SAIDI Explained: Measuring Average Duration of Electric Service Interruptions

SAIDI stands for System Average Interruption Duration Index. It measures the average total duration of sustained electrical service interruptions experienced by customers over a defined reporting period, according to the applicable reliability-reporting methodology. Utilities use SAIDI because it turns a large number of individual outage events into one system-level indicator of how long customers go without power.

SAIDI is one of the most widely used customer-oriented reliability metrics, but it is also easy to misread. It is an average, it depends on how interruptions are defined and counted, and it does not explain causes by itself. This article focuses on SAIDI specifically: the formula, a step-by-step calculation, interpretation, data sources, data quality, limitations and practical use. For wider context, see electric grid reliability, grid reliability vs resilience and how utilities measure grid reliability. Exact definitions and reporting rules vary by jurisdiction and utility.

What Is SAIDI?

SAIDI, the System Average Interruption Duration Index, is the average total time that customers are without electric service during a reporting period.

The key elements are:

• Purpose: to summarize interruption duration at system level.
• What it measures: customer interruption duration, which means how long affected customers were without supply.
• Reporting period: usually a month or a year, depending on the reporting convention.
• Unit: generally minutes per customer or hours per customer.
• Relationship to reliability: it is one of several indices used to describe electric grid reliability.

Definitions are not identical everywhere. Some utilities follow industry guides such as IEEE Std 1366, which defines reliability indices and a way of treating major events, while others follow rules set by their regulators. What counts as a “sustained” interruption, for example, is commonly a period longer than a few minutes, but the threshold depends on the definition used.

SAIDI in Simple Terms

SAIDI answers one question: “On average, how much interruption time did customers experience during the reporting period?”

The word “average” matters. SAIDI spreads total interruption time across the whole customer population. An 18-minute SAIDI does not mean every customer lost power for 18 minutes. Many customers may have had no interruption, and a smaller group may have had much longer ones.

SAIDI Formula

SAIDI = Total customer interruption duration ÷ Total customers served

• Total customer interruption duration is the sum, across the relevant interruption events, of the interruption duration multiplied by the number of customers affected, according to the applicable reporting methodology. It is often expressed in customer-minutes or customer-hours.
• Total customers served is the number of customers served in the area and period being reported, usually the average number over the period or a count at a stated date, depending on the method.

The result is in minutes (or hours) per customer, because customer-minutes divided by customers leaves minutes.

SAIDI Calculation: Step by Step

1. Identify sustained interruption events. Select the interruptions that meet the definition of sustained for the method used.
2. Determine affected customers. Establish how many customers lost supply in each event, often through the network model and customer records.
3. Determine interruption duration. Find the time between the start of the interruption and restoration for those customers.
4. Calculate customer interruption duration for each event. Multiply customers affected by duration.
5. Sum the customer interruption duration. Add the results across events in the period.
6. Divide by total customers served. This gives the average per customer.
7. Apply the applicable reporting methodology and exclusions. This may include treatment of planned interruptions and major event days.
8. Report the resulting SAIDI value with its units, period and the method used.

Hypothetical SAIDI Calculation

Hypothetical calculation. These numbers are invented for illustration and are not real utility data.

A utility serves 1,000 customers. Three sustained interruptions occur in the period.

SAIDI = 18,000 customer-minutes ÷ 1,000 customers = 18 minutes per customer

Notice that the three events have the same customer-minutes even though they differ in size and length. SAIDI reflects the product of customers and duration, not either alone.

How to Interpret SAIDI

A lower SAIDI generally indicates less average interruption duration under the same reporting methodology. A higher SAIDI indicates greater average duration. That does not mean a single SAIDI value proves one utility is better than another. Interpretation depends on:

• Reporting methodology: definitions and exclusions.
• Customer population: density, mix of customer types and size.
• Geography: terrain, climate and accessibility.
• Network design: overhead or underground, radial or meshed, availability of alternate supply.
• Major-event treatment: whether and how severe days are removed.
• Planned and unplanned treatment: which interruptions are counted.
• Data quality: accuracy of events, times and customer counts.

What Does a High SAIDI Mean?

A high SAIDI means average interruption duration is relatively high. Possible contributing factors include long outages, slow restoration, severe events, equipment failures, hard-to-locate faults, access challenges, communication limitations, vegetation or weather impacts, aging infrastructure, network topology and limits on emergency response.

A high SAIDI does not identify the root cause by itself. It signals that interruption duration deserves investigation, for example by looking at which events contributed most customer-minutes and why they lasted as long as they did.

What Does a Low SAIDI Mean?

A low SAIDI generally indicates lower average interruption duration under the same methodology. It does not necessarily mean there were no outages, no momentary interruptions, no localized reliability problems, no resilience concerns or no asset risks. A system can post a low average while a few feeders perform poorly, or while it remains vulnerable to rare severe events. SAIDI should be considered alongside other metrics.

SAIDI vs SAIFI

Two systems can share the same SAIDI but behave differently. One may have many short interruptions and the other a few long ones, which is why frequency and duration are read together.

SAIDI vs CAIDI

SAIDI is the average interruption duration across all customers served. CAIDI, the Customer Average Interruption Duration Index, is the average duration of an interruption for customers who experienced interruptions. The conceptual relationship is:

CAIDI = SAIDI ÷ SAIFI

This holds when all three are calculated on the same basis and methodology. In the hypothetical example, 600 customer interruptions (200 + 300 + 100) occur among 1,000 customers, so SAIFI is 0.6 interruptions per customer. CAIDI is then 18 ÷ 0.6 = 30 minutes per interruption, which equals 18,000 customer-minutes ÷ 600 customer interruptions. CAIDI gives another view of restoration duration, showing how long a typical interruption lasts once it occurs.

SAIDI vs SAIFI vs CAIDI

SAIDI and Momentary Interruptions

SAIDI generally focuses on sustained interruption duration, according to the reporting definition in use. Brief interruptions are usually tracked separately. Under many definitions, a momentary interruption lasts only a few minutes or less and is measured by MAIFI, the Momentary Average Interruption Frequency Index.

Momentary interruptions often come from reclosers and automatic protection that clear a temporary fault and restore supply quickly. A utility can therefore have a low SAIDI and still have a problem with flickering lights or equipment resets for sensitive customers. SAIDI alone does not show that.

SAIDI and Planned Interruptions

Interruptions fall into broad types: planned maintenance interruptions, unplanned interruptions and emergency interruptions. Utilities may treat them differently in reporting. Some include planned interruptions in the index, some report them separately, and some exclude them. Specific treatment and exclusions depend on the reporting methodology, so no single rule applies to all utilities. Any SAIDI figure should state what it includes.

SAIDI and Major Event Days

Severe events can dominate SAIDI. Major storms, flooding, wildfires where relevant, extreme weather and major infrastructure damage can cause very large customer-minute totals in a short time. Reliability frameworks may therefore have specific treatment for major event days. For example, IEEE Std 1366 describes a statistical method for identifying such days, while other jurisdictions use their own thresholds or rules. No one exclusion rule is universal.

When comparing SAIDI values, check whether major events are included, excluded or reported separately. Reported without that information, two numbers may not be comparable.

SAIDI Data Sources

SAIDI relies on outage and customer data from several sources:

• Outage management system (OMS) and outage records.
• SCADA.
• AMI and smart meters.
• Customer information systems.
• GIS and the network model.
• Distribution management systems.
• Field reports.

The data flows like this:

1. Outage event

   Interruption occurs.

2. Customer impact

   Affected customers identified.

3. Start time

   Beginning of interruption.

4. Restoration time

   Supply returns.

5. Customer count

   Customers per event and in total.

6. Reliability database

   Events stored and validated.

7. SAIDI calculation

   Index computed.

Role of OMS in SAIDI

An outage management system can provide outage start time, outage location, affected customers, restoration time, event closure and historical outage records. These are the main ingredients of the calculation, so many utilities use OMS data as a primary source. It is not the only possible source, and its accuracy depends on how well events are recorded and closed. Manual entries and estimated customer counts can introduce errors that carry directly into SAIDI.

Role of AMI and Smart Meters

AMI can improve outage detection, last-gasp notifications, restoration verification, customer-level visibility and event timestamps. For example, restoration messages from meters can confirm that supply returned to specific customers, which can refine restoration times.

AMI communication loss does not automatically mean an electrical outage. A meter can go silent because of a network problem while power is on, and a meter may fail to send a last-gasp message during an outage. Utilities therefore often combine AMI information with OMS, SCADA and field information. For more on smart meter data, see key components of a smart grid.

Role of SCADA

SCADA provides equipment status, alarms, measurements, fault indications and real-time visibility of devices such as breakers and reclosers. It helps identify system-level events and show when protection operated.

See power system automation for how these monitoring and control systems fit together.

Data Quality and SAIDI

SAIDI is only as accurate as the data behind it. Problems that can distort it include missing events, duplicate events, incorrect timestamps, incorrect customer counts, incorrect restoration times, incorrect customer mapping, communication failures, manual data-entry errors and network model errors. For example, if restoration time is recorded when the last customer is restored rather than when each group is restored, the total customer-minutes will be overstated. Accurate SAIDI depends on accurate outage and customer data.

Validating SAIDI Data

Validation steps include event reconciliation between systems, timestamp validation, customer mapping checks, restoration confirmation, AMI cross-checking, field verification, OMS validation and defined data-quality rules. Automated calculation does not remove the need for validation. A system can compute a precise-looking number from flawed inputs, so review of unusual events, large outages and missing records remains important.

SAIDI by Feeder, Area or Customer Group

Utilities may segment SAIDI by feeder, substation, geographic area, customer class, region or urban and rural area where appropriate. System-wide SAIDI can hide local problems. For example, a system may have an acceptable overall SAIDI while one feeder has a much higher interruption duration. Segmentation shows where the customer-minutes come from and where attention is needed. Small groups can swing widely from a single event, so results need careful reading.

SAIDI and Distribution Reliability

SAIDI is especially useful for customer-facing distribution reliability analysis, because most customer interruptions originate in the distribution system. Relevant elements include distribution feeders, distribution transformers, switchgear, reclosers, protection, restoration practices and maintenance. Changes in any of these can show up in interruption duration, which makes SAIDI a practical way to follow the effect of distribution improvements over time. Background on the network is in transmission vs distribution networks.

SAIDI and Transmission Reliability

SAIDI is not a universal metric for every transmission reliability question. Transmission reliability also considers availability, forced outages, contingency performance, system security, stability and adequacy. Transmission events can lead to customer interruptions, but transmission networks are often planned to withstand the loss of single elements, so many such events cause no customer outage at all. Customer-oriented distribution indices and broader transmission reliability analysis answer different questions and should not be treated as interchangeable.

SAIDI and Outage Management

Utilities can use SAIDI trends to investigate long restoration times, repeated problem areas, equipment-related interruptions, weather-related interruptions, crew response and fault location challenges. The improvement loop is:

SAIDI Trend → Identify Problem → Investigate Root Cause → Implement Improvement → Measure Again

Because SAIDI reflects both how many customers were affected and how long, it can point to events whose long duration contributed heavily, which are often the best candidates for process or equipment review.

SAIDI and Asset Management

SAIDI can help identify areas that need deeper asset investigation. It becomes more useful when combined with other information:

SAIDI + SAIFI + CAIDI + Asset Condition + Outage Cause → Maintenance / Capital Planning

SAIDI alone cannot identify which asset should be replaced. It shows where duration is high, and asset condition data and outage cause records are needed to decide what to do about it.

SAIDI and Grid Modernization

Utilities may evaluate reliability initiatives involving AMI, distribution automation, FLISR, ADMS, sensors, smart grid technologies, predictive maintenance and advanced communication using SAIDI among other measures. Results should be compared against appropriate baselines and the same methodology, because changes in definitions, data collection or weather can look like improvement or decline. Better outage detection, for example, can make recorded durations more accurate, which may raise or lower SAIDI for reasons unrelated to real performance. For related technology, see grid modernization and real-time visibility.

SAIDI and Grid Resilience

SAIDI is primarily a reliability-performance metric. It does not fully measure resilience. Resilience also considers preparation, severe events, adaptation, response, recovery, critical loads and restoration capability. A utility could have a favorable SAIDI in ordinary years and still be exposed to rare severe events, and major event day exclusions can hide exactly those events. SAIDI should not be used as a complete resilience score. See grid reliability vs resilience for the distinction.

SAIDI and AI / Analytics

Analytics can support SAIDI work through trend analysis, outage clustering, root-cause analysis, weather correlation, asset correlation, restoration analysis, predictive maintenance and reliability forecasting. AI and analytics can help find patterns, such as repeated outages on similar equipment, but engineering teams must validate causes and decisions. A correlation found in data is not proof of a cause.

Practical Utility Applications

Hypothetical Example: High-SAIDI Feeder

Hypothetical scenario. It describes no real utility or data.

A feeder has a higher SAIDI than the utility’s overall average. The diagnostic process might run:

High SAIDI → Review outage events → Analyze outage causes → Check restoration duration → Review asset condition → Check protection and network configuration → Identify improvement → Monitor future SAIDI

Review of events might show that most customer-minutes came from a few long outages. Cause analysis might point to tree contact in one section, and restoration records might show slow fault location because that section has few sectionalizing devices. Asset review and protection checks might confirm those findings. An improvement, such as vegetation work and added switching points, could then be considered, and future SAIDI for the feeder monitored. All findings here are illustrative.

Hypothetical Example: Before and After Improvement

Hypothetical scenario. Values are illustrative and not from any real utility.

The illustration shows how faster detection, location and switching reduce customer-minutes. No real-world improvement percentage is claimed, and real results depend on the network, the events and the methodology.

Common SAIDI Interpretation Mistakes

• Treating SAIDI as the only reliability metric. It shows duration, not frequency.
• Comparing utilities without checking methodology. Definitions and exclusions differ.
• Assuming low SAIDI means zero outage risk. Risk remains.
• Ignoring SAIFI and CAIDI. They add frequency and restoration views.
• Ignoring momentary interruptions. They are tracked elsewhere.
• Ignoring local feeder performance. Averages hide problem areas.
• Ignoring major-event treatment. It can change the result greatly.
• Assuming correlation means causation. Patterns need engineering review.
• Treating SAIDI as a resilience metric. It does not measure preparation or recovery capability.

Limitations of SAIDI

• It is an average, so it does not describe individual customer experience.
• It can hide local problems.
• It does not directly identify causes.
• It depends on data quality.
• Methodology affects comparability.
• Major events can influence results.
• It does not fully describe momentary interruptions.
• It does not fully measure resilience.
• It does not directly measure power quality.
• It does not capture every customer experience.

How Utilities Can Use SAIDI Effectively

1. Define the methodology, including what counts as sustained and how planned and major events are treated.
2. Collect reliable outage data.
3. Validate events.
4. Calculate SAIDI consistently.
5. Segment results by feeder, area and customer group.
6. Compare trends over time.
7. Investigate root causes.
8. Combine SAIDI with SAIFI and CAIDI.
9. Connect results with asset and network data.
10. Implement improvements.
11. Measure results again.

Current, Emerging and Future Technology

Adoption varies by utility. Future items are possibilities, not current practice.

Summary

SAIDI is the System Average Interruption Duration Index, calculated as total customer interruption duration divided by total customers served, and usually expressed in minutes or hours per customer. It is built from outage events, affected customers and durations under a stated methodology. Read it alongside SAIFI, which shows frequency, and CAIDI, which shows average restoration duration. Data comes from OMS, SCADA, AMI and other sources, and its quality determines how reliable the result is. Utilities use SAIDI for benchmarking, planning and improvement tracking, while remembering that it is an average with limits.

Conclusion

SAIDI is one of the most useful customer-oriented reliability metrics because it converts interruption duration into a measurable system-level indicator. However, SAIDI is most useful when interpreted together with SAIFI, CAIDI, outage causes, asset information, network context and engineering analysis. SAIDI alone does not define grid reliability.

Frequently Asked Questions About SAIDI