AI-Forecast

The problem

Water demand in urban areas is constantly growing, driven by socio-economic factors and climate change, putting increasing pressure on water distribution systems. In this scenario, being able to rely on accurate short-term demand forecasts has become a fundamental requirement for truly efficient network management.

The challenge, however, is not just forecasting: it’s doing so with tools that are genuinely usable in utilities’ daily practice. Scientific research has proposed numerous advanced forecasting models over the years, but there are still few examples of complete tools — capable of starting from raw data, often incomplete or noisy, and arriving at a reliable, ready-to-use forecast.

How it works

AI-Forecast was created to bridge this gap with a complete data-driven chain: it imports historical consumption data, automatically processes it to detect and correct anomalous values and reconstruct any missing data, then applies advanced predictive models to estimate future water demand.

The system provides updated forecasts over a 24-hour horizon, with hourly resolution, giving utilities a dynamic predictive tool rather than a static estimate. The entire chain — from raw data to the final forecast — is designed to be modular: new models and techniques can be integrated over time without having to redesign the whole system.

Forecasts are accessible through a web interface designed to clearly present not only the predicted value, but also information useful for assessing its reliability.

The impact

Knowing expected water demand in advance allows utilities to schedule pumping operations more efficiently, for example by taking advantage of the most favourable time slots in terms of energy costs and tariffs, with a direct impact on operating costs.

A reliable forecast also makes it possible to optimise reservoir volume management, reducing waste and improving the reliability of the service provided to citizens.

AI-Forecast stems from the scientific research activity of the AIAQUA team, a spin-off of the Free University of Bozen-Bolzano, with the aim of showing how academic research results can become a practical tool supporting the daily work of water utility managers.