The data analytics team at Water Corporation has developed two related but distinct optimisation models for the scheduling of water:
- Groundwater bore selection per treatment plant
- Water transfer from treatment plants and other sources to reservoirs and customer distribution within a network.
These describe a decomposition of the problem, and reflects business process of the water scheduling.
We are looking at ways to connect the solving of the two models together in an efficient way, taking the results of one of the models and then applying to the other model. The second set of results then needs to be referenced by the first model, and so on, in a “negotiation” between the two models. At present, the only way to reference the preceding solution is manual import.
Water Corporation is seeking understanding of how to connect up the results of the decomposed problem to produce better results for the overall system. Starting with a literature review, an exploration of the dependencies of the two decomposed models will be valuable before testing proposed methods.
The existing models are relatively mature and use a commercial optimisation solver. Competency with python is an essential skill to combine with mathematical programming and optimisation reasoning.
Site Work: Leederville JTWC and Subiaco WRRF
*APPLICATION: Please nominate on your application, if you are seeking a Full (with vacation work) or 3/4 (without vacation work) project. Please note; If a suitable candidate applies for Full, applications for this project may be closed early.