Thermal Estimation and Analysis of Three-Phase Induction Motors under Voltage Unbalance

Abstract

Voltage unbalance is a persistent power quality issue in industrial and commercial facilities, and induction motors are especially sensitive to its effects. Even moderate levels of unbalance, which are common in practice, can produce disproportionately large increases in motor losses and winding temperatures. These elevated temperatures accelerate insulation aging and reduce motor service life. Although the thermal risks associated with voltage unbalance are well documented, estimating winding temperature under unbalanced supply conditions remains difficult. This dissertation develops a unified framework for thermal estimation and analysis of three phase induction motors operating under voltage unbalance, combining symmetrical component loss analysis, reduced order lumped parameter thermal network modeling, state space formulation, adaptive parameter estimation, and dynamic insulation life prediction.

The thermal modeling approach is based on a minimal order lumped parameter thermal network expressed in state space form. This representation enables the use of linear systems theory for the thermal estimation problem, including observability assessment, optimal state estimation, and systematic model order reduction. Starting from a detailed multi-node MATLAB and Simulink thermal model that captures the transient thermal behavior under both balanced and unbalanced supply, the dissertation derives a compact three node state space model that preserves the essential thermal physics while remaining simple enough for practical identification and implementation. The state space model accepts motor electrical losses, computed from measured voltages and currents using symmetrical component analysis, and produces real time estimates of stator and rotor winding temperatures.

A central contribution of the dissertation is an adaptive parameter estimation method that identifies the thermal network parameters using only nameplate data and online electrical and temperature measurements. The estimation framework combines nameplate based initialization, optimization based refinement from measured operating points, and adaptive updating to track parameter drift as the motor ages. The thermal model is then applied in a large scale Monte Carlo study covering the full range of voltage unbalance conditions encountered in practice, parameterized by the voltage unbalance factor. The simulations, performed over representative twenty-four hour duty cycles, quantify the statistical distribution of temperature rise under unbalance and identify the combinations of VUF magnitude that impose the greatest thermal stress. The predicted temperature histories are combined with Arrhenius based insulation life models to quantify the cumulative thermal aging caused by sustained voltage unbalance. Using the IEEE 101 temperature life tables and the standard Arrhenius activation energy for motor insulation, the analysis shows that even moderate unbalance levels in the 3% to 5% range, when present for extended periods, can reduce the life of Class H insulation by 15% to 38% relative to balanced operation. This establishes a quantitative link between measurable power quality metrics and the economic impact of premature motor aging, providing a basis for evaluating the cost effectiveness of power quality improvements. The contributions of this dissertation are: (1) a state space thermal model for induction motors under voltage unbalance suitable for real time implementation, (2) an adaptive parameter estimation framework that enables deployment without prior thermal testing, (3) a Monte Carlo characterization of motor thermal response across the VUF space, (4) quantitative insulation life predictions under voltage unbalance, and (5) identification of key research gaps that limit practical thermal monitoring in the presence of power quality disturbances. Together, these developments form the foundation for next generation motor protection systems that estimate winding temperature and cumulative thermal damage directly from electrical measurements, enabling predictive maintenance and informed derating decisions under adverse supply conditions.