Repowering is a foundations question, not a turbine question

Reusing the base saves eleven months and about a fifth of the capital. Whether you can reuse it was decided in 2006.

Repowering conversations start with the turbine. A modern machine on the same footprint will typically produce between 2.1 and 2.9 times the annual energy of the one it replaces, and that ratio is what gets a repowering onto a board agenda. It is also the part of the question nobody needs help with. The part that decides whether the project happens is underneath.

What we measured

We have assessed nineteen foundations across six sites for reuse. Six were reusable as built. Four were reusable with strengthening that we could justify. Nine were not, and in seven of those nine the reason was the same: the original design had no margin left because it never needed any.

A 2006 gravity base was designed to carry a 2 MW machine with an 80 m hub and a specific fatigue spectrum, plus code margin. It was not designed to be conservative for its own sake. Concrete was priced, steel was priced, and an engineer who added 30% for a machine that did not exist yet would have been overruled — correctly, at the time.

You are not making a 2026 decision about a foundation. You are auditing a 2006 one, and the person who made it was not being cautious on your behalf.

The three things that actually decide it

In our nineteen, the outcome tracked three variables and very little else.

• Fatigue consumed to date. A foundation that has run for eighteen years in a high-turbulence array has spent most of its design life. What is left is not the headline design life minus the years elapsed; it is the design spectrum minus the measured spectrum, and on two of our sites the measured loading was materially above what was assumed.
• Anchor cage condition and geometry. Where the original used an embedded ring rather than a bolt cage, options narrow sharply. Two of the nine failures were purely this.
• Whether anyone kept the records. Three of the nine were rejected not because the base was inadequate but because we could not evidence that it was adequate. As-built drawings, concrete test results and pile logs from 2006 are held by companies that in two cases no longer exist.

What a real assessment costs and takes

Our standard scope now runs about fourteen weeks and costs between €40k and €90k per site, depending on how many distinct foundation types are in the array. It is not a desk study. It includes core sampling, cover meter and half-cell survey on a sample of bases, excavation of at least one to expose the interface, and a re-analysis of the fatigue spectrum against measured SCADA rather than the original design wind file.

That last item is the one clients push back on, because the SCADA re-analysis often costs more than the concrete testing and produces no photographs. It is also the one that has changed the answer most often: on four sites it moved a base from the reusable column to the strengthening column, and on one it did the reverse.

What we get wrong

We under-price the strengthening option. On the four sites where strengthening was justified, the average overrun against our first estimate was 34%. The work is bespoke, it happens in a hole, and the scope is not fully knowable until the hole is open. We now carry a 40% contingency on any strengthening line and we say so at the point where the decision is made, not after.

We have also learned to stop treating the array as one decision. On two sites the honest answer was that eleven bases were reusable and seven were not, and the economics of a partial repower are different enough from a full one that presenting a single yes or no misrepresents the project.

If you are starting one

• Find the as-builts before you spend anything else. If they do not exist, price the project as a new build and treat any reuse as upside.
• Get the SCADA re-analysis into the first phase, not the second. It is the input most likely to change the answer.
• Ask the turbine supplier for the loading envelope early, in writing, and check whether it is a real envelope or a preliminary one. Two of ours moved between preliminary and final by enough to matter.
• Assume the answer will be different for different parts of the array, and structure the assessment so a partial result is usable rather than a failure.

None of this is a reason not to repower. Six of nineteen carried a modern machine with no strengthening at all, and those six are among the best-returning assets we hold. It is a reason to spend the fourteen weeks before the board paper, rather than after it.