MODELING AND OPTIMIZATION OF CONTRACTED DEMAND FOR ELECTRICITY: APPLICATION IN A PUBLIC INSTITUTION UNDER BRAZILIAN TAX AND REGULATORY RULES

The optimization of the contracted demand for electricity is crucial to reduce costs in public medium voltage consumer units. This work compares three methods (Nonlinear Programming L-BFGS-B, Monte Carlo Simulation and Interior Penalty) to define the optimal demand, using consumption data from 2024 from UFMS. The analysis considers taxes (ICMS, PIS/COFINS), penalties for overtaking and local case law on ICMS in the demand used. Results indicate potential annual savings of up to R$ 248 thousand, with optimal demand between 1300-1450 kW. The stochastic approach proved to be robust, and the method with penalization was more conservative. The proposed methodology is replicable for contract management in public institutions and industries, aiming at cost reduction.