Using DEM-CFD method to model colloids aggregation and

deposition

Florian CHAUMEIL

Supervisor: Dr Martin Crapper

1

BACKGROUND

Why modelling particulate matter deposition ?

It is of great importance in many industrial processes

such as• micro-contamination control of microelectronics • membrane filtration and fouling of heat exchangers • surface deposition in micro-fluidic devices

In nature, micro-particle deposition is of great interest in • microbial pathogen removal through natural granular filtration of surface

water.

2

BACKGROUND

Results from literature

• hydrodynamic drag mitigates deposition and drives re-entrainment of both biological and non-biological colloids.

• re-entrainment was found to be a monotonic function of the extent of deposition.

• Bigger deposited agglomerates increase floc re-suspension. Therefore, particle-particle interactions in a flowing fluid are shown to be as critical as surface interaction.

• Electrostatic interaction of charged colloids is still a challenging problem for researchers. In the present work surface charges are not hindering deposition.

3

AIM

What will we be looking at ?

Mechanisms of deposition

Mechanisms of deposition mitigation:• Hydrodynamic Drag• Cluster Scouring • Impaction

i.e. what drives particles to deposit in the first place and eventually re-suspend?

4

Numerical Tools

• CFD Package– Simulates

hydrodynamics (velocity and pressure field, shear forces, turbulence intensity)

• DEM Package– Simulates particulate

matter dynamics

couplingData computed by one package influence the data of the other

5

Question

how to model colloidal agglomeration ?

1) What force model :

- DLVO

- Brownian forcesGaussian random numbers (Gi) of zero mean and unit variances

2) How to describe surrounding fluid : - CFD -> Drag- Analytic Near the wall retardation

22222

3322

3

32

baRbaR

abRnFlvdw

3

1iiiB eNF

C

Bp

ii Ct

TkrGN

.

2.6

6

colloids aggregation and deposition in a constricted tube

7

METHODOLOGY

100μm

1-2μm 36μm

80μm inw

in

VAC

.0

Particle concentration

8

Results Collection efficiency

9

Results Collection efficiency

gen

dep

in

w

N

N

A

A.

Collection efficiency decreases when:

•Concentration Decreases•Flow rate increases•Particle size decrease

There is a variability between different runs of a same configurations due to the random particle generation 10

Results Effect of Brownian motion

mean collection efficiency for each configuration

Brownian particles have a lower collection efficiency

Random forces have an adverse effect on deposition, because they are applied at each time step in random directions

11

Results Identification of scouring mechanism

1. Particles deposit and form aggregates

2. Aggregates grow bigger by accumulating/catching free flowing particles and flocs

3. Free flowing cluster impacts deposited aggregate

4. Re-suspension

12

Results Effect of scouring on deposition

1. Particles deposit and form aggregates

2. Aggregates grow bigger by accumulating/catching free flowing particles and flocs

3. Free flowing cluster impacts deposited aggregate

4. Re-suspension

13

Results Rolling mechanism

Aggregates loosely attached to the wall roll along the surface until re-suspension or until a larger amount of their particles attach.

Identifiable by the tailed peak on the curve

14

Resultscollisions correlation with collection efficiency

15

Conclusion

• Quantify number of deposited particle

• Observe the effect of Brownian motion

• Identify scouring and impact mechanisms

• Identify of rolling mechanism

• Particle to particle collisions correlation with collection

efficiency

What CFD-DEM help us achieve

16

colloids deposition in membrane filtration spacers

17

What are membrane spacers

They separate membrane sheets in water filtration system in order for the feed to flow through it

18

Spacer modeling

NALTEX-51 commercial range of spacers as this has already been subject to published studies, both experimental and computational and

has therefore a well defined geometry, with specifications readily available

19

Simulations

20

Conclusion

• Initial deposition pattern appear in region of low shear stresses

• preferred deposition patterns that depends on spacer orientation is also predicted

• particle accumulation around Naltex51-1 filament junction

• No agglomeration observed at particle concentration considered

What CFD-DEM help us achieve

21