Enhancement of the extended corresponding states techniques for thermodynamic modeling. II. Mixtures

Abstract

Moving from a former work for pure fluids a new modeling technique has been developed for obtaining a fundamental mixture equation of state in the Helmholtz energy form. This model can be considered an evolution of the extended corresponding states method, which is modified from the conventional analytical mode to a heuristic one through the integration of a general function approximator for the representation of the scale factor functions of a target mixture. The assumed approximator is a multilayer feedforward neural network (MLFN) with two outputs, one for each scale factor. A reference pure fluid, conformal with the components of the studied mixture, is chosen and the independent variables of its dedicated equation of state (DEoS) are distorted by the scale factors, which are individual functions of temperature, density, and composition. The MLFN scale factor functions can be obtained from regression on any kind of thermodynamic data of the target mixture. The model capability to accurately represent the thermodynamic surfaces of five binary and two ternary haloalkane mixtures is studied assuming data generated from the corresponding DEoSs. The obtained prediction accuracies for the mixture thermodynamic properties are competitive with those of the available conventional DEoSs. The proposed modeling technique is then robust and straightforward for the effective development of a mixture DEoS from thermodynamic quantities distributed in the range of interest.

Predicting the Hydrate Stability Zones of Natural Gases Using Artificial Neural Networks

“…Artificial neural networks have large numbers of computational units called neurons, connected in a massively parallel structure and do not need an explicit formulation of the mathematical or physical relationships of the handled problem [5,6,[8][9][10][11]. The most commonly used ANNs are the feed-forward neural networks [11], which are designed with one input layer, one output layer and hidden layers [8][9][10].…”

“…The most commonly used ANNs are the feed-forward neural networks [11], which are designed with one input layer, one output layer and hidden layers [8][9][10]. The number of neurons in the input and output layers equals to the number of inputs and outputs physical quantities, respectively.…”

An Extended Equation of State Modeling Method I. Pure Fluids

“…Lately, the ECS method was revisited by some authors [8][9][10][11][12]17], who independently determined the shape functions through a minimization procedure using experimental data of various properties for the fluid of interest, making the method similar to the classical multiproperty fitting approach [1][2][3]. In such a way both the shape functions become available as continuous analytical functions of temperature and density.…”

“…In particular, in the works of Scalabrin et al [10][11][12]17] an artificial neural network (ANN) was assumed as a general function approximator for the representation of the shape functions. This work was focused on the refrigerant family both as pure fluids [11] and as mixtures [17], always assuming R134a as the reference fluid.…”

“…This work was focused on the refrigerant family both as pure fluids [11] and as mixtures [17], always assuming R134a as the reference fluid.…”