A parallel PIC-MCC simulation of microsecond discharge modeling in EDM

Spark discharge plasma in electrical discharge machining (EDM) is hard to measure due to its small spatial scale, so particle-in-cell with Monte Carlo collision (PIC-MCC) simulation is a powerful method to analyze the discharge channel of EDM. PIC-MCC simulation suffers from traversing particles, which is high time-consuming, and thus, the discharge duration is limited to nanoseconds. However, the nanosecond simulation results cannot reflect the actual machining of EDM (10–100 μs). For this reason, a parallel PIC-MCC simulation architecture based on GPU for EDM is proposed. This architecture parallels the tasks in PIC, greatly reducing the increase of computing time when the number of particles increases, which makes simulation of duration time in microseconds available. Moreover, the simulation process takes into account the working mechanism of the breakdown/discharge coordination of the EDM circuit, which is more meaningful for actual machining. This paper describes the algorithm architecture and some results of EDM in air at atmospheric pressure with open voltage of 150 V. It is shown that micro-peak geometry limits the plasma size, so it does not expand as the pulse duration time increases. The energy transferred into the anode is much higher than that of the cathode, and the energy absorption coefficient of the cathode is less than 4%. And these results are consistent with the existing related reports. Compared with the large pulse duration measurement (> 100 μs) of discharge, this paper provides a method to calculate heat input which can include transient details and conform to the actual machining pulse duration (≈10 μs).