Energy 97 (2016) 350e358

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Dynamic simulation of mixed refrigerant process for small-scale LNG

plant in skid mount packages
Tianbiao He, Yonglin Ju*
Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, Shanghai 200240, China

a r t i c l e i n f o a b s t r a c t

Article history: The utilization of small natural gas reservoirs has attracted more and more attention. Most of the
Received 3 August 2015 optimization studies concentrated on mixed refrigerant liquefaction process at steady-state simulation.
Received in revised form Only a few studies have reported dynamic simulation of mixed refrigerant LNG (liquefied natural gas)
10 October 2015
process. The main purpose of this study was to design a dynamic model of mixed refrigerant LNG process
Accepted 1 January 2016
for small-scale LNG plant in skid mount packages and to investigate the dynamic behaviors of this
Available online 25 January 2016
process. The variations of natural gas temperature, composition, pressure, flow rate and water-cooler
temperature were selected as disturbances to test the stability and dynamic responses of the process.
Keywords:
Dynamic simulation
The dynamic responses of LNG temperature and total compressor energy consumption were the criteria
LNG to investigate the influences of disturbances on the process. Finally, the dynamic responses of distur-
Mixed refrigerant bances were obtained and discussed. The simulation results showed that LNG temperature could go back
Liquefaction process to its set point value after a few minutes, while compressor duty varied with the disturbances. It indi-
Small-scale cated that the mixed refrigerant liquefaction process for small-scale LNG plant in skid mount packages
could overcome some disturbances and operate at stable state.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction research interest regarding the mixed refrigerant liquefaction

With the increasing demand of energy, natural gas has remained
the fastest increasing fossil energy resource in the world recently.
Natural gas is a clean fossil fuel which is widely used and can
reduce the greenhouse emissions. Some agencies have estimated
that the consumption of natural gas in the world will increase by
1.7% per year [1].
Since the gap between natural gas consumption and production
is very large, more and more attentions have been shifted to some
isolated small natural gas reservoirs that were previously consid-
ered to be too remote and costly to develop. Due to the small flow
rate of natural gas, it is a good option to build small-scale LNG
(liquefied natural gas) plant to liquefy the natural gas. Natural gas
can be liquefied when its temperature is below
161  C at
101.325 kPa and its volume reduces by a factor of more than 600 [2].
Mixed refrigerant liquefaction process is considered as a suitable
process for small-scale LNG plant. Many researchers have studied
mixed refrigerant liquefaction process during the past decade. The
* Corresponding author. Tel./fax: þ86 21 34206532.
E-mail address: yju@sjtu.edu.cn (Y. Ju).
process is to obtain the optimal design. Lee et al. [3] developed a
novel method to select mixed refrigerant compositions based on
combination of NLP method and thermodynamic approach. Wang
et al. [4] proposed a thermodynamic-analysis-based study of the
minimization of the energy consumption of C3MR (mixed refrig-
erant process with propane precooling) liquefaction process. Morin
et al. [5] developed a method to optimize the energy consumption
of PRICO liquefaction process by using the evolutionary search. The
results showed that this method had a good match with the liter-
ature results. Khan and Lee [6] applied particle swarm paradigm
algorithm to optimize single mixed refrigerant natural gas lique-
faction process. The optimization results showed that the
compression energy consumption can be reduced by 10% compared
with the base case. They also compared the results with other ap-
proaches in literature. The results showed that PSP could reach
nearly global optimum. Hwang et al. [7] obtained the optimal
operation condition of dual mixed refrigerant (DMR) natural gas
liquefaction process which was applied to LNG-FPSO. Hatcher et al.
[8] proposed a systematic analysis of optimization formulation for
mixed-refrigerant LNG process. They focused on the construction
and testing of eight objective functions with an aim to identify the
most appropriate formulation. Alabdulkarem et al. [9] adopted

http://dx.doi.org/10.1016/j.energy.2016.01.001
0360-5442/© 2016 Elsevier Ltd. All rights reserved.
 T. He, Y. Ju / Energy 97 (2016) 350e358
genetic algorithm to obtain the global optimum design parameters
of C3MR LNG plant. Mortazavi et al. [10] conducted a detailed study
on the energy performance enhancement of C3MR LNG plant by
utilizing expanders to replace throttling valves.
Most optimization studies concentrated on designing mixed
refrigerant liquefaction process with lower energy consumption at
steady state simulation. Only a few studies have addressed dynamic
simulation of natural gas liquefaction process or main equipment in
liquefaction process. Rodriguez and Diaz [11] made a rigorous dy-
namic modeling of countercurrent heat exchangers in large-scale
natural gas processing plant. Singh and Hovd [12] developed a
dynamic model and designing a control structure for SINTEF LNG
plant. Mandler [13] discussed dynamic simulation of cryogenic
separation and liquefaction process. Husnil and Lee [14] proposed
an optimizing control structure for dual mixed refrigerant process
to maintain the DMR process in the optimal compressor duty. Lim
et al. [15] recommended to use dynamic simulation for effective
control design.
Small-scale LNG plant in skid mounted packages was always
built at small gas reservoir which means the LNG plant can be
moved to other gas reservoirs when the old gas reservoir was
exhausted. As a result, the liquefaction process for small-scale LNG
plant in skid mounted packages should be simple, and at the same
time it can handle different kinds of disturbances without changing
any equipment.
The mixed refrigerant liquefaction process in this study was
based on the process proposed by He and Ju [16]. The main aim of
this study was to develop a dynamic model of mixed refrigerant
liquefaction process for small-scale LNG plant and to investigate the
dynamic behaviors. The dynamic simulation results were
compared with steady-state simulation results to verify the accu-
racy of dynamic model. A control structure was designed to
maintain the stability of the process. Then, several different kinds of
disturbances were added to the process to test its flexibility and
dynamic behaviors. The dynamic response of LNG temperature and
total energy consumption were the criteria to investigate the in-
fluences of disturbances on the process.
2. Process design and methodology
2.1. Process design
Mixed refrigerant liquefaction process for small-scale LNG plant
is illustrated in Fig. 1. The compositions and other parameters of
feed gas are listed in Table 1. Natural gas is cooled in HEX-1 to
approximately
60  C firstly and some heavy components are
condensed. Then it enters into vaporeliquid separator (S-1) to
separate heavy hydrocarbons. The vapor natural gas undergoes
HEX-2 and it is cooled to
159  C. The natural gas is liquefied at this
node (NG-6) before it enters the throttling valve.
The mixed refrigerant enters vaporeliquid separator (S-2) to
separate liquid mixed refrigerant (LMR) and then vapor mixed
refrigerant (VMR) is pressurized by the first-stage compressor (C-
1). After that, it is cooled in water cooler (E
101) to 40  C. Then
mixed refrigerant (M8) enters S-3 to avoid liquid entering the
second-stage compressor. After that, it is pressurized by second-
stage compressor (C-2) and cooled in water cooler (E
102). Some
components of mixed refrigerant are condensed at this point and
enter vaporeliquid separator (S-4) to separate vapor mixed
refrigerant (VMR) and liquid mixed refrigerant (LMR). Liquid mixed
refrigerant (M16) is cooled in HEX-1 and then its pressure is
reduced in throttling valve (VLV-3). Vapor mixed refrigerant (M17)
is cooled in HEX-1 and HEX-2 subsequently. Cold mixed refrigerant
reduces its pressure in VLV-2 and then is applied to cool the natural
gas and mixed refrigerant in HEX-2. The vapor mixed refrigerant
351
outlet from HEX-2 (M22) mixes with liquid mixed refrigerant
(M23) in mixer (MIX-1) and then cools the natural gas and mixed
refrigerant in HEX-1. In this mixed refrigerant cycle, two recycles
are designed to avoid the pressures of S-3 and S-4 being higher than
their set point value. The compositions of mixed refrigerant are
shown in Table 2.
2.2. Methodology
The methodology in this study was illustrated in Fig. 2. Firstly,
the steady-state simulation was built. Then, the dynamic simula-
tion which was based on the results of steady-state simulation was
conducted. The results of steady-state simulation and dynamic
simulation needed to be compared to keep the consistency. If the
results were consistent, the degrees of freedom analysis was
applied to the dynamic system. If the results were not consistent,
the dynamic simulation would be adjusted to achieve the consis-
tent. When degrees of freedom analysis was finished, several con-
trol structures were proposed to handle the disturbances. One of
the control structures was selected as the best control structure for
the system. Finally, different kinds of disturbances were added to
the process to investigate behaviors of the process.
In this study, the process simulator ASPEN HYSYS was used for a
rigorous steady-state and dynamic simulation of the mixed
refrigerant process. PengeRobinson equation of state was used to
calculate the vaporeliquid equilibrium.
2.3. Dynamic models of the main equipment
The main equipment in mixed refrigerant liquefaction process
are compressor, multistream heat exchanger and valve. The dy-
namic models of the main equipment are described in the following
sections.
2.3.1. Compressor
This section will present the dynamic model used to describe
the compressor. This model describes the relation between
compressor duty and mass flow rate. The polytrophic efficiency of
compressor is assumed as constant which indicates that the poly-
trophic efficiency does not vary with mass flow rate of mixed
refrigerant. The energy balance of compressor can be shown as
follow:
"
#
n
1
Pin Vin Pout n
Wc ¼ fin ðhout
hin Þ ¼ fin

1 (1)
n
1 Pin
hp n
Polytrophic efficiency



k
1 n
1
hp ¼ (2)
k n
Cp
k¼ (3)
Cv
where Wc is the compressor duty; h is molar enthalpy; f is molar
flow rate; P is pressure; n is polytrophic index; hp is polytrophic
efficiency; k is specific heat ratio; V is molar volume; Cp is specific
heat at constant pressure; Cv is specific heat at constant volume.
2.3.2. Multistream heat exchanger
Multistream heat exchanger is the main heat transfer equip-
ment in LNG process. Plate-fin heat exchanger is usually used as
multistream heat exchanger in small-scale LNG plant. Plate-fin heat
exchanger has several zones and each zone has several layers. Thus,
352 T. He, Y. Ju / Energy 97 (2016) 350e358

Fig. 1. The schematic flow diagram of mixed refrigerant liquefaction process.

Table 1 Table 2
The mole fraction and other parameters of feed gas. The compositions of mixed refrigerant.

Parameters Value Component Mole fraction

The mole fraction components CH4 0.9459 CH4 0.1563

C2H6 0.0343 C3H8 0.0401
C3H8 0.0059 i-C4H10 0.3407
i-C4H10 0.0010 N2 0.0882
n-C4H10 0.0012 C2H4 0.3747
i-C5H12 0.0003
n-C5H12 0.0003
C6H14 0.0005
N2 0.0106 Cold stream
Total 1
Natural gas temperature 40  C dðVhout Þ
Natural gas pressure 4.0 MPa fi  Mi ðhin
hout Þ þ Qsl ¼ ri  Mi (5)
dt
Natural gas flow rate 5  104 Nm3/d (92.95 kmol/h)
Temperature after water cooler 40  C
where

Qsl ¼ Ui Ai ðTh
Tc Þ (6)

each layer is the heat transfer element in plate-fin heat exchanger.
The energy balance of each layer is as follow: Energy balance for the wall
Hot stream
dTw k
Mw  Cpw  ¼  AðTw1
Tw2 Þ (7)
dðVhout Þ dt x
fi  Mi ðhin
hout Þ
Qsl ¼ ri  Mi (4)
dt where Mi is the molecular weight of stream; Qsl is the heat transfer
between hot and cold stream; r is the density of stream; V is the
 T. He, Y. Ju / Energy 97 (2016) 350e358 353

freedom in mixed refrigerant process (Nss) can be obtained by the

following equation [17,18].

Nss ¼ NMV
N0 (10)

where NMV is the dynamic manipulated variables; N0 is the number

of degrees of freedom with no steady-state effect.
Thus, the number of manipulated variables of mixed refrigerant
process in this study is 12 (NMV ¼ 12), the details are shown as follow:

1) One natural gas feed valve (VLV-1)

2) Two throttling valves in mixed refrigerant process (VLV-2, VLV-3)
3) Two heat duties of water coolers (Q-1, Q-2)
4) One compressor speed
5) Two antisurge valves (VLV-7, VLV-8)
6) Three liquid holdup valves (VLV-4, VLV-5, VLV-6)

The liquid holdups have no steady state effect, so the number of

degrees of freedom with no steady state effect is 3 (N0 ¼ 3). Thus,
the number of potential degrees of freedom is 9 (Nss ¼ 9).
The operational variables which could be used in control
structure are among the potential degrees of freedom. To achieve
stable operation of mixed refrigerant process, the discharge tem-
peratures of compressors (stream M5 and M12) should be
controlled at 40  C. Therefore, two heat duties of water coolers (Q1
and Q2) cannot be used as operational variables. Moreover, the
suction pressure of compressor (stream M1) should be controlled.
Thus, the compressor speed should not be used as operational
variables. The antisurge valves (VLV-7 and VLV-8) are always closed
at the normal operation, so antisurge valves also cannot be used as
operational variables. Therefore, the number of actual operational
variables is 3, namely natural gas feed valve (VLV-1) opening per-
centage and two throttling valves opening percentages in mixed
refrigerant process (VLV-2, VLV-3).

3.2. Validation of dynamic simulation results

Fig. 2. The methodology of the dynamic simulation of mixed refrigerant liquefaction
process.
volume holdup of layer; t is time; U is heat transfer coefficient; A is
heat transfer area; Th is temperature of hot stream; Tc is tempera-
ture of cold stream; Mw is the mass of wall; Cpw is heat capacity of
wall; k is thermal conductivity of wall; x is wall thickness; Tw1 and
Tw2 are temperature of two sides of the wall.
2.3.3. Valve
Valve is assumed as an isenthalpic process. This model has an
energy balance equation and an equation relating mass flow rate to
pressure drop across valve. The equations are described as follow:
Energy balance
hin ¼ hout
In this study, dynamic simulation results are verified by
comparing with steady-state simulation results. The comparison of
dynamic simulation results and steady-state simulation results is
shown in Table 3. It is clear from the Table 3 that most of the var-
iables in dynamic simulation is within 0e5% deviation of steady-
state simulation. The comparison indicates that the dynamic
simulation results have a good match with the steady-state simu-
lation results. Therefore, the dynamic simulation can be used to
predict the dynamic behaviors of this process.
3.3. Control structure
The main control objective of small-scale LNG plant is to pro-
(8) duce LNG product with specified LNG temperature in a stable and

Mass flow rate equation

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Table 3
fi  Mi ¼ Cv ri  DP (9) The comparison of dynamic simulation results and steady-state simulation results.

Steady-state Dynamic Deviation

where Cv is the fluid flow rate value of the valve; DP is the pressure
drop across valve. Feed gas flow rate (kmol/h) 92.95 92.95 0%
Feed gas temperature ( C) 40.00 40.00 0%
Feed gas pressure (kPa) 4000.00 4000.00 0%
3. Dynamic simulation description
LNG temperature ( C)
159.00
159.00 0%
NG-3 temperature ( C)
61.80
61.86 0.097%
3.1. Degrees of freedom MR flow rate (kmol/h) 265.00 255.00 3.774%
MR compressor suction pressure (kPa) 230.00 235.00 2.128%
To find the degrees of freedom of the process is an important MR compressor discharge pressure (kPa) 3300 3322 0.667%
Compressor duty (kW) 672.80 661.6 1.665%
issue for control structure. The number of actual degrees of
354 T. He, Y. Ju / Energy 97 (2016) 350e358
low compressor duty state. The control structure also needs to have
a big tolerance to different kinds of disturbances. In this study, feed
gas flow rate is controlled by adjusting the opening percentage of
VLV-1. The feed gas flow rate is controlled at 92.95 kmol/h when
other disturbances are added to the process. However, the feed gas
flow rate is also selected as a disturbance to investigate the dy-
namic response of process. LNG temperature is controlled by
adjusting the vapor mixed refrigerant (VMR) flow rate. The VMR
flow rate increases by increasing VLV-2 opening when LNG tem-
perature is higher than the specified value. Conversely, the VMR
flow rate decreases by decreasing VLV-2 opening when LNG tem-
perature is lower than the specified value. Moreover, heavy hy-
drocarbon in natural gas should be separated at proper
temperature to avoid blocking the channel of heat exchanger at
cryogenic temperature. Thus, NG-3 temperature should be
controlled at approximately
61.8  C to separate heavy hydrocar-
bon. NG-3 temperature is controlled by adjusting the liquid mixed
refrigerant (LMR) flow rate. The LMR flow rate increases by
increasing VLV-3 opening when NG-3 temperature is higher than
the specified value, while LMR flow rate decreases by decreasing
VLV-3 opening when NG-3 temperature is lower than the specified
value.
Tuning parameters of flow, pressure, temperature and liquid
level controllers are recommended by Luyben [19]. The tuning
parameters of controllers in this study are based on Luyben's
recommendation and tuned to keep the process stable and robust.
4. Dynamic responses of disturbances
Small-scale LNG plant in skid mount packages is always built at
small gas reservoirs, which meaning that the LNG plant can be
moved to other gas reservoirs when the old gas reservoir is
exhausted. Thus, liquefaction process for small-scale LNG plant
needs to handle different kinds of disturbances without changing
any equipment. This section will introduce several disturbances
into the dynamic simulation to investigate the dynamic response.
The main disturbances are variations of feed gas temperature,
pressure, flow rate and composition. The variations of water cooler
temperature and mixed refrigerant compressor suction pressure
are also considered as disturbances in this study. LNG temperature,
NG-3 temperature and compressor duty are three important pa-
rameters which need to be observed when the disturbances are
added into the system.
4.1. Disturbance of feed gas temperature
The feed gas temperature affects the cooling duty during the
natural gas liquefaction directly. The feed gas temperature is varied
by 1, 2 and 3  C downward and upward, respectively. Dynamic
responses of controlled variable after the decrease and increase of
feed gas temperature are illustrated in Figs. 3 and 4.
Fig. 3 presents the dynamic responses of LNG temperature, NG-3
temperature and compressor duty after the feed gas temperature
decreases 1, 2 and 3  C, respectively. LNG temperature and NG-3
temperature decrease at first and then increase to a higher tem-
perature. They can reach to the specified values after 400 min.
Compressor duty decreases when the feed gas temperature de-
creases. It is because that the sensible and latent heat of natural gas
decreases when the feed gas temperature decreases. When the
mixed refrigerant flow rate keeps constant at first, LNG tempera-
ture and NG-3 temperature decrease. Then the control system re-
sponds to the disturbance, the mixed refrigerant flow rate valves
turning down and LNG temperature and NG-3 temperature
increasing. Thus, compressor duty decreases due to the decrease of
mixed refrigerant flow rate. At last, the system can go back to stable
state after a while.
Fig. 4 illustrates the dynamic responses of LNG temperature, NG-
3 temperature and compressor duty after the feed gas temperature
increases 1, 2 and 3  C, respectively. LNG temperature and NG-3
temperature increase rapidly and then decrease when the feed
gas temperature increases. Finally, LNG temperature and NG-3
temperature can go back to their set point values. Compressor
duty also increases when the feed gas temperature increases. The
reason is that the cooling capacity which needed to liquefy the
natural gas increases when the feed gas temperature increases. It
means that the mixed refrigerant flow rate must increase to meet
the cooling capacity. Thus, the compressor needs more power to
pressurize the mixed refrigerant.
According to the analysis above, the disturbance of feed gas
temperature can affect the controlled variables, while the
controlled variables can go back to stable state in a quite small time.
Moreover, the bigger disturbance can induce bigger fluctuation.
4.2. Disturbance of feed gas pressure
Feed gas pressure will influence natural gas dew point tem-
perature and latent heat. The feed gas pressure is varied by 100, 200
and 300 kPa downward and upward, respectively. Dynamic re-
sponses of controlled variable after the decrease and increase of
feed gas pressure are illustrated in Figs. 5 and 6.
The dynamic responses of LNG temperature, NG-3 temperature
and compressor duty after the feed gas pressure decreases 100, 200
and 300 kPa, respectively, are shown in Fig. 5. LNG temperature and
NG-3 temperature increase to higher temperature at first. It is
because that the decrease of feed gas pressure reduces the heat
transfer performance in heat exchangers. Thus, the cooling duty is
not enough to cool the feed gas to specified temperature. However,
LNG temperature and NG-3 temperature decrease and then go back
to their set point value. The reason is that the control system turns
up the mixed refrigerant flow rate valves and the cooling duty is
larger than the heat needed to be removed from the feed gas.
Finally, the mixed refrigerant flow rate is adjusted to a proper value
and LNG temperature and NG-3 temperature go back to set point
value. Compressor duty decreases at first and then increases when
the feed gas pressure decreases. The system needs about 500 min to
eliminate the disturbance of decrease of feed gas pressure.
Fig. 6 illustrates the dynamic responses of LNG temperature, NG-
3 temperature and compressor duty after the feed gas pressure
increases 100, 200 and 300 kPa, respectively. When the feed gas
pressure increases, the feed gas flow rate will also increase. That
makes LNG temperature and NG-3 temperature increases at first
and then decreases. Finally, they can reach to their set point values.
Compressor duty decreases when the feed gas pressure increases.
That is because that the increase of feed gas pressure can improve
the heat transfer performance and the mixed refrigerant flow rate
decreases. The system only needs 200 min to go back to a stable
state.
4.3. Disturbance of feed gas composition
Feed gas composition has a significant influence on process
performance. In this study, the change of methane in feed gas
composition is adapted as a disturbance to investigate the
response.
The methane composition is varied by 1%, 2% and 3% downward
and upward, respectively. The other compositions decrease or in-
crease by equal proportion. Dynamic responses of controlled vari-
able after the decrease and increase of methane composition are
illustrated in Figs. 7 and 8.
 T. He, Y. Ju / Energy 97 (2016) 350e358 355

Fig. 3. Dynamic responses of controlled variable after the decrease of feed gas temperature.

Fig. 4. Dynamic responses of controlled variable after the increase of feed gas temperature.

Fig. 5. Dynamic responses of controlled variable after the decrease of feed gas pressure.

Fig. 6. Dynamic responses of controlled variable after the increase of feed gas pressure.

As shown in Fig. 7, the decrease of methane composition can temperature fluctuate for about 300 min. Finally, LNG and NG-3
increase LNG and NG-3 temperature at first. Then LNG and NG-3 temperature go back to their set point values. Moreover,
356 T. He, Y. Ju / Energy 97 (2016) 350e358
Fig. 7. Dynamic responses of controlled variable after the decrease of methane composition.
Fig. 8. Dynamic responses of controlled variable after the increase of methane composition.
compressor duty also fluctuates at first and then stays at stable
value. The final compressor duty is higher than the initial value. The
reason is that the decrease of methane composition will reduce the
viscosity of feed gas, which can enlarge the feed gas flow rate
through the valve. Then the cooling capacity is not enough to cool
the feed gas to specified LNG and NG-3 temperature. Thus, LNG and
NG-3 temperature increase with the decrease of methane compo-
sition. Meanwhile, the opening percentages of VLV-2 and VLV-3
increase to provide more cooling capacity, which decreases LNG
and NG-3 temperature. At last, LNG and NG-3 temperature go back
to specified values after fluctuation. The system spends 500 min to
estimate the disturbance of methane composition decreasing.
However, the increase of methane composition enlarges the
viscosity of feed gas, which decreases the feed gas flow rate through
the valve. So LNG and NG-3 temperature decreases at first. Then the
decrease of mixed refrigerant flow rate makes LNG and NG-3
temperature increasing. Finally, LNG and NG-3 temperature go
back to specified valve. Compressor duty can be reduced along with
the increase of methane composition. What's more, the system
spends 400 min to recover to stable state.
According to the analysis, the decrease and increase of methane
composition have different effects on the process stability. The
decrease of methane composition makes the process more fluctu-
ating than the increase of methane composition.
4.4. Disturbance of feed gas flow rate
Feed gas flow rate may change if the small gas source is not
stable. In this section, disturbance of feed gas flow rate is adopted to
investigate the process performance.
Fig. 9 illustrates the dynamic responses of LNG temperature, NG-
3 temperature and compressor duty after the feed gas flow rate
decreases 1, 1.5 and 2 kmol/h, respectively. LNG and NG-3 tem-
perature decrease at first and then increase. Finally, LNG and NG-3
temperature go back to specified values. Compressor duty de-
creases at first and then reaches stable state. The reason is that the
cooling capacity provided by mixed refrigerant is more than feed
gas needed. Thus, LNG and NG-3 temperature decrease to a lower
value. At the same moment, control structure begins to reduce the
mixed refrigerant flow rate. That's why two temperatures increase
to specified values. Due to the decrease of mixed refrigerant,
compressor duty also can be reduced. The control system needs
300 min to estimate the influence of disturbance.
As shown in Fig. 10, LNG and NG-3 temperature increases at first
and then go back to specified values when feed gas flow rate in-
creases. Compressor duty will also increase when feed gas flow rate
increases. That is because that the increase of feed gas flow rate
needs more mixed refrigerant flow rate to cool it to specified
temperatures. The control system spends approximate 500 min to
reach stable state.
4.5. Disturbance of water cooler temperature
The atmosphere temperature may change from spring to winter.
This will influence the mixed refrigerant temperature after water
cooler. Thus, in this section, the decrease and increase of water
cooler temperatures (stream M5 and M12) are adopted as distur-
bance to test the process's stability.
Dynamic responses of controlled variables after the decrease of
water cooler temperature are illustrated in Fig. 11. The LNG tem-
perature decreases at first and then goes back to the specified value
when water cooler temperature decreases. It is because that vapor
mixed refrigerant has more low boiling temperature compositions.
That makes LNG temperature decrease. NG-3 temperature de-
creases at first and then increases to a high value. Finally, it goes
back to specified value. The reason is that the liquid mixed refrig-
erant flow rate increases as the decrease of water cooler tempera-
ture at first. Then the control system begins to reduce the mixed
 T. He, Y. Ju / Energy 97 (2016) 350e358 357

Fig. 9. Dynamic responses of controlled variable after the decrease of feed gas flow rate.

Fig. 10. Dynamic responses of controlled variable after the increase of feed gas flow rate.

Fig. 11. Dynamic responses of controlled variables after the decrease of water cooler temperature.

refrigerant flow rate. Therefore, the liquid mixed refrigerant flow
rate decreases, which makes the NG-3 temperature increasing.
After 300 min, the process reaches stable state. Compressor duty
also deceases when water cooler temperature decreases. That is
because that the mixed refrigerant flow rate decreases and less
power is needed to pressurize the mixed refrigerant.
As shown in Fig. 12, LNG temperature increases at first and then
decreases to the specified value when water cooler temperature
increases. The reason is that the cold mixed refrigerant needs to
cool more vapor mixed refrigerant. Thus, the cooling capacity can
be used to cool feed gas is reduced. Therefore, LNG temperature
increases at first. For NG-3 temperature, it decreases at first and
then increases to
61.8  C after 500 min. The reason is the total
mixed refrigerant flow rate increases when water cooler tempera-
ture increases. So cooling capacity in HEX-1 increases and NG-3
temperature decreases at first. Compressor duty will increase
when water cooler temperature increases. That is because the
mixed refrigerant flow rate increases.
5. Conclusions
In this study, a dynamic model of mixed refrigerant liquefac-
tion process for small-scale LNG plant was developed to investi-
gate the dynamic behaviors. The dynamic simulation results were
compared with steady-state simulation results to verify the ac-
curacy of dynamic model. The comparison showed good match
between dynamic simulation and steady-state simulation. Then,
five kinds of disturbances were added to the process to test its
flexibility and dynamic behaviors. The dynamic responses of LNG
temperature and total energy consumption were the criteria to
investigate the influences of disturbances on the process. The
results indicated that the process can handle these kinds of
disturbance and go back to the specified values after some time.
The mixed refrigerant process showed good operation flexibility
and can estimate different kinds of disturbance in a wide range.
The dynamic responses of five different disturbances are
concluded as follows:
358 T. He, Y. Ju / Energy 97 (2016) 350e358
Fig. 12. Dynamic responses of controlled variable after the increase of water cooler temperature.
1) The decrease of feed gas pressure and methane composition as
well as the increase of feed gas flow rate have the biggest in-
fluence on the process stable operation.
2) The decrease of feed gas pressure affects the process in two
aspects. The first one is that the decrease of feed gas pressure
can weaken the heat transfer in cryogenic heat exchanger. Thus,
the process needs more mixed refrigerant to liquefy the natural
gas. The second one is that the decrease of feed gas pressure can
also influence the flow rate of natural gas in the process, since
the flow rate in dynamic simulation is based on the pressure
drop of the equipment. As a result, the operation of the process
should try to keep the pressure of feed gas at constant. If the
pressure of feed gas changes, the process needs a long period of
time to go back to stable operation.
3) The decrease of methane composition means that the feed gas
has a heavier molecular weight. The heavier feed gas has
higher heat content and will transfer more heat to the cold
mixed refrigerant, resulting the decrease of temperature dif-
ference between the hot and cold stream. The compressor
duty will increase in order to pressurize more mixed
refrigerant.
4) The increase of feed gas flow rate means that the actual lique-
faction capacity is beyond the design liquefaction capacity. Thus,
the increase of feed gas may cause very big fluctuation to the
process.
5) The variation of feed gas temperature has little influence on the
process. Although it can make the process unstable, the process
can control the key parameters to specified values in quite short
period of time.
6) The variation of water cooler temperature has great effect on
compressor duty, while small effect on process operation. Note
that the compressor duty should not beyond the design shaft
work by 10%.
Acknowledgment
The research is partly supported by SJTU-ENN LNG Engineering
Center.
References
[1] International Energy Outlook. U.S. Energy Information Administration. 2013.
[2] Shukri T. LNG technology selection. Hydrocarb Eng 2004;9(2):71e6.
[3] Lee GC, Smith R, Zhu XX. Optimal synthesis of mixed-refrigerant systems for
low-temperature processes. Ind Eng Chem Res 2002;41(20):5016e28.
[4] Wang MQ, Zhang J, Xu Q, Li KY. Thermodynamic-analysis-based energy con-
sumption minimization for natural gas liquefaction. Ind Eng Chem Res
2011;50(22):12630e40.
[5] Morin A, Wahl PE, Molnvik M. Using evolutionary search to optimise the
energy consumption for natural gas liquefaction. Chem Eng Res Des
2011;89(11A):2428e41.
[6] Khan MS, Lee M. Design optimization of single mixed refrigerant natural gas
liquefaction process using the particle swarm paradigm with nonlinear con-
straints. Energy 2013;49:146e55.
[7] Hwang J-H, Roh M-I, Lee K-Y. Determination of the optimal operating con-
ditions of the dual mixed refrigerant cycle for the LNG FPSO topside lique-
faction process. Comput Chem Eng 2013;49:25e36.
[8] Hatcher P, Khalilpour R, Abbas A. Optimisation of LNG mixed-refrigerant
processes considering operation and design objectives. Comptuters Chem
Eng 2012;41:123e33.
[9] Alabdulkarem A, Mortazavi A, Hwang YH, Radermacher R, Rogers P. Optimi-
zation of propane pre-cooled mixed refrigerant LNG plant. Appl Therm Eng
2011;31(6e7):1091e8.
[10] Mortazavi A, Somers C, Hwang Y, Radermacher R, Rodgers P, Al-Hashimi S.
Performance enhancement of propane pre-cooled mixed refrigerant LNG
plant. Appl Energy 2012;93(SI):125e31.
[11] Rodríguez M, Diaz M. Dynamic modelling and optimisation of cryogenic
systems. Appl Therm Eng 2007;27(7):1182e90.
[12] Singh A, Hovd M. Dynamic Modeling and Control structure design for a Liq-
uefied Natural Gas Process. Conference Dynamic Modeling and Control
structure design for a Liquefied Natural Gas Process. IEEE, p. 1347e52.
[13] Mandler JA. Modelling for control analysis and design in complex industrial
separation and liquefaction processes. J Process Control 2000;10(2):167e75.
[14] Husnil YA, Lee M. Synthesis of an optimizing control structure for dual mixed
refrigerant process. J Chem Eng Jpn 2014;47(8):678e86.
[15] Lim W, Choi K, Moon I. Current status and perspectives of liquefied natural gas
(LNG) plant design. Ind Eng Chem Res 2013;52(9):3065e88.
[16] He T, Ju Y. Design and optimization of a novel mixed refrigerant cycle inte-
grated with NGL recovery process for small-scale LNG plant. Ind Eng Chem
Res 2014;53(13):5545e53.
[17] Jensen JB, Skogestad S. Optimal operation of simple refrigeration cycles. Part I:
degrees of freedom and optimality of sub-cooling. Comput Chem Eng
2007;31(5e6SI):712e21.
[18] Jensen JB, Skogestad S. Steady-state operational degrees of freedom with
application to refrigeration cycles. Ind Eng Chem Res 2009;48(14):6652e9.
[19] Luyben W. Plantwide dynamic simulators in chemical processing and control.
CRC Press; 2002.