Joint production, quality control and maintenance policies subject to quality-dependant demand

Abstract

This thesis is a strive to find a proper solution, using the stochastic optimal control means for an unreliable production system with product quality control and quality-dependent demand. The system consists of a single machine producing a single product type (M1P1) subject to breakdowns and random repairs and must satisfy a non-constant rate of customer demand, which response to the quality of parts received. Since the machine produces with a rate of noncompliant products, an inspection of the products is made to reduce the number of bad parts that would deliver to the customer. It is done continuously and consists of controlling a fraction of the production. Approved products are put back on the production line, while bad products are discarded. The intended objective of this study is to provide optimal quality control and production policy, which maximize the net revenue consisting of the gross revenue, the cost of inventory, the cost of shortage, the cost of the inspection, the cost of maintenance and the cost of no-quality parts. Main decision variables are the sampling rate of the quality control system as well as the threshold of finished product inventory. The demand function reacts to the average outgoing quality level (AOQ) of finished products. In the third chapter of this study, preventive maintenance and dynamic pricing policies are added up to the optimal policy, cited above. To achieve the optimal points of the policy, which maximize our net production revenue, a simulation approach is implemented as an experimental design and its results were used in response surface methodology. To implement the experiment design (simulation approach) which thoroughly reflects model considerations such as its continuous nature and the variety, first, a continuous variable for the probability of defectiveness was introduced, functioning with the age of machine up until its next breakdown maintenance. Second, so as to reflect the effect of quality control process that results in Average Outgoing Quality rather than simple defectiveness possibility, this function (AOQ) was built based on instant behavior of mentioned function above as its independent variable. Third, due to the use of prospect theory assumptions in building a demand function that responds to the level of client delivered defectiveness (AOQ), a responsive continuous function was created for the demand, reacting to the level of product quality by determining it's needed per time amount. Finally. To illustrate the machine’s manufacturing policy based on Hedging Point, finished product inventory variable was introduced in the experiment design. In a nutshell, we have a production system that has been designed in a way that by raising its age (At), leads to more possibility of defectiveness and less demand in time units. This manner continuous up until the next maintenance action of the system, which restores all factors to their initial conditions. By use of the simulation approach of optimization an experiment is designed and implemented to control decision variables of the policy and maximize the objective function of average net revenue (ANR). Decision variables are statistically and practically in the matter of consideration such as finished product inventory threshold (Z), the proportion of inspection (F) and PM thresholds (Mk or Pk).

Date	25 Nov 2019
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Mustapha Ouhimmou (Supervisor) & Ali Gharbi (Co-supervisor)