Physics-based Early Detection Algorithm for Temperature Monitoring Systems in Electrical Equipment

Abstract

Temperature monitoring of electrical equipment is useful to detect deficient conditions like loose connections that cause overheating and could lead to failures. The monitoring systems typically implement fixed threshold-based logic and provide warnings or alarms. However, the industry is starting to leverage additional variables like load (current through equipment), ambient conditions, etc., and implementing pattern recognition or artificial intelligence-based techniques to identify deficiencies at an early stage. The implementation of these advanced techniques generally requires dedicated computational resources and software. This paper presents a simple analytical physics-based model that can be used to provide an early anomaly detection capability for current-carrying conductors in electrical equipment. The analytical model is developed as a second order equation that can be easily included in a monitoring system platform without the need for additional computational resources and external software. This paper includes the theory and simulation results from a finite element model to validate the analytical model. The finite element model was set up in COMSOL to simulate the heat transfer in a current-carrying conductor and validate the proposed analytical physics-based model.