⚡ electrical engineering

Identifying Best Candidates for Busbar Splitting

This paper proposes a set of metrics to efficiently identify and rank promising busbar splitting candidates in large power grids, thereby reducing computational complexity while effectively lowering generation costs through optimized topology without the need to test every substation.

Original authors: Giacomo Bastianel, Dirk Van Hertem, Hakan Ergun, Line Roald

Published 2026-03-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Giacomo Bastianel, Dirk Van Hertem, Hakan Ergun, Line Roald

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the electrical grid as a massive, complex highway system. Cars (electricity) need to get from power plants (generators) to your home (loads). Sometimes, the traffic gets so bad that certain roads get jammed (congestion), forcing the system to use expensive, slow detours or even turn off some cars (curtailment) to keep the system from crashing.

For years, grid operators have tried to fix this by building more roads (new power lines), but that takes forever, costs a fortune, and often faces public opposition.

This paper proposes a smarter, faster solution: rearranging the existing roads.

The Problem: The "Busbar" Traffic Jam

In a substation (a major intersection in the grid), all the wires often connect to a single giant metal bar called a busbar. Think of this busbar as a single, crowded roundabout where every car has to merge. If one road leading to the roundabout is jammed, the whole roundabout gets stuck, even if the other roads are empty.

Busbar Splitting (BuS) is the idea of taking that single roundabout and splitting it into two separate, smaller roundabouts. This allows traffic to flow more freely, bypassing the jam without building a new highway.

The Challenge: Too Many Choices

Here's the catch: A large grid has thousands of these roundabouts. If you want to find the best one to split, you can't just try splitting every single one.

The Old Way: Try splitting Roundabout A, see if it helps. Try Roundabout B, see if it helps. Do this for 3,000 roundabouts.
The Result: This would take a supercomputer years to calculate. It's computationally impossible.

The Solution: The "Traffic Detective" Metrics

The authors of this paper created a set of clues (metrics) to act like a traffic detective. Instead of checking every single intersection, they look for specific signs that tell them, "Hey, splitting THIS one will definitely save money!"

They developed three main clues:

The Price Gap Clue (Metric $\phi$ ):
- The Analogy: Imagine two towns. In Town A, a gallon of gas costs \ $2. In Town B, it costs \$ 10. The difference is huge. If you can build a bridge between them, you save a lot of money.
- The Metric: They look for intersections where the "price" of electricity on the connected roads varies wildly. If the prices are very different, splitting the busbar can help move cheap electricity to where it's expensive, saving money.
The Crowded Road Clue (Metric $\xi$ ):
- The Analogy: Imagine a roundabout where one road is bumper-to-bumper, but the other three roads are empty. That's a bad design.
- The Metric: They look for intersections where some connected roads are totally jammed while others are free. Splitting the busbar can redirect traffic from the jammed road to the empty ones.
The "Stuck" Sign Clue (Metric $\zeta$ ):
- The Analogy: Sometimes, a road isn't jammed with cars, but the speed limit is set to zero because of a broken traffic light (voltage limits).
- The Metric: They check if the intersection is already hitting its physical limits (like voltage). If it is, splitting it won't help because the road is already "stuck" for a different reason.

How They Tested It

The researchers didn't just guess; they ran simulations on grids ranging from a small town (39 intersections) to a massive metropolis (3,374 intersections).

The Test: They used their "Detective Clues" to pick the top 10-20 candidates.
The Result: They found that by only checking these few candidates, they could find the solutions that saved the most money.
The Speed: Instead of taking 15 days to check every single intersection, their method did the job in less than 4 hours.

The Big Picture

This paper is like giving grid operators a GPS for traffic jams. Instead of driving down every street to find the best detour, the GPS instantly points to the three best spots to re-route traffic.

By using these simple metrics, grid operators can:

Save Money: Reduce the cost of generating electricity.
Save Time: Make decisions quickly without waiting for supercomputers.
Avoid Construction: Fix congestion by rearranging wires instead of building expensive new power lines.

In short, they figured out how to find the "golden spots" in the grid where a little bit of rearranging creates a massive improvement, keeping the lights on and the bills lower.

1. Problem Statement

The integration of renewable energy sources (RES) and rising electricity demand are causing severe congestion in transmission grids. While expanding grid infrastructure is a solution, it is often delayed by high costs, supply chain issues, and public opposition. Consequently, system operators are turning to Grid-Enhancing Technologies (GETs), specifically Grid Topology Optimization (GTO).

Two primary GTO methods are:

Optimal Transmission Switching (OTS): Switching transmission lines.
Busbar Splitting (BuS): Reconfiguring substations by splitting a single busbar into two electrically distinct buses, connected by a coupler that can be opened or closed.

The Challenge:
While BuS can significantly reduce generation costs by alleviating congestion, optimizing it is computationally intractable for large grids. A full optimization requires introducing binary variables for every possible busbar split, leading to a combinatorial explosion. System operators cannot exhaustively test every busbar in a large network (e.g., thousands of buses) to find the few that yield cost reductions. There is a lack of systematic, pre-screening methodologies to identify the most promising candidates for BuS without running expensive full-grid optimizations.

2. Methodology

The authors propose a two-stage approach: Metric Development for candidate selection and Validation using a mixed-integer optimization model.

A. Proposed Metrics for Candidate Selection

The paper introduces a set of metrics designed to pre-screen busbars using data from a standard AC-Optimal Power Flow (AC-OPF) simulation. The goal is to identify busbars likely to reduce generation costs if split, without testing every busbar.

Metric $\phi$ (Sum of Absolute LMP Differences):
- Definition: The sum of the absolute differences in Locational Marginal Prices ( $\Delta$ LMP) across all branches connected to a specific busbar.
- Rationale: High $\Delta$ LMP indicates congestion. A busbar with multiple high $\Delta$ LMP connections suggests it is a hub for congestion, making it a prime candidate for splitting to redistribute power flows.
Metric $\xi$ (Number of Congested Branches):
- Definition: The count of branches connected to the busbar that are operating near their thermal limits (congested).
- Rationale: Identifies local bottlenecks where flow redistribution is most needed.
Metric $\zeta$ (Binding Voltage Limits):
- Definition: A binary indicator of whether the busbar is currently at its voltage magnitude or angle difference limits.
- Rationale: If a busbar is already at a binding limit, splitting it may not yield benefits or could violate constraints. This metric filters out busbars where voltage constraints are the primary bottleneck rather than topology.

B. Optimization and Validation Models

To validate the metrics, the authors employ a rigorous testing framework:

LPAC-BuS Model: They use a Mixed-Integer Convex-Quadratic model based on a Linearized AC (LPAC) approximation. This model allows for efficient optimization of the topology (splitting one busbar at a time) by using binary variables for switches and couplers.
AC-OPF Feasibility Check: The topologies generated by the LPAC-BuS model are validated using a full non-linear, non-convex AC-OPF simulation. This ensures the proposed topologies are physically feasible and accurately reflects generation cost reductions.
Testing Strategy:
- Testing Phase: Applied to small systems (39-bus, 118-bus) to refine the metric combinations.
- Validation Phase: Applied to large-scale systems (793-bus, 3374-bus) to test generalizability.

3. Key Contributions

Novel Pre-Screening Metrics: The paper proposes and validates a specific combination of metrics ( $\phi$ , $\xi$ , $\zeta$ ) to identify high-value busbar splitting candidates. Unlike previous works that relied solely on DC-OPF or single-line $\Delta$ LMP, this approach uses AC-OPF data and considers the aggregate impact of all connected branches.
Computational Efficiency: The study demonstrates that pre-screening reduces the computational burden drastically. Instead of optimizing every busbar in a 3374-bus system (which took ~15.6 days), the proposed method selects a small subset of candidates, reducing the optimization time to ~3.77 hours.
AC-Feasibility Validation: The authors bridge the gap between linearized optimization models and reality by validating all results against full AC-OPF simulations, ensuring the cost reductions are real and not artifacts of linear approximations.
Insight into Topological Actions: The research highlights that while BuS is effective, some cost reductions identified by the metrics are actually achieved through transmission line switching (disconnecting a branch entirely) rather than true busbar splitting.

4. Results

The proposed metrics were tested on four IEEE/pglib test cases ranging from 39 to 3,374 buses.

39-Bus Case: The $\phi$ -metric successfully identified buses with high cost reductions. However, it also highlighted that a high $\Delta$ LMP on a single branch is insufficient; the bus must have multiple connections and not be voltage-constrained.
118-Bus Case: Busbar 69 was identified as the top candidate (highest $\phi$ ), resulting in a 0.38% cost reduction. The metrics successfully filtered out busbars with binding voltage limits (which showed 0% reduction).
793-Bus Case: The metrics identified 15 candidates. Splitting these specific busbars yielded cost reductions in 12 cases. Crucially, the metrics identified 3 out of the top 4 busbars that provided the absolute highest cost reductions in the system.
- Efficiency: Full exhaustive search took 19.14 hours; the metric-based selection took 0.98 hours.
3374-Bus Case: The metrics identified 20 candidates. 8 of these led to cost reductions. While the absolute cost savings were smaller (due to the specific grid characteristics), the metrics still successfully identified the busbars yielding the highest reductions (e.g., Busbar 670).
- Efficiency: Full exhaustive search took 15.62 days; the metric-based selection took 3.77 hours.

Key Finding: Busbars with a high sum of absolute LMP differences ( $\phi$ ) and a high number of connected elements, provided they are not at voltage limits ( $\zeta=0$ ), are the strongest candidates for cost-reducing topology optimization.

5. Significance

This paper addresses a critical bottleneck in modern power system operations: scalability. As grids grow and become more complex with renewables, operators cannot afford to run exhaustive topology optimizations.

Operational Impact: The proposed metrics provide system operators with a practical, low-computational-cost tool to prioritize which substations to reconfigure during congestion events.
Economic Impact: By enabling the selection of effective topological actions, the method helps reduce generation costs and curtailment of renewable energy without requiring new infrastructure investment.
Future Planning: The methodology can be applied in grid planning to identify substations that require retrofitting (e.g., adding busbar couplers) to enable future topological flexibility.

In conclusion, the paper establishes that a targeted, metric-driven approach to Busbar Splitting is not only feasible but essential for managing congestion in large-scale, renewable-heavy power grids.