DSO Led-Bilevel Optimization Framework for TSO-DSO Coordination across Active Distribution Networks

TL;DR

This paper introduces a bilevel optimization framework for TSO-DSO coordination that models hierarchical interactions, enabling efficient multi-ADN operation with improved local self-consumption and reduced transmission imports.

Contribution

It presents a novel single-level MISOCP formulation for TSO-DSO coordination that captures discrete and convex aspects without relaxing binary variables.

Findings

Enhanced BESS utilization and local self-consumption.

Reduced imports from the transmission network.

Scalable computational performance with multiple ADNs.

Abstract

This work presents a bilevel coordination model that captures the hierarchical interaction between the transmission and distribution layers under a Distribution System Operator(DSO)-led configuration. In this scheme, multiple DSOs independently optimize the operation of their active distribution networks (ADNs), including photovoltaic (PV) generation, battery energy storage systems (BESS), and peer-to-peer (P2P) energy exchanges both within and across ADNs through the Transmission Network (TN), before the Transmission System Operator (TSO) performs the global coordination. The proposed formulation combines the Second-Order Cone relaxation of the DistFlow model to represent the distribution networks (DNs) with the classical DC optimal power flow (OPF) model for the transmission layer. The DSO-first decision sequence enables the reformulation of the bi-level problem into an equivalent single-level optimization model using the Karush-Kuhn-Tucker (KKT) conditions, resulting in a Mixed-Integer Second-Order Cone Programming (MISOCP) formulation that captures both the discrete and convex characteristics of the problem, while preserving the binary variables associated with DER and P2P operation, which would otherwise need to be relaxed in traditional TSO-led approaches. The model is tested on a hybrid system composed of the IEEE 30-bus transmission network and five IEEE 33-bus DNs. Results show that the DSO-led coordination leads to a more efficient use of BESS, improves local self-consumption, and reduces imports from the TN compared to the conventional top-down scheme. Furthermore, computational results from the case study reveal that the model exhibits near-linear or quadratic growth in problem size as the number of ADNs increases, suggesting its applicability to large-scale multi-ADN configurations.