Thermal Conductivity Prediction of Pure Liquids Using Multi-Layer Perceptron Neural Network

Abstract

H EAT transfer of fluids is important in many industrial sectors, including energy supply, transportation, production, and electronics. Tomodel the heat transfer process, thermal conductivity data are required. Knowing thermal conductivity values of liquids is of great importance in the design of chemical engineering; however, these values are not easily calculable in an accurate manner. Through different investigations, the thermal conductivity has been calculated and estimated for many liquids, and there are two general trends for estimation of thermal conductivity: 1) using relations based on the theory of liquids thermal conductivity; and 2) using correlations relating thermal conductivity to other properties. In the second class, properties can be referred to as critical temperature, critical pressure, and molecular weight; and they are all simply measurable. Widespread application of a neural network has been obvious in many fields of chemical engineering over the last years. The thermal conductivity of liquids has been predicted by using this kind of network and compared with experimental outcomes. The artificial neural network (ANN) has shown successful performance in various fields of modeling, such as engineering systems, mathematics, medicine, economics, and meteorology [1,2]. Based on experimentally measured variables, Krzywanski and Nowak [3] developed an ANN model to predict the local overall heat transfer coefficient for membrane walls in the 260 megawatts of electrical power circulating-fluidized-bed boiler. Kurt and Kayfeci [4] proposed an ANN model to estimate the local overall heat transfer coefficient of ethylene glycol/water solutions at various temperatures and different concentrations. Boniecki et al. [5] investigated the options of applying the ANN as a predictive instrument for modeling ammonia emissions released in composting sewage sludge. The present study is intended to establish a neural network model to predict the thermal conductivities of pure liquids at atmospheric pressure over a wide range of temperatures and many types of substances. In this regard, the properties of a liquid’s temperature, critical pressure, critical temperature, boiling temperature, and molecular weight are set as network inputs.