Modelling of forward osmosis process for performance prediction

Forward osmosis (FO) is an emerging membrane technology that can be used to produce clean water from industrial wastewater, contaminated water sources, and saline solutions by applying utilising inherent osmotic pressure differences between solutions across the membrane. It has advantages in low energy and cost with additional pressure requirement, achieving high osmotic pressure differences, as well as its high recovery rate in recycling wastewater. However, some limitations slow down the development and application of the FO technology, such as low water flux, membrane fouling and draw solution recovery. Modelling provides an alternative solution to experimental work to obtain instantaneous and accurate results without repeating lab experiments and wasting resources. Therefore, modelling was proposed for predicting the process performance accurately and further exploring different system parameters for an improved understanding of the FO processes. In this study, the main objective is to use modelling to gain a better understanding of FO process and provide guidance for performance prediction and process optimisation. Two different modelling approaches including the Solution Diffusion (SD) model and the machine learning model Artificial Neural Network (ANN) were explored as the major modelling methods in this study. After building and optimising the models, both models were validated to generate satisfactory results. By further analysing these results, some unique findings were obtained to enhance FO process efficiency from the fundamental perspective.