Executing quasi-static time-series simulations istime consuming, especially when yearly simulations are required,for example, for cost-benefit analyses of grid operation strategies.Often only aggregated simulations outputs are relevant to gridplanners for assessing grid operation costs. Among them are totalnetwork losses and power exchange through MV/LV substationtransformers. In this context it can be beneficial to explorealternatives to running quasi-static time-series simulations withcomplete input data that can produce the results of interest withhigh accuracy but in less time. This paper explores two methodsfor shortening quasi-static time-series simulations through reducingthe amount of input data and thus the required number ofpower flow calculations; one is based on downsampling and theother on vector quantization. The results show that execution timereductions and sufficiently accurate results can be obtained withboth methods, but vector quantization requires considerably lessdata to produce the same level of accuracy as downsampling. Inparticular, when the simulations consider voltage control or whenmore than one simulation with the same input data is required,vector quantization delivers a far superior trade-off between datareduction, time savings, and accuracy. However, the method doesnot reproduce peak values in the results accurately. This makesit less precise, for example, for detecting voltage violations.
展开▼
