Use casesForecasting and failure prediction

Which asset fails next, when the peak lands and what the storm costs.

Utilities already collect the data to answer the questions that decide their year. It sits in meters, line sensors and outage logs that were never built to talk to each other. Grid Intelligence, our platform on Snowflake, brings them together and turns them into a watch list, a peak call and a dispatch board, each held to a baseline it has to beat.

Start a project Arrange a call

What gets in the way

A distribution transformer now takes about two years to procure, at prices up as much as 95% since 2019. The one that fails next is the one you cannot replace in time.

For some cooperatives, a year of transmission charges is set by demand in four fifteen-minute intervals, the highest of each summer month. Miss the call and you pay for twelve months.

When the storm comes, the question is where to stage crews before it hits, and what every hour without power costs the members living through it.

What we build

Meter, sensor and outage data together on Snowflake, turned into decisions a team can act on.

Every monitored transformer gets a daily score for failing within 90 days, read from loading, oil temperature, arcing and heat stress. Each unit on the watch list carries a plain reason, and dispatching an inspection writes the work order back.
At noon the platform forecasts tonight’s peak and calls demand response only if tonight looks like one of the month’s top three.
A storm replay compares restoration as it happened against crews staged early on the riskiest feeders. A dispatch board recommends the nearest qualified crew.
An outage cost view prices every hour without power for households and businesses from published estimates, so a decision carries a dollar figure.
Anyone can ask in plain English. Snowflake Cortex writes one query, runs it read-only inside the account and answers in two sentences. Every panel prints the query ID behind it.

How we know it works

Every model had to beat a simple baseline to stay, and was graded on months it never saw.

The failure model trained on one stretch of time and was tested on the months after. On the test fleet, a 50-unit watch list caught 58% of failures and the riskiest tenth of units held 6.5 times their share.
The load forecast missed by 8.8% over seven held-out months, against 11.8% for repeating yesterday and 16.3% for last week. Replayed across one summer, the noon rule caught all four monthly peaks.
No cooperative’s internal data went in. The test fleet is 1,200 transformers generated to match public utility records, and every screen says so.
The limits are published. The failure model predicts wear-out, not what a storm takes down. The forecast tests used actual weather, which a production feed would replace with a forecast.

Our accelerator

Ready to pilot

Predict the failure. Call the peak. Price the storm.

Grid Intelligence is the platform we built on Snowflake to answer the three questions that decide a cooperative’s year. Every model result was measured on a test fleet built from public utility data.

58% of failures caught by a 50-unit watch list on the test fleet

8.8% load forecast error on the test fleet, against 11.8% for yesterday’s load

4 of 4 summer peaks caught in an out-of-sample replay on the test fleet

Read the story

Grid Intelligence

Where it fits

Questions

What does a pilot look like?

It starts with a data inventory. We swap the synthetic fleet for your meters, sensors and outage logs, and hold every model to the same baselines it had to beat on the test fleet. If it does not beat them on your data, we say so.

Do we need Snowflake?

Grid Intelligence was built on it, with Snowflake ML FORECAST and ANOMALY_DETECTION running natively in the account. The models and the control room could sit on another warehouse. The plain-English questions through Cortex could not.

What if the tariff changes?

Regulators are reviewing how summer peaks are charged. The lasting value is the forecast and the dispatch, not any one tariff.

Is this only for cooperatives?

It was built for cooperatives because their year comes down to those three questions. The same models serve any utility with transformers to watch, a peak to call and crews to send.

Bring us the project that matters most.

Tell us what you’re building and what’s in the way. You’ll hear back within 24 hours.

Start a project Arrange a call