FOULING PHENOMENON IN CROSS-FLOW MICROFILTRATION PROCESSES 179Influence of the number of reentrairaient N rThis parameter fixes the sensitivity of a suspended particle concern<strong>in</strong>g entra<strong>in</strong>ment by theflow when it makes contact with a previously aggregated object. The simulations drawn <strong>in</strong>Fig.10 present deposit morphology variations for different values of the number ofreentra<strong>in</strong>ment, from branch<strong>in</strong>g out (N r small) to compact structures (N r high).Secondary parameters such as θ v, P <strong>and</strong> R d do not have a significant <strong>in</strong>fluence on themorphology of theoretical deposits.Fig. 10: Simulation for N r =0 (up) <strong>and</strong> 3 (down) with α=20°, θ i =1°, θ v =1°3.3 DiscussionThe ma<strong>in</strong> question is: do such statistical aggregates really represent what happens when afilter surface is clogged by solid particle deposition dur<strong>in</strong>g a cross-flow microfiltration processθ Of course it is very difficult to give an answer because of the difficulty to experimentallyobserve phenomena appear<strong>in</strong>g at a microscopic scale (the pore diameter is about 1µm <strong>and</strong>the particle diameter is supposed to be smaller). Experiments already performed by HOUI(1989) <strong>in</strong> a 2D micromodel for the filtration of dilute well-muds have shown surface depositswhose structures are qualitatively <strong>in</strong> good agreement with some of our statistical simulations.We hope that further <strong>in</strong>vestigations us<strong>in</strong>g the orig<strong>in</strong>al technique of nuclear magnetic resonancewill provide us with useful <strong>in</strong>formation on the time dependant of the thickness all along ahollow fiber.Fig. 11: Horizontal density variations with a=20°, θ i =1°, θ v =1°, N r =4The horizontal density variations, plotted <strong>in</strong> Fig. 11 for the aggregation depicted <strong>in</strong> Fig. 6,show that the statistical simulator ensures a uniform distribution of particles all over thestudied filter surface except at the downstream <strong>and</strong> upstream boundaries. Furthermore itcan be concluded that deposits formed dur<strong>in</strong>g a cross flow microfiltration process have auniform thickness over a filtration length cover<strong>in</strong>g a few pores.It is important to notice that statistical deposits are homogeneous. Bensimon (1983) <strong>and</strong>Meak<strong>in</strong>s (1985) have demonstrated that this can be proved by the follow<strong>in</strong>g relation, sett<strong>in</strong>gε equal to 1:
180 WATER TREATMENTIt has been well verified for each simulated aggregate. Capture mechanisms do not <strong>in</strong>duceheterogeneity.Furthermore, the macroscopic quantity Rms can be used to def<strong>in</strong>e a mean thickness whichtakes <strong>in</strong>to account all the roughnesses of the deposit seen at the microscopic scale.Unfortunatly, it is impossible to calculate the permeability because of the 2D approach, exceptif we admit the validity of the analogy between 2D <strong>and</strong> 3D deposits concern<strong>in</strong>g porosity. Thenthe Br<strong>in</strong>kman method can be used to calculate the permeability of an aggregate of sphericalsolid particles.Further 3D simulations us<strong>in</strong>g the same mov<strong>in</strong>g <strong>and</strong> stick<strong>in</strong>g rules will provide more realisticclusters <strong>in</strong> order to determ<strong>in</strong>e the necessary parameters for flow computation <strong>in</strong> hollow fibersat macroscopic scale.CONCLUSIONAt macroscopic scale, the hydrodynamic field <strong>in</strong> a hollow fiber was obta<strong>in</strong>ed <strong>and</strong> applied toprocess optimisation as for backwash<strong>in</strong>g <strong>in</strong> external sk<strong>in</strong> membranes.The statistical model describ<strong>in</strong>g the formation of a cake consist<strong>in</strong>g of spherical particles enablesus to characterize the porosity <strong>and</strong> the thickness of the cake dur<strong>in</strong>g its formation. To obta<strong>in</strong>the value of the overall permeability, we need to develop a 3D model.These characteristics of the deposit structure give more realistic boundary conditions at the<strong>in</strong>terface between fluid <strong>and</strong> membrane to compute flow <strong>in</strong>side hollow fibers.REFERENCESBELFORT G <strong>and</strong> NAGATA N, 1985, Fluid mechanics <strong>and</strong> crossflow filtration: some thoughts,Desal<strong>in</strong>ation, 53, 57–79BERMAN A.S, 1953, Lam<strong>in</strong>ar flow <strong>in</strong> channels with porous walls, J. Appl. Phys., 24, 1232–1235BRADY J.F, 1984, Flow development <strong>in</strong> a porous channel <strong>and</strong> tube, Phys. Fluids, 27, 1061–1067CHATTERJEE S.G <strong>and</strong> BELFORT G, 1986, Fluid flow <strong>in</strong> an idealized spiral wound membranemodule, J. Membrane Sci., 28, 191–208COX R.G <strong>and</strong> BRENNER H, 1968, The lateral migration of solid particles <strong>in</strong> Poiseuille flow—I. Theory, Chem. Eng. Sc;, 23, 147–173DE GENNES P.G, 1981, Dynamics of concentrated dispersions: a list of problems, Phys.Chem. Hydro., 2, 1, 31–44GALLOWIN L.S et al, 1974, Investigation of lam<strong>in</strong>ar flow <strong>in</strong> a porous pipe with variable wallsuction, AIAA J., 12, 1585–1589GILL W.N <strong>and</strong> BANSAL B, 1973, Hollow fiber reverse osmosis systems-Analysis <strong>and</strong> design,AIChe J., 19, 823–831GUPTA B.K <strong>and</strong> LEVY E.K, 1976, Symmetrical lam<strong>in</strong>ar channel flow with wall suction, J.Fluids Eng., September, 469–474HOUI D <strong>and</strong> RITTER A, 1989, Filtration of muds by a porous medium, 5th I.F.P. ResearchConf. on Expl./Prod., to be editedOVERBEEK J.T.G, 1984, Interparticle forces <strong>in</strong> colloid science, Powder Tech., 37, 195–208QUAILE J.P <strong>and</strong> LEVY E.K, 1973, Pressure variations <strong>in</strong> an <strong>in</strong>compressible lam<strong>in</strong>ar tube flowwith uniform suction, AIAA Paper N72–257SCHMITZ P, 1989, Lam<strong>in</strong>ar flow <strong>in</strong> a dead ended porous tube with wall suction: applicationto hollow fibers crossflow filtration, Int. J. Heat Mass Transfer, submittedSCHMITZ P et al, 1989, Fundamental mechanisms of particle deposition on a porous wall:hydrodynamical aspects, 1st Europ. Conf. on the Math, of Oil Recovery, July, to be editedSPIELMAN L.A <strong>and</strong> GOREN S.L, 1970, Capture of small particles by London forces from lowspeed liquid flows, Env. Sci. Tech., 4, 2, 135–140YUAN S.W <strong>and</strong> FINKELSTEIN A.B, 1956, Lam<strong>in</strong>ar flow with <strong>in</strong>jection <strong>and</strong> suction through aporous wall, Trans ASME, 78, 719–724BENSIMON D, DOMANY E, AHARONY A, 1983, Crossover of fractal dimension <strong>in</strong> diffusion—limited aggregates, Phys. Rev. Letters, 51, 15, 1394MEAKIN P, 1985, Accretion processes with l<strong>in</strong>ear particle trajectories, J. of Colloid <strong>and</strong> InterfaceSci., 105, 1. 240