An assessment of depolarisation models of crossflow microfiltration by direct observation through the membrane

https://doi.org/10.1016/S0376-7388(00)00334-3Get rights and content

Abstract

Direct observation through the membrane (DOTM) has been applied to monitor the deposition of supermicron particles (3–12 μm) on microfiltration membranes and to identify the critical fluxes for cake formation as a function of crossflow. The observed critical fluxes are compared with predictions from various models. It is found that the shear-induced diffusivity model predicts acceptable critical flux for latex particles of 6.4 μm and yeast (5 μm). A modified shear-induced diffusivity model for dilute suspensions predicts closely for yeast and latex (6.4 and 11.9 μm), with the exception of 3 μm latex particles. The shear-induced hydrodynamic diffusivity model over-predicts the critical fluxes by a factor of about 2. The inertial lift model under-predicts the critical fluxes for the microfiltration of all the particles. This study also suggests that for smaller particles (3 μm latex and 4 μm algae), in addition to hydrodynamic effects, other factors such as surface charge may need to be considered for particle back-transport mechanisms.

Introduction

Many theoretical models have been proposed to describe the flux behaviour in crossflow microfiltration (CFMF). The major differences in these models are the mechanisms of particle depolarisation from the membrane region to the bulk flow. Several review papers [1], [2], [3] have summarised the development of the most popular models, and the conditions where each of the models should be applicable.

Flux in CFMF increases with increasing crossflow velocity because cake deposition decreases. There are two main classes of hydrodynamic theory for the reduction of particle deposition. One is based on the enhancement of diffusion due to the interaction of particles in a shear flow [4], [5]. This shear-induced diffusion coefficient has been used by Zydney and Colton [6] in the classical film model and by Davis and Leighton [7] in a model which incorporates axial transport of the cake layer. The other class is based on the production of an inertial lift force away from the wall by the velocity gradient [8]. In addition to the diffusion and inertial lift theories, a third approach is based on an analysis of the drag forces on a particle at the membrane surface which determines whether the particles are going to stick or move along the surface due to the tangential flow [9], [10].

Deposition of particles on the membrane surface may also be influenced by surface interaction, such as electrokinetic effects. Bacchin et al. [11] have developed a surface interaction enhanced migration model that predicts the effect of membrane and surface charge on the critical flux at which deposition occurs. This model is most relevant to submicron particles and those with a substantial surface charge.

Wakeman [12] suggested that the mathematical models in the literature should be regarded as ‘black box’ approximations to the particle dynamics which might exist in the filter close to the membrane surface, and this situation can only be rectified when direct observations of the particle motions are made in typical microfilters using feed suspensions containing particles smaller than about 20 μm.

In this paper the direct observation through the membrane (DOTM) technique has been used to observe the deposition of supermicron particles (3–12 μm) on microfiltration membranes. The DOTM technique can identify the critical flux, here defined as the flux at which deposition commences, for a given crossflow and feed suspension [13]. At the critical flux, convection for particles to the membrane balances depolarising back-transport. The critical flux versus crossflow data are used in this study to assess the two main hydrodynamic theories of depolarisation, i.e. shear-induced diffusion and inertial lift, which should operate prior to cake formation (as well as after cake formation). The drag force models [9], [10] which apply to formed cakes are not assessed here. Apart from the hydrodynamic factors such as shear and particle size, the concentration effect on the critical flux was also examined with our experimental results.

Section snippets

Experimental methods

The experimental set-up including the DOTM facility for critical flux observation is illustrated in Fig. 1. The key to the technique is the use of the Anopore (Whatman, UK) anodised aluminium membranes which are transparent when wet. The majority of the experiments used the 0.02 μm pore membrane. A specially designed membrane module was made of perspex allowing the transmission of light from the feed side of the module. The objective lens of a microscope (Axiolab, Zeiss) was positioned on the

Comparisons between observed critical fluxes and model predictions

The critical fluxes obtained with DOTM are plotted in Fig. 3 against crossflow velocity for the range of particles tested. The trends in the data are that the critical fluxes increase with crossflow velocity and particle size. The two main depolarisation mechanisms assumed in microfiltration models are compared below with our experimental data. The critical fluxes have not been corrected for gravitational effects, which are negligible (see Appendix A).

Conclusions

The critical fluxes for microfiltration of submicron sized particles were observed at various crossflow velocities using the DOTM technique. These results were used to assess the SID models and inertial lift model for their prediction of the critical flux.

It was found that both the SID model which assumes the shear-induced diffusivity D=0.03a2γ0 predicts reasonable critical flux for latex 6.4 μm and yeast, over-predicts latex 11.9 μm. A SID-MOD model which uses the shear-induced diffusivity for

Nomenclature

    a

    particle radius (m)

    Cd

    concentration coefficient

    Cn

    wall correction factor for Stokes law

    dp

    particle diameter (m)

    D

    diffusivity (m2/s)

    Fn

    net drag force (N)

    Fw

    gravitational force (N)

    H0

    height of the membrane channel (m)

    J

    permeate flux (m/s)

    Jcrit

    critical flux (m/s)

    L

    length of the membrane (m)

    Re

    Reynolds number of the feed

    Rep

    Reynolds number of the particles

    U0

    velocity of the fluid (m/s)

    ub

    lateral migration velocity (m/s)

    uw

    settling velocity (m/s)

    vL,0

    maximum lift velocity (m/s)

    y

    dimensionless distance of the

Acknowledgements

The authors wish to acknowledge the help of Georg Hagmeyer who obtained the data for algae. Whatman Ltd. kindly provided the membranes that made the work feasible. The project is funded by the Australian Research Council, who are gratefully acknowledged.

References (23)

  • E.C. Eckstein, D.G. Bailey, A.H. Shapiro, Self-diffusion of particles in shear flow of a suspension, J. Fluid Mech. 79...
  • Cited by (124)

    • Membrane-based separation for oily wastewater: A practical perspective

      2019, Water Research
      Citation Excerpt :

      Either a scaled-down version of the process is studied or the membrane is taken offline for measurement of pore size or membrane autopsy after the operation is shut down. The direct observation through the membrane (DOTM) technique (Li et al., 1998, 2000, 2003; Neal et al., 2003; Tanis-Kanbur et al., 2018; Tanudjaja and Chew, 2018; Tanudjaja et al. 2016, 2017, 2018; Tummons et al. 2016, 2017; Wicaksana et al., 2012; Zamani et al. 2016, 2017c; Zhang et al., 2006a, 2010) and other microscopic techniques, such as Direct Microscopic Observation (DMO) (Subramani et al., 2009) or Direct Visual Observation (DVO) (Culfaz et al., 2011; Mores and Davis, 2001), are based on the usage of light microscopy to directly observe the membrane surface. Firstly introduced by Li et al., (1998) as a novel way to study particle deposition on the membrane, the DOTM uses transparent membranes to enable the visualization of the feed-membrane interface (Fig. 7).

    View all citing articles on Scopus
    View full text