A Markov Chain Monte Carlo Approachto Network Model Calibration Using Tracer Tests

摘要

A preliminary Markov chain Monte Carlo (MCMC) calibration algorithm is presented forrnestimating the demand pattern multipliers of a distribution system network model utilizingrntracer test data. A simple 35 node distribution system network is used to generate a syntheticrntracer study data set over a 55-hour simulation time. The MCMC calibration algorithm isrnable to reproduce the 55-hour demand pattern multipliers from an initial guess absent of anyrntemporal information. This preliminary study for the MCMC calibration algorithm providesrnthe basis for extending the capabilities to include spatial correlation of user demands.