Multi-objective optimization of high-sulfur natural gas purification plant
- 81 Downloads
Abstract
There exists large space to save energy of high-sulfur natural gas purification process. The multi-objective optimization problem has been investigated to effectively reduce the total comprehensive energy consumption and further improve the production rate of purified gas. A steady-state simulation model of high-sulfur natural gas purification process has been set up by using ProMax. Seven key operating parameters of the purification process have been determined based on the analysis of comprehensive energy consumption distribution. To solve the problem that the process model does not converge in some conditions, back-propagation (BP) neural network has been applied to substitute the simulation model to predict the relative parameters in the optimization model. The uniform design method and the table U_{21} (10^{7}) have been applied to design the experiment points for training and testing BP model. High prediction accuracy can be achieved by using the BP model. Non-dominated sorting genetic algorithm-II has been developed to optimize the two objectives, and 100 Pareto optimal solutions have been obtained. Three optimal points have been selected and evaluated further. The results demonstrate that the total comprehensive energy consumption is reduced by 13.4% and the production rate of purified gas is improved by 0.2% under the optimized operating conditions.
Keywords
High-sulfur natural gas purification plant Multi-objective optimization Process simulation model Thermodynamic analysis BP neural network Genetic algorithm1 Introduction
The high-sulfur natural gas purification plant is mainly composed of five units, including the gas sweetening unit (GSU), the dehydration unit (DU), the sulfur recovery unit (SRU), the tail gas treatment unit (TGTU) and the acid water stripping unit (AWSU). In the natural gas purification system, there are a great number of devices and relative couples of operating parameters. The relation of dozens of operating parameters to TCEC is nonlinear and the coupling correlations between some parameters are strong, which makes the optimization problem of TCEC complicated and difficult to be solved by using the traditional optimization method (Yu et al. 1998; Abdulrahman and Sebastine 2013).
In recent years, very few published articles research on the global optimization of operating conditions to reduce TCEC for the whole high-sulfur gas purification plant. Most optimization studies focus on the subsections, such as GSU (Jassim 2016; Peter et al. 2011; Qiu et al. 2014; Behroozsarand and Zamaniyan 2011; Al-Lagtah et al. 2015), DU (Santos et al. 2017; Rouzbahani et al. 2014), SRU (Manenti et al. 2014; Zarei et al. 2016; Adewale et al. 2016) and TGTU (Wahedi et al. 2015). A typical Iranian gas sweetening plant was simulated by HYSYS v3.1 and optimized by NSGA-II method (Behroozsarand and Zamaniyan 2011). Three objectives were minimizing of plant energy consumption, amine circulation rate and maximization of carbon dioxide recovery, respectively. Lekhwair natural gas sweetening plant was simulated by HYSYS v7.3 (Al-Lagtah et al. 2015). The main operating parameters, lean amine circulation rate, temperature and concentration, were optimized based on sensitivity analysis and keeping the device at its best operating range. Moreover, two modifications (conventional split-loop and modified split-loop) are simulated and discussed to meet the required gas purity at lower energy costs. A natural gas DU was simulated by a steady-state flow-sheet simulator, and a sensitivity analysis was performed based on the simulation results (Rouzbahani et al. 2014). For the process optimization, dry gas dew point was assumed as key factor and it was shown that a 10% increase in solvent molar flow rate is applicable in order to reduce dry gas dew point up to 6% without a significant rise in the total energy consumption and volatile organic compounds emission. An optimization of SRU was performed based on the multi-scale model and specific algorithms of BzzMath library (Manenti et al. 2014). The optimization problem has three input variables, furnace pressure, acid gas/air ratio and water temperature of waste heat boiler. The objective function is to maximize the production of sulfur and medium-pressure and low-pressure steam. The optimization result shows steam generation can be improved of more than 6% by preserving the amount of recovered sulfur. An optimization of SRU was carried out to maximize sulfur production and minimize COS emission while maintaining flow ratio H_{2}S to SO_{2} at 2, in which the GA method was used. Three variables in the optimization function are the ratio of air to acid gas flow, the reaction furnace feed temperature and the steam pressure of waste heat boiler (Zarei et al. 2016). It was assessed that the economics of a TGTU rely on a temperature swing adsorption module. The optimization of minimum of net present worth costs (NPWC) of total capital investment, operating and bed replacement costs ensued during a 30-year project life was carried out (Wahedi et al. 2015).
Artificial neural networks have been widely applied to tackle complex issues. Since inception, they have been used in different engineering applications including prediction of key output parameters in the natural gas sweetening and dehydration process (Salooki et al. 2011; Ghiasi et al. 2014; Ahmadi et al. 2014; Darwish and Hilal 2008).
Biology inspired algorithms, which belong to intelligent numerical method, are generally divided into evolution-based algorithm and swarm-based algorithm (Neri and Carlos 2012). Evolution-based algorithms, i.e., evolutionary algorithms, have strong adaptability and self-organization, including genetic algorithm (GA), evolutionary programming (EP), evolutionary strategy (ES), differential evolution (DE) algorithm, etc. (Cui et al. 2017). GA not only has simple, general and strong stability characteristics, but also has the parallel and global search ability, which makes it competent to solve the large-scale and nonlinear optimization problem (Louis et al. 2009; Dai et al. 2000; Azizkhani et al. 2014). Non-dominated sorting genetic algorithm-II (NSGA-II) was first proposed by Deb et al. (2000) based on the NSGA (Srinivas and Deb 1994). In recent years, NSGA-II has been applied in many engineering application to solve the multi-objective problems (Behroozsarand and Zamaniyan 2011; Damavandi et al. 2017; Singh and Das 2016; Deng et al. 2017; Boyaghchi and Chavoshi 2017). An simulation–optimization method based on the NSGA-II for the life cycle optimization of a novel process design for a more cost-effective, greener process for making chemicals from shale gas and bio-ethanol was developed (He and You 2015). New process integration approaches, graphical pinch technique, were proposed to optimize the process of unconventional gas field (Foo et al. 2016).
As mentioned above, there are few papers focused on the global optimization of the whole high-sulfur gas purification plant. In this paper, the objective is to explore an effective method for global optimization of the high-sulfur natural gas purification plants to reduce the TCEC and improve the purified gas production rate (PGPR) of it. Key operating parameters of the natural gas purification plant have been selected on the parametric sensitivity analysis and the energy consumption analysis for each unit. A steady-state simulation model of the high-sulfur natural gas purification process has been set up by using software ProMax to carry out this research. To solve the problem that the process simulation model does not converge in some operating conditions for the existence of nested loops in it, which leads to fail to supply the data required for further optimization, BP neural network has been applied to substitute the simulation model to predict the relative parameters in the global optimization model. The optimization model has been set up based on NSGA-II, and the Pareto optimal results have been achieved. In this work, the prototype natural gas process plant is located in the southwest of China.
2 Process description and simulation
2.1 Process flow of high-sulfur natural gas purification plant
2.2 Operating conditions and specification
Operating data of the natural gas purification plant
Unit | Items | Value | Unit | Items | Value |
---|---|---|---|---|---|
GSU | MDEA temperature into the second absorption tower, °C | 39 | SRU | Claus furnace temperature, °C | 1070 |
MDEA mass concentration, wt% | 50 | Claus furnace air flow, kNm^{3}/h | 38 | ||
MDEA circulation rate, t/h | 570 | Tail gas H_{2}S/SO_{2} ratio | 4 | ||
Regeneration tower reflux ratio, % | 0.89 | LP/MP pressure, MPa/MPa | 0.45/3.5 | ||
DU | TEG temperature into the absorption tower, °C | 55 | TGTU | Combustion air flow, kNm^{3}/h | 16 |
TEG mass concentration, % | 99.66 | Quenched water circulation, t/h | 524 | ||
TEG circulation rate, t/h | 3.8 | AWSU | Stripper column temperature, °C | 106 |
2.3 Process simulation model
Comparison between simulation and plant data
Parameter, mol% | Operating data | Simulation data | AD, % |
---|---|---|---|
H_{2}S content of product gas | 0.00037 | 0.00038 | 2.70 |
H_{2}O content of product gas | 0.0036 | 0.0035 | 2.78 |
CO_{2} content of product gas | 2.03 | 2.03 | 0 |
MDEA content of lean amine | 13.125 | 13.126 | 0.008 |
MDEA content of semi-lean amine | 12.819 | 12.818 | 0.008 |
SO_{2} content of tail gas of Claus reaction | 0.33 | 0.29 | 7.50 |
H_{2}S content of exhaust gas | 0.00 | 0.00 | 0.00 |
SO_{2} content of exhaust gas | 0.021 | 0.0205 | 2.50 |
3 Determination of decision variables for optimization model
3.1 Analysis of energy consumption distribution
Comprehensive energy consumption distribution
Unit | CEC, MJ/10^{4} Nm^{3} | Fuel gas, MJ/10^{4} Nm^{3} | Electricity, MJ/10^{4} Nm^{3} | MPS, MJ/10^{4} Nm^{3} | LPS, MJ/10^{4} Nm^{3} | Water, MJ/10^{4} Nm^{3} |
---|---|---|---|---|---|---|
GSU | 17,136 | 0 | 737 | 917 | 14,861 | 621 |
DU | 172 | 83 | 11 | 78 | 0 | 0 |
SRU | − 16,025 | 0 | 0 | − 13,413 | − 2936 | 324 |
TGTU | 5029 | 6864 | 262 | − 408 | − 2056 | 367 |
AWSU | 410 | 0 | 6 | 0 | 371 | 33 |
Total | 6723 | 6947 | 1016 | − 12,826 | 10,240 | 1345 |
In Eq. (2), E_{C} represents comprehensive energy consumption (MJ/10^{4} Nm^{3}); m_{FNG} is the flow rate of feed natural gas under the standard condition (10^{4} Nm^{3}/h); E_{E}, E_{FG}, E_{LPS}, E_{MPS} and E_{W} represent the energy of electricity (kW), fuel gas (t/h), low-pressure steam (t/h), medium-pressure steam (t/h) and water (t/h), respectively; c_{1}–c_{5} represent corresponding equivalent coefficient of energy transfer medium.
The results show that GSU and TGTU are high comprehensive energy consumption units, and GSU takes about 75% of the TCEC in the four energy-consuming units. In GSU and TGTU, low-pressure steam for amine regeneration and fuel gas for hydrogenation reduction and tail gas combustion are main energy consumption, respectively. SRU is an exothermic unit with a large amount of 3.5 MPa and 0.4 MPa steam as by-product. DU and AWSU are low energy consumption units by contrast. So, the steam used by the purification process mainly relies on the steam derived from heat recovery of exothermic devices. And some pumps are also driven by the by-product steam.
3.2 Decision variables
Selected key operating parameter
Unit | Optimization variable | Range |
---|---|---|
GSU | The temperature of the amine solution into the second absorption tower, V_{1}, °C | 35–45 |
Amine solution circulation rate, V_{2}, t/h | 540–600 | |
Amine solution concentration, V_{3}, wt% | 45–54 | |
Regeneration tower reflux ratio, V_{4} | 0.80–1 | |
SRU | Tail gas H_{2}S/SO_{2} ratio, V_{5} | 2–6 |
TGTU | The combustion air flow, V_{6}, kNm^{3}/h | 18–27 |
Quenched water circulation, V_{7}, t/h | 400–540 |
4 Optimization model based on BP and NSGA-II
4.1 The objective function and constraints
The objective functions of optimization are to minimize the TCEC and maximize the PGPR, which can be expressed mathematically as follows:
In Eq. (4), m_{PNG} and m_{FNG} represent the volume flow rate of purified natural gas and feed natural gas under the standard condition (10^{4} Nm^{3}/h), respectively.
The quality specification (S_{1}–S_{5}) that the treated gas should meet is regarded as the inequality constraint. The input variables are the seven key operating parameters V_{1}–V_{7} in Table 4.
4.2 Frame diagram of optimization process
5 Results and discussion
5.1 Design of training and testing data
The UD method is based on the quasi-Monte Carlo method and proposed by Fang (1994). UD designs the experimental points to be scattered uniformly within the experimental domain. Compared with other conventional statistical experiment methods, UD is an effective approach and can further reduce the number of experimental trials which is proportional to the number of factor level (Fang and Ma 2001). For example, if a case with m factors (e.g., x_{1}, x_{2}, x_{m}) and s levels for each factor, the comprehensive needs m^{s} experiments and orthogonal design method needs s^{2} experiments, while UD only needs number of s experiments. The UD tables for designing different experiment trials have been introduced in the literature (Fang, 1994).
Training and testing data for BP model
No. | Input variable | ||||||
---|---|---|---|---|---|---|---|
V_{1}, °C | V_{2}, t/h | V_{3}, wt% | V_{4} | V_{5} | V_{6}, kNm^{3}/h | V_{7}, t/h | |
1 | 35.0 | 552 | 47.70 | 0.88 | 4.40 | 25.20 | 526 |
2 | 35.5 | 567 | 50.85 | 0.97 | 2.60 | 22.95 | 505 |
3 | 36.0 | 582 | 54.00 | 0.84 | 5.20 | 20.70 | 484 |
4 | 36.5 | 597 | 47.25 | 0.93 | 3.40 | 18.45 | 463 |
5 | 37.0 | 546 | 50.40 | 0.80 | 6.00 | 26.10 | 442 |
6 | 37.5 | 561 | 53.55 | 0.89 | 4.20 | 23.85 | 421 |
7 | 38.0 | 576 | 46.80 | 0.98 | 2.40 | 21.60 | 400 |
8 | 38.5 | 591 | 49.95 | 0.85 | 5.00 | 19.35 | 533 |
9 | 39.0 | 540 | 53.10 | 0.94 | 3.20 | 27.00 | 512 |
10 | 39.5 | 555 | 46.35 | 0.81 | 5.80 | 24.75 | 491 |
11 | 40.0 | 570 | 49.50 | 0.90 | 4.00 | 22.50 | 470 |
12 | 40.5 | 585 | 52.65 | 0.99 | 2.20 | 20.25 | 449 |
13 | 41.0 | 600 | 45.90 | 0.86 | 4.80 | 18.00 | 484 |
14 | 41.5 | 549 | 49.05 | 0.95 | 3.00 | 25.65 | 407 |
15 | 42.0 | 564 | 52.20 | 0.82 | 5.60 | 23.40 | 540 |
16 | 42.5 | 579 | 45.45 | 0.91 | 3.80 | 21.15 | 519 |
17 | 43.0 | 594 | 48.60 | 1.00 | 2.00 | 18.90 | 498 |
18 | 43.5 | 543 | 51.75 | 0.87 | 4.60 | 26.55 | 477 |
19 | 44.0 | 558 | 45.00 | 0.96 | 2.80 | 24.30 | 456 |
20 | 44.5 | 573 | 48.15 | 0.83 | 5.40 | 22.05 | 435 |
21 | 45.0 | 588 | 51.30 | 0.92 | 3.60 | 19.80 | 414 |
Base | 39.0 | 570 | 49.60 | 0.89 | 4.00 | 23.00 | 524 |
To further evaluate reliability of the BP model, different sample sizes such as 50, 100, 300 and 600 have been used. For all the sample sizes, 90% data have been used to train the model, while the left, i.e., 10%, to test the accuracy of the model.
Training and testing data for BP model
No. | Output parameter | ||||||
---|---|---|---|---|---|---|---|
S_{1}, mg/Nm^{3} | S_{2}, V % | S_{3}, °C | S_{4}, % | S_{5}, mg/Nm^{3} | TCEC, MJ/10^{4}Nm^{3} | PGPR, % | |
1 | 5.82 | 2.02 | − 20.45 | 99.883 | 759 | 6206 | 78.097 |
2 | 5.97 | 2.06 | − 20.07 | 99.882 | 787 | 6044 | 78.154 |
3 | 5.87 | 2.09 | − 19.73 | 99.886 | 785 | 6497 | 78.147 |
4 | 5.89 | 2.00 | − 19.06 | 99.877 | 865 | 7177 | 78.024 |
5 | 5.83 | 2.06 | − 18.49 | 99.881 | 750 | 6259 | 78.178 |
6 | 5.97 | 2.10 | − 18.14 | 99.884 | 760 | 6262 | 78.185 |
7 | 5.89 | 1.98 | − 17.29 | 99.875 | 836 | 6941 | 78.021 |
8 | 5.89 | 2.04 | − 16.93 | 99.878 | 838 | 7193 | 78.089 |
9 | 5.82 | 2.10 | − 16.40 | 99.885 | 845 | 5818 | 78.215 |
10 | 5.93 | 1.99 | − 15.86 | 99.869 | 824 | 6915 | 78.062 |
11 | 5.92 | 2.02 | − 15.50 | 99.875 | 819 | 6948 | 78.073 |
12 | 5.90 | 2.06 | − 15.09 | 99.881 | 812 | 7008 | 78.076 |
13 | 5.71 | 1.94 | − 16.45 | 99.871 | 886 | 8092 | 77.945 |
14 | 5.99 | 2.02 | − 15.85 | 99.873 | 797 | 6756 | 78.119 |
15 | 5.97 | 2.06 | − 15.44 | 99.876 | 800 | 6981 | 78.124 |
16 | 5.97 | 1.95 | − 15.22 | 99.864 | 891 | 7655 | 77.983 |
17 | 5.89 | 1.98 | − 15.10 | 99.876 | 858 | 7720 | 77.980 |
18 | 5.87 | 2.06 | − 15.49 | 99.875 | 766 | 6879 | 78.167 |
19 | 5.98 | 1.95 | − 15.43 | 99.861 | 869 | 7504 | 78.022 |
20 | 5.99 | 1.98 | − 15.31 | 99.863 | 881 | 7737 | 78.024 |
21 | 5.96 | 2.01 | − 15.14 | 99.871 | 861 | 7804 | 78.025 |
Base | 5.84 | 2.03 | − 16.42 | 99.879 | 876 | 6722 | 78.081 |
5.2 Prediction of BP neural network model
There are multitudes of different types of artificial neural networks. The multilayer perceptron with the back-propagation of error algorithm is more popular and is used in this paper.
The MATLAB neural network toolbox was used to establish the BP neural network. The main function was set as ‘net = newff (minmax(P), [15,7], {‘tansig’, ‘logsig’}, ‘traingdx’)’. The input data were normalized to [− 1, + 1] by the ‘minmax’ function before training.
There are three layers (input, hidden and output layers) in the training algorithms. After a large amount of practice and comparison for different numbers of neurons, 15 neurons for the hidden layer were recommended. The transfer functions ‘tansig’ for input-hidden layer and ‘logsig’ for hidden-output layer were applied. The ‘traingdx’ was applied as training function. The ‘sim(net, P_test)’ was constructed as the prediction function. The input variables of the BP neural network were the seven operating parameters in Table 5, while the seven output variables were the quality specification (S_{1}–S_{5}), TCEC and PGPR in Table 6.
Comparison between BP prediction and ProMax simulation
Parameter | No. 7 | No. 14 | ||||
---|---|---|---|---|---|---|
Simulation | Prediction | AD, % | Simulation | Prediction | AD, % | |
S_{1} | 5.89 | 5.85 | 0.533 | 5.99 | 5.79 | 3.33 |
S_{2} | 1.98 | 1.96 | 1.20 | 2.02 | 2.01 | 0.363 |
S_{3} | − 17.29 | − 17.75 | 2.66 | − 15.85 | − 15.74 | 0.695 |
S_{4} | 99.875 | 99.869 | 0.006 | 99.873 | 99.875 | 0.002 |
S_{5} | 836 | 819 | 2.04 | 797 | 785 | 1.52 |
Obj. 1 | 6941 | 6961 | 0.289 | 6756 | 6744 | 0.183 |
Obj. 2 | 78.02 | 78.00 | 0.027 | 78.11 | 78.07 | 0.051 |
No. 21 | Base | |||||
---|---|---|---|---|---|---|
S_{1} | 5.96 | 5.99 | 0.461 | 5.84 | 5.98 | 2.38 |
S_{2} | 2.01 | 1.99 | 0.889 | 2.03 | 2.03 | 0.110 |
S_{3} | − 15.14 | − 15.12 | 0.137 | − 16.42 | − 16.01 | 2.48 |
S_{4} | 99.871 | 99.875 | 0.004 | 99.879 | 99.874 | 0.006 |
S_{5} | 861 | 878 | 1.93 | 876 | 840 | 4.13 |
Obj. 1 | 7804 | 7502 | 3.87 | 6723 | 6646 | 1.14 |
Obj. 2 | 78.02 | 77.98 | 0.055 | 78.08 | 78.10 | 0.027 |
5.3 Pareto optimal solutions
The optimization goal is to minimize TCEC and maximize the PGPR. Since two objective functions cannot be optimized simultaneously, the non-dominated sorting genetic algorithm (NSGA-II) has been applied to achieve the Pareto optimal solutions. The real-coded NSGA-II has been implemented in MATLAB. In the NSGA-II model, the values of population size, maximum generations and tournament pool size are 100, 300 and 2, respectively; replace proportion, crossover probability and mutation probability are 0.9, 0.9 and 0.14, respectively; the crossover method and mutation method are two points and selective.
Selected Pareto optimization results for TCEC and PGPR
No. | V_{1}, °C | V_{2}, t/h | V_{3}, wt% | V_{4} | V_{5} | V_{6}, kNm^{3}/h | V_{7}, t/h |
---|---|---|---|---|---|---|---|
A | 35.0 | 540 | 54.00 | 1.00 | 2.35 | 27.00 | 475 |
B | 36.2 | 553 | 54.00 | 1.00 | 2.32 | 27.00 | 474 |
C | 37.9 | 562 | 54.00 | 1.00 | 2.51 | 27.00 | 476 |
No. | NSGA-II | ProMax | ||
---|---|---|---|---|
TCEC, MJ/10^{4} Nm^{3} | PGPR, % | TCEC, MJ/10^{4} Nm^{3} | PGPR, % | |
A | 5819 | 78.2078 | 5699 | 78.2391 |
B | 5822 | 78.2090 | 5721 | 78.2421 |
C | 5826 | 78.2093 | 5780 | 78.2469 |
6 Conclusions
The optimization of a whole high-sulfur natural gas purification plant was investigated to reduce the TCEC and further improve the PGPR. To provide sufficient data and determine key operating parameters, a steady-state simulation model of high-sulfur natural gas purification process including GSU, DU, SRU, TGTU and AWSU was set up by using software ProMax. And the model has good agreement compared with the actual operating data. Seven key operating parameters, as input variables in optimization model, were determined based on the analysis of comprehensive energy consumption distribution. To solve the problem that the process model did not converge in some conditions for the existence of nested loops in it, BP neural network was applied to substitute the simulation model to predict the relative parameters needed in the optimization model. The UD method and the table U_{21} (10^{7}) were applied to design the experiment points for training and testing BP model. The BP model can offer high prediction accuracy compared with the simulation model. NSGA-II was established to optimize the two objectives, and 100 Pareto optimal solutions were achieved. Three optimal points A, B and C were selected and evaluated further. The TCEC was reduced by 13.4% and the PGPR was improved by 0.2% compared with the Base operation condition.
Further investigation will be carried out to verify the stability and reliability by using much more plant operating data. It demonstrates the applicability and feasibility of BP neural network and NSGA-II to resolve global optimization problem in the field of the high-sulfur gas purification plant.
Notes
Acknowledgements
Financial support from National Science and Technology Major Project of the Ministry of Science and Technology of China (Grant No. 2016ZX05017-004) is gratefully acknowledged.
References
- Abdulrahman RK, Sebastine IM. Natural gas sweetening process simulation and optimization: a case study of Khurmala field in Iraqi Kurdistan region. J Nat Gas Sci Eng. 2013;14:116–20. https://doi.org/10.1016/j.jngse.2013.06.005.CrossRefGoogle Scholar
- Adewale R, Salem DJ, Berrouk AS, Dara S. Simulation of hydrogen production from thermal decomposition of hydrogen sulfide in sulfur recovery units. J Clean Prod. 2016;112:4815–25. https://doi.org/10.1016/j.jclepro.2015.08.021.CrossRefGoogle Scholar
- Ahmadi MA, Soleimani R, Bahadori A. A computational intelligence scheme for prediction equilibrium water dew point of natural gas in TEG dehydration systems. Fuel. 2014;137:145–54. https://doi.org/10.1016/j.fuel.2014.07.072.CrossRefGoogle Scholar
- Al-Lagtah NMA, Al-Habsi S, Onaizi SA. Optimization and performance improvement of Lekhwair natural gas sweetening plant using Aspen HYSYS. J Nat Gas Sci Eng. 2015;26:367–81. https://doi.org/10.1016/j.jngse.2015.06.030.CrossRefGoogle Scholar
- Azizkhani JS, Jazayeri-Rad H, Nabhani N. Design of an ensemble neural network to improve the identification performance of a gas sweetening plant using the negative correlation learning and genetic algorithm. J Nat Gas Sci Eng. 2014;21:26–39. https://doi.org/10.1016/j.jngse.2014.07.012.CrossRefGoogle Scholar
- Behroozsarand A, Zamaniyan A. Multi-objective optimization scheme for industrial synthesis gas sweetening plant in GTL process. J Nat Gas Chem. 2011;20:99–109. https://doi.org/10.1016/S1003-9953(10)60153-3.CrossRefGoogle Scholar
- Boyaghchi FA, Chavoshi M. Multi-criteria optimization of a micro solar-geothermal CCHP system applying water/CuO nanofluid based on exergy, exergoeconomic and exergoenvironmental concepts. Appl Therm Eng. 2017;112:660–75. https://doi.org/10.1016/j.applthermaleng.2016.10.139.CrossRefGoogle Scholar
- Cui YF, Geng ZQ, Zhu QX, Han Y. Review: multi-objective optimization methods and application in energy saving. Energy. 2017;125:681–704. https://doi.org/10.1016/j.energy.2017.02.174.CrossRefGoogle Scholar
- Dai XH, Li MQ, Kou JS. Survey on the theory of genetic algorithms. Control Decis. 2000;15(3):263–8 (in Chinese).Google Scholar
- Damavandi MD, Forouzanmehr M, Safikhani H. Modeling and Pareto based multi-objective optimization of wavy fin-and-elliptical tube heat exchangers using CFD and NSGA-II algorithm. Appl Therm Eng. 2017;111:325–39. https://doi.org/10.1016/j.applthermaleng.2016.09.120.CrossRefGoogle Scholar
- Darwish NA, Hilal N. Sensitivity analysis and faults diagnosis using artificial neural networks in natural gas TEG-dehydration plants. Chem Eng J. 2008;137:189–97. https://doi.org/10.1016/j.cej.2007.04.008.CrossRefGoogle Scholar
- Deb K, Agrawal S, Pratap A, Meyarivan T. A fast elitist non-dominated sorting genetic algorithm for multi-objective optimization: NSGA_II. In: Sixth international conference on parallel problem solving from nature 2000. p. 849–58. https://doi.org/10.1007/3-540-45356-3_83.CrossRefGoogle Scholar
- Deng QH, Wang D, Zhao H, Huang WT, Shao S, Feng ZP. Study on performances of supercritical CO_{2} recompression Brayton cycles with multi-objective optimization. Appl Therm Eng. 2017;114:1335–42. https://doi.org/10.1016/j.applthermaleng.2016.11.055.CrossRefGoogle Scholar
- Fang KT. Uniform design and uniform design tables. Beijing: Science Press; 1994 (in Chinese).Google Scholar
- Fang KT, Ma CX. Orthogonal and uniform experimental design. Beijing: Science Press; 2001 (in Chinese).Google Scholar
- Foo DCY, Ooi REH, Tan RR, Lee JY. Process integration approaches to optimal planning of unconventional gas field development. Chem Eng Sci. 2016;150:85–93. https://doi.org/10.1016//j.ces.2016.04.049.CrossRefGoogle Scholar
- Ghiasi MM, Bahadori A, Zendehboudi S. Estimation of triethylene glycol (TEG) purity in natural gas dehydration units using fuzzy neural network. J Nat Gas Sci Eng. 2014;17:26–32. https://doi.org/10.1016/j.jngse.2013.12.008.CrossRefGoogle Scholar
- Guo QL. Evaluation and decision analysis of oil and gas exploration targets. Beijing: Petroleum Industry Press; 2004 (in Chinese).Google Scholar
- He C, You FQ. Towards more cost-effective and greener chemicals production from shale gas by integrating with bioethanol dehydration: novel process design and optimization. AIChE J. 2015;61(4):1184–208. https://doi.org/10.1002/aic.14713.CrossRefGoogle Scholar
- Hu SP. Process modeling and parameter optimization of high-sulfur natural gas purification plant. Beijing: China University of Petroleum; 2013 (in Chinese).Google Scholar
- Jassim MS. Sensitivity analyses and optimization of a gas sweetening plant for hydrogen sulfide and carbon dioxide capture using methyldiethanolamine solutions. J Nat Gas Sci Eng. 2016;36:175–83. https://doi.org/10.1016/j.jngse.2016.10.012.CrossRefGoogle Scholar
- Jia QF, Liu HJ, Zhang L, Tong LZ, et al. Study of process optimization on Qiudong 1# gas treatment unit. Chem Eng Oil Gas. 2009;38(5):386–9 (in Chinese).Google Scholar
- Li Q. Energy analysis and optimization for high sour natural gas purification plant. Beijing: China University of Petroleum; 2012 (in Chinese).Google Scholar
- Li YB, Li XF, Yao YD. Variation laws of H_{2}S concentration in the development process of sour gas reservoir. Acta Petrolei Sinica. 2007;28(6):99–102 (in Chinese).Google Scholar
- Long ZB, Liu J, Wu X. Process schemes selection and simulation evaluation of high sour natural gas desulphurization. Technol Dev Chem Ind. 2007;36(12):28–32 (in Chinese).Google Scholar
- Louis G, Maxime TG, Francois MP. Review of utilization of genetic algorithms in heat transfer problems. Int J Heat Mass Transf. 2009;52(4):2169–88. https://doi.org/10.1016/j.ijheatmasstransfer.2008.11.015.CrossRefGoogle Scholar
- Manenti F, Papasidero D, Bozzano G, Ranzi E. Model-based optimization of sulfur recovery units. Comput Chem Eng. 2014;66:244–51. https://doi.org/10.1016/j.compchemeng.2014.01.019.CrossRefGoogle Scholar
- National Energy Administration, Development Research Center of the State Council and Ministry of Natural Resources of the People’s Republic of China. In: China natural gas development report. Beijing: Petroleum Industry Press; 2016 (in Chinese).Google Scholar
- Neri F, Carlos C. Memetic algorithms and memetic computing optimization: a literature review. Swarm Evol Comput. 2012;2:1–14. https://doi.org/10.1016/j.swevo.2011.11.003.CrossRefGoogle Scholar
- Peter L, Hussain A, Follmann M, Melin T, Hägg MB. CO_{2} removal from natural gas by employing amine absorption and membrane technology—a technical and economical analysis. Chem Eng J. 2011;172:952–60. https://doi.org/10.1016/j.cej.2011.07.007.CrossRefGoogle Scholar
- Qiu K, Shang JF, Ozturk M, et al. Studies of methyldiethanolamine process simulation and parameters optimization for high-sulfur gas sweetening. J Nat Gas Sci Eng. 2014;21:379–85. https://doi.org/10.1016/j.jngse.2014.08.023.CrossRefGoogle Scholar
- Rouzbahani AN, Bahmani M, Shariati J, Tohidian T, Rahimpour MR. Simulation, optimization, and sensitivity analysis of a natural gas dehydration unit. J Nat Gas Sci Eng. 2014;21:159–69. https://doi.org/10.1016/j.jngse.2014.07.025.CrossRefGoogle Scholar
- Salooki MK, Abedini R, Adib H, Koolivand H. Design of neural network for manipulating gas refinery sweetening regenerator column outputs. Sep Purif Technol. 2011;82:1–9. https://doi.org/10.1016/j.seppur.2011.07.015.CrossRefGoogle Scholar
- Santos MGRS, Correia LMS, de Medeiros JL, de Queiroz FAO. Natural gas dehydration by molecular sieve in offshore plants: impact of increasing carbon dioxide content. Energy Convers Manag. 2017;149:760–73. https://doi.org/10.1016/j.enconman.2017.03.005.CrossRefGoogle Scholar
- Singh K, Das R. An experimental and multi-objective optimization study of a forced draft cooling tower with different fills. Energy Convers Manag. 2016;111:417–30. https://doi.org/10.1016/j.enconman.2015.12.080.CrossRefGoogle Scholar
- Srinivas N, Deb K. Multi-objective optimization using non-dominated sorting in genetic algorithms. Evol Comput. 1994;2:221–48. https://doi.org/10.1162/evco.1994.2.3.221.CrossRefGoogle Scholar
- Wahedi YA, Torres AI, Hashimi SA, Dowling NI, Daoutidis P, Tsapatsis M. Economic assessment of temperature swing adsorption systems as claus tail gas clean up units. Chem Eng Sci. 2015;126:186–95. https://doi.org/10.1016/j.ces.2014.12.015.CrossRefGoogle Scholar
- Wang KY. Natural gas purification process-desulfurization, decarbonization, dehydration, sulfur recovery and the tail gas treatment. Beijing: Petroleum Industry Press; 2005 (in Chinese).Google Scholar
- Yu HM, Yao PJ, Yuan Y. Improved genetic algorithm/simulated annealing for large system energy integration. J Chem Ind Eng. 1998;49(6):655–61 (in Chinese).Google Scholar
- Zarei S, Ganji H, Sadi M, Rashidzadeh M. Thermo-kinetic modeling and optimization of the sulfur recovery unit thermal stage. Appl Therm Eng. 2016;103:1095–104. https://doi.org/10.1016/j.applthermaleng.2016.05.012.CrossRefGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.