Year

2010

Degree Name

Master of Engineering - Research

Department

School of Civil, Mining and Environmental Engineering

Abstract

In the 21st century, freshwater scarcity is an emerging worldwide crisis that threatens human activity. Because of this lack of freshwater, an alternative resource using recycled wastewater or seawater has recently become a more attractive option. Pressure-driven membranes play a vital role in water treatment and desalination. Comparing conventional processes, microfiltration (MF) membranes can achieve more stable performance in eliminating colloidal particles due to its high removal efficiency. However, membrane fouling is the major problem which causes a loss of the filtration performance over time and eventually increases energy consumption and plant operating and maintenance costs.

Past studies have widely investigated membrane fouling and its models for MF and Ultrafiltration (UF) membranes. Most models were often evaluated in bench-scale, simplified, non-practical experiments using a flat-sheet membrane and can only expect fouling behaviour under a limited condition. This is because past studies may have focussed on understanding of fouling mechanism under a well-controlled condition. In practice however, hollow fibre membranes are commonly employed in full-scale MF and UF systems which are generally operated under a constant flux, direct-flow mode with regular backwash. Although a few fouling models have been proposed for a hollow fibre membrane, they are still under development.

For better plant design and optimisation, it is necessary to develop a more practical model for the prediction of the full-scale MF performance. Hence, a new analytical fouling model has been developed in this study. This model is able to describe the effects of the major fouling mechanisms such as pore blockage, adsorption and ‗adhesive‘ cake formation during the filtration with regular backwash employing hollow fibre membranes under typical industry operated conditions. The new fouling model developed incorporates both the membrane resistance model and adhesive cake resistance model. The model also takes into account all features that contribute to membrane fouling except the detailed water quality characteristics.

The new model was evaluated under practical conditions by applying it to the operating data through the simulation of 3D Curve Fitting Tool on MATLAB software. The data were obtained from an industry operated MF system treating secondary effluent for industrial reuse, where is located in the Wollongong Wastewater Reclamation Plant (WWRP) in Australia. From the analysis of the operating data, it is found that the MF system is less sensitive to irreversible fouling impact and can maintain the designed filtration performance over a relatively long-term operation without Clean-In-Place (CIP), e.g. 23 weeks including short-term shutdown periods. These findings suggest that it is possible to extend the current CIP interval longer than the manufacturer‘s recommendation, e.g. every 9 weeks. Moreover, the performance degradation in the MF system has a good agreement with observation of the cross-flow filtration by Xie et al., (2008).

On the simulation of the curve fitting, the new fouling model has successfully estimated the fouling behaviour and cake characteristics showing a strong correlation (e.g. R2 =0.893) to the operating data for a relatively long-term operation (e.g. 2016 hours). For example, the model simulated that the clean membrane resistance was a dominant factor over total resistance (e.g. 73%), whereas adhesive cake resistance was the next dominant factor (e.g. 23%) but gradually increased over time. The model also estimated that adhesive cakes forming on membrane fibres were thinner, less concentrated (more porous), and moderately compressible but have lower resistance at the initial filtration stage (e.g. S=0.641, φc=9.632×10-5 and δc(t)=55.27μm on the first week). However, it was predicted that these cakes became thicker, more concentrated, and less compressible but had higher resistance after a long-term operation (e.g. S=0.256, φc=6.136×10-4 and δc(t)=98.34μm after 12 weeks).

Importantly, the simulation results showed that it is possible for the new model to predict the system performance without detailed investigation of foulants. Furthermore, it is estimated that the order of the most sensitive fouling factors affecting the filtration performance were: i) clean membrane resistance, ii) index of adhesive cake compressibility, iii) density of particles presented in feed, iv) flux and overall pressure drop, and v) feed concentration. Finally, it is recommended that further studies should be conducted to investigate other fouling mechanisms such as effects of feed water quality characteristics, effects of regular backwash, and cake characteristics during the filtration with regular backwash.

Share

COinS
 

Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.