Service Restoration Algorithm

The core of any restoration algorithm is the ability to dynamically reconfigure the network's radial or weakly-meshed topology. After a fault is isolated by FDIR systems, the algorithm searches for alternative supply paths by closing normally-open tie switches and opening sectionalizing switches. This is often modeled as a spanning tree search problem on a graph, where the objective is to find a new radial configuration that maximizes load restored while respecting operational constraints. The algorithm must account for the real-time status of all Intelligent Electronic Devices (IEDs) and switching devices to ensure the proposed sequence is physically executable.

A proposed restoration plan is only valid if it does not violate the physical limits of the grid. The algorithm must integrate with a load flow engine to validate each switching step against critical constraints:

• Thermal Limits: Ensure no line or transformer is overloaded beyond its ampacity rating.
• Voltage Constraints: Maintain bus voltages within ANSI C84.1 limits (typically ±5% of nominal).
• Radiality Constraint: The restored topology must remain strictly radial to prevent circulating currents and protection miscoordination. Advanced algorithms use sensitivity analysis to predict voltage drops and line flows before committing to a switching sequence.

Restoration is rarely a single-objective problem. The algorithm must balance competing goals using a weighted sum or Pareto-optimal approach. Key objectives include:

• Maximize Load Restored: Prioritize critical loads like hospitals and emergency services.
• Minimize Switching Operations: Reduce wear on switchgear and minimize transient disturbances.
• Minimize Network Losses: Select the configuration with the lowest I²R losses post-restoration.
• Minimize Customer Interruption Duration: Favor plans that can be executed fastest. The algorithm outputs a ranked list of feasible restoration plans, allowing the operator to select the most appropriate trade-off.

Modern restoration algorithms must treat Distributed Energy Resources (DERs) as active participants, not just negative load. This includes:

• Intentional Islanding: Identifying sections of the network that can be isolated and sustained by local generation and battery storage, forming a stable microgrid.
• Black Start Capability: Coordinating the sequential re-energization of a dead bus using a grid-forming inverter or a diesel generator with black start capability.
• Fault Current Contribution: Accounting for the limited (1.1-1.5 pu) and sometimes unpredictable fault current from inverter-based resources, which can blind conventional overcurrent protection if not modeled correctly. The algorithm must verify frequency and voltage stability within the island before committing to the separation.

A change in network topology invalidates the existing protection coordination study. A sophisticated restoration algorithm includes an adaptive protection module that calculates new relay settings or selects a pre-computed settings group for the new configuration. This prevents nuisance tripping and ensures selectivity. The algorithm communicates new pickup currents and time-dial settings to relays via IEC 61850 GOOSE messaging or by activating alternative settings groups. Without this feature, a restored network may experience sympathetic tripping on inrush current or fail to clear a subsequent fault.

The algorithm generates a step-by-step switching sequence, not just a final topology. This sequence must account for cold load pickup (CLPU) — the inrush current from thermostatically controlled loads (HVAC, refrigeration) that have been off during the outage. CLPU can be 2-10 times the normal load and last for minutes. The algorithm models this delayed exponential decay to ensure that closing a breaker does not immediately trip it again on instantaneous overcurrent. It may stagger the restoration of large feeders to manage the aggregate inrush and prevent a secondary outage.

The overarching automation architecture within which the service restoration algorithm executes. FDIR provides the trigger event—a confirmed, isolated fault—and defines the healthy feeder sections available for reconfiguration. The restoration engine is the final 'R' in the FDIR loop, computing the optimal switching sequence to minimize the System Average Interruption Duration Index (SAIDI).

A distribution network topology that uses the service restoration algorithm as its core decision-making brain. In a self-healing grid, the algorithm continuously monitors real-time topology and load data. Upon fault isolation, it autonomously dispatches IEC 61850 GOOSE messages to open/close tie-switches and sectionalizers, achieving restoration in under 60 seconds without operator intervention.

The mathematical foundation of service restoration. The algorithm solves a constrained optimization problem:

• Objective: Maximize restored load while minimizing switching operations
• Constraints: Radiality, thermal limits, voltage bounds (ANSI C84.1 Range A)
• Methods: Heuristic search, genetic algorithms, or mixed-integer linear programming (MILP) The output is a target switching configuration that respects all operational limits.

The high-speed communication protocol that executes the restoration algorithm's decisions in the physical world. Generic Object Oriented Substation Events (GOOSE) messages are multicast across the substation LAN with less than 4ms latency. The restoration engine publishes trip/close commands for multiple breakers simultaneously, ensuring coordinated, high-speed reconfiguration without point-to-point wiring.

The real-time situational awareness input that feeds the restoration algorithm. Distribution System State Estimation (DSSE) infers voltage magnitudes and angles at unmonitored nodes using limited AMI and SCADA data. The restoration engine uses this estimated state to validate that a candidate switching plan will not cause voltage violations or overload healthy feeders before committing to execution.

A protection philosophy that dynamically adjusts relay settings to match the post-restoration topology. When the service restoration algorithm reconfigures the network, fault current paths and magnitudes change. Adaptive protection automatically recalculates pickup currents and time-current coordination curves for the new configuration, preventing sympathetic tripping and maintaining selectivity during the repair period.