Service Restoration (SR) is the algorithmic and operational process of isolating a faulted section of a distribution feeder and re-energizing the healthy, de-energized downstream customers by transferring them to an adjacent, energized feeder via normally open tie switches. The primary objective is to minimize the System Average Interruption Duration Index (SAIDI) by finding the shortest viable switching sequence that respects radiality constraints and thermal limits.
Glossary
Service Restoration (SR)

What is Service Restoration (SR)?
Service Restoration (SR) is the emergency control process of finding and executing a sequence of switching operations to re-energize de-energized customers after a fault by transferring them to healthy feeders.
Modern SR engines integrate with Outage Management Systems (OMS) and Distribution Automation (DA) hardware to execute restoration plans automatically. The algorithm must account for Cold Load Pickup (CLPU)—the inrush current surge when thermostatic loads restart simultaneously—and verify that the target feeder has sufficient reserve capacity to accept the transferred load without violating voltage bounds defined by the DistFlow equations.
Key Characteristics of Service Restoration
Service Restoration (SR) is the critical emergency control process that finds and executes a sequence of switching operations to re-energize de-energized customers after a fault by transferring them to healthy feeders. The following characteristics define modern, automated SR systems.
Radiality Constraint Enforcement
The restoration algorithm must guarantee the resulting network topology remains a radial tree structure with no closed loops. Distribution systems are designed and protected for unidirectional power flow; a mesh topology would cause protection miscoordination and circulating currents.
- Graph theory models the grid as nodes (buses) and edges (switches/lines)
- The solution must be a valid spanning tree connecting all energized nodes
- Normally Open Points (NOPs) are closed strategically while a downstream sectionalizing switch is opened to maintain radiality
- Violating this constraint triggers cascading relay trips and potential equipment damage
Cold Load Pickup Mitigation
After a prolonged outage, Cold Load Pickup (CLPU) causes a temporary demand surge 2-5x normal load as thermostatically controlled devices (HVAC, refrigerators) start simultaneously. Restoration algorithms must account for this inrush to prevent immediate re-tripping.
- CLPU current decays exponentially over 10-30 minutes
- Model Predictive Control (MPC) staggers restoration steps to manage the inrush
- Load diversity loss means the aggregated peak far exceeds statistical sum
- Ignoring CLPU is the most common cause of failed restoration attempts
Feeder Capacity Constraints
The healthy feeder receiving the transferred load must have sufficient headroom capacity to absorb the new demand without violating thermal limits or causing voltage collapse. This requires real-time loading data from the Outage Management System (OMS) and Distribution Automation (DA) sensors.
- Each candidate feeder is evaluated for remaining ampacity on conductors and transformers
- DistFlow equations calculate the resulting voltage profile along the reconfigured path
- N-1 Criterion may require reserving capacity for a subsequent contingency
- Overloading a healthy feeder cascades the outage to previously unaffected customers
Switching Sequence Optimization
The algorithm must output an ordered sequence of switch operations that minimizes the number of maneuvers while maximizing restored load. Each switching step must be validated for transient stability before execution.
- Mixed-Integer Linear Programming (MILP) formulates switch statuses as binary variables
- Branch Exchange Method iteratively closes a tie switch and opens a sectionalizing switch
- Minimizing switch count reduces wear on aging Intelligent Electronic Devices (IEDs)
- The sequence is simulated in a Digital Twin before field execution to verify safety
Multi-Objective Trade-Offs
Service restoration is inherently a multi-objective optimization problem balancing competing goals: maximizing restored load, minimizing switching operations, equalizing feeder loading, and reducing line losses. There is no single perfect solution.
- Pareto optimal fronts present trade-offs to the operator for final decision
- Conservation Voltage Reduction (CVR) can be applied post-restoration to reduce stress
- Weighting factors prioritize customer count over industrial load criticality
- SAIDI reliability metrics directly penalize every minute of sustained interruption
Distributed Generation Islanding
When Distributed Energy Resources (DERs) are present, restoration may involve intentional islanding — separating a portion of the grid with local generation to maintain supply independently of the main utility system during a wide-area disturbance.
- Microgrid Control Systems manage frequency and voltage within the islanded segment
- IEC 61850 GOOSE messaging enables high-speed peer-to-peer coordination between IEDs
- Synchronization checks are required before reconnecting the island to the main grid
- Soft Open Points (SOPs) provide precise power flow control between islanded feeders
Frequently Asked Questions
Essential questions and answers about the automated process of re-energizing de-energized customers after a fault through intelligent switching operations.
Service Restoration (SR) is the emergency control process of finding and executing a sequence of switching operations to re-energize de-energized customers after a fault by transferring them to healthy adjacent feeders. The process begins when a fault detection mechanism—typically a protective relay or Fault Detection Isolation and Recovery (FDIR) system—identifies and isolates the faulted section. The SR algorithm then analyzes the post-fault network topology using real-time data from Intelligent Electronic Devices (IEDs) and SCADA systems to identify available Normally Open Points (NOPs) on neighboring feeders. The algorithm must respect the radiality constraint, ensuring no closed loops are created, while verifying that the backup feeders have sufficient residual capacity to accept the transferred load without violating thermal limits or voltage constraints. Once a valid restoration plan is computed, the system dispatches control commands to sectionalizing switches and tie switches, reconfiguring the network to restore power to healthy but de-energized sections—often within seconds to minutes.
Enabling Efficiency, Speed & Accuracy
Intelligent Analysis, Decision & Execution
We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.
Talk to Us
Search across company data
Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.
Useful when people spend too long searching or get different answers from different systems.

Automate internal workflows
Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.
Useful when repetitive work moves across multiple tools and teams.

Add AI to products and internal tools
Build assistants, guided actions, or decision support into the software your team or customers already use.
Useful when AI needs to be part of the product, not a separate tool.
Related Terms
Service Restoration relies on a constellation of interconnected grid automation functions. These concepts define the operational constraints, algorithmic approaches, and reliability metrics that govern how utilities re-energize customers after a fault.
Fault Isolation
The critical precursor to restoration. Before re-energizing healthy line sections, protective devices must separate the faulted segment. Modern schemes use peer-to-peer GOOSE messaging between IEDs to achieve isolation in under 100ms, preventing the fault from propagating and minimizing the extent of the outage before the restoration algorithm begins its switching sequence.
Radiality Constraint
The non-negotiable topological rule governing all restoration plans. Distribution networks must operate as a spanning tree—a connected graph with no closed loops. This ensures fault currents flow in a single, predictable path, allowing simple overcurrent protection coordination. Every restoration switching sequence must verify that closing a Normally Open Point (NOP) does not create a mesh.
Cold Load Pickup (CLPU)
A physical phenomenon that can cause restoration attempts to fail. After a prolonged outage, thermostatically controlled loads (HVAC, refrigerators) lose diversity and start simultaneously upon re-energization. This creates a demand surge 2–5 times normal peak load. Restoration algorithms must model CLPU dynamics to avoid tripping the healthy feeder on overcurrent immediately after transfer.
N-1 Criterion
The reliability planning standard that defines restoration capacity requirements. The grid must survive the failure of any single component—a feeder, transformer, or substation bus—without sustained customer interruption. This mandates that every load point has at least one alternative supply path via tie switches, and that the backup feeder has sufficient spare capacity to accept the transferred load without violating thermal or voltage limits.
System Average Interruption Duration Index (SAIDI)
The primary regulatory metric that drives investment in automated restoration. SAIDI measures the total minutes of sustained interruption the average customer experiences annually. Self-healing schemes directly improve SAIDI by reducing the manual patrol and switching time. Utilities target SAIDI reductions to avoid financial penalties from performance-based ratemaking.
Mixed-Integer Linear Programming (MILP)
The dominant mathematical framework for finding globally optimal restoration switching sequences. Switch statuses are modeled as binary integer variables (0 for open, 1 for closed), while power flow physics are expressed as linear constraints using the DistFlow equations. MILP solvers guarantee the solution is the absolute best for minimizing losses or maximizing load restored, unlike heuristic methods.

About the author
Prasad Kumkar
CEO & MD, Inference Systems
Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.
His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.
Partnered with leading AI, data, and software stack.
How We Work
Custom AI workflows for your Business
One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.
01
Review the use case
We understand the task, the users, and where AI can actually help.
Read more02
Pick the right approach
We define what needs search, automation, or product integration.
Read more03
Build the first useful version
We implement the part that proves the value first.
Read more04
Improve from there
We add the checks and visibility needed to keep it useful.
Read moreThe first call is a practical review of your use case and the right next step.
Talk to Us