Guilherme Neves
Department of Industrial Engineering, Pontifical Catholic University of Rio de Janeiro

Madiagne Diallo
Research Productivity Grant of CNPq—Brazilian National Research and Development Centre and Department of Industrial Engineering, Pontifical Catholic University of Rio de Janeiro

Leonardo Junqueira Lustosa
Department of Industrial Engineering, Pontifical Catholic University of Rio de Janeiro

Abstract: There is a consensus that conventional continuous demand distribution methods are not appropriate for forecasting replacement parts. However, many forecasting tools available in market still use them. This work presents an application of the Poisson distribution to forecast the needs of electronic spare parts. Using basic stock management notions and usual concepts of reliability, availability and the Poisson process, an alternative method is proposed for sizing the initial stock of replacement parts to be purchased along with a electronic equipment. The results from the application of the proposed method and its comparison to the SAGA method, which is based on time series and normal distribution, are presented. The analyses of results have shown that it is possible to reduce the forecast errors; hence the stock costs, and the number of stockouts, thus enhancing the operational availability.

Keywords: Replacement parts, spares, forecast, stock management, supply, time series, availability, reliability, Poisson.