Sustainable product design through product modularization / Hayam Gamal Mohamed Wahdan ; Supervised Hisham M. Abdelsalam , Tarek H. M. Abouelenien , Sally S. Kassem

Thesis (Ph.D.) - Cairo University - Faculty of Computers and Artificial Intelligence - Department of Operations Research and Decision Support

Modularity concepts play an important role in the process of developing new and complex products; it is a powerful means in designing Sustainable products. Modularization involves dividing a product into a set of modules, these modules - consisting of a set of components - that are interdependent within and independent between clusters. These modules can be replaced or removed through the production process to create new product or add new function. Removing or replacing modules isn{u2018}t causing significant impact to economic, social and environment aspects that represent sustainability. During this process, a product can be represented using Design Structure Matrix (DSM) that acts as a system analysis tool; it is providing a clear visualization of product elements and the interactions between these elements.This research aims at providing a sustainable product design using modularity concepts. A set of steps are followed to achieve this purpose. First, an efficient optimization algorithm is developed to dynamically divide a DSM into an optimal number and size of clusters in a way that minimizes the total coordination cost. The total coordination cost consists of the interactions inside clusters (modules) and interactions between clusters. Given the complexity and the combinatorial nature of the optimization problem, six meta-heuristic optimization algorithms are selected to solve the problem, these algorithms are mainly used to determine: (1) the optimal number of clusters in a DSM, and (2) the optimal assignment of components to each cluster in order to minimize the total coordination cost and maximize sustainability objective. The six used meta-heuristics are: Cuckoo Search, Modified Cuckoo Search, Particle Swarm Optimization, Simulated Annealing, Gravitational Search Algorithm and Emperor Penguins Colony. Eighty problems with different complexity and dimension are generated and used to examine the proposed algorithms for effectiveness and efficiency. Extensive comparative analyses are conducted. Cuckoo Search outperforms the other five algorithms in most cases

Issued also as CD