Measurement based static load model identification

Accurate load models are necessary to enable multiple power system planning and operation applications. The current day utility practice relies on model parameters that are based on old survey results/tests and often do not accurately capture the behavior of loads across different operating conditions. This results in conservative estimation of loading margin, leading to under-utilization of the assets. Static load models are typically chosen as purely constant power models, resulting in lack of understanding of the voltage dependence of loads. This paper considers the problem of determining the voltage sensitivity of aggregate loads and its impact on voltage stability. The role of voltage sensitivity of loads in determining accurate PV curves is demonstrated. Additionally, a novel recursive least mean squares based algorithm is proposed to estimate a generalized static load model that not only estimates the model parameters but also captures typical patterns in load variation with respect to time, season, temperature, etc., by utilizing measurement data from switching events. The proposed approach eliminates the need to rely on stage tests to obtain the data samples, and instead utilizes available measurements from events recorded over time. The efficacy of the proposed approach is demonstrated using actual field data.