COMPUTATIONAL MODELLING OF SEEBECK COEFFICIENTS OF PT/PD THERMOCOUPLE

This work presents a method in molecular dynamics utilising an artificial neural network designed specifically for the modelling and prediction of thermopower in a Pt/Pd thermocouple. In molecular dynamics simulations, semi-empirical interatomic potentials arc employed to compute some thermodynamic properties, i.e., enthalpy, potential energy, volume and density of the metals. Then the artificial neural network model is utilized to predict thermopower of PtfPd thermocouple, using the computed equilibrium properties. Predicted results are in good agreement with the existing experimental data.