed

w ¼ 1:002ðPwÞPa � Pw

¼ ð1:002ÞðRHÞðPSÞPa � ððRHÞðPSÞÞ ð8Þ

Ha, the enthalpy of the dry air (only sensible heat is involved), iscalculated as follows:

From the 1st law of thermodynamics,

U ¼ Q �W ð9ÞAt a constant pressure, W = PV, thus,

U ¼ Q � PV ð10Þ

Q ¼ U þ PV ð11ÞFrom the definition of enthalpy,

H ¼ U þ PV ð12Þ

Relative Humidity %

Hea

t Flo

w (k

W)

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

Relative Humidity %

Entr

opy

(kJ/

Km

ole-

C

118.0

118.5

119.0

119.5

120.0

120.5

121.0

121.5

(b)

(d)

0 20 40 60 80 100

0 20 40 60 80 100

e thermodynamic properties of air.

Table 3Stream saturator conditions.

Name Dry air Humid air

0% RH 10% RH 90% RH

Vapor fraction 1 1 1Temperature (�C) 25 25 25Pressure (bar) 1.01325 1.01325 1.01325Molar flow (kmol/h) 0.6908 0.6931 0.7107Mass flow (kg/h) 20 20.04 20.36Enthalpy (kJ/kmol) �6.504 �776.7 �6767Entropy (kJ/kmol �C) 118.2 118.6 120.8Heat flow (kW) �1.248e�003 �0.1495 �1.336

Table 4Feed conditions and simulation basis [19].

Property Conditions

Natural gas (Stream-0)Temperature (�C) 32Pressure (bar) 50Flow rate (kg/h) 1.0Composition Mole fractionNitrogen 0.0020Methane 0.9133Ethane 0.0536Propane 0.0214i-butane 0.0046n-butane 0.0047i-pentane 0.0001n-pentane 0.0001

Pressure drops across LNG-exchangersStream-0 to Stream-6 (bar) 1.0Stream-2 to Stream-3 (bar) 1.0Stream-4 to Stream-5 (bar) 0.1Compressors isentropic efficiency (%) 75.0After-coolers outlet temperature (�C) 40.0MITA for air-coolers (�C) [20] 22.0

LNG (Stream-7)Pressure (bar) 1.209Boil-off gas (vapor fraction) 0.08

Fig. 6. Schematic diagram

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 55

Heat transfer at a constant pressure,

Q ¼ CPðDTÞ ð13ÞTherefore,

Ha ¼ U þ PV ¼ Q ¼ CPaðT � Tref ÞHv, which is the enthalpy of water vapor (sensible and latent

heat), is expressed as,

Hv ¼ CPaðT � Tref Þ þ k ð14ÞAt similarity, the molar entropy of humid air is, kJ/kgmole �C,

S ¼ Sa þ ðwÞðSvÞ ð15Þ

where Sa is an entropy of dry air; while Sv is the entropy of watervapor.

From the 2nd law of thermodynamics,

dS ¼ dQrev

Tð16Þ

and,

S ¼ SoðT ¼ 0KÞ þZ T

T¼0

CPðTÞT

dT ð17Þ

where So is the entropy at the reference state.The reference state was chosen at 0 K [5], and

SoðT ¼ 0KÞ ¼def 1 kJ=kg K ð18ÞFor heat flow, Q, kW,

Q ¼ Qa þ Qv ð19ÞFig. 5 shows the changes in enthalpy, heat flow, and entropy by

varying the relative humidity. The observations were examinedusing only the air saturator. These observations imply that airunder varying RH has some key effects on the performance ofLNG operation which is essential to be addressed. Note that thenegative enthalpy and heat flow indicate that increase in RH

of the SMR process.

56 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

(humidification) is an exothermic phenomenon. As seen in Fig. 5d,the higher percentage of RH leads to higher irreversibilities(entropy generation).

Table 3 lists the characteristics of the dry air (0% RH) and humidair at 10 and 90% relative humidity.

In particular, at 90% relative humidity, a negative change inenthalpy is significantly higher compared to that at 10% relativehumidity. Table 3 clearly shows the effect of the percentage ofRH on the thermodynamic properties of the air. The variation ofthe enthalpy, entropy, and heat flow due to an increase ordecrease in RH% provides an opportunity to investigate the ther-modynamic effects on the operational and designing parametersof LNG cycles by varying the RH. The RH of the air has a pro-nounced effect on the performance of air-cooled LNG plants thatare located or currently under construction in the areas of thenorth and south of the equator. To the best of the authors’knowledge, there are no reports that considered the effects ofthe environmental relative humidity on the design of the lique-faction process.

Table 5Decision variables with lower and upper bounds.

Decision variables Lower bound Upper bound

Condensation pressure of MR, P13 (bar) 35.0 85.0Evaporation pressure, P5 (bar) 1.1 3.5Flow rate of nitrogen, mN2 (kg/h) 0.02 0.35Flow rate of methane, mC1 (kg/h) 0.10 0.70Flow rate of ethane, mC2 (kg/h) 0.20 0.85Flow rate of propane, mC3 (kg/h) 0.95 2.5

Fig. 7. MCD optimiz

2. Methodology for studying the effects of relative humidity

The methodology adopted for examining the effects of relativehumidity on the performance of the SMR LNG process is as follows:

1. The SMR process was simulated using the conditions in a previ-ous study [19] with the assumption of negligible heat losses tothe environment.

2. The process optimization was carried out using the modifiedcoordinate descent methodology with the minimum consump-tion of energy as an objective function and MITA constraint of3 �C.

3. The optimal values of UA for air coolers obtained through pro-cess optimization were then used to replace the air-coolers withstream saturators under the assumptions of 95% relativehumidity and a 30 �C dry bulb temperature.

4. The knowledge based-optimization algorithm was used to findthe optimal flow rate of air with respect to the MITA value forair-coolers 22 �C [20].

5. Thermodynamics effects of the environmental relative humid-ity were conducted under various relative humidity and con-stant air flow rates and dry-bulb temperatures.

2.1. Process simulation

The SMR process, which is also known as the PRICO (poly refrig-erant integrated cycle operation) process, was chosen because it isthe simplest small-scale LNG liquefaction processes and one of themost promising candidates for a FLNG-FPSO unit, because it haslow equipment cost and a small area. The SMR process consists

ation algorithm.

(a)

-150 -100 -50 0 500

10

20

30

40

50

60

Temperature ( oC)

App

roac

h te

mpe

ratu

re ( o C

)

Hot composite curveCold composite curve

(b)

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 57

mainly of a cryogenic exchanger, expansion valves, compressors,and air coolers. A simulation model was developed in Aspen HYSYSV9.0 to investigate the thermodynamic effects of the RH on theperformance of the SMR process. Table 4 provides information onthe base case. Fig. 6 shows a schematic diagram of the SMRprocess.

2.2. Process optimization

Minimization of the specific compression power for the SMRprocess was chosen as an objective function that was constrainedto a MITA value of 3 �C. The variables in terms of the flow rate ofthe individual components of the SMR and the operating pressures(evaporation and condensation pressures) have a major impact onthe energy consumption (the objective function). Therefore, thesevariables were selected as the key decision variables in the opti-mization. The decision variables are listed with the lower andupper bounds in Table 5.

Mathematically, the objective function as the specific compres-sion power relative to the key decision variables and constraintsare defined as:

Minimize f ðXÞ ¼ Ws ¼X4i¼1

wi

(ð20Þ

Subjected to:

DTðminÞðXÞ P 3 ð21Þ

T6ðXÞ > T6;DewðXÞ ð22Þ

Xlb < X < Xub ð23Þwhere X is the vector of decision variables, X ¼ ðP13; P5;

mN2;mC1;mC2;mC3Þ, and T6 is the temperature of stream 6 or thetemperature of the recycled mixed refrigerant before entering thecompressor K-1 (see Fig. 6). Note that the temperature of othercompressors associated with intercoolers are reduced up to 40 �C,which is sufficiently higher than the dew point. However, the com-pressor K-1 includes no additional cooling/heating system to adjustthe temperature of inlet stream above its dew point. For this, thetemperature of stream 6 was considered as a required constraint.

2.3. Modified coordinate descent (MCD) optimization algorithm

The SMR process is considered a nonlinear system due to thenonlinear interactions between constraints, objective function,and decision variables. Optimizing a highly nonlinear system is avery difficult and challenging task using commercially availableprocess simulation software. Therefore, external optimizationtechniques need to be linked with process simulators. In this study,

Table 6Optimization results.

KBO optimization[19]

MCDoptimization

Condensation pressure of MR, P13 (bar) 49.97 62.49Evaporation pressure, P5 (bar) – 2.03Post-expansion temperature, (�C) �152.7 –Flow rate of nitrogen, mN2 (kg/h) 0.2276 0.2587Flow rate of methane, mC1 (kg/h) 0.5343 0.4625Flow rate of ethane, mC2 (kg/h) 0.6312 0.7655Flow rate of propane, mC3 (kg/h) 2.728 2.267MITA (�C) 3.0 3.0Specific compression power (kJ/kg-LNG) 1556.64 1344.24Total intercoolers duty (kW) 0.6571 0.5980

Relative compression energy saving (%) – 13.6Relative intercoolers energy saving (%) – 9.0

the MCD algorithm was used to optimize the SMR process. Micro-soft Visual Studio (MVS) was used to develop an optimizer with auser-friendly interface. The algorithm was coded in the MVS envi-ronment and then linked to the ASPEN HYSYS� V9 using COMfunctionality.

The MCD algorithm is based on the idea that the multivariablefunction is optimized by minimizing the objective function alongone coordinate [21]. The significant advantages of the MCD overthe conventional coordinate descent (CD) methodology are thesimplicity of each iteration, easy implementation, and high effi-ciency [22]. Accordingly, this algorithm is one of the best candi-dates to solve nonlinear interaction optimization problems. Fig. 7presents a schematic diagram of the MCD algorithm.

2.4. Optimization results

Referring to Khan’s report [19], the result of using the MCD opti-mization algorithm shows that the specific compression power can

Fig. 8. Composite curves of: (a) Optimized TDCC; (b) Optimized THCC.

Table 7Simulation conditions for stream saturators.

Conditions SAT-1 SAT-2 SAT-3 SAT-4

RH% 95.0 95.0 95.0 95.0Temperature (�C) 30.0 30.0 30.0 30.0Pressure (bar) 1.01325 1.01325 1.01325 1.01325Air mass flow (kg/h) 8.0 8.5 9.0 17.5

Fig. 9. Optimized SMR Process with the addition of the Aspen HYSYS Stream Saturator.

Air mass flow (kg/h)

4 6 8 10 12 14-10

-5

0

5

10

15

20

25Air-cooler-1

Air mass flow (kg/h)4 6 8 10 12 14

-15

-10

-5

0

5

10

15

20

25Air- cooler-2

MIT

A (°

C)

4 6 8 10 12 14-20

-10

0

10

20

30Air- cooler-3

Air mass flow (kg/h)

10 12 14 16 18 20-5

0

5

10

15

20

25

Air- cooler-4

Air mass flow (kg/h)

MIT

A (°

C)

MIT

A (°

C)

MIT

A ( °

C)

Fig. 10. Optimal air mass flow for air-coolers through ASPEN stream saturator.

58 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 59

save up to 13.6% compared to the base case. Table 6 lists the opti-mization results. Furthermore, each hot and cold composite curveof natural gas and the mixed refrigerant should be located as clo-sely as possible to achieve an energy efficient liquefaction processwith low specific compression energy. Fig. 8 presents the compos-ite curves (TDCC and THCC) of the SMR.

The TDCC plot reveals the impact of each individual refrigerantcomponent on the performance of the liquefaction process. Simi-larly, the THCC matching technique is normally used to addressthe energy efficiency of the process. The optimal flow rates of theMR ingredients were optimized to overpower the infeasibilityand reduce the driving force for phase change of natural gas from

Relative Humidity %0 20 40 60 80 100 120

UA(

kJ/C

.h)

15

20

25

30

35

40

45

50

At 32 C dry bulb Temp At 24 C dry bulb Temp At 18 C dry bulb Temp

Relative Humidity %0 20 40 60 80 100 120

UA(

kJ/C

.h)

10

20

30

40

50

60

At 32 C dry bulb Temp At 24 C dry bulb Temp At 18 C dry bulb Temp

Cooler-1

Cooler-3

Fig. 11. Effect of relative humid

Relative Humidity%0 10 20 30 40 50 60

Enth

alpy

diff

eren

ce (k

J/km

ole)

-3220

-3215

-3210

-3205

-3200

-3195

-3190

-3185

-3180(a)

Fig. 12. Effect of relative humidity on (a) the enthalpy

gas to liquid. In Fig. 8, the small gap between the TDCC curves andthe constraint temperature of 3 �C in the temperature interval from�160 �C to �70 �C indicate that enhancing the energy efficiency inthis region is challenging. Nevertheless, optimal flowrates of thelow boiling point components of MR, such as nitrogen, methaneand ethane could further improve the heat transfer performanceby reducing the gap in this region. Meanwhile, the region in thetemperature range from �50 �C to 40 �C (Fig. 8a) leads to a largeapproach temperature inside the main cryogenic heat exchanger.This large approach temperature can be efficiently reduced by pre-cooling the NG feed and MR streams before entering the mainexchanger to improve the efficiency of SMR process.

Relative Humidity %0 20 40 60 80 100 120

UA(

kJ/C

.h)

15

20

25

30

35

40

45

50

At 32 C dry bulb Temp At 24 C dry bulb Temp At 18 C dry bulb Temp

Relative Humidity %0 20 40 60 80 100 120

UA(

kJ/C

.h)

40

60

80

100

120

140

At 32 C dry bulb Temp At 24 C dry bulb Temp At 18 C dry bulb Temp

Cooler-2

Cooler-4

ity on the UA of air-coolers.

Relative Humidity%

0 20 40 60 80 100

Ove

rall

Coo

lers

dut

y (k

W)

0.61

0.62

0.63

0.64

0.65

0.66

0.67(b)

balance of air coolers and (b) the air coolers duty.

Fig. 14. Schematic diagram of the LNG cryogenic exchanger.

60 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

2.5. Addition of stream saturators

To introduce the air under different relative humidity, theASPEN HYSYS� built-in stream saturators (SAT) were used at theinlet of air-coolers in the optimized SMR model. Table 7 lists thesimulation conditions for the SAT. Fig. 9 presents the SMR processwith the addition of SAT. The sensitivity analysis was done to findthe optimal value of air mass flow by keeping the overall compres-sion power and the overall duty of air coolers constant. Fig. 10shows the result of optimal air mass flow for air-coolers throughASPEN stream saturator in terms of MITA value. The MITA valueof 22 �C was chosen as an optimal design criterion [20] for eachair cooler.

3. Effects of relative humidity on the overall liquefactionprocess

To examine the thermodynamic effects of the environmentalrelative humidity on the performance of the SMR LNG process,the optimized SMR process was linked with Excel VBA. The overallperformance was observed in terms of the following main impor-tant parameters of the process:

a. Overall UA of air-coolersb. Overall duty of air-coolersc. Overall compression powerd. LNG-Exchanger dutye. Overall heat transfer coefficient of the LNG-Exchangerf. LNG Production (Natural gas liquid fraction)

3.1. Overall UA of air-coolers

The overall heat transfer coefficient was reported to be affectedby the environmental RH [23]. The fouling resistance parameter ofthe UA depends very strongly on the environmental conditions[24]. RH is one of the environmental parameters that affect theUA of air-coolers largely. Fig. 11 shows the effect of RH on theUA of the inter-stage coolers of the SMR refrigeration cycle. Theanalysis was conducted at different dry bulb temperatures forthe installed four coolers. As a result, there is a linear decrease inthe relationship between the RH and UA of coolers.

3.2. Overall duty of air-coolers

The change in enthalpy is a driving force for heat transferthrough air-cooled heat exchangers [24]. On the other hand, the

RH% at interstage cooler inlet

0 20 40 60 80 100Enth

alpy

at C

ompr

esso

r inl

et (k

J/km

ole)

-85700

-85650

-85600

-85550

-85500

-85450

-85400

-85350(a)

Fig. 13. Effect of RH% (a) on the inlet compresso

steady-state cooling duty of air cooled heat exchanger is calculatedas:

Q ¼ ðnÞð�DHÞ ð24ÞThe overall duty for air-coolers can be calculated as:

Qoverall ¼ ðnÞ �fðDHÞ1 þ ðDHÞ2 þ ðDHÞ3 þ ðDHÞ4g� � ð25Þ

Fig. 12a shows the negative change in enthalpy, which has a lin-ear relationship with a RH of the inlet air. The negative signdenotes the heat removed by the cooling process. Fig. 12b presentsthe effect of relative humidity on overall air-coolers duty, whichshows the direct relationship between the negative decrease inenthalpy with the cooling duty.

3.3. Overall compressors power

The compressor power is strongly dependent on the gas orvapor inlet temperature [25]. As shown in Fig. 9, the heat contentor temperature of the mixed refrigerant stream is reduced by air-coolers before introducing the next consecutive compressors. Onthe other hand, the compression power is affected by the changein enthalpy (indirectly by the RH of the air inlet to air-coolers) ofthe inlet mixed-refrigerant. The compression process is endother-mic. Therefore, the change in enthalpy will be positive and thecompression power can be calculated as

Power ¼ ðnÞðDHÞ ð26Þ

Power ¼ ðnÞðH2 � H1Þ ð27Þ

Relative Humidity%

0 20 40 60 80 100

Com

pres

sion

Pow

er (k

W)

0.374

0.376

0.378

0.380

0.382

0.384

(b)

r enthalpy; (b) on the compression power.

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 61

Accordingly, if the negative value of H1 decreases with increas-ing relative humidity, the compression power increases withincreasing relative humidity of air at the inlet of the pre-cooler ofthe compressor. Fig. 13a shows the effects of the relative humidityon the enthalpy of mixed refrigerant before entering the firstcompressor (through the inlet air to the air-cooler). As a result,the overall effects of the RH on the overall compression power isalso a linear relationship, as shown in Fig. 13b. The higher percent-age of RH requires higher compression energy due to increasedirreversibilities (entropy generation by increased RH as seen inFig. 5d). The trend in Fig. 13b also indicates that the air-fincooler will be a more suitable candidate for the geographic locationof NG reserves where the relative humidity is low in most of theyear.

3.4. LNG-Exchanger duty

Fig. 14 illustrates a schematic diagram of a LNG cryogenicexchanger with the input and out streams. The steady-state heatbalance around the LNG cryogenic exchanger can be expressed as

ðnNGÞðHLNG � HNGÞ þ ðnSMRÞðH4 � H14Þ ¼ ðnSMRÞðH6 � H5Þ ð28ÞThis mathematical expression states that the heat gained by the

mixed refrigerant is related directly to the change in enthalpy ofthe mixed refrigerant. When the RH is varied, it ultimately affectsthe performance of the cryogenic exchanger in terms of the coolingduty, as shown in Fig. 15.

Relative Humidity%0 20 40 60 80 100

(HLN

G-H LN

G) k

J/km

ole

-15000

-14500

-14000

-13500

-13000

-12500

-12000

24600 24800 25000 25200 25400 25600 258000.760

0.765

0.770

0.775

0.780

0.785

0.790

0.795

0.800

(H6-H5) kJ/kmol

LNG

exc

hang

er d

uty

(kW

)

(a)

(c)

Fig. 15. Effect of relative humidity on the enthalpy balance arou

3.5. Overall heat transfer coefficient of LNG-Exchanger

The LNG cryogenic exchanger is considered as a backbone in thenatural gas liquefaction process. The overall heat transfer coeffi-cient, U, (of the LNG cryogenic exchanger) is also dependent onthe enthalpy [26] of the cold fluid (the compressed SMR) and thehot fluid (natural gas) for a constant heat transfer area. The relativehumidity is strongly affected by the enthalpy balance of air-coolers. Therefore, due to the variation of the RH, this change inthe enthalpy of the compressed SMR has an impact on the heattransfer coefficient of the LNG exchanger and the overall perfor-mance of the liquefaction process. On the other hand, to examinethe thermodynamic effects of the RH on the performance of theSMR process, the UA for the LNG cryogenic exchanger wasassumed to be a variable (due to U, i.e. affected by SMR enthalpy).

As shown in Fig. 16b, a polynomial regression demonstrates theeffects on the overall heat transfer coefficient of the LNG-Exchanger. This suggests that the enthalpy of the MR is varied bythe change in RH. This is the fourth-order polynomial effect, wherethe overall heat transfer coefficient (U) is calculated by the follow-ing mathematical relation (from the general energy balanceequation):

UACP

¼ QDHLMHD

ð29Þ

where DHLMHD is the log-mean difference in enthalpy, which isanalogous to the log-mean difference in temperature. Referred inFig. 14, DHLMHD can be expressed as

Relative Humidity%

0 20 40 60 80 100

LNG

exc

hang

er d

uty

(kW

)

0.760

0.765

0.770

0.775

0.780

0.785

0.790

0.795

0.800

Relative Humidity%

0 20 40 60 80 100

(H6-

H 5) k

J/km

ol

24600

24800

25000

25200

25400

25600

25800

(b)

(d)

nd the LNG cryogenic exchanger and LNG exchanger duty.

Relative Humidity%

0 20 40 60 80 100

UA/C

p

150

200

250

300

350

400

450

500

Relative Humidity%

0 20 40 60 80 100 120

(LM

HD) L

NG e

xcha

nger

2000

2500

3000

3500

4000

4500

5000

5500

(a) (b)

Fig. 16. Effect of relative humidity (a) on the LHMD; (b) on the UA of LNG cryogenic exchanger.

Relative Humidity%

0 20 40 60 80 100

(HLN

G–H

NG) k

J/km

ole

-15000

-14500

-14000

-13500

-13000

-12500

-12000

0 20 40 60 80 100-19100

-19090

-19080

-19070

-19060

-19050

-19040

-19030

-19020

Relative Humidity%

(H4–

H 14) k

J/km

ole

0 20 40 60 80 10024600

24800

25000

25200

25400

25600

25800

(H6–

H 5) k

J/km

ole

Relative Humidity%

0 20 40 60 80 1000.65

0.70

0.75

0.80

0.85

0.90

0.95

Relative Humidity%

LNG

pro

duct

ion

(kg/

h)(a) (b)

(c)(d)

Fig. 17. Effect of relative humidity on LNG production.

62 M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63

DHLMHD ¼ ðHNG � H6Þ � ðHLNG � H14Þln ðHNG�H6Þ

ðHLNG�H14Þð30Þ

where H14, is the accumulative enthalpy of a mixed refrigerantbefore entering the main cryogenic heat exchanger and is mostlyaffected by the variation in the RH. There is an inverse relationbetween the log-mean enthalpy difference and the overall heat

transfer coefficient. As shown in Fig. 16a, this relation is also afourth-order polynomial under various RH.

3.6. LNG production

To produce a high liquid fraction of the LNG, according to ther-modynamics, it should satisfy the given condition of a change inenthalpy within the LNG-Exchanger at their higher values.

M.A. Qyyum et al. / Applied Thermal Engineering 128 (2018) 51–63 63

ðDHÞ14 � ðDHÞNG � ðDHÞ5Because DH14 is affected by the relative humidity, it is impor-

tant to investigate that how DH14 affects the enthalpy of the otherstreams (the enthalpy of LNG and the enthalpy of low-pressureMR). Therefore, when the enthalpy of LNG stream changes, the liq-uid fraction of the natural gas (after the end-flash valve) is affected.Ultimately, LNG production decreases when BOG (boil-off gas) rateincreases. The study shows that there is a linear relationshipbetween the relative humidity and the change in enthalpy of amixed-refrigerant. The relative humidity also correlated linearlywith LNG production, as shown in Fig. 17.

4. Conclusion

This study analyzed the thermodynamic effects of varying therelative humidity on the SMR natural gas liquefaction process interms of the design and operational parameters of the process.The efficiency of the SMR process with respect to the overall dutyof the air-coolers, compressors, and LNG cryogenic exchanger, wasalso investigated. The results show that at the regions where therelative humidity is mostly high most of the year, air-fin coolersare not preferred over water-coolers due to the high compressionpower required.

According to the results, the outlet temperature of the air-coolers, flowrate of air, and operating pressures of the process needto be adjusted with respect to the wide variations of RH to main-tain high performance of the liquefaction process. The UA valueof the air-coolers decreases linearly and the UA value of the LNGcryogenic exchanger increases as a fourth order polynomial associ-ated with the increasing percentage of RH. For plants that are builtwith air-fin coolers, a constant UA value of the cryogenic exchangerand air-coolers restrict the improvement of the process efficiencyat low or high relative humidity. This work will help the processengineers to choose the appropriate cooling medium (air/water)for inter-stage cooling of mixed refrigerant compression unitsespecially for offshore applications. This study will also contributeto design the equipment of SMR-LNG plant for the location associ-ated with different humidity.

Acknowledgements

This study was supported by Basic Science Research Programthrough the National Research Foundation of Korea (NRF) fundedby the Ministry of Education (2015R1D1A3A01015621) and by Pri-ority Research Centers Program through the National ResearchFoundation of Korea (NRF) funded by the Ministry of Education(2014R1A6A1031189).

References

[1] H.M. Hudson, J. Wilkinson, K. Cuellar, M.C. Pierce, Integrated liquids recoverytechnology improves LNG production efficiency, in: 82nd Annual Conventionof the Gas Processors Association, 2003.

[2] A. Mortazavi, A. Alabdulkarem, Y. Hwang, R. Radermacher, Development of arobust refrigerant mixture for liquefaction of highly uncertain natural gascompositions, Energy 113 (2016) 1042–1050.

[3] M. Roberts, W. Kennington, A. Weist, Effect of climate on LNG technologyselection, in: World Petroleum Congress, 2014.

[4] BP, BP Statistical Review of World Energy, 2016.[5] ASPEN, Aspentech, ASPEN HYSYS V9., 2016.[6] A.J. Finn, G.L. Johnson, H.L. Tomlison, Developments in natural gas liquefaction,

Hydrocarbon Proces. 78 (4) (1999) 47–59.[7] ExxonMobil, Alaska LNG Project Selects Lead Terminal Location, ExxonMobil,

ExxonMobil News Releases, 2013, pp. 1–3.[8] Average Weather For Kenai, Alaska, USA, Vol. 2017.[9] Department of State Development, Vol. 2017, Government of Western

Australia.[10] Climate Data Online, Vol. 2016, Australian Government Bureau of

Meteorology.[11] Equatorial Guinea LNG Holdings Limited, Vol. 2016.[12] Malabo, Equatorial Guinea Climate Graphs, Vol. 2016.[13] C. Thomas, R. Burlingame direct seawater cooling in LNG liquefaction plants,

in: 15th International Conference & Exhibition on Liquefied Natural Gas (LNG15) Barcelona, Spain, 2007.

[14] K. Park, W. Won, D. Shin, Effects of varying the ambient temperature on theperformance of a single mixed refrigerant liquefaction process, J. Nat. Gas Sci.Eng. 34 (2016) 958–968.

[15] X. Xu, J. Liu, C. Jiang, L. Cao, The correlation between mixed refrigerantcomposition and ambient conditions in the PRICO LNG process, Appl. Energy102 (2013) 1127–1136.

[16] Factors Affecting the Air Temperature, Vol. 2017.[17] A. Bahadori, G. Zahedi, S. Zendehboudi, K. Hooman, Simple predictive tool to

estimate relative humidity using wet bulb depression and dry bulbtemperature, Appl. Therm. Eng. 50 (2013) 511–515.

[18] R. Stull, Wet-bulb temperature from relative humidity and air temperature, J.Appl. Meteorol. Climatol. 50 (2011) 2267–2269.

[19] M.S. Khan, S. Lee, G.P. Rangaiah, M. Lee, Knowledge based decision makingmethod for the selection of mixed refrigerant systems for energy efficient LNGprocesses, Appl. Energy 111 (2013) 1018–1031.

[20] A.K. Coker, Chapter 0 – RULES OF THUMB: SUMMARY A2 – Coker, in: A.Kayode, Ludwig’s Applied Process Design for Chemical and PetrochemicalPlants, fourth ed., Gulf Professional Publishing, Burlington, 2007, pp. xviii–xxvi.

[21] P. Venkataraman, Applied Optimization with MATLAB Programming, Wiley,2009.

[22] N.V.D. Long, L.Q. Minh, T.N. Pham, A. Bahadori, M. Lee, Novel retrofit designsusing a modified coordinate descent methodology for improving energyefficiency of natural gas liquid fractionation process, J. Nat. Gas Sci. Eng. 33(2016) 458–468.

[23] J. Zhang, A. Gupta, J. Baker, Effect of relative humidity on the prediction ofnatural convection heat transfer coefficients, Heat Transf. Eng. 28 (2007) 335–342.

[24] John R. Thome, Engineering Data Book III, in: Chapter 6, Heat Transfer to Air-Cooled Heat Exchangers, Woverine Tube, Inc., 2004.

[25] T.N. Pham, N.V.D. Long, S. Lee, M. Lee, Enhancement of single mixed refrigerantnatural gas liquefaction process through process knowledge inspiredoptimization and modification, Appl. Therm. Eng. 110 (2017) 1230–1239.

[26] J.S. Modelling of Multistream LNG Heat Exchangers, in: Department of Energyand Process Engineering, Vol. Master, Norwegian University of Science andTechnology, Norway, 2011.