Predicting turbidity in water treatment plants

A predictive analytics system anticipates high-turbidity events so water-treatment operators can act before quality problems escalate.

Challenge

Water-treatment plants can experience sudden turbidity peaks caused by process conditions, environmental effects or upstream events. Operators need enough warning to diagnose the situation and adjust the process before the event affects production.

Data

The project combined historical turbidity measurements with process variables such as conductivity and operating signals. External variables, including satellite and environmental information, were considered when they helped explain turbidity events.

Solution

Artelnics built predictive models to identify patterns that precede high-turbidity episodes, studied the most relevant variables and produced visual diagnostics for model testing and event interpretation.

Results

The resulting workflow transforms historical plant data into an operational early-warning capability, making it easier to anticipate abnormal conditions and support decisions with objective evidence.

Illustrations