مقالات رزومه دکتر حسن صالحی فتح آبادی

In this paper the concept of the Minimum Universal Cost Flow (MUCF) for an infeasible flow network is introduced. A new mathematical model in which the objective function includes the total costs of changing arc capacities and sending flow is built and analyzed. A polynomial time algorithm is presented to find the MUCF.

In many real systems, it happens that the existing flow network become inconsistent with the new applications or inputs. This means that some of the applicable structural characteristics have been changed so that the flow network has become infeasible or, in other words, obsolete. Therefore, it has to be adjusted to new applications. It is well known how to use a maximum flow algorithm to determine when a flow network is infeasible, but less known is how to adjust the structural data such that the network becomes feasible while the incurred adjustment cost is minimal. This paper considers an infeasible flow network G= (V, A) in which supplies/demands, arc capacities and flow lower bounds are liable to relax. A minimum cost relaxation model for canceling most positive cuts is constructed. Analyzing the model shows that, in order to make the network feasible, it is sufficient to adjust only one component of the structural data. According to this result, a polynomial time algorithm is developed to cancel all positive cuts and convert the infeasible flow network to a feasible one.

In many real systems in which a state variable should be controlled for being in appropriate range, the length of control (review) intervals is taken to be constant. In such systems, when the cost of reviews and out-of-range values of the state variable are considerable, this method may not be optimal. In this paper we let the length of review intervals to be variable during each operating cycle and construct the related mathematical cost model. Then two scheduled review methods, called U2 and U3, are introduced and the relative annual system costs are analyzed. The model is developed for the case of negative exponential variate as the time between successive consumption points. It is shown that the new methods results a significant reduction in the expected annual cost of the system.