You are on page 1of 7

Applied Thermal Engineering 26 (2006) 898–904

www.elsevier.com/locate/apthermeng

Parameter comparison of two small-scale natural gas


liquefaction processes in skid-mounted packages
Wen-sheng Cao *, Xue-sheng Lu, Wen-sheng Lin, An-zhong Gu
Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, Shanghai 200030, China

Received 20 September 2005; accepted 21 September 2005


Available online 2 November 2005

Abstract

Two typical types of small-scale natural gas liquefaction process in skid-mounted package were designed and simulated. The key
parameters of the two processes were compared, and the matching of the heating and cooling curves in heat exchangers was also
analyzed. The results show that the N2–CH4 expander cycle precedes the mixed-refrigerant cycle on the premise of lacking propane
pre-cooling. Large temperature difference and heat exchange load are the primary reasons of exergy loss in heat exchangers. The
power consumption of compressors is influential to power consumption per unit LNG, so compression with intercooling should
be adopted.
Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: Skid-mounted package; Natural gas; Liquefaction process; Simulation; Parameter comparison

1. Introduction opment of LNG vehicles creates the opportunity of


developing small-scale natural gas liquefaction plants.
Cryogenic liquefaction plants have been applied in Comparing with medium-sized or large-scale liquefac-
the commercial natural gas liquefaction fields which liq- tion plant, the key characteristics of small-scale one
uefaction capacities are very large. For instance, the are simple process, low investment, miniature size and
capacity of single product line of base load LNG plant skid-mounted package [1–5].
is up to 3.4 Mt/a, and the one of the peak shaving The design, simulation and estimation of natural gas
LNG plant is about 0.9 Mt/a. Cryogenic liquefiers are liquefaction process began from 1970. Shell Corporation
commercially available for natural gas liquefaction. has simulated the cascade, mixed-refrigerant and N2
These liquefiers are normally custom-made, permanent expander cycle, moreover analyzed their advantages
large capacity plants for natural gas utility peak shaving and disadvantages [6]. Melaaen set up a dynamic model
and transcontinental natural gas shipping. The develop- for the natural gas liquefaction process of base load
ing market of natural gas vehicles provides an opportu- plant, and carried through the simulation by DASSL
nity for LNG and LNG-CNG fueling stations. Other in 1995 [6]. Terry adopted HYSYS software to calculate
markets for smaller-scale LNG liquefiers include and optimize the typical liquefaction process of peak
onshore gas wells, customer sites that are remotely situ- shaving plant in 1998 [7]. Kikkawa designed the late-
ated from current gas pipelines, and industrial customer model processes of pre-cooling mixed-refrigerant cycle
peak shaving installations. Continued commercial devel- and expander cycle, and calculated them by CHEM-
CAD software in 1997 [8]. Prof. Gu An-zhongÕs research
group of Shanghai Jiao Tong University has carried
*
Corresponding author. Tel./fax: +86 21 629 32602. through the simulation and calculation of natural gas
E-mail address: wscao@sjtu.edu.cn (W.-s. Cao). liquefaction process [6]. The previous work does not

1359-4311/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2005.09.014
W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904 899

concern the small-scale natural gas liquefaction process refrigeration by compression and work expansion of
in skid-mounted package. But it will be helpful to the gas stream.
study presented in this paper. Almost all kinds of lique-
faction process can be adopted in small-scale LNG 2.2. Property methods
plants, so it is not always easy to select the most suitable
process for a certain project. A deep investigation into Property methods are the base of simulation. PR
these processes is needed before any choices are made. equation and LKP equation are the keys of fluid pack-
Based on the thermodynamic analysis, the simulation age in simulation.
and calculation of small-scale natural gas liquefaction The Peng–Robinson equation of state applies func-
process in skid-mounted package were carried through. tionality to some specific component–component inter-
The key parameters of the two sets of liquefaction pro- action parameters, which can be used in the
cess were compared, and the matching of the heating calculation of phase equilibrium [11]. It is written by
and cooling curves in heat exchangers was also
RT a
analyzed. P¼ 
V  b V ðV þ bÞ þ bðV  bÞ
X
N X
N

2. Design a¼ xi xj ðai aj Þ0:5 ð1  k ij Þ ð1Þ


i¼1 j¼1

Two similar processes were designed for comparison. X


N
b¼ x i bi
2.1. Liquefaction processes i¼1

where P is pressure, R is gas constant, T is temperature,


On the basis of the characteristics of small-scale LNG V is specific volume, a and b are the constants relating to
plant, two sets of typical liquefaction process were the gas species, x is mole fraction of a certain compo-
designed in this paper for analysis and comparison. nent, k is binary interaction coefficient.
The liquefaction process of mixed-refrigerant cycle It is rewritten as
(MRC) was adopted in Process 1 shown in Fig. 1, which
removed the common cycle of propane pre-cooling,
Z 3  ð1  BÞZ 2 þ ðA  2B  3B2 ÞZ  ðAB  B2  B3 Þ ¼ 0
making the process simpler and more compact. The
MRC uses a combination of refrigerants in a single aP
refrigeration cycle, which makes it possible to supply A¼
ðRT Þ2
refrigeration at continuously changing temperature.
The liquefaction process of N2–CH4 expander cycle bP

was adopted in Process 2 shown in Fig. 2. It provides RT
ð2Þ

Fig. 1. Process 1: Mixed-refrigerant cycle liquefaction process (MRC).


900 W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904

Fig. 2. Process 2: N2–CH4 expander cycle liquefaction process.

where Z is a constringent factor, A and B are the coeffi- the parameters on the process performance. These fig-
cients relating to the gas state parameters. ures were gained through the optimization of the objec-
The Lee–Kesler–Plocker equation is an accurate gen- tive function on specific power consumption [9,10].
eral method for non-polar substances and mixtures, Assumed the flow rate of natural gas feed was
which can be used in the calculation of enthalpy and 4.0 k mol/h.
entropy of mixed components [11]. It is given by The simulating calculation and optimization on Pro-
x cesses 1 and 2 were done using PR equation of state and
Z ¼ Z ð0Þ þ ðrÞ ðZ ðrÞ  Z ð0Þ Þ ð3Þ
x LKP equation through HYSYS software. PR or LKP
where x is an acentric factor, 0 and r denote the relevant equation is the one of the most important Fluid Pack-
parameters of simple and reference liquids. ages that is the base of the simulation by HYSYS. Many
factors influence the performance of a certain process.
2.3. Parameter optimization For instance, they are high pressure and low pressure
of the mixed-refrigerant, temperature of the refrigerant
Before the simulation, the required parameters were before expansion, and mole fraction of nitrogen, meth-
specified in Tables 1–3 based on the effect analysis of ane, ethane, propane, etc. These factors are described

Table 1
Mole fraction of components
N2 CH4 C2H6 C3H8 iC4H10 nC4H10
Natural gas 0.7 82.0 11.2 4.0 1.2 0.9
Mixed-refrigerant 1.0 40.0 40.0 19.0
N2–CH4 56.0 44.0

Table 2
Specified pressure and temperature in Process 1
Natural gas: Node 14 High-pressure refrigerant: Node 4 Low-pressure refrigerant: Node 13 LNG store: Node 19
Pressure (MPa) 5.00 2.60 0.29 0.20
Temperature (°C) 32.0 32.0 29.0 151.3

Table 3
Specified pressure and temperature in Process 2
Natural gas: Node 14 High-pressure refrigerant: Node 4 Low-pressure refrigerant: Node 13 LNG store: Node 19
Pressure (MPa) 5.00 4.40 0.60 0.20
Temperature (°C) 32.0 32.0 29.0 150.7
W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904 901

as X = [x1, x2, x3, . . . , xN]T. In this case, the optimization From the data analysis of above Table 4, it can be seen
problem is finding out the optimum parameter values to that the key parameters of the N2–CH4 expander cycle
make the power consumption lowest. The objective defeat the ones of the mixed-refrigerant cycle except liq-
function is uefaction rate, thanks to the latter lacking the propane
min f ðX Þ ¼ W =F LNG ð4Þ pre-cooling cycle for meeting the requirement of the nat-
ural gas liquefier in skid-mounted package.
where W is the total power consumption of compres- The mixed-refrigerant cycle with propane pre-cooling
sors, FLNG is flow rate of LNG. In fact, it is necessary (C3/MRC) is the same with large-scale LNG plant of
to take all the factors such as initial cost, power con- base load, but not fits small-scale one in skid-mounted
sumption, plant area, simplicity of the process, etc., into package. There are few advantages left for the mixed-
consideration. But many of these factors are not pure refrigerant cycle running short of the propane pre-cool-
technical ones. In this paper, only specific power con- ing, competing with the N2–CH4 expander cycle. Differ-
sumption (power consumption per unit LNG) is taken ing from the C3/MRC which are complex process and
as the optimization aim. The constraints are as below: low-power consumption, the N2–CH4 expander cycle is
(A) Sum of the mole fractions of mixed-refrigerant is seasoned with mobile LNG plant of medium-sized or
1. (B) The temperature of mixed-refrigerant at the inlet large-scale style, due to its compact process, flexible
of compressor is higher than its dew point. (C) The tem- operation and extensive applicability. It should be con-
perature difference between the hot and cold areas of sidered integrating the technical characteristics of above
fluid cannot be negative. both to develop new type of liquefaction process in skid-
HYSYS contains a multi-variable steady state Opti- mounted package, which caters to the development tide
mizer. Once the flowsheet has been built and a con- of natural gas liquefaction process currently pursuing a
verged solution has been obtained, the Optimizer can brief and efficient one.
be used to find the operating conditions which minimize
(or maximize) the objective function. The Optimizer 3.2. Heat exchange load and temperature distribution
owns its own spreadsheet for defining the objective func-
tion, as well as any constraint expressions to be used. The main energy loss of a natural gas liquefaction
HYSYS has five modes of Optimizer: Original, Hypro- process exists in compressors and heat exchangers, but
tech SQP, MDC Optim, MDC DataRecon and Selec- not neglecting the loss in throttles.
tion Optimization. Any mode of Optimizer can be Compression with intercooling was used in this case,
used in this case, but the Original mode was selected in order to decrease the irreversible degree and the power
for its common optimization in HYSYS. The following consumption of compression process. The reclaimed
sections describe the optimization schemes for the Origi- power of the expander in Process 2 was used to drive
nal Optimizer: Function Setup, BOX Method, SQP the retropack, which reduced the power consumption
Method, Mixed Method, Fletcher Reeves Method and of compressors. In this paper, the matching of the heat-
Quasi-Newton Method, which are the types of built-in ing and cooling curves between the feed gas and the
algorithm. mixed-refrigerant in heat exchangers was analyzed. The
temperature difference and heat exchange load in heat
transfer contribute to the exergy loss, so large tempera-
3. Results and analysis of simulation ture difference and heat exchange load are the primary
reasons of the exergy loss in heat exchangers. Sure the
The simulation and calculation of the two processes heat flow is not the only one affected by the changing
were done. temperature, the enthalpy and pressure also are the ones.
The matching of the heating and cooling curves in heat
3.1. Results of simulation exchangers in Processes 1 and 2 was analyzed. It can be
seen from the results shown in Figs. 3–6.
The key parameters of the two liquefaction processes It can be found from Figs. 3–6 that, the heat
were compared, and the comparison of the optimization exchange load of the pipelines in the heat exchangers
results is presented in Table 4. are climbing along with the ascending temperature.

Table 4
Comparison of the optimization results of the two liquefaction processes
Liquefaction Flow rate of Flow rate of Load of water-cooling (kW) Power consumption of Power of Liquefaction Power per
process natural gas refrigerant compressors (kW) expander rate unit LNG
(k mol/h) (k mol/h) (kW) (kW/mol/s)
Process 1 4.000 60.25 145.95 129.23 0.951 122.3
Process 2 4.000 50.00 109.47 105.30 12.03 0.905 93.2
902 W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904

Fig. 3. Heat exchange load variation as a function of temperature of the pipelines in heat exchanger 1 of Process 1.

Fig. 4. Heat exchange load variation as a function of temperature of the pipelines in heat exchanger 2 of Process 1.

Generally, under the condition of the same temperature, The largest temperature difference between the high-
the heat exchange load of the natural gas pipelines (14– pressure and low-pressure refrigerant pipelines in heat
15, 15–16) is lowest, next is the load of the high-pressure exchanger 1 and 2 of Process 1 is about 45 °C, in the
refrigerant pipelines (Process 1: 4–7; Process 2: 4–5, 8– condition of lacking propane pre-cooling. The tempera-
9), and the load of the low-pressure refrigerant pipelines ture difference between the same ones in heat exchanger
(Process 1: 8–9, 12–13; Process 2: 10–13) is highest. This 1 of Process 2 is comparatively even, which is about
is because that the low-pressure refrigerant pipelines are 30 °C. But the temperature difference in heat exchanger
charged with cooling capacities to the natural gas pipe- 2 of Process 2 is about 35 °C. It will cause more exergy
lines and the high-pressure refrigerant pipelines. More- loss in heat exchangers if the temperature difference and
over, the heat exchange load of exchanger 1 is more heat exchange load are not limited. So it is necessary to
than that of exchanger 2. The emergence of the inflex- take some measures of strengthening heat transfer, such
ions on the curves indicates the phase change from as increasing the heat transfer areas and reducing the
two-phase to single-phase and vice versa. temperature difference, which may cut down the exergy
W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904 903

Fig. 5. Heat exchange load variation as a function of temperature of the pipelines in heat exchanger 1 of Process 2.

Fig. 6. Heat exchange load variation as a function of temperature of the pipelines in heat exchanger 2 of Process 2.

loss in heat exchangers. Although the temperature differ- very important to the mixed-refrigerant cycle. It
ence between the natural gas and low-pressure refriger- should be considered for energy saving in the
ant pipelines is very large, the exergy loss of the development of small-scale natural gas liquefac-
natural gas pipelines is still small owing to their low-heat tion process in skid-mounted package.
exchange load. (2) It should be paid more attention to the high-tem-
perature sections of heat exchanger 1 in the design
of heat exchanger, since the heat exchange load of
4. Conclusions exchanger 1 is more than that of exchanger 2. The
exergy loss of the natural gas pipelines is small
(1) It is found that the N2–CH4 expander cycle pre- because of their low-heat exchange load.
cedes the mixed-refrigerant cycle on the premise (3) The temperature difference and the heat exchange
of lacking propane pre-cooling from the parameter load between the pipelines in exchangers are large.
comparison. The role of propane pre-cooling is They are the key factors that cause the exergy loss,
904 W.-s. Cao et al. / Applied Thermal Engineering 26 (2006) 898–904

so the measures of strengthening heat transfer [2] A.J. Finn, G.L. Johnson, T.R. Tomlinson, Development in
should be carried through. natural gas liquefaction, Hydrocarbon Processing (4) (1999) 47–
59.
(4) The power consumption of compressors is more [3] G.L. Johnson, A.J. Finn, T.R. Tomlinson, Offshore and smaller-
influential to specific power consumption, so com- scale liquefiers, LNG Journal (July–August) (1999) 19–22.
pression with intercooling should be adopted to [4] L.A. Wenzel, LNG Peakshaving Plants—a comparison of cycles,
lessen the irreversible degree of compression pro- Advances in Cryogenic Engineering (20) (1973) 90–102.
cess. The reclaimed power of the expander in Pro- [5] K.J. Vink, Comparison of baseload liquefaction process, in:
Twelfth International Conference and Exhibition on Liquefied
cess 2 was used to drive the retropack for reducing Natural Gas, Perth, Australia, 1998, 3.6.
the power consumption of compressors. [6] A. Gu, X. Lu, R. Wang, Y. Shi, W. Lin, Liquefied Natural Gas
Technology, China Machine Press, 2004.
[7] L. Terry, Comparison of liquefaction process, LNG Journal 21 (3)
Acknowledgements (1998) 28–33.
[8] K. Yoshitugi, N. Moritaka, Development of liquefaction process
for natural gas, Journal of Chemical Engineering of Japan 30 (4)
We gratefully acknowledge the funding support of (1997) 626–630.
Natural Science Foundation Project of Fujian Province [9] Y. Shi, The thermodynamic research of the mixed refrigerant cycle
in China (Item no.: E0440002). to liquefied natural gas, Doctoral dissertation, Shanghai Jiao
Tong University, 1998.
[10] G. Zhu, Study on transport properties of natural gas and
References optimization on liquefaction processes of LNG peakshaving
plants, Doctoral dissertation, Shanghai Jiao Tong University,
[1] F. Wolfgang, B. Wilfried, S. Rudolf, et al., A new LNG baseload 2000.
and the manufacturing of the main heat exchangers, in: Twelfth [11] C.R. Robert, The Properties of Gases and Liquids, fourth ed.,
International Conference and Exhibition on Liquefied Natural McGraw-Hill Book Company, 1987.
Gas, Perth, Australia, 1998, 2.6.

You might also like