Understanding fouling and optimizing cleaning for membranes used for rainwater harvesting

Sammanfattning

This master degree thesis project focuses on assessing the performance of flat-sheet ceramic silicon carbide (SiC) microfiltration membranes used for rainwater purification. The rainwater roof runoff samples were collected from a sedimentation tank installed in the Pilot Hall at Kemicentrum, Faculty of Engineering, Lund University, Sweden. Two membranes were tested in multiple varying fouling and cleaning sequences obtained through a design of experiments (DoE). A suction pressure applied through a peristaltic pump installed on the permeate side created a driving force for this membrane filtration. Experiments assessed the influence of sodium hydroxide and citric acid as cleaning agents chosen through a literature review. The theoretical part of the project covers information on rainwater harvesting systems, including contaminants commonly found in the roof runoff. Pressure driven membrane processes and water quality requirements for both potable and non-potable applications have been described. Moreover, different fouling mechanisms and cleaning approaches divided into physical and chemical methods were discussed. The experimental part of the project involves examining the membrane performance through a series of fouling and cleaning tests. The fouling layer was analysed through selected analytic methods: scanning electron microscopy combined with energy dispersive X-ray spectroscopy (SEM-EDS), atomic force microscopy (AFM), Brunauer-Emmett-Teller (BET). Visual differences between the active layer and the porous support of the membrane were described, along with elemental characterisation. Membrane roughness and area were assessed. Conducted project shows that the optimal cleaning of membranes is achieved through the application of sodium hydroxide at the highest used concentration of 1.5 wt.%. Citric acid exhibits a negative influence on the process, with the highest cleaning effectiveness achieved in the absence of the agent. Time has little influence, with medium values of approximately 61 minutes exhibiting the best results. Although the applied cleaning strategy was successful at improving the flux, there is still room for improvement, and future recommendations have been proposed.