IBM machine learning brings sewage cost and performance benefits for Aqualia

  • Research


Cost and performance benefits for Spanish utility Aqualia’s Lleida wastewater treatment plant have been revealed following work by IBM researchers to implement use of machine learning algorithms at the plant.

The pilot project involving Israel-based IBM Research was undertaken as part of the IT giant’s First-of-a-Kind programme to deliver innovative real-world solutions for end users. The company and utility have announced that it has brought a 13.5% cut in electricity use, a 14% cut in chemical use for phosphorus removal, and a 17% reduction in sludge production at the 96,000 m3/day plant, along with improvements in total nitrogen removal, particularly at low temperatures.

The approach draws on a range of sensors and information sources and applies machine learning algorithms to learn and predict the impacts of factors such as weather, plant malfunctions and equipment maintenance in order to achieve a dynamic optimisation of the plant. The system delivers recommendations every two hours to plant operators, allowing them to adjust operational settings to deliver efficient operation. The researchers have reported that gains have included an improvement in total nitrogen efficiency of more than 20% in some months.

The project was co-financed under the INNPRONTA programme of Spain’s Centre for Industrial Technology Development, with funding from the EU European Regional Development Fund.

The project forms part of IBM’s wider efforts in the development of cognitive systems, which use natural language processing and machine learning ‘to enable people and machines to interact more naturally to extend and magnify human expertise and cognition’, according to the company. These systems will learn and interact to provide expert assistance in many arenas. ‘Far from replacing our thinking, cognitive systems will extend our cognition and free us to think more creatively,’ is IBM’s promise.


  • IBM, Aqualia, smart utilities