On Friday, July 5, 2019, Joost de Kruijff will defend his PhD thesis “High-tech Low-volume Production Planning”. This thesis has been supervised by prof.dr. A.G. de Kok, dr.ir. C.A.J. Hurkens and dr.ir. N.P. Dellaert. The ceremony will take place in room 0.710 of the Atlas building, Eindhoven University of Technology at 11:00 hrs.

Abstract

Production planning decides on the timing and sizes of production orders while aiming to utilize the available resources and materials efficiently and taking into account current and future customer demand. This thesis focusses on an under-represented application area of the production planning literature: High-tech low-volume industries. Commercial airplanes and photolithographic machines are two examples of items that are produced by at most a few hundred per year. These industries often have long lead times, large bills of materials and complex resource requirements. Models for high-volume production planning, which are common in literature, often do not capture these complexities and output continuous production quantities. In low-volume environments, rounding these continuous production quantities leads to infeasible or far from optimal production plans. Therefore, models should output integer production quantities, which makes the models much harder to solve. This thesis aims to develop models and solution methods for high-tech low-volume production planning, while focusing on three research subjects.

The first research subject considers mid-term planning. A mixed integer linear programming model is introduced, which utilizes semi-flexible capacity constraints and allows different production modes to capture the complexities of the high-tech low-volume industries. Benders’ decomposition is applied to get an alternative formulation and a faster solution method. The second research subject investigates the effect of the choice of integer variables on the solvability of low-volume production planning models. The integrality of the inventory variable has a significant impact on the solvability without changing the optimal solution. Furthermore, a preprocessing algorithm is proposed, which sets the integrality of variables based on the LP-relaxation. The third subject considers long-term production planning. A multi-stage stochastic programming model is proposed, which can be used when an uncertain structural demand change is expected. The model decides on production quantities and capacity (dis-)investments. It takes the long lead times of production and capacity (dis-)investments as well as the uncertainty in the demand and the reaction on future demand information, into account. A number of insights are gained by experimenting with a system closely resembling a real-life situation.