SAIDI — System Average Interruption Duration Index — is the metric that reliability engineers live with every day and that regulators review every year. The minutes add up: tree contact, equipment failure, overloading, animal intrusion, third-party damage. Each cause code gets a share of the annual total. For many Texas investor-owned utilities, one cause code quietly accumulates a larger share than most field operations teams intuitively expect: distribution transformer failure.
We have reviewed publicly filed reliability reports from ERCOT-area investor-owned utilities covering 2021 through 2023 — three years that include a highly unusual weather event, two standard summer peak seasons, and accelerating load growth in the high-growth corridors of Central Texas. Across this data, distribution transformer failures consistently fell in the 18–23% range of sustained SAIDI contributions from equipment failure causes (excluding major event days). That is not a small number for a single equipment category.
Why Transformer Failures Punch Above Their Weight in SAIDI
A distribution transformer failure almost always causes a sustained outage — defined as an interruption lasting more than five minutes. Unlike a recloser operation, which may restore service automatically on the first or second reclose attempt, a transformer failure removes service from all customers on that secondary circuit until either the transformer is replaced or customers are switched to alternate supply. In dense suburban areas, that may be 20–60 customers. In rural areas with single-phase lateral service, it may be fewer customers but longer restoration time because of crew travel distance.
The restoration time for a transformer failure depends heavily on whether a spare unit of the correct size and configuration is in inventory and whether the field crew can reach the location during the failure event (which, in the case of a summer heat wave, may occur at 3:00 PM on a 105°F day when every crew is already committed to other restoration activities). SAIDI contribution per event for transformer failures is typically in the range of 90–180 minutes per affected customer — meaningfully higher than the average for equipment failures overall, which includes many shorter-duration switching events.
The 2021 Winter Storm Effect on the Data
The 2021 winter storm event (February 2021) is explicitly excluded from major event day SAIDI calculations in most utility reliability filings, following standard IEEE 1366 methodology. However, the transformer failures that occurred or accelerated during that event are worth understanding separately because they represent an important pattern: thermal cycling stress from a cold-load pickup event accelerates insulation aging in ways that show up as early failures in the one-to-three-year period following the event.
Utility distribution engineers in Texas reported elevated transformer failure rates in 2021 and 2022 that were plausibly connected to mechanical and dielectric stress from the February 2021 cold-load pickup. Oil-immersed transformers that were already thermally aged took cold-start energization current at temperatures well below their design operating range. Repeated energization attempts as sections were restored added through-fault thermal stress. Some of these units did not fail immediately — they failed 6–18 months later during normal summer loading, with no DGA indication preceding the event because the degradation mechanism was primarily mechanical (winding deformation from through-fault forces) rather than thermal decomposition.
Segmenting the Failure Contributions
Transformer failures are not uniformly distributed across the fleet or across cause types. Based on utility reliability data patterns, three segments dominate the transformer SAIDI contribution:
- Aged units in high-load-growth corridors: Transformers installed in areas that have seen significant load growth (commercial development, EV charging, large HVAC installations) are operating at higher percentages of nameplate rating than when originally installed. IEEE C57.91 thermal modeling shows that sustained operation above 100% nameplate loading under Texas summer ambient temperatures accelerates insulation thermal aging at a rate well above the baseline Arrhenius model design assumption. These units have shorter residual life than their installation year implies.
- Single-phase laterals on rural feeders with long restoration paths: These failures may affect fewer customers in total, but SAIDI contributions are disproportionately high due to restoration time. A three-hour outage affecting 12 customers contributes 36 customer-hours to SAIDI, the same as a one-hour outage affecting 36 customers.
- Pad-mount units in commercial applications with poor ventilation clearances: Pad-mount transformers installed in enclosures or near structures that restrict airflow operate at higher ambient temperatures than their ratings assume. Thermal aging accelerates. These failures tend to be concentrated in commercial corridors where load growth and installation density have both increased.
What the ROI Calculation Looks Like
This is where the SAIDI data becomes actionable for a predictive maintenance investment case. The calculation has four inputs:
- Average SAIDI contribution per transformer failure event (customer-minutes per event, from the utility's own outage data)
- Number of transformer failure events per year
- Percentage of those failures that a monitoring program could reasonably provide advance warning for (not all failures are predictable — sudden through-fault failures without prior mechanical degradation are not detectable in advance)
- Cost of a planned preventive replacement versus an emergency restoration replacement
On the last point: the cost difference between a planned transformer replacement (scheduled outage, crew pre-positioned, spare unit staged in advance) and an emergency restoration replacement (emergency crew dispatch, extended customer interruption, potential for secondary damage to cable or switchgear from the failure event) is typically in the range of 2x to 4x in fully loaded cost. A utility that executes 50 emergency transformer replacements per year and converts 40% of those to planned maintenance actions through early anomaly detection has material cost savings — before counting SAIDI-related regulatory exposure or customer satisfaction impacts.
We are not suggesting that continuous vibration monitoring on every transformer in a large fleet is economically justified on these numbers alone. The economics work best when monitoring is concentrated on the highest-consequence assets — transformers with the highest customer counts, longest restoration paths, or least available spare-switching capacity — combined with a prioritized screening approach to identify which assets in the broader fleet show elevated risk signals worthy of DGA acceleration or physical inspection.
The SAIFI Side of the Equation
SAIFI — the frequency index — is less directly affected by transformer failures than SAIDI, because most transformer failure events result in a single sustained interruption rather than multiple momentary events. However, there is an interaction worth noting: transformers that fail during high-load events often cause protection system operations upstream (feeder breaker trips, lateral fuse operations) that increase momentary event counts even when the sustained outage is limited to the transformer's immediate secondary circuit. SAIFI from equipment failures is harder to directly attribute to transformer condition, but the correlation is real in high-failure-density periods.
Reading the Numbers for Your Territory
Texas IOU data is not universal. A utility in the Pacific Northwest with different load profiles, different transformer fleet age distributions, and different storm exposure will have different SAIDI contribution patterns. The 18–23% figure for transformer failures in the Texas IOU context is specific to that climate, that era of load growth, and that fleet age profile.
What generalizes is the methodology: pull your own cause-code-segmented SAIDI data for the last three to five years, isolate equipment failure causes, and see what fraction of sustained outage minutes are attributable to distribution transformer failures specifically. If that number is above 15%, it warrants serious examination of whether your current transformer maintenance program is providing adequate early warning for the failure modes that are driving your reliability costs.
