Scheduling optimization of combined heat and power units with multiple degrees of freedom based on the superposition principle

This work proposes a revised version of the Mixed Integer Linear Programming (MILP) model developed by the authors in [1-3] for optimizing the short-term and long-term scheduling cogeneration units’ networks. Compared to the previous model formulation, the revised version allows to save computational time and to efficiently deal with cogeneration units with more than two degrees of freedom (i.e., independent control variables). To this end, the computationally expensive n-dimensional PieceWise Linearization (PWL) of the off-design performance curves of the units is avoided by better exploiting the mathematical properties of the polynomial best-fit curves and applying the superposition principle. In the analyzed test cases with two degrees of freedom, compared to the PWL approximation used in [1-2], the saving of computational time is up to almost 90% and almost 80% in average, with the same accuracy as the PWL. Besides, the saving of computational time is expected to increase exponentially with the number of degrees of freedom. A three degrees of freedom unit test case with superposition principle has also been performed, reaching a 0.001% error, MILP-gap, in an average time of 40 seconds while preserving accuracy.