SENSITIVITY ANALYSIS FOR OPTIMIZING LARGE SOLAR THERMAL SYSTEMS

摘要

Two techniques have been applied for the determination of sensitivities of the performance of a large solar thermal system on its dimensioning and operation parameters: Differential Sensitivity Analysis (DSA) and Monte-Carlo Analysis (MCA). Since the sensitivities have to be determined for an optimal designed system, automatic optimizations with classical and evolutionary algorithms have been carried out. For this, a TRNSYS system model of the solar thermal system has been implemented and coupled with the optimization algorithms. However, the determination of the sensitivities depends strongly on the chosen range of parameter variations, which is different for each individual parameter. Furthermore, since the influence of the parameters on the system performance is not linear and correlation between parameters exist, the results from DSA has to be interpreted carefully. Especially a comparison of sensitivities of different parameters is difficult. However, the application of MCA for a detection of sensitivities in a minimum of the objective function leads to difficulties concerning the distribution of parameter variation and resulting objective function values, too. Only by using quantiles, a determination of a overall sensitivity is possible depending on the parameter variations carried out.