Racem Mellouli

A new realistic modeling approach for two-echelon logistics network design

In this paper, we deal with a single-commodity location-allocation problem for a two-echelon logistic network. We present novel models with a new reading of the problem. Indeed, some classical models in literature seem in certain cases simplistic and then generate gaps between theoretical formalisms and the reality. Concretely, these gaps arise when setting up and calibrating these models, thus the identification of robust parameter values could prove difficult. The reason is the aggregation process of data, since it is naturally used in modeling, lacks to provide sufficient accuracy. In a pragmatic framework, the issue of logistic network design is rich and presents several possibilities for data apprehension in terms of details to be modeled. This motivates to construct a more realistic modeling approach. In this context, we propose a way for integrating a maximum precision degree without leaving the context of modeling a strategic issue. Actually, the question of designing a logistic network is substituted by the need to compute a logistic master plan mixing the tactical with the strategic. This becomes a major concern of businesses and this observation fits the importance of location-allocation problems. Two mixed integer programming (MILP) are proposed. A derived linear programming (LP) is identified and used to describe a perspective of a cooperative hybrid resolution approach based on Genetic Algorithm. Experimental study provides a comparison of the models results.

Profitable Vehicle Routing Problem with Multiple Trips: Modeling and constructive heuristics

In this paper, we study a new variant of the Vehicle Routing Problem (VRP) which combines two known variants: the Profitable VRP (PVRP) and the VRP with Multiple Trips (VRPMT). This problem may be called as (PVRPMT). The problems objective is to maximize the sum of collected profit prior to the customers visit minus the transportation costs. This problem has a very practical interest especially for daily distribution schedules with a short course transportation network and a limited vehicle fleet. A discursive method for profit quantification is suggested so as to make it more realistic. A mixed integer programming formulation of the problem is proposed. In addition, two constructive heuristics and a set of benchmark instances which include small and large size data are provided. Finally, a computational study evaluates and compares the performance of the different proposed methods.

A Column Generation Method for the Parallel-Machine Scheduling Problem with availability constraint

In this paper, we study the problem of scheduling jobs on identical parallel machines with the objective of minimizing the total completion times of jobs where each machine is subject to one unavailability time interval. We provide a column generation based method to obtain a lower bound for the problem. Computational experiments show the interest and the versatility of the proposed method

Impact of means and resources pooling on supply-chain management: Case of large distribution

The aim of this study is to find the impact of means and resources pooling on the logistics networks. In fact, the pooling of resources and means, allows sharing resources and means in the logic of quality improvement and cost reduction, it will ensure a better functioning of the various operations management. In this context, we evaluate the difference between the actual situation and future situation of a firm by adopting the phenomenon of mass and that of the pooling. These results will be evaluated on the various costs associated with the company. However, our goal is to identify the different gains induced by the change of the organization of the logistics network of the firm. By this, we will try to optimize the various costs and more specifically transportation costs.

Branch-and-Bound Method to solve the P/sub m/, h/sub j1/||/spl Sigma/C/sub i/ problem

In this paper, we consider the parallel-machine scheduling problem with a maintenance scheduled period on each machine, to minimize the sum of completion times of jobs. We propose a branch-and-bound method to solve this problem. We define a new separation scheme for this method and we compare it with the first scheme that we preconized for this method at the beginning. Dominance properties, upper and lower bounds are described. A performance study, propped up by experimental tests, shows an important reduction of the search space and a better convergence delay by using the new separation scheme