Optimal Flexibility Allocation in Electrical Distribution Grids (PDF)

Master's Thesis from the year 2019 in the subject Energy Sciences, grade: 1.0, Technical University of Munich, language: English, abstract: With the rising adoption of Electric Vehicle (EV) technology and Renewable Energy Sources (RES), electric distribution grids are facing new challenges regarding congestion management. The present work steps into the topic of controlled charging mechanisms to reduce physical grid extension by utilizing flexible loads from EV. Although, existing research concludes a positive impact on congestion relief, less attention is given to a holistic and light system that is implementable under current circumstances. This thesis develops a novel system towards micro-auctions for local flexibility allocation amongst EVs to reduce grid congestion. A functional software prototype simulates a virtual market and grid environment. Each EV acts as an autonomous agent submitting bids to the local flexibility market, offering 15-minute charging breaks. Based on individual risk preference and state-of-charge, bidprices vary amongst EVs.

The Distribution Grid Operator (DSO) constantly asses grid status and contracts positive capacity during critical phases by accepting current bids. It can be shown, that regardless of the penetration rate of EVs, the proposed model balances the tested grid topology below the maximum workload and within a predefined range. According to simulation assumptions, a ninefold increase of EVs can be accommodated with the proposed model. Although, with monotonically increasing penetration rate, average charge-increase converges to zero. Due to the proposed intervals, EVs are grouped to continues batches with demandresponse latency. Once contracted, EVs remain charging or not-charging for 15 minutes. The assignment to certain batches does not change over simulation time. Based on the proposed request control mechanism, critical grid conditions can be reduced by 49%. Whereas quantitative results are limited to the proposed simulation assumptions, qualitative effects are generalizable to a certain extend.