A portfolio assembled by acquisition is a portfolio of separate reporting systems. Fleet analytics normalizes the operating record so a turbine at one site can be compared with a turbine at another, and so the fleet total is built from assets rather than from spreadsheets.
Comparability first, because every other question a portfolio owner asks depends on it.
Different manufacturers report different channels under different names at different intervals. The record is normalized once, on ingestion, so that a comparison between two machines from two suppliers means what it appears to mean.
Lost production is separated into wind resource, availability, curtailment, wake interaction, and machine underperformance. A fleet that missed budget for the first reason requires a different response from one that missed it for the last.
Findings are ordered across the whole fleet rather than within each site, which is what allows engineering time and capital to move to the assets where the next unit of either returns the most.
Owner, lender, and board reporting is produced from the same figures the operations team works from, with the method stated, rather than assembled separately at the end of a quarter from a different set of sources.
A fleet delivering ninety-six percent of budget is describing an average across assets whose individual performance may range from full delivery to a persistent double-digit deficit. The average is the figure that reaches the board, and it is the figure least capable of directing anyone to a turbine.
The deficits that matter are rarely concentrated in a machine that has failed, because a failed machine is visible to every system already in place. They are distributed across turbines that are running, available, and quietly producing below the units beside them. At fleet scale a two percent shortfall spread across a third of the assets is a larger quantity of energy than any single outage in the same period, and it is the only part of the shortfall that no existing report is looking for.
Two turbines of the same rating from two manufacturers do not record the same channels, do not sample at the same interval, and do not use the same operating state codes. A fleet view assembled without reconciling those differences produces comparisons that are arithmetically valid and physically meaningless, which is worse than no comparison because it is acted on.
Normalization also has to account for what the machines cannot control. A site with a higher long-term mean wind speed will produce more energy than a better-run site with less resource, and a comparison that does not adjust for that ranks the terrain rather than the operation. The same applies to wake position within a site: an interior turbine is not underperforming because it stands behind three others.
The output is an asset-level record on a single scale. Everything a portfolio owner wants to know afterward depends on that record existing, which is why it is the first piece of work rather than a reporting layer added at the end.
Which assets are underperforming relative to comparable machines, and by how much in annual energy. Where the next unit of engineering time returns the most. Whether a site's shortfall is resource, availability, curtailment, or degradation. Which assets justify lifetime extension, which justify repowering, and which are worth more to a different owner.
Each of those is a capital decision, and each is currently answered with a judgment supported by whatever evidence could be assembled in the time available. Asset-level evidence does not make the decision, but it changes what the decision is made from.
Comparability across a mixed fleet is the precondition for allocating anything by return rather than by site.
See how OpenTurbine can help your team understand performance, anticipate operational issues, and make better decisions across your wind assets.