{"stable_id":"63b51d1a931c9f58","slug":"google-wind-commitment-optimization","company":"Google and Google DeepMind","workflow_name":"ML-recommended hourly wind-energy delivery commitments","function_code":"logistics","pattern_codes":["continuous_decisioning","step_before_decision"],"changed_assumption":"That variable wind generation cannot be committed to the grid on a useful day-ahead schedule.","evidence_strength":"mixed","publication_tier":"showcase","freshness":"current","reviewed_at":"2026-08-23","updated_at":"2026-08-22","source_quality_summary":"1 independent, 1 primary; publication outcomes are verified and reported.","caveat_summary":"Google-reported value metric; monetary definition, forecast error distribution, and independent audit were not published.","freshness_caveat":null,"workflow_summary":{"before":[{"actor":"Wind operator","action":"Estimate uncertain output without the ML forecast and avoid firm hourly commitments.","actor_type":"human"},{"actor":"Grid","action":"Treat wind as less schedulable and therefore less valuable.","actor_type":"control"}],"hinge":"That variable wind generation cannot be committed to the grid on a useful day-ahead schedule.","after":[{"actor":"Neural forecasting model","action":"Predicts wind output 36 hours ahead from weather forecasts and turbine history.","actor_type":"ai"},{"actor":"Optimization system and operator","action":"Recommend and submit hourly delivery commitments a day ahead.","actor_type":"human"}],"decision_mode":"moved","decision_marker":"Selection moves from a fixed rule to the model"},"before":[{"order":1,"actor":"Wind operator","action":"Estimate uncertain output without the ML forecast and avoid firm hourly commitments.","handoff_to":"Power grid","control":"No time-based commitment baseline"},{"order":2,"actor":"Grid","action":"Treat wind as less schedulable and therefore less valuable.","handoff_to":"Market settlement","control":"Grid balancing rules"}],"after":[{"order":1,"actor":"Neural forecasting model","action":"Predicts wind output 36 hours ahead from weather forecasts and turbine history.","handoff_to":"Commitment optimizer","control":"Forecast uncertainty"},{"order":2,"actor":"Optimization system and operator","action":"Recommend and submit hourly delivery commitments a day ahead.","handoff_to":"Power grid","control":"Operator and market rules govern commitments"}],"decision_rights":"The model recommends commitments; accountable operators and market processes determine submitted schedules.","exception_path":"Operators update positions and manage imbalance when weather or turbine availability diverges from forecast.","removed_work":["Manual combination of weather and turbine history at portfolio scale","Default treatment of wind as entirely unscheduled supply"],"outcomes":[{"metric":"Value of wind energy versus no time-based commitments","baseline":"No time-based delivery commitments to the grid","result":"Roughly 20% higher value","period":"Results through 2019-02-26 after deployment beginning in 2018","scale":"700 MW of Google wind capacity in the central United States","attribution_caveat":"Google-reported value metric; monetary definition, forecast error distribution, and independent audit were not published.","evidence_label":"reported"}],"executive_lesson":"Prediction mattered because it was connected to a changed commitment decision; forecast accuracy alone would not create grid value.","anti_pattern":"Reporting 20% more energy generated; the source says value, not production.","questions_for_leaders":["Which forecast changes a binding operating commitment?","How is imbalance risk allocated?"],"collections":[],"bundle_version":"1.0.0","bundle_fingerprint":"sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93","canonical_url":"https://brianletort.ai/transformations/google-wind-commitment-optimization"}