Stochastic Predictions of Solar Cooling System Performance

Abstract

The use of computerized system simulations for sizing and performance predictions of various solar systems requires some form of weather input to act as a system stimulus. When actual weather data are used, simulations run on an hourly basis are expensive and require considerable data handling. For many design procedures, however, hourly information is not needed, and simpler methods are desirable. One such method employs a probabilistic approach. This method involves the use of an algorithm that generates a probabilistic matrix, and an analytical formulation which is used to generate synthetic weather data. The approach has been found to be satisfactory. This work uses the stochastic (probabilistic) method to produce representative weather for five geographic regions in the U.S. for the summer months. Parallel runs are conducted with real and stochastic weather. A comparison of the results clearly shows that the probabilistic approach can satisfactorily substitute for real weather for the purpose of system simulation, at reduced cost and data handling.