Presentation given at 6th IWA Specialist Conference on Membrane Technology for Water & Wastewater Treatment 4-7 October 2011, Aachen, Germany

1. Numerical modelling of non-Newtonian
fluid in a rotational cross-flow MBR
T. R. Bentzen1, N. Ratkovich1, S. Madsen2, J. C.
Jensen2, S. N. Bak2 & M.R. Rasmussen1
1Department of Civil Engineering, Aalborg University – Denmark
2Grundfos BioBooster – Denmark
6th IWA Specialist Conference on Membrane Technology for Water
& Wastewater Treatment
October 5, 2011, Aachen - Germany
No. 1 of 17

2. Outline
Introduction & Objectives
• Fouling
• Rotational Cross-Flow MBR
Background
• Non-Newtonian behavior (viscosity)
• Wall shear stress in rotating systems
Methodology
• Experimental measurements & CFD model
Results & discussion
• Wall shear stress
Conclusions
No. 2 of 17

3. Introduction
Membrane fouling
• Fouling is the main bottleneck of the widespread of MBR systems.
• Decrease permeate flux
• Increase trans-membrane pressure
Control/reduction of fouling
• Process hydrodynamics can decrease and/or control fouling…
• by increasing liquid cross-flow velocity.
• Increase permeate flux
• Surface shear stress → scouring effect
• Increase mass transfer (cake layer → bulk region)
No. 3 of 17

4. Rotational cross-flow (RCF) MBR (Grundfos
BioBooster®)
• It operates…
• between 20 – 40 lmh
• pressurize system (~5 bar)
• up to 5 times higher sludge
concentration than in
conventional MBR systems
(TSS up to 50 g l-1).
• Rotating impellers between
filtration membrane discs
prevent fouling.
• Impellers ensures low viscosity
in the reactor biomass due to
the non-Newtonian behaviour
of activated sludge (AS).
• energy consumption and
flux. No. 4 of 17

5. Background (I)
No. 5 of 17

6. Background (II)
No. 6 of 17

7. Methodology (I)
Tangential velocity measurements
• RCF MBR operates between 50 to 350 rpm
• Experimental tangential velocity measured at 59, 119 and 177 rpm
with water
• Measured with Laser Doppler Anemometry (LDA)
• LDA is an optical technique to measure velocity field in transparent
media and cannot be used with activated sludge.
No. 7 of 17

8. Methodology (II)
CFD model
• Star CCM+ v6.04
• Single phase
• Rigid body motion
• Turbulent model
• k- SST (water) - validation
• Laminar (AS)
• Enhanced wall treatment (y+ < 1)
No. 8 of 17

9. Results (I)
Reynolds number for water
Reynolds number for AS
No. 9 of 17

10. Results (II)
Tangential velocity measurements for water
• A good agreement between the experimental measurements and the
CFD simulation results, with an error up to 8 %.
No. 10 of 17

11. Results (III)
Shear stress (for water)
50 rpm 150 rpm
250 rpm 350 rpm
No. 11 of 17

12. Results (IV)
Disk type k
Shear stress for water Infinite smooth
0.313
disks
• Shear stress at the membrane
Smooth disks 0.450
• Theoretical solution:
Disk with vanes 0.650 – 0.840
• k: velocity coefficient which depends on disk geometry (vanes)
Grundfos (14
0.795
vanes)
2000 rpm
*Torras et al, 2009
No. 12 of 17

13. Results (V)
Area-weighted average shear stress at the membrane
No. 13 of 17

14. Results (VI)
250 rpm
Shear stress for AS
30 g/l 40 g/l 50 g/l
No. 14 of 17

15. Results (VII)
250 rpm
No. 15 of 17

16. Results (VIII)
Area-weighted average shear stress
(Pa)
AS AS AS
Water
(rpm) (30 g l-1) (40 g l-1) (50 g l-1)
50 1.8 13.5 22.6 34.3
150 8.2 19.9 29.0 40.7
250 13.8 25.5 34.6 46.3
350 19.2 30.9 40.0 51.7
No. 16 of 17

17. Conclusions
• A proper validation of the CFD model was made in terms of
tangential velocity measurements using a LDA system with water.
• RCF MBR operates with AS and LDA measurements cannot be
made.
• CFD model was modified to account for the viscosity of AS.
• Local shear stress at any place of the membrane surface and area-
weighted average shear stress was determined.
• An empirical relationship was made, to determine the area-
weighted average shear stress in function of the angular velocity (in
rpm) and the TSS.
No. 17 of 17

18. Thank you for your attention
No. 18 of 17

19. Consultancy services on CFD for MBR
• CFD has become a frontline tool for virtual simulations (conceptual design
& performance prediction)
• CFD not only gives qualitative data very exact quantitative predictions.
• CFD tools and techniques are extensively validated with experimental and
analytical results allowing more robust models for R&D.
• Software expertise:
• CAD: Rhino
• Meshing: GAMBIT, ICEM, Star CCM+
• Solver: Star CCM+, CFX, Fluent
• Domain expertise
• CAD repair for meshing
• Moving mesh
• Turbulence modelling
• Multiphase modelling
No. 19 of 17