We developed an algorithm that increased the success rate of detecting electricity theft from 15% to 50%.

Thanks to this, the client significantly reduced costs and gained a system capable of identifying even atypical energy fluctuations in the grid.

Assignment

The largest electricity distributor in Spain faced issues with non-technical losses – in other words, electricity theft. Typically, this involved cases where a household had a legal connection but also set up an unofficial second one, for example to power air conditioning in the summer. The original control method relied only on voltage measurements at distribution points, but its success rate was just around 15%.

Our Solution

We developed an algorithm that uses a topological network model and voltage data. It can:

• estimate missing voltage data,
• determine the actual energy flow,
• identify suspicious sections of the network.

The algorithm repeatedly evaluates current and voltage readings, continuously refining its estimates and enabling the localization of electricity theft with much greater accuracy.

Result

Our method increased the success rate of detecting illegal electricity consumption to 50%, compared to the original 15%.

This meant a significant reduction in financial losses and operational costs for the client. Moreover, the system is not limited to detecting theft – it can also recognize atypical energy consumption fluctuations, such as locations in the distribution grid where losses occur due to poor infrastructure conditions.

Iberdrola is one of global energy leaders with over 170 years of history. It is the number one producer of wind power, and one of the world’s biggest electricity utilities in terms of market capitalisation.
https://www.iberdrola.com