Test Case 9

9

Tran The Hoang, Luigi Pellegrino, Quoc Tuan Tran

1

Erigrid 2.0

02/03/2021

Evaluation of congestion management in distribution grid

The distribution network has been becoming congested because of the introduction of bi-directional power flow (due to the increasing penetration of DERs), unpredictable and increased power demands for consumption by the residential consumers (due to the introduction of new forms of loads such as Heat Pumps, EVs, etc.). As a result, the distribution network operators (DSOs) need to focus on the challenge of balancing power supply and demand. Congestion in the distribution network refers to an overvoltage at the connection points as well as overloading of the network components.

On the one hand, to mitigate the network strains, conventionally DSOs focus mainly on network development by installing new cables and transformers to meet the increasing power flows. Nonetheless, the distribution loads are spread over large geographically areas and in a distributed manner, making the upgrade of the network more financially infeasible in a short term. Another alternative solution is to develop grid congestion management approaches so that the network infrastructure can be utilized in a better way. There are two types of congestion management methods namely direct as well as indirect. The former technique is realized by performing load curtailment, local generation reduction, network re-configuration, new installation of Battery Energy Storage System (BESS). In contrast, the latter approach focuses on solving the optimization of electricity cost with the constraints ensuring the transformers/feeders not to be overloaded.

The direct congestion management method in this test includes two stages. The first stage consists of using a machine learning method, such as support vector machine, multi-class classification, decision tree, ANN..., in order to build congestion classification models. Once congestion is detected, it has to be labeled to one of the following statuses: normal, alert, emergency, and critical depending on the output of the trained models. In the second step, DSOs will use the congestion labeling to calculate the expected flexibility portfolio. With the expected procurement cost, the flexibility available in the feeders/households can be used to solve the congestion problem. After comparing the results with different conditions, the best setting for the congestion management can be chosen.

On the other hand, an indirect congestion management needs to be based on an online learning technique to emulate the demand flexibility of a network. As for emulating the demand flexibility, the concept of price elasticity of demand can be considered. Accordingly, demand flexibility during all time-periods of a day shall be treated as a commodity that can be substituted or complemented to each other.

The objective of this Test Case is to evaluate different congestion management methods in distribution grid under the circumstance of high penetration of DERs and other active loads such as EVs, HPs ...

Congestion management of the DMS controller

• Direct approach: mitigating congestions by curtailment of load and local generation and by influencing the voltage level at the secondary side of a MV/LV transformer
• Indirect approach: motivating individual prosumers with dynamic prices through intermediate market entities such as aggregators and retailers. DSOs also incentivize customers by providing compensations for their load reduction when needed to solve network congestions.

• DMS controller

• Electrical domains
• Control and ICT domain

• Characterization and comparison of different congestion management methods.

In electric power domain:

• DMS controller
• DER (PV system)
• Household appliances
• Distribution transformer
• Aggregator/consumer/prosumer controllers
• Household controllers
• Household appliance controllers
• DER controllers

In ICT domain:

• Communication network

• DMS congestion management functionality
• DER power output control
• Household appliances control of the aggregator/consumer/prosumer controller
• Communication via ICT

• Performance of the congestion management algorithm under realistic conditions
• The transformer and feeders should not be overloaded
• Reduction in the cost of flexibility procurement

• Accuracy of congestion prediction
• Transformer overloading/loss of transformer life/Hot spot temperature of the transformer/transformer loss
• Feeder overloading
• Cost of congestion management
• DER power curtailment
• Household voltage profiles
• Flexibility procured by DSO
• Reduction in peak demand

• Household consumption profiles
• DER generation (weather condition)
• Packet loss
• Communication delay

• Transformers/feeders are not overloaded
• Voltage deviation within ±10% (typically for LV networks)
• Reduction in DER power curtailment