Data-driven congestion management for LV distribution networks

With the presence of Distributed Energy Resources (DERs) such as heat pumps (HPs), electric vehicles (EVs) and photovoltaics (PVs) within the modern distribution networks, conventional power flow is being significantly altered [1]. This situation becomes worse during the winter time, when the HPs with large rating consumption is running for a long period. The stress on the grid infrastructure is tackled conventionally by network reinforcement (building or/and upgrading new cables and new transformers to meet the loads requirements). However, the fact that load locations in distribution networks spread widely over geographical areas makes the network reinforcement more financially infeasible in a short term [2]. Another alternative for this solution can be the procurement of demand flexibility offered by owners’ DERs so that the network infrastructure can reach a better utilization [3].

In support with the European Union's efforts for proactive, advance, secure and intelligent digital grids [4], the H2020 project UNITED-GRID aims to develop technical solutions to serve needs and opportunities for DSOs in their electricity grids [5]. In this paper, under the UNITED GRID project objectives, a mechanism is developed to support the DSOs with a better grid management scheme regarding congestion management by incorporating data-driven congestion scheme and thus reducing the congestion. The objective of the work is to avoid/reduce thermal overloading of the transformer while reducing cost [6]-[7]. An algorithm is developed, which can help DSOs to solve congestion problems in the LV distribution network. A demand response is applied for procurement of flexibility, using available flexibility in order to mitigate the congestion in the distribution network.  Once the congestion occurs, DSO can refer to the data driven classifier model and mitigate the congestion with the minimized cost. A typical Dutch LV distribution network, is considered for analysis assuming every house is equipped with at the most three types of appliances: buffer appliance, time-shifter appliance and uncontrollable appliance. Different evaluation scenarios are considered for flexibility procurement and developed model is tested in a simulation environment. It is observed that the DSO have significant savings and reduction of congestion occurrence in LV distribution network. Performance and accuracy of the developed data driven models is also found to be well within the satisfactory limits. Detailed methodology and results analysis would be shared in full paper.

 References 

[1]  M. J. M. Al Essa and L. M. Cipcigan, “Integration of Renewable Resources into Low Voltage Grids Stochastically,” ENERGYCON, Leuven, 2016, pp. 1-5. 

[2]  L. P. M. Gutierrez et al., “Smart Design is more Effective than Smart Grids”, IEEE Eindhoven PowerTech, Eindhoven, 2015, pp. 1-5. 

[3]  Pudjianto et. al., “Smart Control for Minimizing Distribution Network Reinforcement Cost due to Electrification”, Energy Policy, Elsevier, vol. 52(C), pages 76-84, 2003.

[4]  https://ec.europa.eu/energy/en/topics/energy-strategy-and-energy-union/2020-energy-strategy.

[5]  https://united-grid.eu/

[6]  E. Veldman, “Power Play: Impacts of Flexibility in Future Residential Electricity Demand on Distribution Network Utilisation”, 2013. doi: 10.6100/IR761384.

[7]  A. Haque et. al, “Real-Time Congestion Management in Active Distribution Network based on Dynamic Thermal Overloading Cost,” in Power Systems Computation Conference (PSCC), 2016.