Prediction of the Liquid Vapor Pressure Using the Artificial Neural Network Group Contribution Method (original) (raw)
Related papers
Journal of Mathematics and Computer Science
A new method to estimate vapor pressures for pure compounds using an artificial neural network (ANN) is presented. A reliable database including more than 12000 data point of vapor pressure for testing, training and validation of ANN is used. The designed neural network can predict the vapor pressure using temperature, critical temperature, and acentric factor as input, and reduced pressure as output with 0.211% average absolute relative deviation. 8450 data points for training, 1810 data points for validation, and 1810 data points for testing have been used to the network design and then results compared to data source from NIST Chemistry Web Book. The study shows that the proposed method represents an excellent alternative for the estimation of pure substance vapor pressures and can be used with confidence for any substances.
Industrial & Engineering Chemistry Research, 2012
In the present study, a group contribution model is developed for determination of the vapor pressure of pure chemical compounds at temperatures from 55 to 3040 K. About 42 000 vapor pressure values belonging to around 1400 chemical compounds (mostly organic ones) at different temperatures are treated to propose a reliable and predictive model. A three-layer artificial neural network is optimized using the Levenberg−Marquardt (LM) optimization algorithm to establish the final relationship between the functional groups and the vapor pressure values. The obtained results indicate the average absolute relative deviation (AARD%) of the calculations/estimations from the applied data to be about 6% and a squared correlation coefficient of 0.994. Furthermore, the outliers of the model are detected using the leverage value statistics method.
Expert Systems With Applications, 2011
Vapor pressure and liquid density of 20 pure alcohols were correlated using an artificial neural network (ANN) system and statistical associating fluid theory (SAFT) equation of state. The SAFT equation has five adjustable parameters as temperature-independent segment diameter, square-well energy, number of segment per chain, association energy and association volume. These parameters can be obtained by a non-linear regression method using the experimental vapor pressure and liquid density data. In continue, the vapor pressure and liquid densities of pure alcohols were estimated by using an artificial neural network (ANN) system. In the neural network system, it is assumed that thermodynamic properties of pure alcohols depend on temperature, critical properties and acentric factor. The best network topology was obtained as (4-10-2). The weights connection and biases were obtained using batch back propagation (BBP) method for 611 experimental data points. The average absolute deviation percent (ADD%) for vapor pressure of pure alcohols for ANN system and SAFT equation of state are 3.593% and 3.378%, respectively. Also, the average absolute deviation percent (ADD%) for liquid density of pure alcohols for ANN system and SAFT equation of state are 0.792% and 1.367%, respectively. The results emphasized that the artificial neural network can more accurately predict thermophysical properties of pure alcohols than the SAFT equation of state.
Brazilian Journal of Chemical Engineering, 2009
Equations of state are useful for description of fluid properties such as pressure-volumetemperature (PVT). However, the success estimation of such correlations depends mainly on the range of data which have originated. Therefore new models are highly required. In this work a new method is proposed based on Artificial Neural Network (ANN) for estimation of PVT properties of compounds. The data sets were collected from Perry's Chemical Engineers' Handbook. Different training schemes for the backpropagation learning algorithm, such as; Scaled Conjugate Gradient (SCG), Levenberg-Marquardt (LM) and Resilient back Propagation (RP) methods were used. The accuracy and trend stability of the trained networks were tested against unseen data. The LM algorithm with sixty neurons in the hidden layer has proved to be the best suitable algorithm with the minimum Mean Square Error (MSE) of 0.000606. The ANN's capability to estimate the PVT properties is one of the best estimating method with high performance.
IRJET, 2022
In the chemical industry, vapor-liquid equilibrium (VLE) data are essential for the design, analysis, control, and modeling of process equipment such as distillation columns, absorbers, and reactors. The usually employed feed-forward neural network, the multi-layer perceptron (MLP), was used to develop MLPNN-based models for VLE prediction in this study. The MLPNN approximates nonlinear relationships that exist between the variables in an input data set and the output data set associated with it. Experimental data was used to develop MLPNN-based VLE models for predicting the vapor phase composition of a ternary system (benzene + cyclohexane + anisole). A physical property of pure components (acentric factor) and thermodynamic parameters (equilibrium temperature, liquid phase composition) are included as the input space for the model development. An error-back propagation (EBP) approach is used to train the proposed MLPNN-based model. The experimental data was split at random, with 75% of the data used as the training set for constructing the models and 25% of the data used as the test set for evaluating the models generalization ability. The predicted values are in good agreement with the corresponding experimental values, indicating that the proposed models have good prediction accuracy and generalization ability.
Brazilian Journal of Chemical Engineering, 2008
Artificial neural networks are applied to high-pressure vapor liquid equilibrium (VLE) related literature data to develop and validate a model capable of predicting VLE of six CO 2-ester binaries (CO 2-ethyl caprate, CO 2-ethyl caproate, CO 2-ethyl caprylate, CO 2-diethyl carbonate, CO 2-ethyl butyrate and CO 2isopropyl acetate). A feed forward, back propagation network is used with one hidden layer. The model has five inputs (two intensive state variables and three pure ester properties) and two outputs (two intensive state variables).The network is systematically trained with 112 data points in the temperature and pressure ranges (308.2-328.2 K), (1.665-9.218 MPa) respectively and is validated with 56 data points in the temperature range (308.2-328.2 K). Different combinations of network architecture and training algorithms are studied. The training and validation strategy is focused on the use of a validation agreement vector, determined from linear regression analysis of the plots of the predicted versus experimental outputs, as an indication of the predictive ability of the neural network model. Statistical analyses of the predictability of the optimised neural network model show excellent agreement with experimental data (a coefficient of correlation equal to 0.9995 and 0.9886, and a root mean square error equal to 0.0595 and 0.00032 for the predicted equilibrium pressure and CO 2 vapor phase composition respectively). Furthermore, the comparison in terms of average absolute relative deviation between the predicted results for each binary for the whole temperature range and literature results predicted by some cubic equation of state with various mixing rules and excess Gibbs energy models shows that the artificial neural network model gives far better results.
In this study, existing experimental vapour-liquid equilibrium (VLE) data covering a wide range of temperature, phase composition and pressure for ethane-n-butane-n-pentane was correlated using MATLAB (Matrix Laboratory) software. To increase the reliability of correlations, neural network was trained using existing vapour-liquid equilibrium data with the aid of Levenberg Marquardt algorithm. Network parameters are fine-tuned until the output generated by simulation are checked and observed to match with pre-determined experimental V L E data. It was found that there is high degree of coherence between the chosen targets from experimental data and predicted values. This confirms that correlations and predictions of V L E data using neural network is efficient and significant.
Application of artificial neural network for prediction of halogenated refrigerants vapor pressure
Afinidad Revista De Quimica Teorica Y Aplicada, 2010
Accurate prediction of stock market returns is a very challenging task because of the highly nonlinear nature of the financial time series. In this study, we apply an artificial neural network (ANN) that can map any nonlinear function without a prior assumption to predict the return of the Japanese Nikkei 225 index. (1) To improve the effectiveness of prediction algorithms, we propose a new set of input variables for ANN models. (2) To verify the prediction ability of the selected input variables, we predict returns for the Nikkei 225 index using the classical back propagation (BP) learning algorithm. (3) Global search techniques, i.e., a genetic algorithm (GA) and simulated annealing (SA), are employed to improve the prediction accuracy of the ANN and overcome the local convergence problem of the BP algorithm. It is observed through empirical experiments that the selected input variables were effective to predict stock market returns. A hybrid approach based on GA and SA improve prediction accuracy significantly and outperform the traditional BP training algorithm.
Prediction of Vapor-Liquid Equilibrium for H2S-DGA-H2O System Using Neural Network
Vapor-liquid equilibrium (VLE) data play a significant role in the designing and modeling of the separation processes. In this study, an artificial neural network (ANN) model has been proposed to predict the VLE data of H 2 S-H 2 O-DGA system. Experimental data needed for development of training, testing, and evaluating of the single layer perceptron network were taken from the literature. The applied perceptron neural network contains one hidden layer with ten neurons. Mean absolute error (MAE) of the perceptron network to predict mole ratio of H 2 S in the amine (α H2S ) was 0.0036. The system was also modeled using a reported thermodynamic model. The results of the thermodynamic model were compared with those of ANN model. The comparison indicated that the ANN model yielded a better fit of experimental data with a good accuracy and rapidity as compared to the thermodynamic model. Therefore, the ANN method can be applied as a tool for modeling phase equilibrium of chemical processes with a reasonable accuracy.