Transient thermal modelling of an Axial Flux Permanent Magnet (AFPM) machine with model parameter optimisation using a Monte Carlo method

This paper presents the development of a transient thermal model of the EVO Electric AFM 140 Axial Flux Permanent Magnet (AFPM) machine based on a hybrid finite difference and lumped parameter method. A maximum deviation between simulated and measured temperature of 2.4°C is recorded after using a Monte Carlo simulation to optimise model parameters representing a 53% reduction in temperature deviation. The simulated temperature deviations are lower than the measurement error on average and the thermal model is computationally simple to solve. It is thus suitable for transient temperature prediction and can be integrated with the system control loop for feed forward temperature prediction to achieve active thermal management of the system. Keyword: Axial Flux Permanent Magnet (AFPM) machine, Monte Carlo simulation, transient thermal modelling Nomenclature A Area of heat transfer (m) Subscript  Thermal capacitance (J/°C) amb Ambient  Deviation (%) bdry Boundary  Current (A) cd Conduction  Thernal conductivity (W/m) cv Convection  Thermal conductance (W/m) Cu Copper  Thermal conductance (W/°C) exp Experimental  Number of samples link Interface  Spread in fraction max Maximum  Heat flow rate (W) min Minimum  Heat generation rate (W) o Central value  Thermal resistance (°C/W) sim Simulated  Thickness (m) tot Total  Temperature (°C)  Simulation time step (s)  Electrical resistance ()