An energy-based global model for predicting HVDC relay explosions

Abstract High voltage direct current relay explosions pose significant risks to the power systems in electric vehicles due to rapid energy deposition during short-circuit conditions. This paper addresses a crucial research gap by developing and validating a global energy-threshold model for predicting explosion events in relays based on measured electrical waveforms. Principal parameters, including peak current, peak voltage, and the rate of energy deposition (power), are used to construct a regression framework that accurately estimates critical energy thresholds leading to relay explosions. Prediction uncertainty is quantified and propagated through the model to provide conservative safety margins suitable for circuit design and protection coordination studies. Experimental validation on the Panasonic AEV14012 relay, using a series of high-speed imaging tests and independent datasets, demonstrates robust predictive accuracy. The methodology is presented in detail so that the approach can be re-applied to other relay types through re-testing and re-calibration. This physically interpretable yet computationally efficient framework provides a practical tool for explosion prediction, enabling enhanced safety and reliability in modern high-voltage power applications.