4-6 months master training position
Subject:Uncertainty quantification in new power systems elements and model order reduction
Keywords : decentralized production, loads, simulation models,stochastic methods, model order reduction
Description:The current representation of the European power system for dynamic studies consists in the interconnection of a dynamic model for each geographic zone (which, in most cases, corresponds to a country). The model of each zone itself consists in the interconnection of major elements of the zone (generators, transmission lines, transformers, loads). This methodology should be now revisited due to several facts:
-the renewable energy was systematically neglected in the previous models. It reaches today and in perspective a level which should be taken into account. However, its integration into a large-scale dynamic model is a challenging task since, in general, this production consists in small units which are geographically distributed,
-the distributed generation (low power injections in manylocations of the grid) is more and more present under the form of cogeneration. It has the same characteristics mentioned at the previous point,
-an increasingvolume of new types of loads like the electric cars is expected.
One difficulty for the incorporation of these elements is related to the size of the model obtained in fine. Aggregation or model order reduction methodsfor dynamical systems should be investigated.One challenge is here to provide generic projection-based order reduction methods that are able to preserve some physical structures (ultimately, we would like a reduced order model in which the lines, generators… are correctly modeled). A strong investment in the specific field of power system and transmission grids is expected since ad-hoc order reduction methods may arise from a good understanding of the specificities of these models.
Another challenge is to provide equivalent models able to capture several phenomena of different nature (oscillations of various types, fast transients...). This kind of objectives should be taken into account a priori in the new proposed methodologies of reduction or aggregation (i.e. should be integrated into the specifications)and not only checked a posteriori on a limited number of cases.
Finally, one should also manage the random nature of some dynamics. This is the case of renewable generation (wind power) and particular loads like the electric cars. It is therefore necessary to find ways to quantify and take into account uncertainties in the models for both the simulation and the synthesis of control laws. In particular, model order reduction methods for stochastic dynamical systems will have to be proposed.
This work requires a strong background in applied mathematics, in particular in numerical analysis, differential equations, stochastic modeling and model reduction. This work alsoinvolves a system-based approach of the electric network. Therefore, a background in the analysis of dynamical systems will also be appreciated.
Framework andcontext:thisworkisproposed in theframeworkofthechairbetweenEcoleCentrale Nantes (IRCCYN andGeMlaboratories) andRéseau de Transport d’Electricité (RTE) R&D Division. RTE isthe French transmissionsystemoperator. Providedwith an R&D divisionabout 120 people, ithas an activerole in theinvestigationofnewmethodologiesfortheanalysis, control, simulationanddeveloppementofthe power transmissiongrid in France and Europe. Thepositionisbased in Nantes, EcoleCentrale Nantes.This position is intended to be extended by a PhD thesis. The candidates should take this into account.
Contacts:
B. Marinescu, ECN-IRCCYN, , 02 40 37 69 46
A. Nouy, ECN-GeM, anthony., 02 40 37 16 76