HK DATACREST

Energy and utilities

Ranking 180 transformers by condition, from data the utility already had.

For a regional power utility running around 180 substation transformers, many past 25 years in service, the evidence is scattered: dissolved gas results arrive from the lab as PDFs, online monitors feed SCADA on a few critical units, and inspection notes live in spreadsheets kept by each zone.

Industry
Energy and utilities
Solutions
Know which assets need attention first · Put all your numbers in one place
Engagement
Pilot, then build and run

The challenge

Every zone has a list of worries. Nobody has one list for the whole fleet.

Outage windows are booked a year ahead and replacement budgets are fought for every quarter. Without a common view, the asset management team plans by age and by whoever raised the loudest concern, and a deteriorating unit can sit unnoticed between tests.

The approach

How it works.

  1. Collect the evidence

    Lab results, monitor exports, inspection sheets and work orders are loaded for every transformer, with gaps and stale tests flagged rather than guessed.

  2. Agree the rules

    The utility's engineers review the scoring ruleset: which tests matter, how much, and the thresholds that match their own standards.

  3. Score and explain

    Every unit gets a 0 to 100 index, a band and a recommended action, with each lost point traced to a specific reading.

  4. Plan from one list

    The ranked worklist feeds outage and budget planning, and alarms flag any reading that moves the wrong way between reviews.

The solution

What gets built.

The solution is our asset health application, configured with a transformer ruleset for this fleet: a fleet overview, a per-unit breakdown, a diagnostic map of the physical unit, and printable reports for each asset and for the board.

Fleet overview of transformer health: average score, units needing action, health against age, a risk matrix and a ranked worklist.

Outcomes

What it delivers.

  • One ranked list for the whole fleet so outage slots and budget go to the units most likely to need them.
  • Every score explained so engineers can challenge a ranking with evidence, not opinion.
  • Missing and late tests made visible so the testing programme targets the gaps that matter.
  • Board-ready reports from the same scores the engineers use.

Technology

Typical stack for this.

  • Python
  • FastAPI
  • Scoring rulesets in YAML
  • PostgreSQL
  • AWS
  • or whatever you already run

Related solutions

All use cases

Know which assets need attention first

Condition scoring for transformers and grid assets.

Put all your numbers in one place

Pipelines, data checks and dashboards.

Start with a conversation

Tell us what's slowing your operation down.

A 30-minute call with the people who would build it. You leave knowing whether a pilot makes sense, and what it would cover.