Wavelet neural network-GCM based for density prediction of liquefied and compressed natural gas (LNG and CNG) over a wide range of temperature and pressure

Natural gas is a gaseous fossil fuel consisting primarily of methane but including significant quantities of ethane, propane, butane, and pentane. It is a major source of electricity generation through the use of gas turbines and steam turbines. Also, it is used as fuel in cooking on many barbecues, portable stoves and in motor vehicles. Natural gas burns cleaner than other fossil fuels, such as oil and coal, and produces less carbon dioxide per unit energy released. Liquefied natural gas (LNG) and compressed natural gas (CNG) are different natural gas forms used as fuel. Therefore, the volumetric properties of them and also their temperature and pressure dependence are of particular interest in petroleum and other chemical industries. One important characteristic of them that it is central to chemical processing design in chemical industries is density. Although experimental data can be very accurate, it is difficult to provide all the data needed for every compound at all thermodynamic condition of temperatures and pressures. For this reason, several methods have been developed for predicting liquid densities such as Equation of states, group contribution method or correlation functions. But these methods which are often used to predict thermodynamic properties require pure fluid parameters as input. Also, derivation of their relationship is a difficult task. Hence, predictive methods which are relatively simple to use and have as few possible parameters are desirable. In recent years artificial neural networks (ANN) and combination of wavelet theory with neural networks, namely wavelet neural network (WNN), have provided a high performance nonlinear analysis tool that may be used to avoid the shortcomings involved in prediction methods In this work, we modeled wavelet neural network (WNN) [1] for density prediction of CNG and LNG based on group contribution method for the first time.