Enedis (France's main electricity DSO) and Mines Paris Tech are working on a new method called MOSAIC that aims at assessing the impact of long term local development projects on the electrical grid. The MOSAIC method uses a bottom-up simulation tool able to determine the current and future consumption and production load curve of an area. The consumption and production simulators are based on a multi-level data model that makes it possible to run a simulation even if some data are missing. The simulation parameters are calibrated by comparing the simulated load curves with observations on MV feeders (for the consumption simulator) and on 100 renewable energy producers (PV and WP). Once the simulators are calibrated for the current situation, the load curve of each development scenario of the area is estimated (the evolution scenarios are proposed by the local government, based on infeed growing, building projects …). Enedis and Mines Paris Tech are now working on linking the load curve simulators with more traditional electrical study tools. From the simulated load curve, we estimate the most likely maximal power of the grid infrastructures and use these values in the load flow tool called ERABLE to evaluate the impact of the scenarios on the grid. Initial tests were successful and this method will be further developed in 2017 on ten experimentation projects in France. Each project will help to develop new functionalities such as integrating Demand Response or electrical vehicle infrastructures.
This paper presents a novel approach of an electric load curve simulator using a set of grey box models that results to an efficient trade-off between complete and complex physical models and fast simplified statistical models. The input parameters are macroscopic data coming from large databases such as national census, DSO's client information and meteorological data such as temperature or irradiation data. The problem of matching between the different databases is investigated to assess comparable load curves. Validation is performed using load measurements at the medium voltage level. Once the model is calibrated it can be turned into a good prediction tool useful for planning studies since it permits easily to incorporate the evolution of usages, the characteristics of consumption devices, as well as the evolution of the building's characteristics.
In the last decades, renewable energy sources have been increasing their shares in the world energy market. In addition to the ecological benefits, this trend can have adjunct benefits, for example for distribution system operators: a gain in their grid sizing. Indeed, installation of decentralized production, when used in a selfconsumption approach, can lead to reduction of the consumption peaks. This work is willing to quantify what grid sizing reduction a distribution system operator can expect, knowing the renewable energies penetration rate on a MV feeder. To do so, a description of the actual sizing strategy is first described. Estimation of electricity demand is performed using a bottom-up simulation method while photovoltaics and wind power production are evaluated with reanalysis data coupled with a new method to inject variability to the smooth curves. This procedure leads to a new sizing power which can be used, guaranteeing an equivalent quality of supply for consumers. For the tested MV feeders, a maximum reduction of about 4 % of the sizing power is observed. Lastly, an analysis of the under-sizing risk is carried out, characterizing the error in the new sizing power estimation with the number of scenarios taken into account.
The objective of the paper is to provide a clear and comprehensive analysis of probabilistic load flow using Point Estimation Method, and its accuracy in the evaluation of the quantiles of the state variables of real electrical distribution networks. Three Points Estimation Method (TPEM) has been implemented to evaluate the first four moments of output variables (voltages, currents and active power flows), and several methods to reconstruct the probability density function from moments and calculate the quantiles have been compared, including Generalized Lambda Distribution and Gram-Charlier Development. Monte-Carlo PLF has been taken as reference to evaluate the accuracy of aforementioned TPEM results and the scope of application of the method in a real electrical distribution system.
This paper presents a method that permits to match customer information from the French DSO Enedis and housing information from the French population census institute INSEE. Our method allows having a list of housings linked to each customer in order to add household and building information to customers. We show with our method improvements in predictions of aggregated load curve indicators compared to the traditional method that averages socio demographic indicators from housing information of the zone covered by measurements. Our results indicate that the proposed algorithm is able to capture efficiently the information of housings in some feeders. This permits to combine the databases of the DSO with external databases that exist from census or other processes. Enriching the information at the level of clients through the proposed automated way is a cost effective approach given the number of customers served by a DSO. This enhanced information can be then the basis to model, analyse and simulate demand in a bottom up approach which can be useful for planning purposes of the distribution networks.
This paper participates in the challenging data science opportunity offered by the growing number of databases made available to public institutions. It presents an innovative method to match household-scale databases using address information. The developed algorithm authorizes different matching qualities, depending on the reliability of the link between the paired elements. This work was carried out in collaboration with the French DSO Enedis, which provided valuable customer information that was matched with a national database describing dwellings. The matching algorithm performances are analyzed, and adjustments are proposed to improve the matching quality in urban, suburban and rural contexts. Lastly, two basic characterization analyses were made to highlight the potential of these consolidated databases.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.