My current research interests lie in mathematical optimisation and uncertainty management, especially through robust and stochastic programming techniques.
These developments started with easing the modelling of industrial processes with generic block-diagram models and their associated mathematical formulation. Thanks to this kind of models, those processes can be used in mathematical-programming-based solutions for optimising plants. Currently, a seemingly good candidate for this modelling is the reservoir, which can model many kinds of processes.
Once the process models are developed, they should be integrated within a complete plant model to actually exploit the flexibility. Outside the processes, two elements must be considered: the order book, but also the workers, as they are at the heart of the production. Industrial sites tend to use shift work to be able to produce twenty-four hours a day; however, this requires to organise the shifts according to legal constraints. Our approach considers the two faces of the coin: production and shift work. The first one exploits the flexibility, while the second optimises the schedule for the workers' well-being.
The source code of the integrated models will be freely available on GitHub. The production-HR coupling has been presented at the COMEX workshop.
Formal optimisation can take into account the uncertainty in two main ways: either stochastic or robust programming. Both paradigms have their own strengths, based on quite different theoretical notions: stochastic optimisation considers the uncertain parameters are described by their probability density functoin, while robust optimisation regards them as belonging to an uncertainty set.
Those two approaches are quite different, but have rarely seen a direct comparison, be it for their actual performance when solving the resulting optimisation problems, but also for the robustness of their solution against unforeseen uncertainty.
Last modification: July 08 2017 21:02:56.