Challenging issues of modeling load injections in state estimation

Abstract

Uncertainty of load models affects state estimation (SE) solution robustness and quality at EMS control center, particularly at the locations where unmeasured loads are outside EMS footprint or when load measurements suddenly become unavailable. It is quite common that loads in external network are not measured and their model uncertainties based on scheduled or forecasted MW values are much higher than loads in internal network. The concept of using pseudo measurement injections is applied to these unmeasured loads for our SE solution. Operation engineers at EMS control center are very interested in the load modeling techniques as well as pseudo load injection techniques used in SE solution in order to quickly identify the causes of an invalid SE solution or a SE solution with non-trivial residuals. One common approach of applying pseudo load injections is to allocate individual load MW values from their parent load areas via the load schedule evaluation or system load forecast function. Enhancement of using combination of measurement and schedule load data along the hierarchical load structure is added to this approach to indirectly allocate individual load MW values. Another approach is to use neighboring measurements to directly calculate individual load MW values. These two approaches as well as our experience working with operation engineers will be described in this paper.