The predictions of condensation pressure drops of R12, R22, R32, R125, R410A, R134a, R22, R502 and R507a flowing inside various horizontal smooth and micro-fin tubes are made using the numerical techniques of Artificial Neural Networks (ANNs) and non-linear least squares (NLS). The National Institute of Standards and Technology’s (NIST) experimental data and, Eckels’ and Pate’s experimental data, as presented in Choi et al.’s study provided by NIST, are used in our analyses.
In their experimental setups, the horizontal test sections have 1.587 m, 3.78 m, 3.81 m and 3.97 m long countercurrent flow double tube heat exchangers with refrigerant flowing in the inner smooth (8 mm, 8.01 mm and 11.1 mm i.d.) and micro-fin (5.45 mm and 7.43 mm i.d.) copper tubes as cooling water flows in the annulus. Their test runs cover a wide range of saturation pressures from 0.9 MPa to 2.9 MPa, inlet vapor qualities range from 0.19 to 1.0 and mass fluxes are from 8 kg m−2s−1 to 791 kg m−2s−1. The condensation pressure drops are predicted using 673 measured data points, together with numerical analyses of artificial neural networks and non-linear least squares.
The input of the ANNs for the best correlation are the measured and the values of the test sections are calculated, such as mass flux, tube length, inlet and outlet vapor qualities, critical pressure, latent heat of condensation, mass fraction of liquid and vapor phases, dynamic viscosities of liquid and vapor phases, hydraulic diameter, two-phase density, and the outputs of the ANNs as the experimental total pressure drops in the condensation data from independent laboratories. The total pressure drops of in-tube condensation tests are modeled using the artificial neural networks (ANNs) method of multi-layer perceptron (MLP) with a 12-40-1 architecture.
The average error rate is 7.085%, considering the cross validation tests of the 867 condensation data points. A detailed model of f(MLP) is given for direct use in MATLAB. This explanation will enable users to predict the two-phase pressure drop with high accuracy. As a result of the dependency analyses, dependency of the output of the ANNs from 12 sets of input values is shown in detail, and the pressure drops of condensation in smooth and micro-fin tubes are found to be highly dependent on mass flux, all liquid Reynolds numbers, the latent heat of condensation, outlet vapor quality, critical pressure of the refrigerant, liquid dynamic viscosity, and tube length. New ANNs based empirical pressure drop correlations are developed separately for the conditions of condensation in smooth and micro-fin tubes as a result of the analyses.
Application of artificial neural networks for gradient elution retention modelling in ion chromatography
