A learning observer approach for fouling detection and localization in direct contact membrane distillation systems

As the need for freshwater is continuously growing, seawater desalination technologies are increasingly used to meet these demands. The high solutes’ rejection factor and the low energy consumption of Membrane Distillation (MD) technologies make them high potential desalination techniques. However, they suffer from membrane fouling which deteriorates the system’s performance and causes high maintenance costs. This paper presents a novel approach based on a learning observer for detecting and localizing fouling in Direct Contact Membrane Distillation (DCMD) systems which allows to significantly decrease the maintenance costs. This approach has the advantage of being flexible and computationally inexpensive. Before the architecture and the design steps of the proposed observer are presented, the DCMD module layout in the fault free scheme and in presence of fouling is explained. To illustrate the effectiveness of the proposed fouling monitoring approach, several tests were simulated and two scenarios were considered: homogeneous and non-homogeneous fouling evolution along the membrane. The various simulations demonstrated very encouraging results in both fouling estimation and localization.