Content area
Abstract
Cette these aborde le probleme du dimensionnement des stocks tampons dans une ligne de production constituee de n processeurs ou postes de travail. En l'absence de ces stocks, toute defaillance d'un processeur entrai ne l'arret de la ligne de production. Un stock tampon dont la taille est correctement choisie, permet de decoupler les differents postes de travail et rend la ligne moins vulnerable a la defaillance de ses processeurs.
Le projet de recherche traite dans cette these est issu d'une problematique industrielle dont l'examen a permis de realiser la complexite du probleme et les limitations des modeles proposes dans la litterature. Ainsi, il a fallu rappeler les specificites des differents systemes de production et la pertinence du stock tampon. Une revue complete des differentes approches analytiques a ete effectuee. Une attention particuliere a ete accordee aux approches quantitatives conduisant a l'optimisation de la taille du stock tampon et a l'evaluation de la performance d'une ligne de production.
Outre les extensions apportees aux divers modeles proposes dans la litterature, des contributions importantes ont ete introduites au niveau de l'exploitation de la programmation lineaire comme modele d'analyse d'une ligne de production concue pour traiter des produits differents.
Plusieurs alternatives ont ete developpees pour prendre compte la fiabilite et la maintenabilite de chaque processeur. Ces facons de faire exploitent la structure lineaire du modele tout en le rendant plus proche de la realite industrielle.
Pour tous les modeles analyses dans cette these, des outils informatiques robustes ont ete mis au point. Des indications sont fournies sur le processus de modelisation et d'exploitation des outils developpes pour traiter efficacement le probleme de deploiement des stocks tampons dans les lignes de production.
Une extension du modele de programmation lineaire a ete developpee pour traiter le probleme de conception d'une ligne de production lorsqu'un temps de cycle est impose. Le modele genere la taille du stock tampon a prevoir pour atteindre le temps de cycle vise.
Plusieurs outils developpes dans cette these pourront etre utilises efficacement pour la conception et l'exploitation d'une ligne de production traitant un ou plusieurs produits et utilisant des processeurs dont la duree de vie et de reparation, en cas de panne, sont aleatoires.
Une modele de simulation utilisant un logiciel commercial a ete developpe. Ce modele a permis de valider les extensions developpees. Il a ete concu de maniere a permettre a l'utilisateur de configurer l'unite a simuler et de choisir les parametres caracteristiques des differents processeurs et produits traites.
Alternate abstract:
You are viewing a machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
This thesis addresses the problem of sizing buffer stocks in a production line made up of n processors or workstations. In the absence of these stocks, any failure of a processor will lead to the shutdown of the production line. A buffer stock whose size is correctly chosen makes it possible to decouple the different workstations and makes the line less vulnerable to the failure of its processors.
The research project treated in this thesis comes from an industrial problem whose examination made it possible to realize the complexity of the problem and the limitations of the models proposed in the literature. Thus, it was necessary to recall the specificities of the different production systems and the relevance of the buffer stock. A complete review of the different analytical approaches was carried out. Particular attention was paid to quantitative approaches leading to the optimization of the size of the buffer stock and the evaluation of the performance of a production line.
In addition to the extensions made to the various models proposed in the literature, important contributions have been introduced at the level of the exploitation of linear programming as a model for analyzing a production line designed to process different products.
Several alternatives have been developed to take into account the reliability and maintainability of each processor. These ways of doing things exploit the linear structure of the model while making it closer to industrial reality.
For all the models analyzed in this thesis, robust computer tools have been developed. Indications are provided on the process of modeling and exploitation of the tools developed to effectively deal with the problem of deploying buffer stocks in production lines.
An extension of the linear programming model has been developed to deal with the problem of designing a production line when a cycle time is imposed. The model generates the size of the buffer stock to be expected to achieve the targeted cycle time.
Several tools developed in this thesis can be used effectively for the design and operation of a production line processing one or more products and using processors whose lifespan and repair life, in the event of a breakdown, are random.
A simulation model using commercial software was developed. This model made it possible to validate the extensions developed. It has been designed to allow the user to configure the unit to be simulated and to choose the characteristic parameters of the different processors and products processed.
