^{1}, M. C. Martins

^{2,a}, D. G. Teixeira

^{2,b}, and P. F. Frutuoso e Melo

^{2,c}

^{1}Department of Nuclear Engineering, Polytechnic School, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil.

^{2}Graduate Program of Nuclear Engineering, COPPE, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil.

^{a}max@lmp.ufrj.br

^{b}dteixeira@nuclear.ufrj.br

^{c}frutuoso@nuclear.ufrj.br

We consider a plant equipped with a single protection channel whose demand, failure, and repair times follow exponential distributions. However, we may not know the parameters of these distributions precisely; this is due to variability on them and parameter ranges may be found. This lack of plant specific data gives rise to plant-to-plant data variability. In this sense, the available information may be translated as a probability distribution and it is necessary to perform an uncertainty propagation on the plant accident rate to assess the impact of the variability of the parameters used. We use a lognormal distribution for this purpose because all rates may be written as m × 10^{n}, where 0 < m < 10 is a real number and n is an integer and the involved ranges vary by factors of 10^{±p} around a median (p is an integer). In this sense, n follows a normal distribution and the rate itself follows a lognormal distribution. However, it is not reasonable to consider that rates may assume values from zero up to infinity, so it is necessary to consider a truncated probability distribution. We discuss the truncated lognormal distribution, its properties and perform the Monte Carlo uncertainty propagation by the inverse transform method for generating random numbers. Considerations on the mean accident rate and its standard deviation are presented.