ACER faults Portugal's adequacy study over its assumptions
Serguey Asael Shinder3 min readTwo studies asked whether Portugal will have enough electricity in the 2030s. The European one, prepared by ENTSO-E and reviewed by ACER, found no adequacy concern. The Portuguese national assessment, published as part of the country's security-of-supply monitoring, found a risk in every target year from 2028 to 2035 — with the expected hours of unserved demand exceeding the national reliability standard of 1.46 hours a year slightly in 2028, around fivefold by 2030, and close to fiftyfold by 2035.
Same country, same decade, opposite conclusions. ACER's opinion, published this month, is worth reading precisely because it is not about who is right — it is about where the two models part company.
The national study assumes a harder world. It expects demand from new energy-intensive consumers, data centres in particular, to grow faster than the resources meeting it. It assumes renewables arrive more slowly. It derives pumped-hydro availability from historical performance rather than nameplate capability. It takes a conservative view of how much battery capacity will be built in the later years. And it treats the Tapada do Outeiro gas plant as an out-of-market reserve, which removes roughly a quarter of Portugal's thermal capacity from the assessment — the kind of plant that is normally very much available in exactly the hours that scarcity is measured.
ACER's objection is that several of those choices are not symmetrical. Its recommendations are to apply the economic viability assessment the European methodology prescribes, to model the new market-based investment that higher prices would call forth — batteries and demand response in particular — and to reflect what the out-of-market resources would actually contribute. The agency also notes the treatment of imports during scarcity hours. Taken together: the national study may be counting the ways the system gets tighter without counting the ways a tight system attracts supply.
What it means
An adequacy number is a model output, and the assumption list is the story. "Fifty times the reliability standard by 2035" is the sentence that travels; the sentence that matters is that a quarter of thermal capacity was set aside, batteries were held flat, and hydro was derated to history. Anyone quoting the first number without the second paragraph is quoting a headline, not a finding.
Data centres are now load-bearing in national planning, and that is new. Portugal's divergence is driven substantially by assumed demand from a class of consumer that did not appear in adequacy studies a decade ago, whose siting decisions are commercial, fast and reversible. A forecast that depends on where a handful of companies decide to build is a forecast with a much wider error bar than the tidy hours-per-year output suggests.
And the interesting regulatory question is who gets to be conservative. A national operator has every incentive to assume the worst — the cost of being wrong the other way is a blackout. A European methodology has every incentive to assume markets respond, because assuming otherwise builds capacity payments into the continent's bills. Both incentives are legitimate, and neither is visible in the final number. ACER's call for better modelling is really a call to make the disagreement explicit rather than to settle it.
The opinion and the underlying assessment are linked from the announcement above.