Anda di halaman 1dari 8

MODELING AND OPTIMIZATION OF A C3MR LNG PLANT EFFICIENCY BY CHANGE OF MIXED REFRIGERANTS COMPONENTS

Hamid Saffari*, Masoud Fasihbeiki *LNG Research Laboratory, School of Mechanical engineering, Iran University of Science and Technology, Narmak, 16846, Tehran, Iran *saffari@iust.ac.ir

ABSTRACT
In this paper, we have optimized the energy efficiency of an industrial C3MR LNG base load plant by changing its refrigerants components and their mole fractions in liquefaction and subcooling cycles. The process is modeled by using the Hysys software. The PRSV equation of state is used for thermodynamic properties calculations both for the natural gas and the refrigerants. Two methods for modeling and optimization are explained and the results are compared. The first optimization method is done by a try and error method, which is based on the use of temperature vs. enthalpy diagrams or composite curves. In the second method, Hysys optimizer is used for optimization. The results show that by optimization of mixed refrigerants, it is possible to decrease the energy demand about 10.4 MW (5.36 %.) Key words: Liquefied Natural Gas; C3MR Process; Efficiency; Hysys software; Cooling Curve

1. INTRODUTION Natural gas is set to become one of the most important primary energy sources for the 21st century. Compared with other fossil fuels, gas is relatively clean with regards to air pollution and greenhouse gas emissions and has larger proven reserves. It is expected that the natural gas would account for about 30% of total electricity generation by 2020 compared with 17% in 2000 [1]. In this paper, we would like to optimize a propane precooled mixed refrigerant process refrigerants where precooling is achieved by a multi-stage propane cycle while liquefaction and subcooling are accomplished by a two-stage mixed-refrigerant cycles, which is so far the most common process used since 1972 in 8 different countries. The propane pre-cooled mixed refrigerant LNG process (C3MR) has been applied in LNG plants producing from 1 to 5 Mtpa of LNG per train using steam, gas turbine or electrical drivers, sea water and air cooling, rich and lean feeds containing a broad range of nitrogen with and without LPG extraction. The process has proven to be efficient, flexible, reliable, and costcompetitive [2]. Propane precooling mixed component refrigerant process (C3-MR) represents 80 % of that used in the current base load plants. Large-scale liquefaction of natural gas takes a large amount of energy. That is why optimization is necessary to the process at steady state operation. Because of low efficiency of conversion of fossil fuels to electricity and consumption of approximately 200MW in electrically driven compressors and main pump of each train, optimization is advantageous in this case. There are some differences between the proposed cycle and a mixed fluid cascade cycle. In a MFC process, A Mixture of ethylene, ethane, propane and butane is used for precooling, while almost a pure propane refrigerant is utilized for precooling in the cycle studies. In addition, expander will not be used in a C3MR Process and throttling achieves by throttling valve and precooling is achieved in three or four stages, while in a MFCP, expander usually is used for throttling and precooling happens in two stages. A Mixture of Ethylene, ethane, methane is used for liquefaction and subcooling in MFCP, while ethylene will not be used in C3MR Process. The design, simulation and estimation of natural gas liquefaction process began from 1970. Shell Corporation has simulated the cascade, mixed-refrigerant and N2 expander cycle and analyzed their advantages and disadvantages [3]. Melaaen set up a dynamic model for the natural gas liquefaction process of base load plant, and carried through the simulation by DASSL in 1995 [3]. Kikkawa designed the late model of pre-cooling mixed-refrigerant processes and expander cycle, and he used CHEMCAD software in his calculations in 1997 [4]. Terry adopted Hysys software to calculate and optimize the typical liquefaction process of peak shaving plant in 1998 [5]. Gu and his associates have carried through the simulation and calculation of natural gas liquefaction process [3]. They also compared the key parameters of two small-scale natural gas liquefaction processes using Hysys software in 2006 [6]. For simulation of a LNG Production process, a reliable equation of state (EoS) is needed for thermodynamic data predictions. Several review articles and books cover the equations of state published in the literature. Most of them concentrated on cubic EoS's and their mixing rules [7]. Simulation of the process has been conducted using Hysys simulator version 3.2 due to many fluid properties data, Binary coefficients and suitable equations of state. In an

industrial system in which there are many exchangers requiring refrigerants at many temperature levels, the simulation of such a system is much simplified if the PFD contains only real equipment. The elimination of controllers makes the try and error method simulation much easier than that of using numerical convergence standpoint (using Hysys optimizer). The PFD consists of only real equipment other than a cluster of nonequipment items such as controllers and calculators [8]. Large temperature difference and heat exchange load are the primary reasons of exergy loss in heat exchangers. In this method, by comparing the consumed work of cycle of each mole fraction with previous composition, the preferred composition will be cleared and by comparing the changes in composite curves formed in main heat exchangers, in each iteration, the mole fraction of next step will be predicted. Easily made modeling, rapid countering and close answers to optimal operation (which will be obtained by Hysys optimizer) are the advantages of this method.

But there are some differences between the real simplified process modeled here and the standard C3MR process. For example in a C3MR Process a mixture of refrigerants is used for liquefaction and subcooling. In this process, after the refrigerant passes through the separator, the liquid is used for liquefaction and the gas is used for subcooling after dropping its pressure while in the modeled process two distinct refrigerants are used for liquefaction and subcooling.

3. MODELING AND OPTIMIZATION: Before the simulation, the required parameters from main process are specified in Tables 1. As the flow rate of feed gas is constant, we have used the consumed power instead of specific power which is normally used in optimization (for example [6]). Assumed flow rate of natural gas feed is 46 170.25 kg mol/h. Table 1. Mole fraction of components

2. PROCESS DESCRIPTION: The process which will be model and optimize is an industrial process. But, because of some industrial limitation, the detailed description of simplified process has been presented in figure 1. Propane is evaporated at four pressure levels in precooling cycle (LNG-100, LNG-101, LNG-102 and LNG-103) to desuperheat the natural gas feed (stream 1) and cool and to partially condense the main refrigerants mixture (streams 28 and 36 for liquefaction and subcooling respectively) before they enter to main heat exchangers(LNG-104 and LNG-105). The heated refrigerant then passes through the separators (V100, V-101 and V-102). The remained liquid is used for next stage of precooling after it passed the valve. The vapor collects at each pressure levels and enters four compressors (K-100, K-101, K-102 and K-103). In liquefaction cycle, mixed refrigerant first enters four propane precooling heat exchangers identified before. After cooling to about -30.4C at the last heat exchanger (LNG103), it passes through the liquefaction main heat exchanger (LNG-104) and liquefies. After dropping its pressure and temperature in a throttling valve (VLV-100), the refrigerant comes back to the heat exchanger and gets heat from hot composite (natural gas, Subcooling refrigerant and inlet liquefaction refrigerant). Then it comes back to the compressor (K-104). In subcooling cycle, its mixed refrigerant like liquefaction mixed refrigerant passes propane precooling heat exchangers (LNG-100, LNG-101, LNG-102 and LNG-103) and liquefaction main heat exchanger (LNG-104). Then passes through main subcooling heat exchanger (LNG-105) and is depressurized in VLV-101 and comes back through the subcooling heat exchanger to subcooling compressors (K105 and K-106). The natural gas is cooled in the propane precooling cycle through four heat exchangers (LNG-100, LNG-101, LNG102 and LNG-103) to about -30.4C. Then by passing through two main heat exchangers, namely LNG-104 and LNG-105, the temperature falls to about -162.9C and finally after dropping its pressure to about 0.116 bar releasing most of its nitrogen contents in separator V-104, it is pumped to receivers.

Methane % Natural gas Precooling Refrigerant 89.88% 0.00%

Ethane % 4.86% 1.81%

Propane and heavier % 0.18% 98.19%

Nitrogen % 5.08% 0.00%

Different equations of state (EoS) have been evaluated at the cryogenic conditions to choose the most suitable one to be used in the simulation of the process. Literature says that PRSV EoS is superior to other examined EoS's [7, 9, 10, 11, and 12]. Details about these EoS are mentioned in other works (for example [6]) and are beyond the scope of this paper. The simulation, calculation and optimization of Processes were done using PRSV equation of state through Hysys software. PRSV is a modified Peng-Robinson EoS and has the best adaptation and the least average absolute deviation with experimental data at cryogenic condition. [7] This equation is one of the most important Fluid Packages that is the base of the simulation by Hysys. Many factors influence the performance of a certain process. For instance, they are pressures and temperatures of the mixed-refrigerants at each stage, temperature of the refrigerant before expansion, and mole fraction of nitrogen, methane, ethane, propane, temperature approach between hot and cold composites in main heat exchangers, etc: As we would not like to change any physical parameters in the real process such as thermal surface of heat exchangers by changing temperature approaches and so on. In this case, the optimization problem is finding out the optimum mole fraction of liquefaction and subcooling refrigerant component values to make the power consumption to its lowest level. This is some of powers consumed by 7 compressors and 1 pump (see figure 1) and we neglect the power used for air cooler and other utilities. We also assume to have 80% adiabatic efficiency in all units of compressors and pump and this value will not be changed by changing the mass flow rate of refrigerants. The process described in the previous section was simulated using Hysys flow-sheeting program using PRSV EoS property package.

Fig.1. The Modeled simplified C3MR Process 4. TRY AND ERROR METHOD FOR OPTIMIZATIOM: 4-1. Concept: Many parameters can be used for optimization. Suction and discharge pressure of compressors, the temperature and pressure of natural gas and refrigerants at each stage may be uses in an optimization process. As we would liked to apply the optimization into the main cycle without any physical in physical components, only composition of mixed refrigerants and their mass flow rates have been selected as variables in this work and other properties of refrigerant and natural gas have been assumed to be constant. Appreciating the variation of temperature of the high and low pressure streams along the length of the heat exchanger is the first step in understanding the reason for low exergy efficiency of the cryogenic refrigerators and liquefiers operating with pure fluids. Figure 2 shows the temperature profiles of the hot streams (main refrigerant mixture, natural gas feed) and the cold stream (propane refrigerant) in four precooling heat exchangers. The large temperature difference between the hot and cold streams results in a higher exergy loss in the propane-precooled process compared to the mixed refrigerant precooled process (Figure 3) [13].

Figure 3. Temperature profiles in the two main heat exchangers of a typical precooled mixed refrigerant process [13] Figure 2. Temperature profiles of the hot and cold (propane) streams in four precooler heat exchangers of a four-stage precooler of propane precooled mixed refrigerant natural gas liquefaction process. [13]

The indicative cooling curve of a natural gas with the temperature showed in figure 4 profiles two superimposed

routes to liquefaction with pure and mixed refrigerants. For commercial liquefaction process, the intent is to minimize the difference between the hot and cold composites in order to lower the exergy loss and improving thermal efficiency [14]. As the temperature difference between two curves declines, the entropy generation and exergy loss decrees. That is why the designers use mixed combination of hydrocarbons instead of pure fluids as refrigerant in recently appeared cycles. Here we are going to optimize the refrigerants using these cooling curves based on this concept. In each stage, the preference of selected mixture is understood by observation of consumed work uses by compressors and pump and the mole fraction of components has been changed by observation of the change in cooling curve produced in main heat exchanger and the predicted mole fractions were obtained. This operation has been done for subcooling and liquefaction refrigerant. It is important to say that in all stages of optimization, the predefined temperature and pressure of natural gas and refrigerant do not change. In optimization of a process by changing its refrigerants some points should be considered. In this process the components of refrigerants are the products of this plant. So availability of refrigerants components should be observed. At each cycle, selection of components of refrigerants corresponds with low and high pressure and temperature of the cycle. For optimal operation all components of refrigerant must be in two phase at the evaporator and in gas phase at the inlet of compressors to prevent corrosion. That is why propane is used for precooling and it is not used for subcooling and for the same reason nitrogen for subcooling and not for precooling in the main non-optimized process.

noteworthy to mention that if we liked to start optimization from liquefaction refrigerant, we should have optimized composition of Subcooling refrigerant and its optimum flow rate to have the whole plant optimized. After changing the refrigerant mole fraction in each stage, first the mass flow rate of refrigerant modifies to (a) have the least flow rate and to lower the duty of compressors, and (b) avoid temperature cross in main heat exchanger. Then the required mass flow rate for liquefaction and precooling refrigerants obtained on the basis of two cited constraints and total duty estimated. Other constraints are as below: (A) Sum of the mole fractions of mixed-refrigerant is 1 and (B) The temperature of mixed-refrigerant at the inlet of compressor is higher than its dew point. It is clear that as the number of iterations increases, the optimization would become better and results will converge with by Hysys optimization results. In the diagrams shown in Figure 5 to Figure 13, cold composite is the backward low-temperature low-pressure refrigerant and hot composite is sum of other curves formed in the heat exchanger (natural gas and forward subcooling refrigerant in subcooling heat exchanger and natural gas and forward subcooling and liquefaction refrigerant in liquefaction heat exchanger). 4-3. Results and Analysis of Optimization for Subcooling Cycle: As cited before, we could not start optimization from liquefaction cycle from the beginning of the process. After specifying the optimized composition and mass flow rate of refrigerant in Subcooling cycle, and assuming of constant condition for natural gas and subcooling refrigerant at the inlet and outlet of main liquefaction heat exchanger, we can optimize the liquefaction refrigerant as it was done for the subcooling cycle. In this refrigerant, we do not have nitrogen because it stays in the form of gas anywhere in the cycle in operating pressures and temperatures. Some samples of examined combination of mole fractions are presented in Table 2. The relative cooling curves are presented in figures 5-9.
Table 2. Some samples of examined combination of mole fractions
Iteration Methane Ethane Nitrogen Required no. % % % subcooling mass Flow rate(kg/h)

Power MW

Figure 4. The indicative cooling curve of natural gas and two superimposed route of liquefaction with pure and mixed refrigerant [14]

Based on the thermodynamic analysis (first and second law of thermodynamics), the simulation and calculation of natural gas liquefaction process in skid-mounted package were carried through. 4-2.Optimization: Optimization starts from subcooling refrigerant because we assumed that the conditions of natural gas is constant at the inlet and outlet of main Subcooling heat exchanger; and only the natural gas and subcooling refrigerant passes through this heat exchanger and we can study the effect of changing mole fraction and mass flow rate of refrigerant directly on cooling curve and consumed work. It is

1 2 3 4 5

60 65 55 54 56

35 30 40 39 41

5 5 5 7 3

501 134 515 000 500 000 510 000 495 000

193.8 202.4 188.7 191.0 187.1

Note that at iteration no. 5 the temperature approach is about zero, and these two curves are similar and the consumed work is minimum value. Because of high demand for heat transfer area and lack of heat exchanger that could work at this condition, this iteration will be neglected and iteration no. 3 is used for continuance of optimization.. More attempts showed that it is not possible to gather more than two curves and to decrease the total duty. Optimization in this case is finished. By using wide range of refrigerants we might reduce the total consumed power further.

Figure 5. Cooling curve for subcooling heat exchanger before optimization Iteration no. 1

Figure 7. Cooling curve for subcooling heat exchanger Iteration no. 3

Figure 6. Cooling curve for subcooling heat exchanger Iteration no. 2

Figure 8. Cooling curve for subcooling heat exchanger Iteration no. 4

Figure 9. Cooling curve for subcooling heat exchanger Iteration no. 5

4-4. Results and Liquefaction Cycle

Analysis

of

Optimization

for

The concept of optimization is as explained before. Effect of changing the concentration of an element on cooling curves and total duty guided us to modify the mole fraction of that is the next iteration.
Table 3. Some samples of examined combination of mole fractions Required Iteration Methane Ethane Nitrogen subcooling no. % % % mass Flow rate(kg/h)

Power (MW)

It is clear that if we like to have a different temperature approach we can select the related mole fraction of mentioned cooling curve. The first iteration in this section is the optimized condition achieved in the previous part. (Iteration no. 3). Some samples of examined combination of mole fractions are presented in Table 3. The relative cooling curves are presented in figures 10 and 11. At iteration 11, two curves are too close and this result is not acceptable like iteration 5. As it is obvious in the previous figures, when two curves come closer, the demand for work in this plant will come down. 5. OPTIMIZATION USING HYSYS OPTIMIZER: Hysys contains a multi-variable steady state Optimizer. Once the flow sheet has been built and a converged solution has been obtained, the Optimizer can be used to find the operating conditions which minimize (or maximize) the objective function. The Optimizer owns its own spreadsheet

6 7 8 9 10 11

10 15 10 25 10 10

70 70 80 50 60 55

20 15 10 25 30 35

790 000 795 000 795 000 815 000 820 000 820 000

188.7 196.3 195.9 198.9 184.0 181.7

Table 4.Optimum component mole fraction obtained by Hysys optimizer

Methane % Subcooling refrigerant (before optimization) Subcooling refrigerant (optimized) Liquefaction refrigerant (before optimization) Liquefaction refrigerant (optimized) 60 53.7 10 11.93

Ethane % 35 41.37 70 55.22

Propane % 0 0 20 32.85

Nitrogen % 5 4.93 0 0

Consumed work (MW) 193.8 183.4 193.8 183.4

Figure 10. Cooling curve for liquefaction heat exchanger Iteration no. 6

Figure 12. Cooling curve for subcooling heat exchanger optimized by Hysys optimizer

Figure 11. Cooling curve for liquefaction heat exchanger Iteration no. 10

Figure 13. Cooling curve for liquefaction heat exchanger optimized by Hysys optimizer

for defining the objective function, as well as any constraint expressions to be used. HYSYS has five modes of Optimizer: Original, Hyprotech SQP, MDC Optim, MDC DataRecon and Selection Optimization. As it is not possible to define the mole fraction of components in each refrigerants, we defined the mass flow rate of components as optimization variable, which they mix and generate the refrigerants. For each heat exchanger, an adjust controller determines the cooling stream flow rate based on predefined temperature approach (about 3C in main and precooling heat exchangers). An imaginary heating process which consists of a liquid flow controller and a heater prevents the existence of liquid in compressors intervals. It is expected that at optimized condition, no heat should be added to the streams. Optimum subcooling and liquefaction refrigerants' components mole fraction obtained by Hysys optimizer and predefined values are presented in Table.4. Relative cooling curves are also present in figures 12 and 13. Optimization using Hysys optimizer results in better answers to optimization. Optimization using this method could redound in less power consumption and more convergence of composite curves. By comparing figures 12 and 13 to figures 7 and 11 respectively, the concept of exergy losses due to definite temperature difference will be clear.

4- While using EoS in some steam properties, there are some errors in data prediction. These uncertainties affect the data and may not lead to real optimum answer. Using restricted deviation that is more compatible with experiment data, especially for low temperatures, helps us to become assured of these results. Surrendering heat leakage to the process, assuming constant adiabatic efficiency of compressors (neglecting the effect of stream flow on it) and assuming constant pressure drop in all streams also lead to less assurance.

7. REFERENCES 1. United Nations Framework Convention on Climate Change, Proposal on Cleaner or Less Greenhouse Gasemitting Energy, Subsidiary body for scientific and technological advice, 16th session, Bonn, June 5-14, 2002. 2. M. J. Roberts and Y. N. Liu and J. C. Bronfenbrenner, Reducing LNG Capital Cost in Today's Competitive Environment, 14th International Conference & Exhibition on Liquefied Natural Gas (LNG14), PS2 -6.1-12, Doha Qatar, 2004. 3. A. Gu, X. Lu, R. Wang, Y. Shi and W. Lin, Liquefied Natural Gas Technology, China Machine Press, 2004. 4. K. Yoshitugi and N. Moritaka, Development of liquefaction process for natural gas. Journal of Chemical Engineering of Japan, Vol. 30, pp. 626630, 1997. 5. L. Terry, Comparison of liquefaction process, LNG Journal Vol. 21, No. 3, pp. 2833, 1998. 6. W. Cao, X. Lu, W. Lin and A Gu, Parameter Comparison of Two Small-Scale Natural Gas Liquefaction Processes in Skid-Mounted Packages, Applied Thermal Engineering, Vol. 26, pp. 898-904, 2006. 7. I. Ashour and T. S. Sayed-Ahmed, Modeling and Simulation of a Liquefied Natural Gas Plant", the Fourth Annual U.A.E. University Research Conference, pp. 48-51, 2002. 8. K. H. Pang, A Novel Use of a Computer Simulator to Design an Industrial Refrigeration System, Chemical and Materials Engineering, pp. 117-124, 1999. 9. C. Newton and L. Gaumer, Process for Liquefying Methane, US Patent No. 445916, 1984. 10. Y. S. Wei and R. J. Sadus, Equations of State for the Calculation of Fluid Phase Equilibria. AIChE Journal. Vol. 46, pp. 169- 196, 2000. 11. A. Firooozabadi, Thermodynamics of Hydrocarbon Reservoir. McGraw-Hill, 2000. 12. S. Malanowski and A. Anderko, Modeling Phase Equilibria: Thermodynamic Background and Practice Tools. John Wiley & Sons, Inc. 1992. 13. G. Venkatarathnam, Cryogenic Mixed Refrigerant Processes. Springer, New York, 2008. 14. www.nexant.com, the expanding Horizons of Liquefaction Technology and Project Execution Strategies, 2007, Available online 2008.

6. CONCLUSIONS 1- As we expected to happen, by closing of two curves in cooling curve diagrams, we could reduce total duty and increase the efficiency of LNG production. By changing the percentage of refrigerants' components and selecting the optimized ones, we could reduce 5.06 % of preliminary power (Trial and error method). Besides, we could increase about 10.53% of base load by spending the original power. The advantage of this method is that we could achieve these results without any change in any part of this factory. 2- The mole fraction of refrigerants' composition might be obtained either by trial and error method based on cooling curves appeared in the main heat exchangers or using Hysys optimizer. In this method, it is possible to decrease the energy demand about 10.4 MW (5.36 %). Optimization by Trial and error method needs a simpler flow sheet and gives leads to rapid and fairly good results, but using Hysys optimizer needs some controller elements and time to converge, even the answer is more accurate. 3- As we know, in pure refrigerants if we assume that the upstream condition is determined before throttle valve, the pressure of refrigerants will be calculated when the temperature is known and vice versa. But while using mixed refrigerants, A point which should be considered during modeling is that the pressure is a function of the composition or the temperature is a function of pressure and composition and during changing the composition of refrigerant, the pressure might fall down under atmosphere pressure if the temperature assumed to be fixed. So fixing the appropriate variable should be considered.

Anda mungkin juga menyukai