Optimization of Membran Unit Location in

full-scale Membrane Bioreactor Using Co

mputational Fluid Dynamics
Maratus Sholihah ( 施曼拉 )
10770516
 Introduction & Review
MBR (Membrane Bioreactor)
It is combination of membrane separation technology and conventional activated sludge process,
that widely used in WWTP to provides stable effluent with high quality. MBR has problem in the
high energy consumption from blowers in membrane tanks for membrane fouling control.

CFD (Computational Fluid Dynamic)

It is software that uses numerical analysis and data structure to analyze and solve
problems involve fluid flows. CFD have been applied to study the hydrodynamics
of MBR, and is of great importance to membrane fouling control and energy
saving.

Aim this study

Utilized an integrated model that in multiphase turbulent flow, porous media and non-
Newtonian effect into consideration in predict the hydrodynamics of the membrane tank
in a full-scale MBR to control membrane fouling and energy consumption.
Research Case

Variables in this research :

 Consider factors such as multiphase turbulent flow, porous media and non-
Newtonian effect into consideration to predict the hydrodynamics in
integration model.

 The fifth percentile value of the liquid velocity (V 0.05) in the membrane unit
region was used to judge membrane fouling potential.

 The location was indicated by five indexes, each of which was adjusted
independently to evaluated its individual effect. Finally an optimized
configuration was proposed and compared with the original design.
 Methods

2.1 Numeric
Method

2.2 Experimental
Validation

2.3 Modeling
Process
2.1.1 Basic Model
Equation for modeling
Solve the governing In a bubbly flow (water and dispersed phased)
equations on the discrete
mesh with certain algorithms • The Continuity Equation :

2 • The Momentum Equation :

To discrete the
continuous fluid
into grid (or mesh) 1 3 • The values of k and Ɛ could be derived
this equation :

Yielding the spatial

hydraulic information
2.1.2 Non-Newtonian Model Development
The activated sludge
system was behave as non-
Newtonian fluid whose
viscosity was found to be
related to a number of
factors such as:
• The shear rate
• Temperature
• Mixed liquor suspended
solids concentration
Relationship between shear stress
(MLSS)
and shear strain rate of a non-newton
• Extracellular polymeric
fluid
substances (EPS)
concentration
• Soluable microbial
product (SMP)
concentration et al
2.1.3 Porous Media Model Development
Treat the whole membrane
unit as a porous media that
would cause resistance in
the fluid region. By this
method, an additional
momentum source should
be added as a force per unit The membrane unit has similar geometry
volume acting on the fluid. characteristics of tube banks, so the
viscous loss term could be neglected and
only inertial loss term that calculated :
2.2 Experimental Validation

A bench-scale reactor was constructed to

corfirm the reliability of the modeling. Water It was compared with the water velocity profiles at
velocity was measured among a certain line in the same location obtained by CFD simulation.
the reactor (x=[−20, 20]mm, y=120mm, and
z=15mm).
A good agreement was found between the water
velocity profiles obtained from particle image
velocimetry (PIV) and the simulation data
(relative error of 1.4%), which guaranteed the
accuracy of the modeling.
2.3 Modeling Process

• iLu – Lu/H, the membrane unit – liquid surface

Computational studies (one
interval index, Lu = distance between the
single unit) The entire unit was
membrane unit top and liquid surface, H = height of
simplified as a cuboid of
membrane unit (2000mm).
1250×2700×2000mm3. • iLa – La/H, the membrane unit – aeration pipe
interval index, La = distance between membrane
unit bottom and aeration pipe.
The aeration pipe was set • iLb –Lb/H, the aeration pipe – tank bottom interval
with an air velocity inlet of index, Lb = distance between aeration pipe and
3 m/s, bubbles were tank bottom.
consumed to have a uniform • iLint – Lint/W, the membrane unit interval index,
diameter of 6 mm. Lint = interval between neighboring units, W = width
of the membrane unit (1250mm).
• iLw – Lw/L, the membrane unit – tank wall interval
index, Lw = interspace between unit and tank wall,
L = length of membrane unit (2700mm).
Results and Discussion
3.1 Standart vs Integrated Model
a. The standart MBR, there was streamline shrink to be
denser in the middle of the unit region, which liquid
velocity as high as 3 m/s.
b. In the integrated model, the streamline was found inflated
in the unit region with a much lower velocity (0.4 m/s) and
shrank in the space above the unit. The inflation could be
ascribed to the flow resistance and energy loss caused by
membrane unit (porous region).

• In standart model, liquid velocity had a wide range

from 0-3 m/s. A comparatively high frequency was
found between 0 and 1 m/s with a peak frequency
of 6.9 % between 0.184-0.214 m/s.
• Integrated model had liquid velocity distribution
between 0-0.704 m/s with highest frequency 0.25
% was found between 0.245 m/s and 0.275 m/s.
• The integrated model could produce results
closer to the real motion, thus it was applied
throughout the overall optimizing.
3.2 Effects of Membrane Location on Fluid Dynamics

• The value of liquid velocity V 0.05 among membrane unit region was selected instead to evaluated
the fouling potential. Among different MBRs a higher V 0.05 indicates a lower risk of fouling.
• The location of membrane unit the could be determined by 5 indexes (i Lu , iLa , iLb , iLint , iLw ). It
gave result of the effects of each location index’s dependent effect on the risk velocity (V 0.05) with
a step increase of 0.2 separately to evaluate its individual effects on fluid motion.
3.3.1 Effects of iLu

• Highest velocity about 1 m/s was observed near aeration pipe. As liquid
V0.05 increased from 0.093
continuosly moved upward through the membrane unit, it dramatically
m/s to peak value of 0.155 slowed down to about 0.25 m/s. It due to by sudden drop in resistance
m/s (almost 66.7% higher) loss causing a re-acceleration of liquid by released the bubbles.
as increased from 0.2 to 0.6 • The vortex was gradually formed above the unit with the increase of i Lu. It
then dropped to above 0.6
could be concluded that channel could neither be too narrow, which
until again to the minimum
would result in poor circulation and sever flow resistance, nor could
value of 0.093 m/s.
it be too wide, which may allocate too much aeration energy.
3.3.2 Effects of iLa

V0.05 gradually decreased with the

• The distance was too small the “hyperbola” would be compressed and
increase of iLa, except for a sudden even and eliminated, made the high-velocity which would end up with
rise at iLa = 0.6 (0.132 m/s). The stagnant zones near bottom corners of the unit.
overall trend could be ascribed to • When iLa was 0.6 the “hyperbola” was fully developed and covered
the distance between membrane most of the unit bottom, quite uniform circulating loop was formed.
unit and the aeration pipe, because
energy would decay along the jet • When iLa was above 0.6 the unit would overpass the best effect zone of
orientation. the hyperbola causing stagnant zones again near baffles.
3.3.3 Effects of iLb

• When iLb was 0.2, stagnant zones were found near the bottom corner of the
V0.05 of iLb had a bulk circulation on
membrane unit, when it increased the stagnant zones was reduced.
the top flow of the channel. The risk
• When iLb reached 0.6, the stagnant zones could be hardly observed, the
velocity rose with iLb from 0.110 m/s
overall tank formed a relative uniform circulation with a considerable
to a peak value of 0.228 m/s at i Lb =
increase in the bulk velocity.
0.6 then slowly dropped to 0.203 • When iLb was above 0.6 another huge vortex was formed beneath the
m/s as iLb continued increasing to 1.
aeration pipe, with a small stagnant zones near the unit bottom corner.
3.3.4 Effects of iLw

iLw had a much smaller efect of the fluid

motion. It had a narrow fluctuation around
0.1 m/s. It due to the space surrounding • When iLw was 0.2 the stagnant zones mostly occured at the
the membrane unit made the downcomer bottom corner of the membrane unit, when i Lw was 0.4 it was
for bulk circulation, yet there would be a found at both top and bottom corners of the unit, when iLw
higher resistance for liquid flowing was above 0.6 the flow field tented to be stable.
through the channel between unit and
lateral tank wall that between neighboring
units.
3.3.5 Effects of iLint

iLint decides most portion of the

downstream flow for the bulk It could be observed that the with the unit interval increasing,
[Link] the increase of the vortex above the unit gradually expanded. When iLint was
iLint the risk velocity firstly rose below 0.6 some upflow could be observed in the downcomer,
from 0.095 m/s to a peak value which was not expected for a uniform bulk circulation. When
of 0.159 m/s at iLint = 0.6 then iLint was above 0.6 the upflow in the downcomer dissapeared.
slowly dropped to 0.139 m/s.
3.4 Performance of Optimized Tank

• An optimal design with

• The original design risk
all the location
zones mainly appeared in
indexes equaling at
the up corner near baffle.
0.6 was proposed.
• The optimized design, a
• Compared with the
risk zone were completely
original design, the
removed, a significant
optimized one had a
improvement of the bulk
risk velocity of
motion was observed.
0.242 m/s, which was
This is indicated a succesful
146,9 % higher than
optimization of the unit
the original design
location.
(0.098 m/s)
 Conclusions

 The integrated CFD model was applied to simulate the hydrodynamics of

the membrane tank in full scale MBR.
 The fifth percentile values of the liquid velocity (V 0.05) in membrane unit
was used as the risk velocity to judge the membrane fouling potential.
 An optimal design for the unit location was proposed (i Lu = 0.6, iLa =0.6, iLb
= 0.6, iLint= 0.6, iLw = 0.6) who had an elevation of 146.9% for the
risk velocity (V0.05).
 What I have learned?
o Know some variables that can used to predict the
hydrodynamics in integration model of MBR, there are
multiphase turbulent flow; porous media; non-Newtonian fluid
(V0.05); and the location of membrane unit the by 5 indexes (i Lu,
iLa , i Lb , iLint , iLw )

o Understand severals formulas or equations and steps to

simulation the integration model of MBR.

o Can analyze of each variables and the best result for optimal
design in the full scale MBR unit.
Thank you
谢谢