Energiforsk 2026-1151 Effektauktioner med Värmepumpar (2026)

Full citation: Månborg, V., Mazzotti Pallard, W., Vautrin, A. et al. (Profu AB + Bengt Dahlgren AB). Effektauktioner med värmepumpar. Energiforsk Report 2026-1151. Stockholm: Energiforsk AB, January 2026. Program: Elnätens hållbara teknikutveckling och digitalisering. (Authors per the raw: Vanja Månborg, Profu; Willem Mazzotti Pallard and Adrien Vautrin, Bengt Dahlgren — an earlier version of this page named a “Samuel Månborg” as lead and assigned the modelling to Bengt Dahlgren; the raw gives no task split.)

Audit note (2026-09-20): an earlier version of this page reversed the VCG-vs-marginal-price relationship, mislabelled the price results, and invented an “aggregator revenue problem”, several “challenges”, a 300,000-heat-pump threshold and a Nobel detail. Corrected below against the raw report.

Summary

The report tests a VCG (Vickrey-Clarke-Groves) auction algorithm as a procurement mechanism for heat pump demand response in DSO local flexibility markets. It simulates 137 heat pump datasets for a real area of 137 properties in Kristianstad (about 0.5% of the municipality’s small and multi-family houses), built from UK (and some Danish) real-world monitoring data adapted to a Swedish normal year, and calculates VCG prices for flexibility needs, comparing VCG with marginal-price auctions. Its stated context: local markets have few auctions per year, and VCG is meant to draw in many small bidders.

Key claims

The VCG mechanism

• Each winner is paid the externality it imposes on others; truthful bidding is the dominant strategy; the mechanism maximises system welfare (not necessarily minimising cost — raw §3.3: VCG “inte nödvändigtvis minimerar kostnaderna”).
• The raw says applied auction theory of this kind won the 2020 Nobel Prize (the earlier “Vickrey 1996 / awarded posthumously in 2020” wording was wrong and partly outside knowledge).

VCG vs marginal pricing (reversed in the earlier version)

Per the raw summary and §4 (Figur 12): when available flexibility is large relative to the demanded volume, the VCG price approaches the marginal price; otherwise the VCG price exceeds the marginal price — marginal price is lower than VCG when demanded flexibility is high relative to maximum available. §5.1: marginal-price auctions “har potential att leverera flexibilitet till en lägre kostnad”. When the price without a bidder would be infinite, it is capped at the assumed maximum price of 4 kr/kW (the earlier “VCG = 0 for the last bidder” row was invented). VCG is vulnerable to coordinated manipulation (cartels) by jointly misreporting costs; the report’s mitigations are many bidders, monitoring/restricting bids, and a blockchain audit trail.

Simulation design

• Real area of 137 properties in Kristianstad; the 137 datasets come from only about 80 unique installations (multi-year series split into annual datasets under separate IDs).
• Data mainly from the UK (HeatpumpMonitor), including Denmark; adapted to a Swedish normal year with an SVM using outdoor temperature, hour of day and a weekend flag as inputs (building envelope is not modelled); pump-to-house matching used the Hungarian algorithm on estimated floor area.
• Simulated need: 50 kW, chosen as relevant for the area’s size (from a 10–50 kW share of C4’s 2–10 MW estimate); maximum available flexibility is about 170 kW; the need arises over 3,951 hours, which the report says does not reflect a typical grid.
• Bid curves: dimensionless curves scaled by predicted hourly power with an 80% safety margin.

VCG price results

For a 50 kW need the average VCG price varies between 0.8 and 1.6 kr/kW across ten bid-curve flexibility levels (not “sufficient vs tight supply” scenarios). Averaged over all needs and bid curves the range is 0.2–4 kr/kW; above about 150 kW demanded the price is capped at 4 kr/kW. Example: level 6 gives about 0.58 kr/kW, 29 kr in total for 50 kW, with 123 of 137 bidders taking part. The low price depends heavily on assumptions about bid curves and the maximum price.

4 kr/kW is the assumed maximum price, taken from Hartvigsson & Steen 2024 as the capital cost of substation strengthening; the report also cites Power Circle at 330–451 kr/kW for capacity reinforcement and aFRR prices of roughly 6–23 kr/kW as far higher comparators.

Total annual payout for a 2–10 MW need is a range of 1,600–16,000 kr/year (not a per-endpoint mapping; the earlier “limited hours of activation” explanation is not in the raw).

Challenges and open issues (per raw §5)

• Cartel/manipulation risk (above); a large number of manipulated heat pumps could threaten the power system, so cyber-protection matters (§5.2.3 — no “300,000 heat pump” threshold and no ER 2025-35 citation appears in the raw).
• Complexity for participants: VCG is hard for participants to understand — they find it difficult to know their expected payment (§5.2.2, citing Palm et al.); complexity is itself a participation barrier, not something VCG relieves.
• Agreement is needed on the prediction model and training data (§5.2.3); ON/OFF pumps need stepped bid curves; rebound effects; flexibility demand may disappear.
• Aggregators: §5.3 says the algorithm’s nature gives very low incentives to act as an aggregator because it handles bids from individual bidders directly — “inget direkt behov av en aggregator”. The report’s vision is a fully automated auction with a very low participation threshold. (The earlier “no revenue incentive → must bundle with other services” framing was invented.)

National heat pump flexibility potential

Raw §1.1: 5.75 GW of hourly flexibility by 2030 versus 0.3 GW today, in line with the 1–6 GW found by Averfalk and Lindahl’s literature study. The wiki’s own comparison: Sweden’s DSO-reported need at 6–10 years is 1,387–2,523 MW (Source - Ei PM2025-03 DNDP Sammanställning (2025)), so 5,750 MW is about 2.3–4.1× larger (wiki arithmetic, not a report claim).

Relationship to Palm et al. (2023)

The report cites Source - Palm et al LFM Drivers and Barriers (2023) for the finding that complexity is a barrier — including complexity of VCG itself. (The quoted “I have no idea what price to set on flexibility” does not appear in the raw.)

Relevance to the wiki

This source directly informs or strengthens:

• Flexibility Market — introduces VCG as an alternative auction mechanism to pay-as-bid/pay-as-cleared marginal pricing; provides design argument for thin/infrequent markets; updates heat pump potential to 5.75 GW
• Demand Response — 5.75 GW national heat pump flexibility potential; bid curve concept for continuous DR
• Aggregation — the report argues the auction needs no aggregator (individual bids handled directly)
• Distribution Network Development Plan — ~4 kr/kW is the report’s assumed price cap (from Hartvigsson & Steen 2024); VCG as alternative to grid investment for recurring substation-level congestion
• Baseline Methods — bids are based on predicted power; the report notes agreement on the prediction model and training data is still needed

Internal links to other Energiforsk 2024–2026 program sources:

• Source - Energiforsk 2024-1043 DNDP Analys och Flexibilitet (2024) — cited; same ~4 SEK/kW benchmark; both analyze the DSO-side case for flexibility over grid investment
• Source - Energiforsk 2026-1157 Nationell Metod Effekt och Kapacitetsprognoser (2026) — complementary: 2026:1157 addresses how DSOs forecast what they need; 2026-1151 addresses how they procure it once the need is defined
• Source - Palm et al LFM Drivers and Barriers (2023) — cited; VCG design addresses the pricing difficulty barrier identified by Palm et al.

Data gaps

• The UK-to-Sweden adaptation has not been independently validated; Swedish heat pump profiles may differ
• Results come from a small real area (137 properties, ~80 unique installations); generalisation to other densities is untested
• Whether an automated, aggregator-free auction with very low participation threshold is practically achievable is left open by the report