Modelling of biological fouling propensity by inference in a side stream membrane bioreactor

It has been found by recent researchers that the dominating fouling agents causing membrane pore blocking are extracellular polymeric substances (EPS) which seem to be produced by micro-organisms which make up the biomass. The other form of fouling caused by cake biomass build up on the membrane surface is usually easily removed by standard cleaning and sparging techniques. Consequently, two versions of two modified activated sludge models (ASM), which take into account the production of EPS as soluble microbial products, were tested using reallife data produced from a full-scale side stream configuration membrane bioreactor (MBR) plant. This study was carried out in order to see which version better predicted substrate removal efficiency and by inference the production of EPS which would directly impinge upon membrane fouling rates by subsequent pore blocking. The modified ASM3 with MP model gave better overall results. In conclusion, fully calibrated predictive biological ASM models that include EPS production should give the operator advance warning of potential extreme fouling events, especially when critical flux is exceeded.

The modified ASM3 with MP model gave better overall results although actual measured values of unbound EPS need to be determined so that the models can be properly calibrated for this plant. This would allow fully predictive models to be developed so that a better comparison could be made. Since EPS is composed of proteins and polysaccharides [1], standard chemical assays to measure these components would need to be carried out. However these assays which are time consuming could not be performed regularly. They are also not standardised in wastewater measurement protocols. In conclusion, fully calibrated predictive biological ASM models that include EPS production should give the operator advance warning of potential extreme fouling events especially when critical flux is exceeded. The ultimate aim of this type of modelling research would be to produce a fully integrated predictive model of activated sludge biology (with unbound EPS) coupled to a good membrane filtration model.