Regular Article
Predictive models for density correction factor of natural gas and comparison with standard methods
Faculty of Petroleum and Chemical Engineering, Razi University, Kermanshah 6714967346, Iran
^{*} Corresponding author: f.bashipour@razi.ac.ir
Received:
14
October
2018
Accepted:
24
January
2019
Two intelligentbased models which do not require complete gas compositions are presented to estimate natural gas density correction factor using comprehensive datasets (nearly 60 000 instances) originating from the AGA8DCM (Detail Characterization Method) standard: (1) NGDCANN model (Natural Gas Density Calculator based on Artificial Neural Network) and (2) AGA8GCMD model (Gross Characterization Method Developed by applying genetic algorithm technique). In the suggested models, only five input variables (specific gravity at base condition, operating temperature and pressure and molar composition of CO_{2} and N_{2}) are employed. The experimental datasets obtained from this work (68 instances) and literature (505 instances) are applied to validate the developed model showing a very good agreement between experimental and estimated data. Simplicity, improving accuracy and satisfactory results of the suggested models over a wide range of operational conditions show that these models would be excellent alternatives for the traditional standard methods, so that, the NGDCANN model prediction besides of its simplicity to use show the highest accuracy over a wide of operational range in comparison to similar models.
© F. Bashipour & B. Hojjati, published by IFP Energies nouvelles, 2019
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1 Introduction
In recent years, significant attention has been paid to developing an easy and accurate measurement methods of natural gas mass flow rate in gas industries especially in European and Asian countries. As an example, Compressed Natural Gas (CNG) is sold at the marketing level by mass that would lead to noticeably economic loss [1]. Thus, density metering along with volume flow metering which needs less cost is essential in natural gas marketing. Devices such as multiple ultrasonic transienttime meters and conventional orifice plates are employed to measure the volume flow rate of natural gas [2]. Volumetric meters cannot account with changing gas composition, pressure, and temperature. Moreover, direct measurement of natural gas density is difficult as it needs highly expert staffs and costly instruments such as Coriolis density meters, gas chromatographs, etc. [3, 4]. Also these instruments have their own problems; the potential of erosion, sensitivity to pulsation and vibration close to operating frequency, and requirements of regular calibration [5–8].
Physical and thermodynamic properties of natural gas mixture could be obtained by employing Equation of State (EOS) with acceptable accuracy and without spending much time. To estimate these properties in gas industries, real time natural gas properties such as temperature, pressure, gas compositions, specific gravity, etc. are needed [9, 10]. Though, composition measurements are costly and may not be applicable, it is necessary to develop correlation based on real time measurable property. Various EOS or correlations have been suggested for computing natural gas density over a wide range of temperature, pressure and composition; either in terms of accuracy such as GERG2008 and AGA8DCM (Detail Characterization Method) or with respect to simplicity in calculation such as NX19 and AGA8GCM (Gross Characterization Method) [4]. Dranchuk and AbouKassem [11] developed a gas density calculator correlation employing 1500 data points. Londono et al. [12] reported simplified correlations for estimating the density of natural gas, while AlQuraishi and Shokir [3] reported a new equation for computing the density of hydrocarbon gases using Alternating Conditional Expectations (ACE) algorithm. FarzanehGord et al. [13] and FarzanehGord and Rahbari [14] developed novel correlations for calculating density of natural gas based on AGA8 EOS.
AGA8DCM is mostly employed to predict density or compressibility factor of natural gas at high accuracy [15, 16]. Though, it is complex and needs the natural gas mixture temperature, pressure, and composition. Pressure and temperature are measurable, while as mentioned before, the compositions could be measured by utilizing experimental facilities such as gas chromatography and it depends on each measurement site and compositions of crude oil differ from region to region which resulting erroneous density.
The complexity of AGA8DCM EOS makes it difficult to apply, especially for mixtures with large number of components. On the other hand, NX19 and AGA8GCM EOSs are simpler to implement and need less input information; temperature, pressure, specific gravity at base condition, and only compositions of N_{2} and CO_{2} of natural gas. By using these EOSs, density and compressibility factor cannot be calculated as high accurate and wide ranges of temperature and pressure as using AGA8DCM EOS.
Representing of a novel model and modifying other methods for natural gas density calculation implementing intelligentbased techniques is the goal of this work. These models are more efficient in such cases which involve nonlinear mathematical modeling along time consumption, and when there is not any significant relation between input and output of a system [17–19]. The intelligent models developed in this study are Artificial Neural Network (ANN) and Genetic Algorithms (GAs) Using Matlab (8.1.0.604).
This work tries to present a new Natural Gas Density Calculator based on ANN (NGDCANN) in order to keep high accuracy of AGA8DCM model for natural gas density prediction over a wide range of operational variables. Moreover, the required natural gas data for density estimation is reduced to those applied for less accurate models such as AGA8GCM and NX19 containing specific gravity at base condition, operating temperature and pressure and molar composition of CO_{2} and N_{2}. Because of less required variables, the less accurate models are widely applied in estimation software of natural gas mass flow rate. Therefore, by the help of Genetic Algorithm and introducing a tuning coefficient, this work is trying to increase the prediction accuracy of AGA8GCM EOS close to accuracy and operational range of AGA8DCM EOS.
The performance accuracy and effectiveness of proposed models are tested against reported experimental measurements from this work and literature by statistical techniques for error expression like the coefficient of determination (R ^{2}), Mean Square Error (MSE) and Mean Absolute Error (MAE).
2 Theory
2.1 Artificial Neural Network (ANN) model
ANN is an informationprocessing paradigm that takes numeric inputs, performs computations on those, and outputs one or more numeric values while no need any preassumptions about the relationships between them. ANNs are inspired by biological nervous systems such as brain and consist of elements that receive inputs and generate a single output, where the output is a function of the inputs. ANN has large numbers of computational units connected in a massively parallel structure and do not require a mathematical formulation or physical relationships of the handled problem [20, 21]. This property is a significant advantage of the flexible ANN model to predict complicated systems. ANN structure consists of several parallelinterconnected units called neurons in one or more hidden layers. A neuron sums inputs multiplied by their respective weight and then applies a transfer function to the sum. The connection of each neuron with other neurons makes the next layer [22, 23]. The pattern of interconnection between neurons is called the network architecture. MultiLayer Feed Forward (MLFF) networks are the most common architecture of ANN for static regression applications, consisting of one input layer, one output layer, and one or more intermediate or hidden layer(s) [24–28]. Figure 1 shows the MLFF network involving an input layer with five neurons, an output layer with one neuron, and two hidden layer. An ANN is considered an adaptive system, i.e., each parameter is changed during its operation and is trained to recognize the process under investigation.
Fig. 1 Architecture of the fourlayered MLFF network. 
An MLFF network can be represented as follows:
where is the neuron j’s output from layer k, b _{ jk } is the bias weight for neuron j in layer k, and w _{ ijk } (the modelfitting parameters) are the randomly selected connection weights. F _{ k } is the nonlinear activation transfer function, which is one of the main characteristic elements of an ANN with the most common type of sigmoidal transfer functions. According to equation (2), input and output values were employed as normalized values in the range of 0–1 to gain higher ANN performance and consistent results [29, 30]:
where , , and are the normalized, actual, maximum, and minimum values of variable X, respectively.
In this study, NGDCANN model is presented as a replacement for AGA8DCM EOS. Natural gas density (as an output of this model) is estimated by requiring less input variables. Table 1 presents the input variables range of NGDCANN model equaled with normal ranges of input variables of AGA8DCM EOS.
Employed ranges of input variables of NGDCANN model.
As shown in Figure 1, the output of the MLFF ANN model, dimensionless density correction factor (C) (an estimated value by AGA8DCM called C _{DCM}) can be related to natural gas density as:
where ρ is natural gas density at operating temperature (T) and pressure (P). ρ _{b}, , and SG_{b} are natural gas density, air density, and specific gravity at base temperature (T _{b} = 60 °F) and pressure (P _{b} = 14.73 psia), respectively.
2.2 Genetic Algorithms (GAs)
Genetic Algorithm (GA) is a search heuristic in computer science which was first introduced by J. Holland in order to solve optimization problems [31]. Based on Darwin’s theory, the species of organisms have evolved over a long period of time through natural selection while all of them share a common ancestor [18, 32]. GA is a particular class of evolutionary algorithms such as inheritance, mutation, selection, and crossover (also called recombination) [33]. GA starts from an initial population of randomly generated individuals and proceeds in an iterative process resembling the genome evolution. In each generation, the fitness of every individual in the population is evaluated, multiple individuals are selected from the current population (based on their fitness), and modified to form a new population using in the next iteration of the algorithm. The process terminates when either a maximum number of generations has been produced, or a satisfactory fitness level has been reached for the population.
By utilizing GA technique, a tuning coefficient γ is introduced. This coefficient is a function of input variables used for AGA8GCM EOS. By the help of this tuning coefficient, this modified EOS can be applied over a wider range of input variables of AGA8GCM EOS with higher accuracy:
where, unknown variables V _{ j } are obtainable by regression of data resulted from the solution of equations of AGA8DCM and AGA8GCM. The unknown variables can be estimated using GA technique and MSE as a fitness function.
3 Material and methods
3.1 Materials
The natural gas applied in the experiments was supplied from Gas Company of Isfahan Province’s (Isfahan, Iran). The normalized molar composition of the natural gas was determined by a gas chromatograph (Elster EnCal 3000): 90.81% methane, 3.20% ethane, 0.93% propane, 0.22% nbutane, 0.16% isobutane, 0.03% npentane, 0.04% isopentane, 3.93% Nitrogen, 0.67% Carbon dioxide, and 0.01% C_{6+}. SG_{b} of the natural gas was determined 0.609 at T _{b} of 60 °F and P _{b} of 14.73 psia.
3.2 Density correction factor measurement
The 68 experimental datasets containing the density correction factor of natural gas with above composition measured at various levels of five model input variables of T, P, SG_{b}, and (X _{ i }, i = 1–5). X _{ i }s were selected within the ranges presented in Table 1. The experimental datasets (68 instances) along with other experimental datasets from literature (505 instances) [34–37] were only employed to validate suggested models.
4 Results and discussion
4.1 NGDCANN model
To predict the density correction factor of natural gas, AGA8DCM model was replaced by a proposed NGDCANN model, which requires less input variables like AGA8GCM and NX19 EOSs. So that, NGDCANN model can keep high accuracy prediction of C and applicability over a wide range of operational variables like AGA8DCM EOS. The model was optimized by specific properties of ANNs including number of hidden layers (N _{L}), number of neurons in the hidden layers (), transfer function, and BackPropagation Learning Algorithm (BPLA) as described in the following section. Performance of optimized model regarding to prediction results were evaluated and then compared with corresponding results obtained from other estimation methods such as AGA8GCM and NX19 as presented in Section 4.1.2.
4.1.1 Optimization of NGDCANN model
The optimization of MLFF NGDCANN model was carried out in seven sectors: (1) determining the architecture of NGDCANN model, (2) determining the transfer functions, (3) determining , (4) determining the BPLA, (5) estimating the initial values of weights and biases for equation (1), (6) updating the weights and biases to obtain optimum ANN performance up to the minimum network error like MSE and complete the training process of ANN, (7) validation and test of ANN trained. In the ANN model constructed based on above seven steps, approximately 60 000 datasets resulted from AGA8DCM EOS were applied according to Figure 2. Seventy percent of the datasets were randomly utilized in the training process and the rest was used in the processes of test and validation. As mentioned previously, the experimental datasets (68 instances form this study and 505 instances from literature [34–37]) were applied to investigate the performance of NGDCANN model, and were not employed in the construction of the suggested models.
Fig. 2 Data gathering and grouping to apply in intelligent models. 
According to the universal approximation theory, an ANN model with a single hidden layer and a sufficiently large number of neurons can interpret any inputoutput neurons [38, 39]. However the results show that applying one hidden layer for this research data cannot adequately decrease MSE. Therefore as an optimum architecture of NGDCANN model, a MLFF network with one input layer, two hidden layers, and one output layer was considered in NGDCANN model as shown in Figure 1.
In order to select appropriate transfer function, the tansigmoid (tansig) and logsigmoid (logsig) transfer functions were applied at two hidden layers. According to the highlighted row of the Table 2, applying “tansig” transfer function at both of hidden layers and a linear transfer function (purelin) at the output layer resulting estimates with having less MSE. Tansig, logsig and purelin transfer functions are defined by the following equations:
Performance comparison of 4 NGDCANN model by applying different applied transfer functions in two hidden layers using = = 5 and LevenbergMarquardt backpropagation (trainlm) as a BPLA.
As mentioned before, the adjustable is one of the features of ANN architecture. Note that there is no individual method to select appropriate . Even though the neuron numbers can be determined by trial and error; in this study, Kolmogorov theory was employed to determine the neuron numbers for hidden layers, due to faster response and higher accuracy. The theory creates an upper limit for , so that it generates required to be less than two times of neuron numbers utilized for input layer of MLFF ANN. In fact, using more neurons in hidden layers leads to over fitting problems and weak performance in network predictions [20, 40]. Considering five neurons used in input layer of NGDCANN model, the network performance was evaluated by applying two to ten neurons for each hidden layer to obtain minimized MSE. Figure 3 compares performances of MLFF NGDCANNs by applying different and using constant conditions: tansig transfer function for hidden layers, purelin transfer function for output layer, and trainlm as BPLA. It can be inferred that NGDCANN model with eight and five neurons in the first (L _{1}) and the second (L _{2}) hidden layers, respectively, has the least MSE indicating the best performance.
Fig. 3 Effect of different in both of hidden layers (L _{1} and L _{2}) on performance of MLFF NGDCANN model to predict C _{DCM}. 
Another feature which plays an important role in enhancement learning rate of MLFF ANNs is the selection of an appropriate BPLA. The criteria to select a proper BPLA for MLFF ANN were to minimize MSE and maximize R ^{2}. Considering the criteria for eleven different BPLAs, the performance of MLFF ANN were compared with each other by employing tansig transfer function for first and second hidden layers, with eight and five neurons, respectively, and purelin transfer function for output layer. The results indicate that the maximum learning rate and minimum MSE can be achieved by using learning algorithm. Thus, in the present work, the MLFF ANN by applying BPLA function of trainlm was known as an optimized NGDCANN model to predict density correction factor of natural gas, according to the highlighted row of the Table 3.
Comparison of ANNs performance with 11 different BPLAs to predict C _{DCM}.
Ability of optimized NGDCANN model to predict C _{DCM} was investigated through comparisons between the results obtained from AGA8DCM EOS and corresponding data predicted by the model for three sectors (learning, validation, and test) in Figure 4. Based on the calculated R ^{2} (0.9995), it concluded that the NGDCANN model could accurately describe C _{DCM}.
Fig. 4 Comparison of predicted results by optimized NGDCANN model with (a) training set and (b) all data. 
As shown in Figure 5, for all three sectors of learning, validation, and test, the best compatibility of optimized NGDCANN model observes at epoch 479 with MSE value 5.715 × 10^{−5}.
Fig. 5 Data MSE of predicted C _{DCM} with optimized NGDCANN model. 
4.1.2 Investigation of NGDCANN model performance
To estimate natural gas properties such as compressibility factor, density, C, etc., NGDCANN, AGA8GCM, and NX19 models need similar input variables (T, P, SG_{b}, , ). Therefore, the validity of these models applied in their particular operating conditions was evaluated by comparing with experimental data from this work (68 instances) and Refs. [34–37] (505 instances) through applying the statistical analyses like R ^{2}, MSE, and MAE presented in Table 4. Note that the experimental data utilized in Table 4 were not used in construction of NGDCANN model. In comparison to other models, the results show that in addition to keeping the required number of input variables (e.g. T, P, SG_{b}, , ), the NGDCANN model shows the least error (MSE and MAE) in prediction of C within a wider range of operational conditions indicating the priority of this model. The NGDCANN model represents R ^{2} 0.996, MSE 102.0, and MAE 374.4 for 891 experimental data. The outcomes indicate that the NGDCANN model has satisfactory agreement with experimental data.
4.2 AGA8GCM EOS development
To improve the accuracy and to increase the applicability operational range of AGA8GCM EOS regarding to natural gas density prediction, a tuning coefficient was determined by GA technique, multiplied by C _{GCM} (Eq. (4)), and then tuned. The coefficient was presented as a power law function of required input variables of AGA8GCM EOS due to produce the least MSE in this mathematical formulation. The developed AGA8GCM EOS by applying the tuning coefficient has been called AGA8GCMD. The unknown coefficients of V _{ j } in equation (4) were calculated using GA technique considering MSE 7.9 × 10^{−6} (between C _{DCM} and γ _{GCM} C _{GCM}) as the fitness function to be minimized by using values in equation (8):
GA parameters applied in this technique were characterized as follows:
Population type: double vector; population size: 30; creation function: constraint dependent; generations: 1000; scaling function: rank; selection function: stochastic uniform; crossover function: scattered; crossover fraction: 0.8; mutation function: constraint dependent; hybrid function: fminsearch; other GA parameters were considered as Matlab (8.1.0.604) software defaults.
It is obvious that required input variables for AGA8GCMD model are similar to NGDCANN, AGA8GCM, and NX19 models. The predicted results of physical property C of natural gas resulted from these four models were compared in Table 4. For this comparison, various natural gases introduced at this work and Refs. [34–37] at different operational conditions specified to each predictive model were evaluated to estimate C by applying R ^{2}, MSE and MAE. Observing results indicate that, however AGA8GCM EOS applying lead to the higher R ^{2} and lower MSE and MAE, AGA8GCMD model has applying capability within wider operating ranges of T (225–390) and P (up to 70 MPa) with acceptable values of R ^{2} 0.980, MSE 365.85 and MAE 4.575. For example, if AGA8GCM EOS is utilized, it produces R ^{2} 0.702, MSE 4175.349 and MAE 10.388 which are not satisfactory values. These results represent that the AGA8GCMD model can predict more precisely C within wider operational temperature and pressure ranges than AGA8GCM and NX19 EOSs. However, the NGDCANN model still has the higher accuracy and applicability operational ranges rather than other models represented in Table 4.
Investigation results indicate that the presented intelligentbased models (NGDCANN and AGA8GCMD) have more advantages than EOS introduced previously (AGA8DCM, AGA8GCM and NX19): (1) simplicity of use, (2) covering wider ranges of operating conditions, (3) higher accuracy of natural gas properties estimation especially at high pressure or low temperature in other word higher C.
5 Conclusion
Prediction of compressible fluid density plays a key role in production, equipment design, processing, storage, marketing, and transportation of natural gas industries. Some methods and devices have been utilized to measure the gas mass flow rate; however, they have both their pros and cons. Based on available artificial intelligence methods and comprehensive datasets (nearly 60 000 instances), the present work studies NGDCANN and AGA8GCMD models to predict natural gas density in order to keep simplicity, high accuracy, less required input variables, and wider range of operational variables in comparison to other methods such as AGA8DCM, AGA8GCM, and NX19. The models are evaluated using comparison of experimental and predicted C data.
Using MLFF NGDCANN model, the number of required input variables are reduced from 23 to 5 items and the C prediction accuracy is improved (R ^{2}, MSE, and MAE of 0.989, 396.25, and 1.999, respectively, for 871 experimental data) over a wide range of operational variables as AGA8DCM EOS.
By applying a tuning coefficient to C _{GCM} of AGA8GCMD model, the accuracy of gas density prediction over wider operational ranges of temperature (225–390 K) and pressure (up to 70 MPa) is improved (R ^{2}, MSE, and MAE of 0.980, 365.85, and 4.575, respectively). However, in comparison to AGA8GCMD, AGA8GCM and NX19 models, the results show that, besides of its simplicity to use, the NGDCANN model prediction has the highest accuracy over a wide of operational range.
Acknowledgments
This work has been supported by Razi University grants.
References
 FarzanehGord M., Farsiani M., Khosravi A., Arabkoohsar A., Dashtie F. (2015) A novel method for calculating natural gas density based on joule Thomson coefficient, J. Nat. Gas Sci. Eng. 26, 1018–1029. doi: 10.1016/j.jngse.2015.07.029. [Google Scholar]
 Froysa K.E., Lunde P. (2005) Density and calorific value measurement in natural gas using ultrasonic flow meters, 23rd International North Sea Flow Measurement Workshop, 18–21 October 2005, Norway. [Google Scholar]
 AlQuraishi A.A., Shokir E.M. (2009) Viscosity and density correlations for hydrocarbon gases and pure and impure gas mixture, Pet. Sci. Technol. 27, 1674–1689. doi: 10.1080/10916460802456002. [CrossRef] [Google Scholar]
 Saeedi Dehaghani A.H., Badizad M.H. (2017) A soft computing approach for prediction of PρT behavior of natural gas using adaptive neurofuzzy inference system, Petroleum 3, 4, 447–453. doi: 10.1016/j.petlm.2016.12.004. [CrossRef] [Google Scholar]
 Hammond R.H. (2001) Ultrasonic measurement system with molecular weight determination, Patent US6216091B1. [Google Scholar]
 Watson J.W., White F.A. (1982) Acoustic measurement for gas BTU content, Oil Gas J. 80,14, 217–225. [Google Scholar]
 Dell’Isola M., Cannizzo M., Diritti M. (1982) Acoustic measurement for gas BTU content, Measurement 20, 2, 75–89. doi: 10.1016/S02632241(97)00016X. [CrossRef] [Google Scholar]
 Buonanno G., Carotenuto A., Dell’Isola M. (1998) The influence of reference condition correction on natural gas flow measurement, Measurement 23, 2, 77–91. doi: 10.1016/S02632241(98)000098. [CrossRef] [Google Scholar]
 Nishiumi H., Saito S. (1975) An improved generalized BWR equation of state applicable to low reduced temperatures, J. Chem. Eng. Jpn. 8, 5, 356–360. doi: 10.1252/jcej.8.356. [CrossRef] [Google Scholar]
 FarzanehGord M., Rahbari H. (2011) Developing novel correlations for calculating natural gas thermodynamic properties, Chem. Process Eng. 32, 4, 435–452. [CrossRef] [Google Scholar]
 Dranchuk P., AbouKassem J. (1975) Calculation of Zfactors for natural gases using equations of state, J. Can. Petrol. Technol. 14, 3, 14–34. doi: 10.2118/750303. [Google Scholar]
 Londono F., Aicher R., Blasingame T. (2002) Simplified correlations for hydrocarbon gas viscosity and gas density validation and correlation behavior using a large scale database, SPE Gas Technology Symposium, 30 April–2 May, Calgary, Canada. doi: 10.2118/75721MS. [Google Scholar]
 FarzanehGord M., Khamforoush A., Hashemi S., Pourkhadem N. (2010) Computing thermal properties of natural gas by utilizing AGA8 equation of state, IJCEA 1, 20–24. doi: 10.7763/IJCEA.2010.V1.4. [CrossRef] [Google Scholar]
 FarzanehGord M., Rahbari H. (2012) Numerical procedures for natural gas accurate thermodynamics properties calculation, J. Eng. Thermophys. 21, 4, 213–234. doi: 10.1134/S1810232812040017. [CrossRef] [Google Scholar]
 Kunz O., Wagner W. (2012) The GERG2008 widerange equation of state for natural gases and other mixtures: An expansion of GERG2004, J. Chem. Eng. Data 57, 11, 3032–3091. doi: 10.1021/je300655b. [Google Scholar]
 AGA8 (1992) Compressibility and super compressibility gas and other hydrocarbon gases, Transmission Measurement Committee Report, Number 8, AGA Catalogue No. XQ 1285, Arlington. [Google Scholar]
 Rostami A., Shokrollahi A., Ghazanfari M.H. (2018) New method for predicting ntetradecane/bitumen mixture density: correlation development, Oil Gas Sci. Technol. – Rev. IFP Energies nouvelles 73, 35, doi: 10.2516/ogst/2018037. [Google Scholar]
 MohamadiBaghmolaei M., Azin R., Osfouri S., MohamadiBaghmolaei R., Zarei Z. (2015) Prediction of gas compressibility factor using intelligent models, Nat. Gas Ind. B 2 2, 4, 283–294. doi: 10.1016/j.ngib.2015.09.001. [CrossRef] [Google Scholar]
 Rostami A., KalantariMeybodi M., Karimi M., Tatar A., Mohammadi A.H. (2018) Efficient estimation of hydrolyzed polyacrylamide (HPAM) solution viscosity for enhanced oil recovery process by polymer flooding, Oil Gas Sci. Technol. – Rev. IFP. Energies nouvelles 73, 22. doi: 10.2516/ogst/2018006. [Google Scholar]
 Bashipour F., Nouri Khorasani S., Rahimi A. (2015) H_{2}S reactive absorption from offgas in a spray column: Insights from experiments, Chem. Eng. Techol. 38, 12, 2137–2145. doi: 10.1002/ceat.201500233. [CrossRef] [Google Scholar]
 Mohammadi A.H., Richon D. (2007) Use of artificial neural networks for estimating water content of natural gases, Ind. Eng. Chem. Res. 46, 1431–1438. doi: 10.1021/ie060494u. [Google Scholar]
 Chouai A., Laugier S., Richon D. (2002) Modeling of thermodynamic properties using neural networks: Application to refrigerants, Fluid Phase Equilib. 199, 1–2, 53–62. doi: 10.1016/S03783812(01)008019. [Google Scholar]
 Karpash O., Darvay I., Karpash M. (2010) New approach to natural gas quality determination, J. Petrol. Sci. Eng. 71, 133–137. doi: 10.1016/j.petrol.2009.12.012. [CrossRef] [Google Scholar]
 AlQuraishi A.A., Shokir E.M. (2011) Artificial neural networks modeling for hydrocarbon gas viscosity and density estimation, JKSUES 23, 123–129. doi: 10.1016/j.jksues.2011.03.004. [Google Scholar]
 Sipöcz N., Tobiesen F.A., Assadi M. (2011) The use of artificial neural network models for CO_{2} capture plants, Appl. Energy 88, 7, 2368–2376. doi: 10.1016/j.apenergy.2011.01.013. [Google Scholar]
 Shahsavand A., DerakhshanFard F., Sotoudeh F. (2011) Application of artificial neural networks for simulation of experimental CO_{2} absorption data in a packed column, J. Nat. Gas Sci. Eng. 3, 3, 518–529. doi: 10.1016/j.jngse.2011.05.001. [Google Scholar]
 Fast M., Assadi M., De S. (2009) Development and multiutility of an ANN model for an industrial gas turbine, Appl. Energy 86, 1, 9–17. doi: 10.1016/j.apenergy.2008.03.018. [Google Scholar]
 Akcayol M.A., Cinar C. (2005) Artificial neural network based modeling of heated catalytic converter performance, Appl. Therm. Eng. 25, 2341–2350. doi: 10.1016/j.applthermaleng.2004.12.014. [Google Scholar]
 Fausett L.V. (1993) Fundamentals of neural networks: Architectures, algorithms and applications, PrenticeHall, Englewood Cliffs. [Google Scholar]
 Bashipour F., Rahimi A., Nouri Khorasani S., Naderinik A. (2017) Experimental optimization and modeling of sodium sulfide production from H_{2}SRich offgas via response surface methodology and artificial neural network, Oil Gas Sci. Technol. – Rev. IFP Energies nouvelles 72, 9. doi: 10.2516/ogst/2017004. [Google Scholar]
 Holland J.H. (1975) Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence, University of Michigan Press, Michigan, USA. [Google Scholar]
 Darwin C., Bynum W.F. (2009) The origin of species by means of natural selection: Or, the preservation of favored races in the struggle for life, Penguin Classics, London. [Google Scholar]
 Xiucheng D., Shouchun W., Renjin S., Suoqi Z. (2010) Design of artificial neural networks using a genetic algorithm to predict saturates of vacuum gas oil, Pet. Sci. 7, 118–122. doi: 10.1007/s121820100015y. [Google Scholar]
 Čapla L., Buryan P., Jedelský J., Rottner M., Linek J. (2002) Isothermal pVT measurements on gas hydrocarbon mixtures using a vibratingtube apparatus, J. Chem. Thermodyn. 34, 5, 657–667. doi: 10.1006/jcht.2001.0935. [Google Scholar]
 Atilhan M., Aparicio S., Ejaz S., Cristancho D., Hall K.R. (2011) PρT behavior of a lean synthetic natural gas mixture using magnetic suspension densimeters and an isochoric apparatus: Part I, J. Chem. Eng. Data 56, 2, 212–221. doi: 10.1021/je100676j. [Google Scholar]
 Hwang C.A., Simon P.P., Hou H., Hall K.R., Holste J.C., Marsh K.N. (1997) Burnett and pycnometric (p, V_{m}, T) measurements for natural gas mixtures, J. Chem. Thermodyn. 29, 12, 1455–1472. doi: 10.1006/jcht.1997.0258. [Google Scholar]
 Patil P., Ejaz S., Atilhan M., Cristancho D., Holste J.C., Hall K.R. (2007) Accurate density measurements for a 91% methane natural gaslike mixture, J. Chem. Thermodyn. 39, 8, 1157–1163. doi: 10.1016/j.jct.2007.01.002. [Google Scholar]
 Tampe S.S., Kulkarni B.D., Desphande P.B. (1996) Elements of artificial networks with selected applications in chemical engineering and chemical and biological sciences, Simulation and Advanced Controls Ltd., Louisville. [Google Scholar]
 Hornik K., Stinchcoombe M., White H. (1989) Multi layer feed forward networks are universal approximators, Neural Netw. 2, 359–366. doi: 10.1016/08936080(89)900208. [Google Scholar]
 Kolmogorov A.N. (1957) On the representation of continuous functions of many variables by superposition of continuous functions of one variable and addition, Dokl. Akad. Nauk USSR 114, 5, 953–956. [Google Scholar]
All Tables
Performance comparison of 4 NGDCANN model by applying different applied transfer functions in two hidden layers using = = 5 and LevenbergMarquardt backpropagation (trainlm) as a BPLA.
All Figures
Fig. 1 Architecture of the fourlayered MLFF network. 

In the text 
Fig. 2 Data gathering and grouping to apply in intelligent models. 

In the text 
Fig. 3 Effect of different in both of hidden layers (L _{1} and L _{2}) on performance of MLFF NGDCANN model to predict C _{DCM}. 

In the text 
Fig. 4 Comparison of predicted results by optimized NGDCANN model with (a) training set and (b) all data. 

In the text 
Fig. 5 Data MSE of predicted C _{DCM} with optimized NGDCANN model. 

In the text 