Digital twin

A digital twin combines a physical model of the plant with live data. For a solar plant, it takes measured irradiance, module temperature and the plant's layout, and computes the expected output of each string, inverter and block. For a wind farm, it takes the wind speed and direction at each turbine, applies the turbine's power curve and allows for wake losses from its neighbours. The model is calibrated against the plant's own history so that its expectations match a healthy plant.

Every few minutes, the twin compares its expected output with what SCADA and the meters report. A persistent gap points to a cause. One string reading low suggests a local fault or shading. A whole inverter low suggests a trip. A slow drift across the plant suggests soiling or degradation. The twin can also run forward on forecast weather to predict output, or test a change, such as a new tracker setting, before it is made on site.

An asset manager running many plants cannot inspect every inverter every day. A twin turns raw data into a ranked list of energy lost and where it was lost. That sends field crews to the faults that cost the most. It also gives a fair baseline for warranty and contractor claims. A plant losing 2% to undetected faults loses 2% of its revenue for as long as nobody notices.