Cloud Manufacturing platforms have recently been proposed to develop more agile and flexible Supply Chains, while improving utilization of manufacturing tools. This subject has therefore drawn a number of studies in the past years, but still requires further work before being successfully implemented on a large scale. The process of Service Selection in particular is still presenting many challenges. Service Selection consists in finding the best combinations of suppliers among those registered on the platform in order to complete a given production order. Previous research has suggested different algorithms to find an optimal or near-optimal combination of suppliers, but there is no consensus yet on which one to use, as none are entirely satisfying for real-world implementations.
The first step of this thesis is to investigate the concept of Cloud Manufacturing platforms in the literature, to learn about the state of the art in this field and in particular on the Service Selection Process and thus identify precise gaps in the research. It also allows to understand the steps that are needed during normal operations of Cloud Manufacturing platforms, which is necessary to define inputs and outputs of algorithm for Service Selection.
Then, the development of three algorithms - two of them being heuristics, and their testing performed on a simplified cased confirmed the need to use heuristics to perform Service Selection. Their benefices include the reduction of computational time, the possibility to return near-optimal results to large and complex problems (involving many production steps and a great number of potential suppliers).
Additionally, the influence of transportation on the algorithms performing the Selection Process has been studied. It is shown that in order to find the optimal combination of suppliers in term of time and cost, it is necessary to take transportation into account. Moreover it is demonstrated that considering this data also impacts heavily the performances of the algorithms, as it increases the complexity of the problem. Finally the performances of a specific heuristic, an Ant Colony Optimization (ACO) to perform the Service Selection process of a Cloud Manufacturing platform have been studied. The results show that near-optimal results can be found thanks to this heuristics with good computational times. However, the absence of a systematic method to tune the parameters of the algorithms reduces its applicability.
To reach the objectives, a literature study as well as algorithms experimentation have been led. The literature study allows to gain deeper knowledge in the field of Cloud Manufacturing platforms and on the Service Selection process, and to build on actual knowledge to investigate the research questions. The implementation and testing of algorithms have allowed to understand how various parameters influence their performances, why are heuristics needed to perform the Service Selection process, and how adapted ACO are in particular.