The concept of energy support is heading more and more towards the idea of a distributed production, which points to a gradual replacement of the standard production concept centered on single large units. In this respect, for example, uncertainty associated to the energy demand of each unit of a district becomes a fundamental point to take in account: thus, a probabilistic approach to the analysis of distributed generation systems is highly recommended. In this sense, an off-design steady-state model of a micro-gas turbine (mGT) has been employed: it is built on the configuration of a Turbec T100 actually installed at the research laboratory of the Thermochemical Power Group. In particular, elements of uncertainty linked both to operating parameters of the machine and to the load demands during the year were present. Uncertainty analysis has been treated with two different methods. The first and most famous is Monte Carlo method, which, however, requires a large time consuming in order to achieve the necessary number of samples. It is clear that its application for analyze such complex systems would be computationally inefficient and expensive so, this paper proposes as alternative an approximated method called Response Sensitivity Analysis (RSA). It is based on Taylor series expansions and promises good accuracy still retaining acceptable computational time. This work has led to the following results:1. estimations of probabilistic distributions for electrical power production at the outlet of generator, fuel consumption and net efficiency of the system during the entire working period; and in addition2. an evaluation of the most probable return on investment time and the range within it could fall, depending on the requested interval of confidence based on the standard deviation of the output distribution probability.3. Moreover, RSA has shown a wide compatibility with different types of problems, promising an easy applicability and scalability in different fields of study.
