From Metering Data to Dispatch in Under an Hour
eZap ingests live and historical grid data, forecasts demand per feeder, and produces a schedulable generation plan. Here is how it works.
Data ingestion
eZap uses data your grid already generates. No new sensors or hardware are needed.
Metering data
15-minute smart-meter aggregates, distribution feeder SCADA readings, and at least 12 months of load history for seasonal training.
Weather feeds
Per-zone forecasts for temperature, wind, cloud cover, and solar irradiance from ECMWF and Copernicus-compatible feeds.
Calendar signals
Day, holiday, and event signals, with Portuguese and EU holidays pre-loaded and custom calendars configurable per network.
Formats supported: Modbus TCP, IEC 61968/61970 CIM, REST JSON, and CSV for historical backfill.
Probabilistic Demand Forecast
Each feeder model learns from historical load patterns and updates with new meter data, producing demand curves with probabilities, not one prediction.
P10, P50, and P90 bands are generated every 15 minutes for up to 72 hours. P50 is expected demand, P90 guides reserve sizing, and P10 highlights over-commitment risk.
The model adjusts for seasonal patterns and event-driven spikes, including industrial shutdowns and heatwaves.
Dispatch optimization
Using the P50 forecast and procurement constraints, the scheduler builds a least-cost generation plan for the next 72 hours.
Generation unit assignments
Generation units are assigned to each 30-minute block by contracted cost and ramp-rate feasibility.
Reserve margin vs forecast band
A visual view compares committed reserves with P90 demand, making over-procurement visible before the procurement window closes.
Flagged risk intervals
The system flags intervals where capacity is below P90 demand or reserves significantly exceed P50.
Real-Time Monitoring and Deviation Alerts
During dispatch, eZap compares actual demand with the live forecast and flags deviations above configured thresholds.
Alerts flag load tracking above P90 or high divergence from the dispatch schedule. They arrive before the short-notice procurement window closes, giving your team time to act.
Alerts go to the operator console, email, or a webhook connected to your incident management system.
Questions Utilities Ask During Evaluation
With a standard REST API connection to SCADA or a metering aggregator, initial data flow is usually live in 2 to 4 weeks. Historical calibration takes 2 more weeks. First forecast onboarding takes 4 to 6 weeks, depending on data and IT coordination.
Customer data is processed and stored in EU data centers. EU customer data is not transferred outside the EU. eZap is a GDPR data processor, with a data processing agreement in every pilot and commercial agreement.
Beta and early-access pilots have no SLA-backed accuracy guarantee. Each pilot includes monthly forecast-versus-actual reporting for comparison with your baseline. Internal beta results show 22% average improvement in forecast error versus persistence models. Results vary by feeder topology and data quality.
Each feeder is calibrated with its topology in mind. For unusual profiles, including large industrial customers, embedded generation, or seasonal patterns, our team configures model parameters with your operations team. Grid and Regional customers receive dedicated integration engineering support.
Pilot is a 90-day fixed-term agreement at EUR 1,200/month for up to 3 feeders. Continue with Grid or Regional, or exit without penalty. At pilot end, we provide a full export of forecast and accuracy records.