PS

-P-9

9-05

0

BACKGROUND• Development of high flux (> 4 kg/m2hr) and high

selective (> 200, 5% H2O-nBuOH) inorganic

pervaporation membranes and modules.• Experiments: existence of strong mass and heat

transfer resistances decrease the pervaporationefficiency.

OBJECTIVEIncrease the separation efficiency by improving theflow characteristics and design of the (lab scale)module.

APPROACHOptimisation of the lab scale module by:• Chemical engineering calculations (Poster part I).• Computational Fluid Dynamic calculations

(This poster).1. understand physical transport phenomena in the

module,2. calculate pervaporation efficiency of proposed

module designs.

INORGANIC MEMBRANEPERVAPORATION MODULE DESIGNOptimisation by Computational Fluid Dynamics

CFD MODEL• 2D and 3D modelling of feed side (concentration,

temperature, flow, membrane length, and annularspace).

• Low Reynolds turbulence modelling (no wallfunction necessary).

• Heat and mass transfer.• Membrane included as boundary condition on feed

side: flux is function of concentration andtemperature on membrane surface.

• Transversal flow near inlet and outlet stronglyincreases local PV flux.

CONCLUSIONS• Lab scale module: strong influence of temperature

and concentration polarisation.• Bench scale module: need for e.g. baffles to

enhance transversal flow/turbulence.• Detailed physical insight in heat and mass transport

leads to better design of bench and pilot scalemodules.

• Local flux strongly depends on local flowconditions.

2D CFD RESULTSFlux decrease Concentration Temperature Combined

due to polarisation polarisation effectat Re = 2400 27% 8% 31%at Re = 4700 n.a. n.a. 21%at Re = 9400 n.a. n.a. 14%

Figure 1: Schematic view lab scale module

Figure 4:PV flux on top, side and bottom of membrane surface.

Figure 5: 2D/3D PV flux as function of axial distance

• Increasing flux with feed flow rate, however stillpolarisation effects at high flow rates.

3D CFD RESULTS

Figure 3: PV flux as function of axial distance

Figure 2: 3D CFD mesh,approx. 300,000computational cells

Y.C. van Delft

H.M. van Veen

P.P.A.C. Pex

Netherlands Energy

Research Foundation, ECN,

P.O. Box 1, 1755 ZG Petten,

The Netherlands

Tel. +31 224 564640,

E-mail: pex@ecn.nl

Work performed by:

J.A. Lycklama à Nijeholt

C.J.J. Beemsterboer

NRG,

P.O. Box 25, 1755 ZG Petten,

The Netherlands

E-mail: lycklama@nrg-nl.com

S. Sommer

B. Klinkhammer

T. Melin

Institut für Verfahrenstechnik,

RWTH Aachen,

Turmstrasse 46, Aachen,

Germany

This work is partially financed bythe European Union and byNovem, the Dutch organisationfor Energy and Environment.