18 TH INTERNATIONAL CONFERENCE ON COMPOSITE MATERIALS BUBBLE SHAPE AND TRANSPORT DURING LCM PROCESSES: EXPERIMENTAL MODELING IN A T-JUNCTION TUBE M. A. Ben Abdelwahed 1 *, Y. Wielhorski 1 , L. Bizet 1 , J. Bréard 1 1 Laboratoire Ondes et Milieux Complexes (LOMC), FRE 3102 CNRS, 53 rue de Prony, Université du Havre – BP 540, 76058 Le Havre cedex France *Corresponding author (mohamed.ben-abdelwahed@univ-lehavre.fr) Keywords : Bubble, LCM process, capillary number, void formation, void transport. In the present work, we attempt by an experimental 1 Introduction simple modeling to study bubble formation and Long fiber composite materials can be elaborated by transport mechanism for low Reynolds numbers ( Re Liquid Composite Molding (LCM), a family of << 1). processes where fibrous preforms are injected by a low viscosity resin. During this process, we have to 2 Experimental procedures pay particular attention to the void formation (Fig. 1a) inside the preform because it could modify the In order to perform an advance-delay effect involved material final characteristics. Indeed, a preform in bubble production phenomenon in LCM presents two different porosity scales: between yarns processes, we carry out an experiment which called macropores and inside yarns, namely consists of converging two flows perpendicularly micropores. Due to this double porosity, bubbles of with different flow rates. Liquid is injected in the T- different shapes and lengths can be created inside shaped junction by two syringe compressors in composite materials [1]. It is observed setting two different flow rates: Q 1 , corresponding to experimentally that at low Capillary number Ca , the the cross flow and Q 2 , related to the gas injection capillary flow is favored inside the yarns, i.e. (Fig. 2a). The break-up mechanism during the between fibers, and leading to create inter tow voids bubble formation is represented on Fig. 2b. The two or macrovoids. However, at higher Ca , the flow flows merging at the junction create regular spaced occurs between yarns is faster than the one between bubbles. Bubble length L and distance between two successive bubbles are measured (Fig. 2c). fibers at such velocities, the Stokes flow is more important and the associated regime included tow Glass capillaries are used to allow bubble voids or microvoids [2]. Numerous numerical visualization by a monochromic Dalsa M1024 approaches [3, 4] attempted to simulate these voids camera. Accuracy of length measurement is about creation by coupling Laplace forces and Stokes law. one pixel on recorded images. Thus precision of However, the experimental visualization of the void obtained values is around 40µm . Images are formation and transport through the flow inside a analyzed with Aphelion 3.2 software. Three fibrous preform remains delicate. Consequently, we different liquids are used in our experiments: two have chosen to investigate the bubble generation and silicone oils, Rhordorsil 47V100 and 47V1000 given motion by a modeling device as a cylindrical by Rhodia with viscosities η of respectively 0.1Pa.s capillary T-junction. It may represent for instance and 1.0Pa.s and a water-glycerol mixture in two convergent pores (Fig. 1b). proportion (15-85%) with a viscosity of 0.1Pa.s. The Many microfluidic flow-focusing devices are liquid surface tensions γ L were measured by a K100 developed in order to study bubble creation. One of SF Krüss tensiometer for both silicone oils and the these devices is a rectangular T-shaped junction, water-glycerol mixture. The values obtained are which is used to create and characterize drops and close to 21mN/m for the both silicone oils and bubbles by converging flows [5, 6]. Some studies 47mN/m for the mixture of water-glycerol. Two attempt to build a flow phase pattern diagram capillary tubes are used with two radii R c (0.5 and linking the liquid capillary number and the gas flow 1.0mm). This choice is governed by the capillary rate [7]. Bubble velocities in capillary tube are also length with is close to 1.5mm for the silicone oils investigated [8, 9, 10, 11]. and about 2mm for the mixture water-glycerol. 1
BUBBLE SHAPE AND TRANSPORT DURING LCM PROCESSES: EXPERIMENTAL MODELING IN A T-JUNCTION TUBE 3 Results and discussion Theoretical values are compared with experimental ones for both silicone oils and for R c = 1mm. The 3.1 Bubble length model is based on the determination of the liquid Fig. 3 shows the normalized bubble length as a volume entrapped between two successive bubbles function of flow rate ratio Q 2 / Q 1 . As the result, three and the gas (bubble) volume, respectively noted V l different regimes [7] are distinguished according to and V b . The elementary volume V λ corresponds to the bubble length and the range of flow rate ratio the whole volume delimited by two bubble centers values. Squeezing regime is defined for L/2R c >2.5 in noted by λ (Fig. 2c), i.e. the sum of V l and V b . For which long slug bubbles are obtained. In this regime, transition and squeezing regime, the following the interfacial force is much higher comparing to the equation is obtained: cross-flow shear force and the dynamics of break-up Q 1 K ' Q is dominated by the filling pressure. Then, a 2 1 1 (3) transition regime where short slug bubbles are 3 2 R Q 3 Q 2 R c 1 2 c created, was observed for relative bubble length Where K ’ [m 3 ] is linked to the bubble time growth. 1< L/2R c <2.5. Here, the break-up mechanism is Note that Eq. (3) is defined for Q 2 / Q 1 ≥1.5. dominated by a balance between both forces. Dripping regime, corresponding to the dispersed bubbles, is the third regime where relative bubble For dripping regime, the λ -model gives: length L/2R c <1. 3 B Q Q 2 K ' 2 1 A 1 (4) Two different laws are found: for the squeezing and 3 2 R 3 Q Q 2 R c 1 2 transition regimes (Eq. (1)) and for the dripping c regime (Eq. (2)): Eq. (4) is valid for the condition Q 2 / Q 1 ≤0.064 . L Q 2 To conclude, we can say that the -model is in a (1) 2 R Q good agreement with the experimental values (Fig. c 1 4). Main result is that the distance between two B Q L successive bubble centers reaches a minimal value in 2 A (2) 2 R Q the transition regime. c 1 with α, β, A and B are fitt ing coefficients. Eq. (1) is a linear model where α and β depend on 3.3 Bubble pattern the liquid proprieties. However, for both silicone oils This part attempts to show the influences of different the Garstecki’s model for square capillary parameters on the squeezing-to-dripping diagram microdevice [6] is very close to the experiment data. plotted with the cross flow capillary number Ca 1 and Eq. (2) is a power law relation as a function of flow the gaseous phase flow rate Q 2 as coordinates. Ca 1 is rate ratio. defined as η U 1 / γ L where U 1 is the main velocity for the cross flow Q 1 . This study deals with regime limits according to the variations of bubble length. 3.2 -model Regime limits was proposed by Fu and al. [7] but To investigate the bubble frequency inside the T- adapted to our cylindrical geometry. junction device, which is strikingly related to both The limit between transition and squeezing regimes flow rates, distance between two successive bubbles TS ( Ca 1 for L/2R c = 2.5) which is combined with Eq. noted λ is measured experimentally. This parameter (1) is given by: could be quite important to understand bubble 1 TS Ca Q (5) formation because it can determine the gas quantity 1 2 2 . 5 2 R L created as a function of time. In the following, a c By the same reasoning, Eq. (2) is arranged to theoretical model named λ -model is proposed to determine the limit between the dripping and determine the distance between two successive bubble centers by a geometrical approach. 2
Recommend
More recommend
