GridOps
the right model for the right horizon
Wind and solar are nonlinear dynamical systems. AI forecasts perform well 1–4 hours out and degrade sharply at 24–48 hours. Grid operators using AI for 24-hour schedules are routinely operating past their model's Lyapunov horizon — and paying for it. GridOps matches every scheduling window to the model most reliable at that horizon.
27.2%reduction in grid instability penalties for 24-hour wind dispatch (peer-reviewed result).
The pain point
A grid operator using one AI forecast across all horizons accrues balancing penalties that dwarf any software budget. The ROI of horizon-aware selection is measured in millions per GW per year.
A 1.2 GW onshore wind portfolio running day-ahead dispatch into an ENTSO-E balancing market.
Without Rebound
- Single transformer forecast model used for 1h, 4h, 12h, 24h, 48h windows.
- Balancing penalties pile up in long-horizon windows during storm fronts.
- No automated regulatory export for FERC, NERC, or ENTSO-E reviews.
- Scheduling continues through regime transitions without operator awareness.
With Rebound
- Reservoir models handle 1–4h; CNN-LSTM covers 12h; transformer only wins at 24h.
- Dispatch Horizon Optimizer selects per-window at bid time.
- Regime Alert triggers conservative scheduling when λ spikes.
- Conformity reports for FERC, NERC, ENTSO-E generated from the audit ledger.
Baseline vs Rebound · per-window balancing penalty on a 1.2 GW portfolio.
gridops.Rebound.io / portfolios / onshore-NL-1 / dispatch1.2 GW wind portfolio · €/MWh penalty
1h
Baseline
transformer-v3
14.2 €/MWh
Rebound
reservoir-net
8.9 €/MWh
−5.3
4h
Baseline
transformer-v3
18.5 €/MWh
Rebound
reservoir-net
11.1 €/MWh
−7.4
12h
Baseline
transformer-v3
27.3 €/MWh
Rebound
cnn-lstm
19.8 €/MWh
−7.5
24h
Baseline
transformer-v3
41.7 €/MWh
Rebound
transformer-v3
30.4 €/MWh
−11.3
36h
Baseline
transformer-v3
58.9 €/MWh
Rebound
— blocked —
—
refused
| Window | Baseline model | Penalty €/MWh | Rebound choice | Penalty €/MWh | Delta |
|---|---|---|---|---|---|
| 1h | transformer-v3 | 14.2 | reservoir-net | 8.9 | −5.3 |
| 4h | transformer-v3 | 18.5 | reservoir-net | 11.1 | −7.4 |
| 12h | transformer-v3 | 27.3 | cnn-lstm | 19.8 | −7.5 |
| 24h | transformer-v3 | 41.7 | transformer-v3 | 30.4 | −11.3 |
| 36h | transformer-v3 | 58.9 | — blocked — | — | refused |
Portfolio
1.2 GW wind (NL)
avg penalty price €50/MWh
Monthly savings vs baseline
€ 1.14M
avg −27.2% across windows
Regulatory export
FERC · NERC · ENTSO-E
one-click conformity report
Ideal customer profile
- Utilities managing renewable portfolios (wind, solar) ≥ 500 MW
- Grid operators: PJM, CAISO, ERCOT, National Grid, TenneT, 50Hertz
- Energy trading desks at BP, Shell, EDF Trading, RWE
- Smart-grid software vendors integrating AI dispatch
Buyers we work with
VP Grid OperationsHead of TradingChief Digital OfficerRegulatory & Compliance
Typical deployment
Regional Utility
1–5 GW portfolio
$300K–$800K / yr
ISO / RTO
Large grid operator
$1.5M–$5M / yr
Energy Trading Desk
Per desk
$500K–$2M / yr
OEM Integration
Smart-grid software OEMs
8–15% rev share
Ready to scope a pilot?
A 60-day Predictability Audit on your real workload, no commercial commitment.