Dynamic Line Rating Based on Phasor Measurements: An Uncertainty Analysis

Abstract

Rising electricity demand from energy-intensive technologies has heightened the need for reliable power transmission. Conventional static line ratings, based on conservative seasonal assumptions, often underutilize available capacity. Dynamic Line Rating (DLR) overcomes this by estimating ampacity from real-time conditions, but most implementations rely on extensive weather monitoring that limits scalability. Phasor-based DLR offers a scalable alternative by using synchronized voltage and current measurements to infer conductor temperature and power losses, enabling higher utilization without additional infrastructure. This thesis investigates the theoretical performance of phasor-based DLR for short 130 kV overhead lines, with a focus on error propagation. Statistical inference methods are used to estimate conductor temperature from synthetic measurements while accounting for sensor accuracy. The impact of multiple measurements on estimating both static and dynamic estimation line parameters is examined. The results show that single-measurement temperature estimates are fundamentally limited by amplified uncertainty in power loss calculations, requiring measurement accuracy improvements of at least two orders of magnitude over current standards. Multiple measurements improve accuracy and precision but require 103 -105 samples, depending on prior assumptions and sensor quality. Dynamic methods perform well under most conditions, achieving RMSEs of 0.8-2.6 °C for high-accuracy measurements, but abrupt current changes degrade accuracy while maintaining high model confidence, indicating model over-confidence. These findings suggest that phasor-based DLR can reach accuracy levels comparable to weather-based approaches under ideal conditions, but practical deployment is constrained by measurement uncertainty, inference assumptions, and fundamental information limits. Further development of phasor-based DLR will require more robust uncertainty modelling, and stronger integration of physical priors in both static and dynamic estimation frameworks.