ABSTRACT In this work, the presence of cut nodes in a network is exploited to propose a competitive method for the multi-terminal maximum flow problem. The main idea of the method is based on the relation between cut-trees and cut nodes, which is observed in the context of sensitivity analysis on the variation of edges capacities. Computational experiments were conducted with the proposed algorithm, whose results were compared with the ones of Gusfield, for randomly generated and well-known instances of the literature. The numerical results demonstrate the potential of the method for some classes of instances. Moreover, the proposed method was adapted for the single maximum flow problem, but failed to improve existing running times for the very same classes of instances.

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.