Prediction of the circulation velocity in a membrane bioreactor

In this study, the hydrodynamics of a pilot scale membrane bioreactor (MBR) with different configurations for air scouring were investigated experimentally and numerically. The results were analyzed with the aim to derive a correlation between the aeration flow rate and the circulation velocity in the reactor, as it is well known for airlift loop reactors without internals. To achieve a fouling reduction by air scour at a minimal energy input, an airlift loop configuration was chosen because higher cross flow velocities and therefore higher shear rates can be obtained on the membrane surface. The experimental investigations were carried out with water and air in a quasi 2-dimensional model with 2.1 m height, 1.2 m width and 0.1 m depth. The bubble distributions were optically analyzed by video imaging through the transparent walls of the tank. Numerical simulations were used to perform parameter studies by varying geometrical values or operating conditions (e.g. channel width, bubble diameter). A first approach to calculate the circulation velocity depending on the geometry of the flat sheet membrane module and the aeration intensity was derived.

A modified approach according to the procedure given by Chisti et al. [7] was derived to calculate the circulation velocity in a membrane bioreactor which was designed as an airlift loop reactor with a flat sheet membrane module in the riser. Based on the assumption that the pressure drop of the membrane module is caused by the liquid phase only, the cross flow velocity can be determined depending on geometric parameters like length and thickness of the membrane plates and the distance between them. This mathematical model can be applied to perform a design optimisation to balance the aeration flow rate (effort) with the circulation velocity (benefit). Further investigations have to be carried out to adjust this approach to reach a better agreement with the experimental data. Especially the estimation of the gas holdup in the downcomer needs to be adapted to real conditions. To predict the downcomer gas holdup more exactly in comparison to the linear Eq. (3) given by Bello et al. [9] the bubble size and the liquid velocities in the downcomer have to be considered [12]. Numerical simulations are helpful to compare the impact of geometric variations and different operating conditions to the flow field and the bubble distribution. This concerns especially the arrangement of the aeration tubes above the recirculation zone at the bottom of the riser which is important to obtain a homogeneous aeration of the entire membrane module. It was shown, that a satisfactory agreement with the experimental data strongly depends from the selected bubble size.