Aannamedossier

Status: FIRST TRANCHE APPROVED 2026-07-27 (assumption set v0.1.0-partially-verified). Ev approved: import prices 21/25/31 ct · self-use 0.35/0.28/0.19 · the export-economics era pairs (§5+§6) · the PVGIS profiles + yields (§1) · the MFFBAS E1A profile + 0.6465 evening share (§2) · salderen end date (§0). These are verified: true in lib/engine/assumptions.ts with sources and approval dates. Still open: capture factor (§8, harness), degradation curve + warranty PDFs (§7), heat-pump profile (§2 remainder), trading (§9, kept low-confidence), self-use bonuses, input defaults, and the policy constants (founder sign-off pending) — the recommendation gate stays closed until all are resolved and the 0.6 independent review signs off. Method: multi-agent web research (106 agents, 25 sources fetched, 120 claims extracted, 22 surviving 3-vote adversarial verification). Claims below cite only verified sources. Items marked OPEN did not yield a verified primary value yet and block freezing.

Verdict legend: ✅ confirmed · ✏️ correction proposed · 🔬 method verified, numbers still to extract · ⚠️ OPEN

---

0. Statutory basis — end of salderen ✅

• Spec assumption: salderen ends 1-1-2027 (statute).
• Finding: the Wet beëindiging salderingsregeling (bill 36.611) is enacted law, not a pending proposal: adopted by the Tweede Kamer 14-11-2024 and the Eerste Kamer 17-12-2024; enters into force 1-1-2027. Sources: eerstekamer.nl bill page, EK news 17-12-2024. Vote 3-0.
• Action: freeze as verified. Confidence high. This also validates the site's core positioning copy.

1. pvFactor[12] per orientation + yield factors ✏️ EXTRACTED — corrections proposed

• Spec assumption: zuid ×1.0 / oost-west ×0.85 / noord ×0.6; monthly shapes were hand-drawn placeholders.
• Extraction done 2026-07-26 — PVGIS v5.2 PVcalc API, De Bilt (52.10, 5.18), 35° tilt, 14% system loss, 1 kWp; oost-west = mean of east (az −90°) and west (az +90°) runs. Yearly totals: zuid 1032.8 kWh · oost-west 817.5 kWh · noord 541.7 kWh per kWp.
• Corrections proposed:
  • Yield factors: oost-west 0.79 (spec guessed 0.85) and noord 0.52 (spec guessed 0.60).
  • Monthly shape: oost-west is more summer-peaked than zuid on a monthly basis, not flatter — "flatter" describes the daily curve (already modeled via the +0.03 self-use bonus), not months.
• Status: raw monthly kWh now live in lib/engine/assumptions.ts (normalized at export, still verified: false). All 24 engine tests pass on the real data. Awaiting your approval to freeze. Confidence high (deterministic extraction from the authoritative source).

2. consFactor[12] (standard + heat-pump variant) 🔬

• Spec assumption: winter-weighted standard profile; heat-pump variant strongly winter-weighted.
• Findings:
  • The authoritative source is the official Dutch standaardprofielen (used for market allocation and reconciliation), published on Energiedatawijzer (MFFBAS points there; its own documenten page does not carry them). Votes 3-0.
  • Heat-pump winter weighting is directionally confirmed: CBS (2024) reports registered heat-pump homes consume much more electricity in cold winter months (CBS seizoenspatronen); Energysense panel data (2022) shows 60% of annual net grid consumption in Jan/Feb/Nov/Dec for heat-pump + PV homes — note the caveat that net-of-PV data overstates the winter skew of gross consumption.
• Extraction done 2026-07-27: monthly factors computed from the official MFFBAS Standaardprofielen elektriciteit 2026 v1.00 CSV (35,040 quarter-hours, E1A AZI afname — households without PV feed-in, so gross consumption; annual sum verified = 1.0): Jan 0.0995 · Feb 0.0848 · Mar 0.0892 · Apr 0.0800 · May 0.0762 · Jun 0.0739 · Jul 0.0754 · Aug 0.0758 · Sep 0.0746 · Oct 0.0814 · Nov 0.0893 · Dec 0.0999. Now live in lib/engine/assumptions.ts (still verified: false). Remaining OPEN: the heat-pump variant — no official appliance-specific profile exists; current winter-weighted shape follows CBS seizoenspatronen qualitatively and needs a quantitative anchor (CBS microdata or Energysense gross figures). Standard profile confidence high (authoritative source, deterministic extraction).

3. Self-use fractions s0 by presence profile ✏️ RESEARCHED 2026-07-26 — correction proposed

• Spec assumption: thuis_overdag 0.40 / gemengd 0.30 / werkend_buitenshuis 0.22 (+ bonuses, cap +0.10).
• Findings: the NL battery-free average of ~30–35% is confirmed by three independent chains: TNO 2022 P10584 (~30%, Dec 2022) · CE Delft & TNO feitenbasis saldering (kernrapport 230470, Sep 2024, 30% modeling standard) · CE Delft/Vereniging Eigen Huis measured field study (Jan 2026, 13 households Oct 2024–Oct 2025: typical 30–35%, spread 6–49%). Milieu Centraal (updated 30-06-2026) popularizes the same 30% and confirms the east-west bonus at +3–6 pp and EV-at-home uplift. The only occupancy-differentiated quantified source is the UK MCS MGD 003 standard (Loughborough profiles): home-all-day 30–35% / half-days 24–28% / out-all-day 17–18% at PV ≈ load.
• Correction proposed: the current triple sits at the optimistic edge of every verified range (~2–4 pp high), which understates battery benefit. Recommend 0.35 / 0.28 / 0.19. The +0.03 east-west bonus is well supported (source range 3–6 pp); daytime-EV bonus direction confirmed.
• Confidence: average anchor high; occupancy split medium (UK profiles, NL sources only qualitative). Source PDFs archived locally. Founder decision required.

4. Import price scenarios (ct/kWh, marginal all-in) ✏️ correction proposed

• Spec assumption: conservative 25 / base 30 / favourable 38.
• Findings (all vote 3-0):
  • CBS StatLine 85592NED: average variable delivery tariff June 2026 = 13.97 ct/kWh incl. VAT (down from 15.2 ct in Dec 2025); energy tax 2026 = 11.085 ct/kWh incl. VAT (down from 12.286 ct in 2025) → marginal all-in ≈ 25.1 ct/kWh (June 2026). Source.
  • CBS energierekening Jan 2026: total bill €1,993/yr (−€52 vs 2025); variable electricity supply cost declined into 2026. The €902/1747 kWh ≈ 51.6 ct/kWh figure includes fixed charges and is not a marginal price — do not use it for the scenarios. Source.
• Correction proposed: current reality sits at the spec's conservative value, and tariffs are drifting down, not up. Propose conservative 21 / base 25 / favourable 31 (favourable ≈ 2025 tax level + tariff upswing). Monotonic direction preserved. Founder decision required — this lengthens every payback vs the old base and materially affects recommendation distribution. Confidence in the underlying data high.

5. Post-2027 export compensation (ct/kWh) ✏️ RESEARCHED 2026-07-26 — structural correction required

• Spec assumption: exportComp conservative 8 / base 4 / favourable 2, with feed-in charges (§6) as an independent triple 0/2/4 — combined net export value +8 / +2 / −2 ct across scenarios.
• Legal findings (primary sources): the floor is in primary law, not an AMvB: the Wet beëindiging salderingsregeling (Stb. 2025, 17, Wet van 18-12-2024) amends Energiewet art. 2.34: lid 7 — the vergoeding voor kleine aansluitingen may not be negative on a monthly weighted average (permanent); lid 9 — ≥50% of the contracted kale leveringsprijs until 1-1-2030 (from amendement 36611 nr. 17, Grinwis/Rooderkerk/Erkens). The floor governs the bruto vergoeding; separate per-kWh terugleverkosten remain permitted (ACM), which is how suppliers net it down.
• Announced 2027 supplier tariffs: bruto vergoedingen cluster at 6.5–7.25 ct (exactly at the ~50% floor); per-kWh terugleverkosten 4.4–6.9 ct; net for fixed contracts ≈0.2–1.8 ct (Vattenfall 0.24, Essent 0.25, Greenchoice 0.25, Budget 0.5–1.8). Dynamic suppliers: no terugleverkosten, net ≈ spot ±1–2 ct (Zonneplan spot+2ct+10%, Frank spot+1.8ct+bonus, Tibber spot−1.8ct).
• Why the spec triple fails: (a) bruto 4 and 2 ct are legally impossible before 2030 (50% floor); (b) the favourable scenario's combined net −2 ct conflicts with the non-negativity rule as implemented; (c) conservative net 8 ct exceeds anything announced.
• Correction proposed — set §5 and §6 as coherent per-scenario PAIRS (2027–2030 window):

  Scenario	bruto vergoeding	terugleverkosten	net export value
  conservative	7.0	0	~7 ct (dynamic-with-bonus, no kosten)
  base	6.5	4.5	~2 ct (floor-compliant fixed contract)
  favourable	6.5	6.25	~0.25 ct (announced Vattenfall/Essent level — never negative)
  Post-2030 the 50% floor lapses (non-negativity stays) — net values trend toward spot-minus-margin,
  strengthening the battery case; disclose as a methodology note rather than a constant.

• Engine implication: §5 and §6 constants must be chosen jointly per scenario (the current independent triples can produce the impossible −2 net). Confidence high on the legal floor (statute text), medium on supplier numbers (announcements, may shift before 2027). Founder decision required.

6. Feed-in charges (terugleverkosten) ✏️ RESEARCHED 2026-07-26 — correction proposed

• Spec assumption: avoided charges 0 / 2 / 4 ct/kWh-equivalent.
• Findings (supplier primary sources, retrieved 2026-07-26): for vast/variabel customers at the major suppliers, current terugleverkosten are far higher than the spec triple — ct/kWh-equivalent at 2000–3500 kWh export: Vattenfall ~11.5–12.7 (staffel, lowered 1-5-2026) · Essent ~12.2–12.5 (monthly staffel) · Eneco ~11.4–13.3 per-kWh (13.33 ct until 31-12-2026, 3.87 ct from 1-1-2027 per their own rekenvoorbeeld) · Greenchoice ~11.6 (variabel) · Budget Thuis ~8.5–10 (lowered Dec 2025). Dynamic contracts are exempt at Essent/Vattenfall/Eneco; Zonneplan charges €0; Frank ~0 with Slim Terugleveren; Tibber ~2.5 ct/kWh (bonus flipped to charge 1-1-2026). No supplier waives for battery ownership per se — the exemption route is contract type.
• Post-2027 signal: announced per-kWh terugleverkosten drop sharply (Eneco 3.87 ct, Greenchoice ~5.2 ct, Essent folds into near-zero net vergoeding) — so the charge a battery avoids decays after the saldering era ends.
• Correction proposed: the calculator simulates years 1–10 (mostly post-2027), so a lifetime- weighted triple of 0 / 3 / 6 ct is defensible (conservative 0 = dynamic contract; base ≈ announced 2027 per-kWh levels; favourable = 1–2 saldering-era years at ~11 ct blended with ~4–5 ct after). Alternative if modeling year-1 only: 0 / 4 / 10–12 with an explicit decay disclosure. Methodology page must note staffel schemes make marginal value lumpy. Confidence medium-high (supplier primary sources; 2027 values are announcements). Founder decision required.

7. Round-trip efficiency + degradation ✅ (RTE) / ⚠️ OPEN (degradation)

• Spec assumption: AC-coupled RTE 0.85–0.90 (default 0.88); retention at yr 10: 75/80/85%.
• Findings:
  • Sessy (manufacturer, interested source): fixed home battery RTE ≈ 85% — supports the default range, at its lower end. Vote 2-1. Source.
  • A claim that plug-in batteries run ~70% RTE was killed 0-3 — do not use it; but flag: our single 0.88 default applied to plug-ins is unvalidated in either direction.
  • No manufacturer warranty retention curve survived verification.
• Action: consider default RTE 0.86 (or keep 0.88 pending more data); collect warranty PDFs (BYD, Sessy, HomeWizard, Zendure, Marstek, EcoFlow, Growatt) for retention floors — these are also per-battery DB fields, so this doubles as M2 data verification. Confidence medium (RTE) / low (degradation).

8. Capture factor ✅ CALIBRATED + APPROVED 2026-07-27 (0.55/0.70/0.85)

• Spec assumption: 0.80 / 0.88 / 0.95.
• Calibration: Phase C chronological harness (scripts/calibration/, report: docs/calibration/phase-c-report.md) — real E1A load chronology × measured KNMI 2024 daily radiation × verified PVGIS monthly totals, 81-cell grid, energy balance exact. Shift-solved capture distribution: p10 0.52 / p50 0.79 / p90 0.90; typical NL cell (3,500 kWh, ratio 1, 5.2 kWh battery) 0.65–0.80.
• Correction proposed: the current triple is too optimistic (its conservative value equals the calibrated median). Propose 0.55 / 0.70 / 0.85. Engine v2: replace the constant with a function of battery size ÷ median daily surplus (curve available from the harness). Side findings: oversizing guardrail confirmed empirically; self-use ratio-dependence (P1.5) confirmed directionally. Confidence medium-high (single weather year; limitations in report). Founder decision required.

9. Dynamic-trading uplift ✏️ structure confirmed, numbers volatile

• Spec assumption: gross 0/20/35 €/kWh-usable/yr, 30% revenue share, net cap €40/kWh-usable/yr.
• Findings (all vote 3-0):
  • RVO/DNV report (May 2025) — the best independent source found: rapport: passive imbalance trading averaged ~€97/kWh-installed/yr gross (onbalansmarkt.com, Dec 2024, 216 batteries) but DNV explicitly calls this not sustainable; day-ahead-only arbitrage yields just €5–7/kWh-usable/yr (CE Delft); self-consumption remains the dominant value model.
  • Frank Energie (interested source) claims €65/kWh-usable/yr; live onbalansmarkt telemetry (26-07-2026, 283 Frank systems) shows daily results from €3.25 down to negative on a 100 kWh system (€0.03/kWh that day) — revenue is highly volatile and shrinking as participation grows.
• Assessment: the spec's 0/20/35 gross triple sits sensibly between the unsustainable imbalance peak and the meagre day-ahead floor; the €40 net cap is doing exactly its job against interested claims like Frank's €65. Keep the triple and the cap; label confidence low (already enforced in the engine); revisit quarterly against onbalansmarkt/DNV.

10. Installed cost bands ✅ (reference data)

• Finding (vote 3-0): RVO/DNV market analysis (Dec 2024, 60 batteries, 16 suppliers, incl. installation): ~€1,005/kWh ≤5 kWh · €723/kWh 5–10 kWh · €565/kWh 10–15 kWh · €499/kWh >15 kWh (roughly linear decline); CE Delft (NREL 2023 basis): €1,000/kWh @5 kWh, €800/kWh @10 kWh.
• Action: use as sanity bands for M2 offer verification and G14 quote-band widths (not an engine constant — netCost always comes from the actual offer). Confidence high for late-2024; prices are falling, so re-check at launch.

---

Summary for Ev — what needs your decision

Freezing order that unblocks the most: §1+§2 extractions → §4 decision → §3 → then G1–G15 numeric freeze and the 0.6 independent review.