Climate Intelligence to Reduce DEC and FEC in Power Distribution
Mateus Lima
CEO
··10 min read
DEC and FEC drop when the distribution utility anticipates the weather event that will take down the grid, not when it reacts after the outage has already been logged. Wind, lightning, extreme rainfall and wildfire account for a significant share of the occurrences that penalize collective indicators. An alert calibrated per asset changes the field decision before the outage becomes a number in ANEEL's bulletin.
DEC and FEC measure the duration and frequency of outages per consumer; they are collective continuity indicators defined by ANEEL, calculated for a group of consumer units (gov.br/aneel).
Breaching DEC or FEC does not generate direct compensation to the consumer: it triggers a tariff effect via the Q component of the X Factor, regulatory oversight, and evaluation of the concession contract. Individual compensation comes from DIC, FIC, DMIC and DICRI.
Weather causes (wind, lightning, extreme rainfall, wildfire) account for a significant share of DEC, which in some concessions exceeds 20%.
PRODIST excludes Emergency Situation Interruptions and Critical Day from the DEC and FEC calculation. That is why the indicator suffers more from moderate, recurring events, which do not qualify for exclusion.
INMET runs the COSMO model at 7 km resolution for South America (174h horizon) and 2.8 km for parts of Brazil (48h); the gain from per-asset calibration lies in translating that into an operational threshold by grid segment.
When a distribution utility calls me, it is rarely to talk about weather. It is to talk about an indicator: DEC went up, management wants an explanation, ANEEL wants justification. But at the root of nearly every one of these conversations there is a weather event that already had a signal before it happened.
Continuity indicator is the technical name for something simple: how long and how often the consumer goes without power. DEC measures the equivalent outage duration per consumer unit. FEC measures the frequency. Both are collective indicators, calculated per electrical grouping, not per individual consumer.
What few address is the origin of a significant part of these outages: strong wind bringing down a branch onto the grid, lightning inducing an overvoltage on a nearby segment, extreme rainfall flooding a substation, wildfire approaching the line right-of-way. This is not an isolated equipment failure. It is a weather decision made, or not made, hours before the event reaches the grid.
What DEC and FEC are and why they penalize the distribution utility
DEC and FEC are collective continuity indicators defined by ANEEL: DEC measures the average outage time per consumer in an electrical grouping, FEC measures how many times that consumer lost power during the period. Breaching the grouping's limit does not compensate the consumer directly: it triggers a tariff effect, regulatory oversight, and evaluation of the concession contract.
The financial compensation the consumer actually receives comes from another family of indicators, the individual ones: DIC, FIC, DMIC and DICRI, each with its own limit, calculated per specific consumer unit. Confusing the two families is the most common mistake I see in conversations with regulatory teams.
DEC and FEC are calculated every month, per group of consumer units, and ANEEL's open data show the history of every distribution utility in the country. The common mistake is treating both as resulting only from grid maintenance, old cables, weak poles. Part of it is that. But a significant fraction stems from a weather event that already had a signal hours earlier.
Which weather events most raise DEC and FEC in Brazil
Strong wind, lightning strikes and extreme rainfall account for most of the weather-related outages that raise DEC and FEC at Brazilian distribution utilities. Wildfire comes next: it is mostly a transmission problem, due to the exposure of the right-of-way over long stretches, and weighs less on distribution than wind, lightning and rainfall.
Wind brings down branches and trees onto the grid, especially in segments with vegetation close to the right-of-way. In distribution, lightning tends to affect the grid not through a direct hit on the transformer, but through the overvoltage induced when the discharge strikes near the line, without physically hitting it. Extreme rainfall floods substations and hampers maintenance crew access.
At i4sea, when monitoring energy assets across Latin America, we treat these hazards in an integrated way: strong wind usually comes with rain from the same front, and a severe gust almost always originates from a convective cell that also carries lightning risk. Looking at each hazard in isolation from the week's context means missing the pattern that precedes the outage.
Why generic weather forecasting does not prevent the outage
The Brazilian public forecast, run by INMET with the COSMO model, covers South America at 7 km resolution and a 174-hour horizon, and part of Brazil at 2.8 km and 48 hours. It is useful for saying whether it will rain in the region, at a resolution comparable to the one we use.
What it does not deliver is calibration for the specific asset, and that calibration is what allows you to say whether the forecast wind exceeds the threshold that would bring down that particular grid segment.
Weather forecasting is not climate intelligence. The difference lies in translating the weather condition into an operational threshold for the asset: it is not about promising a gust on a specific tower at an exact hour, it is about estimating the probability that segment will exceed the threshold that compromises operations, within a defined time window.
This matters because much of the gusts that bring down the grid originate from convective phenomena (downbursts, squall lines), which have only probabilistic predictability between 12 and 48 hours in advance. The exact location only becomes clear in nowcasting, using radar and lightning detection, within a window of zero to two hours before the event.
How climate intelligence anticipates risk before the outage
i4cast monitors 18 hydrometeorological hazards at a resolution of 1 to 3 km, calibrated for the specific asset (the substation, the line segment, the feeder), not for the surrounding city. This calibration delivers the probability that the segment will exceed the operational threshold within a window of hours, refined through nowcasting as the event approaches.
In practice, the method cross-references each asset's occurrence history with the weather conditions that preceded each past outage. That way it gets to know the safe operating limits of that specific segment better: from what gust speed, what lightning density, what rainfall volume the outage becomes likely within that window.
Where most people see the unexpected, we see a manageable variable. The same reasoning we apply to critical assets in other sectors applies to the power grid: the weather event has a probabilistic signal before it becomes an occurrence. Anyone following topics like lightning-related outages in the power sector knows that a few hours of lead time already changes the decision to deploy a crew.
For wildfire, even though it is a challenge more concentrated in transmission, monitoring heat sources near the right-of-way enables preventive vegetation clearing before the fire approaches the line, a topic we cover in detail in wildfire and transmission and distribution outages.
The financial impact of reducing DEC and FEC with weather anticipation
Reducing outages through weather anticipation shows up on three cost fronts: compensation avoided on individual indicators (DIC, FIC, DMIC and DICRI), tariff effect avoided through the Q component of the X Factor, and operational cost avoided by sparing crews and equipment.
PRODIST excludes Emergency Situation Interruptions and Critical Day from the DEC and FEC calculation. In practice, the most extreme event, the one that makes headlines, tends to be excluded from the indicator. What actually weighs on DEC and FEC is the moderate, recurring event, precisely the one the distribution utility cannot exclude.
This distinction changes where it is worth investing in anticipation. There is no point targeting only the catastrophic event: the indicator gain lies in reducing the volume of moderate outages that repeat month after month, because those are what make up the average calculated by ANEEL.
The EPE study, cited by Agência iNFRA in November 2025, points out that factoring climate change into power sector planning reduces costs. It is a sign that the sector is moving in that direction, even though the study addresses generation planning, not distribution utility operations.
There is a second, less discussed use: the DEC and FEC exclusion report requires weather evidence to support the alleged Emergency Situation or Critical Day. Georeferenced data per asset, with an auditable history, serves as technical evidence in that regulatory defense, reducing the risk of ANEEL challenging the exclusion for lack of evidence.
From alert to action: what changes in the distribution utility's operation
When the operations team receives the risk alert in advance, two distinct levers come into play. Positioning crews before the event reduces restoration time, which is what makes up DEC. Reducing FEC, in turn, requires avoiding the outage altogether, which calls for prior action such as targeted pruning, segment reinforcement, or load transfer switching.
Positioning crews before the event seems obvious, but it depends on knowing where and when the risk concentrates. Without that data, the crew stays centralized and moves only after the call comes in, which is already a reaction. With a georeferenced alert, the crew moves closer to the segment with the highest probability before the gust or the lightning strike.
One caveat is worth noting: a preventive shutdown also counts toward FEC, because it is an outage like any other for calculation purposes. The decision to shut down for safety may protect the equipment, but it does not improve the indicator by itself.
That is why planned restoration, decided with weather conditions in hand, tends to take less time than improvised restoration in the middle of the event.
How does a weather alert system reduce DEC and FEC in practice?
There are two separate levers. Positioning crews before the event reduces restoration time, which makes up DEC. Reducing FEC requires avoiding the outage altogether, through targeted pruning, segment reinforcement, or load transfer switching before the event arrives.
What lead time and granularity does an alert need for operations to act before the outage?
It depends on the hazard: wind associated with a cold front usually gives one to three days of notice, while convective gusts have only probabilistic predictability between 12 and 48 hours, with precise location only available through nowcasting, zero to two hours beforehand. Calibration needs to be done per asset, specific line segment, or substation, not for the surrounding metropolitan area.
Does i4cast serve energy distribution utilities, or is it just for transmission?
i4cast monitors energy assets across generation, transmission and distribution, applying the same 18 hydrometeorological hazards to the risk profile of each asset type. Substations, feeders and distribution lines receive the same calibrated data treatment we already apply to port and mining assets across Latin America.
How does the AI Climate Agent support the operations team's day-to-day decisions?
The Agent answers natural language questions such as "what is the probability of strong wind exceeding the threshold on segment X in the next 24 hours" or "what is the lightning risk in this window", without requiring the reading of a technical bulletin. The team checks the risk for its own asset before deciding whether to deploy a crew or schedule restoration.
What changes for decision makers
For those accountable for the continuity indicator, the decision point shifts from the occurrence report to the risk alert, hours or days in advance. This changes the conversation with management: instead of justifying DEC and FEC after the fact, operations shows the decision made beforehand, backed by data that supports operational efficiency and regulatory defense.
If you want to measure your grid's real climate risk before the next event, i4sea offers a free 7-day diagnostic with the AI Climate Agent, applied to the assets you already operate today.
We use analytics cookies to understand how the site is used and marketing cookies to measure our campaigns. They are only enabled with your permission.
