Probabilistic method to estimate the need of nuclear power plant units in Romania in the years 2020 - 2025

The paper presents a probabilistic method dedicated to the estimation of future electricity needs and applicable solutions. The preliminary tests suggested that this probabilistic method is a complementary tool and may be used together with conventional deterministic techniques, rather than instead of them. Theoretically, the electricity planning is based on conventional analysis methods, able to provide an acceptable accuracy. However, all these methods are based on a deterministic approach. They require a large amount of input data. As the uncertainties related to these input data are significant, a finite number of scenarios must be selected to be analyzed. When the area covered by the output data is narrow enough, the deterministic analysis is really successfully. The decision entity gets the needed data. There is a high chance to derive a finally selected decision having a sound scientific basis. However, the deterministic investigations often prove that the output data area is much too broad. Further hypotheses are needed. The necessary operation of scenario elimination is practically based on the 'analyst's experience'. In this case, the decision entity must be fully aware of the hidden output data uncertainties. In the period of 1990 - 2000 years, Romanian energy analysts faced and still face a very difficult problem. The uncertainties of the needed input data are very high. Consequently, the output data were not enough useful. In addition, a significant public suspicion was related to the quality of the management and strategy of the Romanian Authority of Electricity (RENEL). The actual role of the conventional analyses abruptly, diminished. The decision entity was practically forced to increase the extend of the 'personal feeling' of a very limited number of 'selected persons'. The paper suggests that the probabilistic method is able to offer an objective and well balanced picture of the more or less realistic solutions. However, interpretation of the output data is no longer a very simple one and the expectations of the decision entity are not always satisfied. As a matter of fact, the output data of the probabilistic method reflects not only the analyst's knowledge, but also the analyst's lack of knowledge at a given moment. (author)