Optimality, flexibility and efficiency for cell formation in group technology

PhD ceremony: Mr. D. Krushynskyi, 14.30 uur, Academiegebouw, Broerstraat 5, Groningen

Dissertation: Optimality, flexibility and efficiency for cell formation in group technology

Promotor(s): prof. R.H. Teunter, prof. B. Goldengorin, prof. J. Slomp

Faculty: Economics and Business

The thesis focuses on a development of optimal, flexible and efficient models for cell formation (CF) in group technology. By optimality is meant guaranteed quality of the solutions provided by the model, by flexibility - possibility of taking additional constraints and objectives into account, by efficiency - reasonable running times. The main aim is, thus, to provide a reliable tool that can be used by managers to design manufacturing cells based on their own preferences and constraints imposed by a particular manufacturing system.

Though the CF problem has been extensively studied for more than 50 years, there have been very few attempts of solving the problem to optimality and almost all the proposed models for CF problem are either of intuitive (heuristic) nature or are solved by heuristic procedures. This means that the obtained solutions incorporate two types of errors: an intrinsic error of modelling and a computational error induced by a heuristic solution procedure.

The author proposes two models based on the p-Median and the minimum multicut problems, respectively. The first model has very short running times (usually less than 1 sec.) at a cost of a small modelling error. The second approach excludes the modelling error but has substantially larger (yet, practically acceptable) running times. Both models are expressed in terms of Mixed Integer Linear Programs and require a general-purpose MILP solver, like CPLEX or Xpress. Several realistic constraints and objectives, as well as the ways of introducing them into the proposed modes, are discussed.