Model-Free Dynamic Operating Envelopes to support increased penetration of distributed energy resources and reduced curtailment

(PV), battery energy storage systems (BESS) and electric vehicles (EV) are leading to network congestion where high penetrations of these assets exhibit coincident generation or load profiles. Solar PV inherently exhibits a coincident profile within a geographic region and other

DER, particularly price-responsive resources (PRR), can also display coincident behaviours, which may breach network constraints with high enough penetrations. Traditionally, to mitigate this, fixed limits such as an export limit on a PV system have been applied to avoid distribution network augmentation.

A Dynamic Operating Envelopes (DOE) specifies a varying power range, typically at the connection point, for exports and imports that are within the operational limits of the network.

In comparison to fixed limits, they enable increased flows between DER and the network as they only restrict flows in periods and locations when congestion occurs rather than always applying to manage worst case conditions. This reduces the need to curtail generation or consumption from flexible DER in constrained network areas with high DER penetrations and supports ongoing connections. The first quasi-DOE implemented uses allocated firm export capacity, customer numbers and seasonal load and generation profiles to determine a seasonal schedule and is referred to as a scheduled operating envelope (SOE). The SOE typically restricts exports in the middle of the day (due to solar) and imports in the evening, when residential load profiles typically peak. The model-free (MF) approach to DOEs can determine more optimal envelopes within network constraints in areas with poor model quality (such as the low voltage network) using grid visibility from network monitors and Advanced Metering Infrastructure (AMI) combined with a basic low voltage (LV) network hierarchy, without requiring a detailed network impedance model. This project demonstrated that MF DOE result in significantly lower levels of curtailment than SOE, better address voltage constraints and can allocate DOE among dynamic connections more optimally than a simple equal allocation. We compare performance on the Energex and Ergon Energy Network in Queensland, Australia, under baseline and 100% solar uptake scenarios, with equal and optimal allocation policies, with and without voltage constraints.

Simulations across 52 LV networks on three medium voltage (MV) feeders demonstrate substantial reductions in curtailment with the model-free DOE relative to SOE, particularly under optimal allocation as shown in Table 1. The increased curtailment by the MF DOE with equal allocation was due to the network voltage management equitably constraining neartransformer customers. The modelled SOE only resolves thermal constraints and does not address voltage constraints so underestimates the curtailment necessary when autonomous inverter voltage response modes apply to maintain acceptable voltages.

Table 1 – Summary of curtailed energy results Scenario

Baseline 100% solar Equal SOE

Curtailed Energy 3.9 % 5.1 % Equal MF DOE

Curtailed Energy 1.1 % 6.6 % Optimal MF DOE

Curtailed Energy 0.2 % 1.9 % Voltage prediction accuracy improves with grid visibility (AMI coverage), with transformer voltage root mean square error (RMSE) decreasing markedly above 20% penetration.

These findings indicate that model-free DOEs can reduce curtailment and increase DER hosting capacity, provided sufficient near real-time data (>20% AMI coverage).