Application of information theory for the analysis of cogeneration-system performance

Successful cogeneration system performance depends critically upon the correct estimation of load variation and the accuracy of demand prediction. We need not only aggregated annual heat and electricity demands, but also hourly and monthly patterns in order to evaluate a cogeneration system's performance by computer simulation. These data are usually obtained from the actual measurements of energy demand in existing buildings. However, it is extremely expensive to collect actual energy demand data and store it over a long period for many buildings. Here we face the question of whether it is really necessary to survey hourly demands. This paper provides a sensitivity analysis of the influence of demand-prediction error upon the efficiency of cogeneration systems, so as to evaluate the relative importance of various demand components. These components are annual energy demand, annual heat-to-electricity ratio, daily load factor and so forth. Our approach employs the concept of information theory to construct a mathematical model. This analysis provides an indication of the relative importances of demand indices, and identifies what may become a good measure for assessing the efficiency of the cogeneration system for planning purposes.