All case studies
Energy
AI Load Forecasting for a State Discom
Probabilistic load and renewable-generation forecasts at 15-minute granularity across 4.7M consumers.
Client
State Power Distribution Company
Timeline
10 months
Location
Southern India
2.8%
Forecast MAPE
−₹118 Cr / yr
Exchange spend
−46%
Renewable curtailment
+3.2×
Operator decisions / day
The challenge
Forecast errors above 9% were forcing expensive last-minute power purchases on the open exchange.
Our approach
- Ingested SCADA, weather and rooftop-solar telemetry
- Trained ensemble probabilistic forecasting model
- Built a what-if scheduling cockpit for grid operators
- Integrated automated bid-suggestion module
The solution
A forecasting + scheduling cockpit that materially reduces exchange exposure and improves renewable absorption.
