Loading...
AI-based remaining useful life prediction and modelling of seawater desalination membranes
Al Ali, Fajer
Al Ali, Fajer
Files
Description
A Master of Science thesis in Engineering Systems Management by Fajer Al Ali entitled, “AI-based remaining useful life prediction and modelling of seawater desalination membranes”, submitted in November 2024. Thesis advisor is Dr. Hussam Alshraideh. Soft copy is available (Thesis, Completion Certificate, Approval Signatures, and AUS Archives Consent Form).
Abstract
The increasing demand for freshwater has heightened the reliance on desalination plants as a vital resource, particularly in the context of the Sharjah Electricity, Water and Gas Authority (SEWA) in the United Arab Emirates (UAE), which integrates different water desalination plants, including reverse osmosis (RO) to produce water. This thesis focuses on the development of an artificial intelligence-based predictive model for estimating the remaining useful life (RUL) of RO membranes. It addresses the critical operational challenge of membrane fouling caused by particle accumulation, which can lead to significant efficiency losses and system damage. The concept of RUL is defined as the anticipated time until the RO membrane reaches a specified performance threshold, guiding maintenance actions to enhance system longevity and efficiency. The predictive model developed in this study utilizes data from SEWA's operational database and laboratory records to forecast the RUL. R Software was employed as the primary tool for building and testing the predictive models, including Linear Regression, Decision Tree, Random Forest, and XGBoost. The Random Forest algorithm demonstrated the best performance, achieving an R² coefficient of 0.984, an RMSE of 0.136, and an MAE of 0.0997. These results highlight the exceptional accuracy and reliability of the model in predicting the RUL. Additionally, the findings from the variable importance analysis revealed that the most significant features influencing the RUL were SDI, water temperature, pump speed, and the age of the membrane. Understanding these key variables can provide valuable insights into optimizing operational conditions, thus extending membrane lifespan. By accurately predicting RUL, the model aims to reduce pressure changes that contribute to fouling and mitigate potential membrane damage. The implementation of this AI-driven model is expected to optimize clean-in-place (CIP) scheduling, ultimately maximizing the longevity and performance of RO membranes. The findings of this research will not only enhance understanding of the necessary operating conditions for effective seawater treatment but will also contribute to the broader literature on predictive maintenance in desalination, supporting more efficient and sustainable water production practices.