Flow management in production and in supply chain networks

Producing goods is the aim of manufacturing companies. Controlling and optimizing the production tool is necessary but not sufficient for a virtual enterprise. It is a fact that real competition no longer takes place between companies but between Supply Chain Networks (SCN). A SCN is a set of plants connected by means of transport assuming the functions of providing raw materials, their transportation, their transformation in components then in finished goods and the distribution of the latter to the customer. Traditionally, the various actors of a network used to operate independently; each of them aiming to maximize one?s own interests. So were the different services of a given company. The reactivity induced by the current context requires an integrated management.

Design and management of supply chain networks

(L. Dupont, M. Dupuy, J. Lamothe, M. Lauras, J. Mahmoudi)

On top of the classical, still topical problems of production management, one has to add the problems of design and management of such networks. A supply chain behaves thanks to mechanisms of cooperation between the partners. By nature, these mechanisms evolve in time due to unstable contexts and/or environments. In order to support the dynamic of the network?s evolution, it seems judicious to try and define the phenomenon inducing such a change, to analyze the effects of this situation on the management of networks and to propose suiting management tools.

In the thesis of M. Lauras, this question is tackled in the context of the particular sector of pharmaceutical and cosmetic industry. The first objective of the study is to analyze the current functioning of various supply chains, to determine their forces and weaknesses and to bring out the best practices in a given context. The final objective is to use this knowledge to advocate the future management of a supply chain. Therefore, several practical tools are being proposed and developed. These tools are organized around five main points:

• Observe (1) to understand the existing ways of functioning,
• Classify (2) to target and categorize the actors and their relation modes,
• Model (3) and evaluate (4) to represent the cooperation modes and estimate their performance,
• Propose (5) to define and test the various possible solutions.

The thesis of J. Mahmoudi deals with the electronic sector (mobile phone, computer, etc). In this sector, the management of information appears to be the main cause of the problems observed. The customer?s final demand being not very easy to forecast, the information coming back to component manufacturers are uncertain and later on, hardly or not taken into account.

Consequently, partners have lots of difficulties to take up planning decisions, in particular on the increase or decrease of their manufacturing capacity. To help them in such a situation, we aim to simulate the behaviour of the supply chain in time, integrating:

• uncertainties about the market behaviour and on the production process reliability,
• an aggregated formalization of planning processes; Sales and Operation Plan (SOP), Manufacturing Production Schedule (MPS), short term stocks and production schedule,
• the procedures for information sharing between the different partners of the network,
• the behaviour of decision-makers facing capacity decisions?

The tool defined will have several uses. First, it will help an actor evaluate the risks in connection with his decision, for himself and for the network. Then, it has an educational vocation: help the actors of a supply chain network collectively define: in the scope of a ?win ? win? relation 1) the nature of the information that should be exchanged 2) the information sharing procedures. A first model validated this approach. A prototype of more important size is currently in progress.

Production management

(L. Dupont, M. Dupuy, J. Lamothe, P. Gaborit, F. Fontanili)

The thesis of M. Dupuy deals with the improvement of reactivity in the production of medicine and allied products by optimizing the control over materials and products. Through the last decade, the pharmaceutical industry underwent extremely important changes that made it compulsory for pharmaceutical companies to gather and to optimize their management. Product manufacturing, up to then, mainly focused on the respect of the good manufacturing practices. Moreover, it now has to take into account rigorous economic criteria. Production sites have been specialized according to their galenic form and progressively acquired an international dimension. Their management has been modernized, thanks to the setting up of software, generally Enterprise Resource Planning (ERP).

Activities of quality control have been little considered, although they commonly represent between one fourth and one third of the production cycle. Therefore, rationalizing their management would allow important time gains.

The objective of the PhD thesis (CIFRE), in collaboration with the Pierre FABRE company, was to carry out a business process reengineering of the control activity, to adjust the approaches of continuous improvement developed in other industrial sectors to this activity, in accordance with the specificities of the pharmaceutical industry. This work includes a first methodological part, which deals with modelling the quality control laboratory within a plant of pharmaceutical production, integrating the human, organizational, technical and of information system dimensions.

The second part aims to define a scheduling tool for chromatography chains that appeared to be a bottleneck. This problem generalizes various classical problems of the scheduling theory (parallel machines with set-ups, operator disponibility constraints, secondary resources), all known to be NP-hard. Two approaches have been experimented: priority rules and simulated annealing.

The last part deals with the global organization of a laboratory. The main difficulty concerns the diversity of analyses that have to be done and the multiplicity of the operations required per analyze. Simulations on the WITNESS software, using data that were collected thanks to the industrial case, are in progress to measure the various organizational possibilities.