Frequently asked questions

With an export of existing SCADA history for the sites in scope. The data is validated for coverage and quality, the models are fitted per turbine, and the first findings are returned ranked by the annual energy and revenue at stake. The quickstart in the documentation sets out each stage.

Days rather than a deployment project, because the analysis runs on history that already exists. The elapsed time is usually governed by how quickly the export can be produced and transferred, not by the modeling.

No. There is no hardware, no change to the control system, and no software installed in the operational technology environment. The platform reads exported operational data and returns findings.

Someone who can produce the SCADA export, and someone who knows the fleet well enough to explain events in the record. The second is more valuable than it appears: a machine that was down for a retrofit is an anomaly to the model and an obvious fact to the person who scheduled it.

Active power, wind speed and direction, nacelle position, and pitch angle where recorded, plus the alarm and status log. Temperature and vibration channels are used where a turbine model exposes them and are what allow a finding to name a component rather than a deficit.

Twelve months is the working minimum and twenty-four months is preferred, because two winters separate a seasonal pattern from a developing fault in a way that one year cannot.

No. Pre-cleaning is the most common cause of a weak assessment. Gaps, outliers, and curtailed periods are informative, and filtering them removes exactly the evidence the models use to separate a fault from a period of low wind resource.

They are reported as out of scope with the reason stated, rather than modeled at low confidence and presented alongside the rest. The validation report lists every exclusion before any modeling runs.

Yes, and that is often the reason to adopt the platform. Expectations are fitted per turbine from each machine's own history, so a portfolio running several manufacturers and rotor diameters is assessed on one scale instead of through separate portals.

The turbine and site, the period over which the deviation was observed, the class of loss, the evidence behind the classification, a confidence level, and the annual energy and revenue at stake. Every field is documented in the reading a finding reference.

Check it against what the operations team already knows. Findings that correspond to events already understood are the control for the rest of the list, and the evidence on each finding is published precisely so that an engineer can disagree with it.

That is the class of loss the platform exists to identify. Underperformance and alignment losses do not raise faults and do not affect availability, which is why they persist through conventional reporting.

Confidence is exposed rather than hidden behind a fixed threshold, because the right threshold depends on the cost of acting. Findings closed out as incorrect are fed back, which is one of the reasons close-out matters operationally.

Three operator decisions govern it: the confidence level at which a finding raises a work order, who receives findings for each site, and how a finding is closed out once the intervention is complete. We help set all three during onboarding.

Close-out records what intervention was performed and when, which turns the estimated value on a finding into a measured recovery over the following quarters. It is the step most often skipped and the one that compounds.

Yes. Support is staffed by engineers who work on the modeling, so a question about a result is discussed with someone who can explain the expectation behind it rather than escalated.

No. A finding identifies a deviation and its probable cause from operational data. It narrows where to look and establishes what looking is worth; it does not diagnose a component the way a physical inspection does.

Per turbine under management, on one axis, with no per-seat component. Assessments, continuous monitoring, and portfolio engagements are set out on the pricing page and quoted directly.

Yes. An assessment runs on your own historical data over a period whose outcome your team already knows, which is the only form of evaluation that proves anything about a specific fleet.

It is processed to serve your fleet. It is not sold, not pooled with other operators' data, and not used to train models serving anyone else. The data processing addendum is available for review before any transfer.

For the term of the agreement, so that expectations remain fitted on the fullest history available. Retention and deletion terms are set in the commercial agreement rather than left to a policy page.