Back to Projects

Model Free Operating Envelopes at NMI Level

Project Partners: University of Melbourne

Project in progress

Dynamic operating envelopes (DOE) are a key enabler for Distribution Network Service Providers (DNSPs) to integrate more distributed energy resources (DER) on to the electricity grid. These “envelopes” are contractual and operational limits that define how much electricity a customer can import or export to the electricity grid, using their DER such as rooftop solar, electric vehicle, or other storage device. In most cases these limits are fixed, typically at conservative levels, to protect network assets, customer devices, and maintain compliance obligations.

However, as more DER comes onto the network at the low voltage (LV) level – which is where homes and most businesses receive their power – applying DOE can allow electricity consumers to maximise their DER investments while allowing the networks to manage operations more efficiently and with less investment requirements.

Instead of static export constraints at all times, DOE allow for more export at times of the day where the system can handle additional export load and only constraints when it is a danger to the network.

For this to occur, DNSPs need detailed visibility into LV networks which typically, has required employing electrical models of LV circuit models including; impedances, phase connectivity, nodal placement, etc. This can be complex, time consuming, and sometimes inaccurate.

This project provides the technical foundations and specifications (algorithms and computing processes)  for  DNSPs to estimate the maximum power exports or imports of individual customers without employing electrical models of low voltage (LV) circuits (hence, model-free) and using primarily historical Advanced Metering Infrastructure (AMI) data. 

Model-free calculation of operating envelopes remove the challenges associated with production of LV circuit models  (which requires knowledge of impedances, phase connectivity, etc.) but will also make the operating envelope calculations many times faster than model-based techniques as there is no need for conventional power flows analyses. Crucially, the model-free calculations from this project can also be used for network planning purposes, i.e., to estimate the hosting capacity of different DER, from solar PV to electric vehicles, as well as the effects of demand growth.

At the heart of the model-free approach is the process to capture the physics of a given three-phase LV circuit are captured using a Neural Network (NN) and historical AMI data. The NN is trained based on net demand values (active and reactive power) of individual customers and their corresponding voltage magnitudes. Once the training is successful (i.e., accurate enough), the NN is ready to calculate voltages for any net demand values of that specific LV circuit. In other words, by capturing the physics of the LV circuit, the trained NN becomes a proxy for conventional power flow analyses used to calculate voltages. Once the trained NN is ready, it can be used in combination with an operating envelope algorithm to check if the values of operating envelopes that will be explored by the algorithm that are within statutory limits. 

Operating envelopes, calculated at the customer connection point (where the meter is – identified with a National Meter Identifier [NMI]) in near real time or, with forecast data, for the horizon of interest, can help DNSPs ensure network integrity, i.e., ensure that all voltages and flows are within limits. Once calculated, the operating envelopes can be passed on to aggregators (or individual customers) who manage behind-the-meter DER, such as batteries, so they can take appropriate actions. 

Given the thousands of LV circuits that would require offline computational processes, this project is investigating the production of a single NN that covers all LV circuits per distribution transformer or, potentially, per high voltage (HV) feeder.

This could make the approach even more scalable and facilitate the future deployment of more DER on LV and HV networks.

Categories: Energy Data Resources

Return to Top