The world faces an increasingly urgent challenge of water scarcity, driven by population growth, climate change, and inefficient resource management. Innovative solutions are critical to ensuring sustainable access to and usage of water. Among the most promising technologies is Seawater Reverse Osmosis (SWRO) desalination, which has emerged as a pivotal tool in addressing this crisis. However, SWRO is energy-intensive due to the substantial energy required to pressurise seawater and overcome osmotic pressure to produce fresh water.
To optimise SWRO operations, real-time smart management of operating parameters is essential. By reducing energy consumption and minimising chemical use, these innovations help adjust water production based on demand and prevailing water conditions. The adoption of real-time monitoring and advanced simulation tools, such as digital twins, is crucial to achieving these goals.
By definition, reverse osmosis is the process where water molecules are forced to move from a region of high solute concentration (high osmotic pressure) to a region of low concentration (low osmotic pressure) across a semipermeable membrane, against natural osmotic flow, by applying external pressure greater than the osmotic pressure.
Research Spotlight: Aissam Daaboub’s Master’s Thesis
Aissam Daaboub, a graduate of CIHEAM Zaragoza’s MSc programme “Sustainable Water Management and Governance in Natural and Agricultural Environments”, has provided significant insight into improving desalination processes. His Master’s thesis, titled “Development of Predictive Solutions for Drinking Water Generation and Distribution”, explores the potential of data-driven models to simulate SWRO behaviour and optimise processes.
This research was conducted under the supervision of Edgar Rubión Soler, Head of Water and Climate Change AI Solutions at Eurecat, a leading technology centre in Catalonia, Spain.
“The second year of the master's programme was a rewarding learning experience for me”, recalls Aissam. “The choice of topic came from my desire to strengthen the knowledge I acquired during the first year, particularly in the area of digital transformation in the water sector. This motivated me to get in touch with Eurecat through the programme coordinator, Maite Aguinaco. This project has helped me a lot in developing skills in mathematical modeling, programming, machine learning, simulations, and solving real-world problems”.
Predictive solutions for generating drinking water
The primary goal of this research was to develop predictive solutions for drinking water generation in response to the challenges posed by water scarcity and climate change. The study focused on enhancing Seawater Reverse Osmosis (SWRO) desalination through the integration of digital twins and machine learning (ML), which enable real-time monitoring, ‘what if’ scenario simulations, process optimisation, and autonomous control. By leveraging these tools, the research aimed to improve energy efficiency, reduce operational costs, and ultimately minimise the environmental impact of desalination technologies.
Applicability of machine learning models
The research yielded several significant findings that advance the understanding and optimisation of the seawater reverse osmosis (SWRO) process. A robust mechanistic model was developed, accurately replicating the complex behaviour of the reverse osmosis process and demonstrating high precision in predicting key performance indicators, including permeate salinity, water recovery, and specific energy consumption (SEC). Additionally, data-driven machine learning models, such as CatBoost and XGBoost, were tested for their ability to simulate desalination processes. Analysis using SHAP (SHapley Additive exPlanations) confirmed these models’ effectiveness in capturing system mechanisms.
Furthermore, the study underscored the real-world applicability of these machine learning models, demonstrating their potential to enhance system performance, improve energy efficiency, and balance trade-offs among crucial variables like permeate salinity, SEC, and permeate flow. This work highlights the transformative role of machine learning in advancing SWRO technology, contributing to the development of digital twins to enable more efficient and controlled operations.
About the author
Aissam Daaboub graduated as an agricultural engineer specialising in plant and environment protection from the National School of Agriculture in Meknes, Morocco. He later earned an MSc in Sustainable Water Management and Governance in Natural and Agricultural Environments at CIHEAM Zaragoza.
Aissam’s research interests span sustainable water management, climate change, droughts, desalination, artificial intelligence, remote sensing, hydrological modelling, and simulation. He is particularly focused on developing data-driven solutions for desalination, wastewater treatment, and hydrological systems. Currently, he works as an Applied Artificial Intelligence Researcher in Water and Climate Change at Eurecat in Catalonia.