Experimental Thermal and Fluid Science 31 (2006) 97–110 www.elsevier.com/locate/etfs Phenomena of droplet–surface interactions Š. Šikalo *, E.N. Ganić Faculty of Mechanical Engineering, University of Sarajevo, Vilsonovo setaliste 9, Sarajevo, Bosnia and Herzegovina Received 11 September 2005; received in revised form 9 December 2005; accepted 8 March 2006 Abstract The fluid flow associated with impinging droplets is rather complicated and not understood in detail. Depending on the circumstances, different characteristic features are observed. Various phenomena can appear when a droplet impacts a surface. The outcome of an impact depends on the droplet properties and of the impacted surface, which may be, for example, a dry or liquid surface. Following a dimensional analysis, the number of independent parameters can be reduced to a set of dimensionless groups that governs the process considered. An important dimensionless parameter for impact processes is the impact Weber number. However, this number alone is not sufficient to classify different types of droplet impact. Different outcomes of droplet–surface interaction will be discussed in the present paper. Our consideration is limited to the interaction of single droplets with different horizontal and inclined surfaces. 2006 Elsevier Inc. All rights reserved. Keywords: Droplet impact; Droplet rebound; Partial rebound; Wettability 1. Introduction In most practical processes, the fluid-flow phenomena during droplet impact are followed or accompanied by heat transfer or phase change, liquid evaporation in spray cooling, and solidification during spray forming and spray coating. Detailed knowledge of the transport phenomena involved in practical operations that employ droplet impingement on solid materials is critical for the overall process development and further advancement. A comparison of the time scales of the fluid dynamics of droplet spreading to the time scales of the associated heat transfer processes shows that the former scales are markedly smaller [1]. Hence, it appears that the droplet spreading occurs largely, first, and the heat transfer follows. It is then relevant to investigate initially the fluid dynamics of a droplet impinging on a surface to provide a better understanding of the transport processes taking place in engineering applications. * Corresponding author. Tel.: +387 33 656562; fax: +387 33 653055. E-mail addresses: sikalo@mef.unsa.ba (Š. Šikalo), ganic@mef.unsa.ba (E.N. Ganić). 0894-1777/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermflusci.2006.03.028 In industry, the impact of droplets on walls and films occurs in a variety of forms; however, the desired effect may vary according to application. For instance coating and painting applications strive for a total deposition (spreading) of droplets, whereas in fuel injection or respirations, the impact is used to generate smaller droplets through splashing, albeit sometimes unintentionally. Although most industrial situations involve a polydispersed ensemble of droplets or a spray, a basic understanding of the impact process is better obtained using simple droplets of monodispersed size. This allows systematic variation of the influencing parameters and permits submodels to be formulated for the numerical treatment of the problem in computational fluid dynamics codes. Recently there has been significant research directed towards this phenomenon, empirically, numerically and analytically, leading to a rather broad set of conditions for which the outcome of impact can be predicted. A noteworthy exception is the impact of droplets on highly inclined surfaces. This condition occurs in heat exchangers and on airplane wings (icing). Numerous reviews of the subject are available, dealing with a wide span of impact conditions and surface properties [2–4]. Correspondingly, the desire and need to predict 98 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 Nomenclature Ca d D Oh Ra Re t t* u We x y impact capillary number, Ca = We/Re spreading diameter, m droplet diameter before impact, m Ohnesorge number, Oh = We1/2/Re average surface roughness, lm impact Reynolds number, Re = quD/l time after impact, s non-dimensional time after impact, t* = tu/D impact velocity, m/s impact Weber number, We = qu2D/r spread factor in forward/backward direction apex height, m h l q r static contact angle dynamic viscosity of the liquid, Pa s density of the liquid, kg/m3 surface tension of the liquid, N/m Subscripts h horizontal n normal v vertical a advancing r receding s secondary t tangential Symbols a droplet impact angle d liquid film thickness these phenomena are also large, and both theoretical and numerical approaches have been taken. The most recent theoretical work [5] is valid up to moderate impact Weber numbers, We = qDu2/r with q and r denoting the density and surface tension of the liquid and D and u the diameter and impact velocity magnitude of the droplet. This theory has also been extended to the case of two-drop impacts [6], reflecting applications associated with predicting the results of spray impingement. More numerous are the numerical predictions of drop impact, typically using a finite volume and volume-of-fluid (VOF) approach [7–12]. 2. Normal droplet impact onto dry and wetted surfaces 2.1. Experimental technique To investigate the droplet–surface interaction a series of experiments were conducted with individual droplets impacting onto solid, dry and wetted surfaces. The experimental method has been previously described in [13]. A high resolution charge-coupled device (CCD) camera (Sensicam PCO, 1280 · 1024 pixels) equipped with a long distance microscope is used to observe the spreading droplet in detail. The magnification was manipulated so that the image could accommodate the maximum spread of the droplet. From the side-view images, the spread factor and apex height are measured. In the present study experimental data on droplet spreading on the surfaces are presented for a range of Weber numbers (We = 50–1063) based on droplet velocity and the liquid density. Three liquids were selected for their characteristic liquid properties, distilled water, isopropanol and glycerin (85 vol.% glycerin/water solution), to study the effect of liquid surface tension and viscosity. Two target surfaces of glass, one smooth (with roughness of Ra = 0.003 lm amplitude) and one rough (Ra = 3.6 lm), and one surface of smooth wax (Ra = 0.3 lm) were used. The droplet’s initial diameter, the physical properties of the liquids and the wettability of the surfaces (advancing ha and receding hr contact angles) are shown in Table 1. The surface tension, density and viscosity values were collected from standard tables of physico-chemical liquid properties. The thickness of the liquid film was measured (with accuracy of ±5 lm) directly by comparing an image of the film surface to a reference image of the dry plate. The film thickness was in the range of 40–800 lm, depending on the inclination and liquid used in the experiment. The deposition–rebound limit of a single droplet was studied for dry and wetted surfaces. The inclination angle of the impact plate was measured with an accuracy of ±0.1 (95% confidence). The impact velocity was derived from two consecutive images immediately before impact with an accuracy of 0.02 m/s (95% confidence) at Weber numbers of 50 and 0.06 m/s at Weber Table 1 Properties of the liquid droplets and wettability of the surfaces Liquid Water Isopropanol Glycerin D (mm) 2.7 3.3 2.45 r (N/m) 0.073 0.021 0.063 l (mPa s) 1.0 2.4 116 q (kg/m3) 996 786 1220 ha–hr () Smooth glass Rough glass Smooth wax 10–6 0 17–13 78–16 0 62–12 105–95 0 97–90 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 numbers of 1063. Considering also the ±6 lm uncertainty in the droplet diameter determination, the uncertainty in the calculated Weber number ranged from ±2 at Weber numbers of 50 and ±40 at Weber numbers of 1063. In this case the results of a droplet impact can be spreading (deposition), splashing or recoiling. The term deposition indicates impact without the production of any secondary droplets. In contrast to this, the term splash is characterized by the formation of secondary droplets. The transition from deposition to splash happens with increasingly energetic impact onto smooth surfaces or with the influence of surface roughness. A qualitative comparison of the impact of water and glycerin droplets on the smooth glass (with average surface roughness Ra = 0.003 lm) is illustrated in Fig. 1. On impact, onto a dry smooth surface, a thin liquid film (lamella) begins to jet radially outward over the solid surface. The spreading velocity of a glycerin droplet is lower. The lamella thickness is much higher and no visible perturbations appear on the rim during the spread of a glycerin droplet. The spread factor (d/D) and apex height (y/D) of droplets of different liquids are compared, as a function of nondimensional time after impact (tu/D), to examine the effect of droplet viscosity. Experimental data, characterized by nearly the same Weber numbers (about 90) and Reynolds numbers of 4200, and 36.3 for water (D = 2.7 mm) and 4 d/D 3 Water, We=90 Glycerin, We=93 2 1 0 0.01 0.1 (a) 1 10 1 10 tu/D 1 0.8 0.6 Water, We=90 Glycerin, We=93 y/D 2.2. Spread of a droplet 99 0.4 0.2 0 0.01 0.1 (b) tu/D Fig. 2. Effect of liquid viscosity (a) spread factor and (b) apex height (water Re = 4200 and glycerin droplet Re = 36.3) on smooth glass. glycerin (D = 2.45 mm) droplets, respectively, on smooth glass, are shown in Fig. 2. An increase of viscosity decreases the spread factor, as Fig. 2a shows. The glycerin droplet recoils earlier than water droplet on smooth glass, as Fig. 2b illustrates. After tu/D = 0.75, the apex height of glycerin droplet decreases slowly and the minimal apex height is larger for the glycerin droplet than that for the water droplet (Fig. 2b). 2.3. Splash of a droplet Fig. 1. Impact of a water and a glycerin droplet onto glass for We = 391: (a and b) spreading phase, (c and d) receding phase (t is time from impact). As the literature review [14,15] demonstrate, the related phenomena of fingering and splashing during the impact of a droplet onto a dry surface are not well understood. Although correlations exist to predict the onset of splashing, these have not proven universal. The mechanism that indicates the perturbation at the leading edge of the expanding lamella has not yet been fully understood. The onset of crown formation, its time evolution and the nature of the lamella instability (see below) that leads to splash have received scant attention. The influence of surface roughness and wettability on crown instability has not been satisfactory explained. Depending of a number of factors, the leading edge of the lamella may detach. The lamella may become unstable soon after it appears, resulting in the emergence of very regular azimuthal undulations. At the critical Weber number, that defines the boundary between the splashing and Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 Fig. 3. The splash/deposition limit: (a) adhesion, (b) break up and (c) droplet ejection. (Isopropanol droplet/smooth glass, D = 3.3, We = 287, Dt is the time between two exposures.) non-splashing regimes, on the smooth surface the lamella may fall down. Viscous drag due to the underlying solid surface plays a role in damping the undulations. The fluid will continue to spread while remaining in contact with the solid surface. Some droplets may be ejected from the lamella just before it falls down, as Fig. 3c shows. At the larger We, the expanding lamella lifts up off the surface and forms a crown, similar to the impact of a droplet onto a liquid surface. If the lamella grows far enough beyond the contact edge, its instability leads to the pinching off of secondary droplets, as seen in Fig. 4 at t = 0.959 ms and final disintegration at t = 1.607 ms. From this time the deposited part of droplet spreads, slowly approaching the equilibrium state. When the same droplet impacts the rough glass, the expanding lamella tends to lift off of the surface, and fluctuations at the microscopic scale translate into large fluctuations in the region of the lamella far from the contact line and it disintegrates. This outcome is called the prompt splash [16]. The detachment of the lamella and formation of the crown at the impact onto a dry, smooth surface depend on the droplet surface tension. The surface tension is embedded in the Weber number. The detachment of the lamella on smooth surface was observed with an isopropanol droplet at a Weber number larger than about 287 (Fig. 3). At Weber numbers about this value (We from 280 to 300) the detached lamella of an isopropanol droplet simply falls and adheres to the surface. The critical Weber Fig. 4. Splash of an isopropanol droplet on dry, smooth glass (We = 391). number is the one at the deposition/splash threshold. At larger values of the Weber number the difference between the contact diameter and the detached lamella diameter increases up to the time of break-up (splash) of the lamella, as Fig. 4 indicates. The lamella diameter is taken here as diameter of the expanding liquid film before the splash, as Fig. 5a shows. The splash occurs due to instability of the lamella, after it reaches a certain diameter (Fig. 5b). Fig. 6 shows the impact sequence of a 3.3-mm isopropanol droplet (We = 544) on smooth glass. In the first phase of the lamella evolution, it seems smooth and stable, and no secondary droplet detaches. Then liquid accumulates in the periphery of lamella and it become unstable (t = 0.472 ms). The instabil- d lam. dcont. (a) 4 break up We=544 3 Contact d/D 100 Lamela 2 1 0 0.01 (b) 0.1 tu/D 1 Fig. 5. Spread factor of the lamella and contact line for an isopropanol droplet impact on the smooth glass: (a) typical view (t = 0.254 ms) and (b) its time evolution (We = 544). Fig. 6. Splash of an isopropanol droplet on dry smooth glass, We = 544. Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 ity increases large enough for droplets to separate from the periphery of the lamella (t = 1.010 ms). The lamella disintegrates gradually into secondary droplets as the sequences (t = 1.235 ms and t = 1.443 ms) illustrate. The effect of surface roughness on deposition/splash limit for an isopropanol droplet was not observed in our study. No splash was observed for a water droplet on a smooth surface for Weber numbers up to 1080, the maximum value considered in our experiment. Surface roughness affects the deposition/splash limit for a water droplet. A water droplet (D = 2.7 mm) splashes at Weber number large than 390 on the rough glass. After the lamella is ejected from the base of the droplet, a thin liquid film has not full contact with the rough surface. The lamella contacts the roughness peaks and the air is entrapped in the roughness valleys. The lamella/solid contact is random, the adhesion forces are weak and, as the result of the early lamella instability, it breaks up just after it appears (prompt splash), as Fig. 7 shows. The splashing mass due to formation of secondary droplets increases with increase of the impact velocity. A comparison between the splash of an isopropanol droplet on the smooth (Ra = 0.003 lm) and rough glass (Ra = 3.6 lm) in Figs. 6 and 7, respectively, shows that the roughness does not increase greatly the amount of secondary droplets. At higher Weber number impact of an isopropanol droplet onto a smooth surface, the lamella forms a corona-like shape. The corona increases with increase of impact velocity up to the instant of secondary droplet ejection and the final breakup, as Fig. 8 shows. This outcome is termed corona splash [16]. A classical corona splash at a higher-impact Weber number is shown in Fig. 9. The lamella appeared shortly (t = 0.070 ms) after the droplet contacted the surface. After the lamella expands a certain distance, small droplets eject from its periphery and the very thin lamella becomes unstable (t = 0.610 ms), it breaks up into filaments (frames from t = 0.878 ms to t = 1.160 ms) and the filaments disintegrate into fine droplets (spray) (t = 1.661 ms). The spray disappears (t = 2.433 ms) and the remaining part of the droplet continues to spread on the surface, as Fig. 9 shows. Fig. 7. Prompt splash of an isopropanol droplet, We = 544, on dry rough glass (Ra = 3.6 lm). 101 Fig. 8. Time evolution of the corona for an isopropanol droplet impact onto dry smooth glass: (a) We = 391, (b) We = 544 and (c) We = 786 (Dt = 0.160 ms for all). Fig. 9. Corona splash of an isopropanol droplet on dry, smooth glass (We = 1063). A selection of images of the corona evolution (growth and subsidence) at the impact of an isopropanol droplet (D = 3.3 mm, We = 1063) on a shallow liquid layer (film thickness d = 0.8 mm) is shown in Fig. 10. The liquid lamella is formed shortly after the base of the droplet reaches the liquid layer. It then propagates radially outward, forming a thin cylindrical liquid film (corona). Initially the upper edge of the corona is bend down (from t = 0.182 ms to t = 1.212 ms). Small droplets begin to eject from its rim and the bended part disintegrates (t = 1.212 ms). Later when the expansion of the corona slows down and the bent edge is shattered, the upper corona edge become unstable and then small jets rise up and elongate at its periphery (t = 3.524 ms). Liquid accumulates on the periphery of the corona and jets become large (t = 13.193 ms). After the corona base reaches its maximum it starts to subside. The wall of the corona bends inward (t = 14.284 ms) and the diameter of upper edge of the corona reduces, while the diameter of the base remains fixed for a short time. Then the shape of the corona changes from cylindrical to conical (t = 26.216 ms). The peripheral 102 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 Fig. 11. Effect of film thickness on splash of an isopropanol droplet (We = 1063): (a) d = 0.1 mm, (b) d = 0.8 mm. Fig. 12. Corona without splash, of an isopropanol droplet, We = 287, d = 0.14 mm. Fig. 10. Corona evolution at the impact of an isopropanol droplet on liquid film (We = 1063, film thickness d = 0.8 mm). (a) Growth of the corona, (b) subside (collapse) of the corona. jets increase and droplets separate from their tips. The jets leave the corona edge forming large droplets (t = 33.895 ms). Droplets detached from the jets of the corona fall down on the liquid. This produces a small number (between 10 and 15) of larger droplets (about 1.8 mm in diameter). The corona subsides into the liquid layer (t = 43.737 ms) and the large secondary droplets deposit. Subsequently, as results of liquid layer retraction, a central jet forms (t = 61.282 ms) from the center of the cavity in a liquid layer that disintegrates in a few droplets. Capillary waves can be seen propagating down the corona (t = 26.216 and 33.895 ms). Droplet impact onto dry smooth surface forms a very thin lamella, which collapses in spray. The disintegration process of the lamella on a dry smooth surface is fast; it finished at t = 1.160 ms (Fig. 9). A droplet impacting onto a liquid layer forms a thick liquid corona with larger droplets pinching off from the rim jets while the corona settles down. The impacting droplet kinetic energy is converted to corona potential energy and an increase of surface tension energy. The corona consists of both liquid of droplets and target liquid. Thus, the disintegration process of the corona is longer and the corona disappears later (t = 43.737 ms in Fig. 10) than that at impact onto a dry surface (t = 1.160 ms in Fig. 9). Retention of liquid in the cavity in the target layer remaining after the corona collapse will result in a central liquid jet (t = 61.282 ms). Small variations of the liquid film thickness d, influence the splashing behavior, the corona morphology, the secondary droplet size and the time scale of the splashing process. The height of the corona is higher at the impact onto a thicker liquid film owing to the formation of a cavity immediately after impact, the displaced material of which is incorporated into the corona wall (Fig. 11). A splash includes a liquid film rising from the periphery of the crater. The splash on wetted surfaces occurs at a smaller impact velocity than on dry surfaces. Furthermore, in contrast to impact on dry surfaces, the corona formation does not always lead to splash, as Fig. 12 shows, for We = 287. 3. Impact of a droplet onto inclined surfaces The geometry of the droplet impact is shown in Fig. 13a, where the droplet velocity (u) is composed of a normal (un) and tangential (ut) component, determined by the impact angle (a). The side view of a droplet on an inclined surface is shown in Fig. 13b. For applications related to heat transfer in annular dispersed flows, typical droplet impact angles are low (<5) and normal Weber numbers ðWen ¼ qDu2n =rÞ based on Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 Fig. 13. The geometry of droplet impact (a) and side view of a droplet on an inclined surface (b). the wall-normal velocity un and the droplet diameter D are less than 5. For instance [17] studied the rate of deposition of 50 lm water droplets injected centrally into a downward flowing pipe flow at a gas Reynolds number of Re = 30,600. The drop impact angle was found to lie in the range 0.5–3.5 with a normal Weber number in the range 0.007 < Wen < 0.34. Šikalo et al. [18] worked with an upward flowing configuration at Re = 40,000–95,000 and found impact angles of 0.5–1.5. The drop mean diameter ranged from 7 lm to 30 lm, yielding a normal Weber number range of 0.025 < Wen < 1.80, similar to [17]. Besides the application in heat exchangers, the oblique drop impact in this range of Weber number is of considerable interest in agricultural spraying, where low impact angles on leaves can lead to diminished deposition on the surface. Fuel injection sprays impacting obliquely on combustion walls or droplets impacting onto airplane wings, considered for icing studies [19], exhibit considerably higher Weber number and will not be considered in the present paper. Oblique droplet impact at low Weber numbers and at low impact angles was investigated in [20–26]. 3.1. Phenomenological observations 3.1.1. Dry surface When a droplet impacts onto an inclined surface, the shape of the droplet distorts and it spreads asymmetrically relative to the point of impact, as shown in Fig. 13b. Elongation and back-to-front asymmetry increases with time. The front edge of the droplet spreads forward, while the back edge spreads backwards or slips forward. The difference in the spread velocity of the lamellas in the forward and back directions increases with a decrease of impact angle. They are equal for the normal impact. The maximum spread of the rear lamella is reached when the radial velocity in that direction becomes close to zero. When this occurs, the back edge remains fixed for a glass surface (high wettability), while the back edge slides down the wax surface (low wettability). The footprint given by 103 contact line upon impact is usually neither circular nor parallel-sided, but lies somewhere in-between. Note that both the xfront and xback values have been defined positive from the point of impact. The first phase of impact involves the initial deformation of the droplet similar to the case of normal impact. The subsequent phase may have widely varying outcomes, for example: spread (Fig. 14a), spread and slide (Fig. 14b), slip and spread (Fig. 14c), splash in all directions (Fig. 15a) or only in the forward direction (Fig. 15b), rebound, partial rebound (Fig. 16), depending on the impact angle, liquid and surface properties as well as the Reynolds and Weber numbers. At the impact onto an inclined smooth surface the lamella extends in forward (front) direction, its rim become unstable and it disintegrates in droplets (splash), Fig. 15b. The spread factor of the lamella is taken here as the distance of the expanding film, before the splash, from the point of impact, as Fig. 17a shows. The contact spread factor in forward and backward direction is measured, as a Fig. 14. Deposition of a water droplet (We = 391; a = 5; Wen = 3.0) with: (a) spread on smooth glass, (b) spread and slide on wax (t = 0, 1, 3 ms), and (c) and slip and spread on smooth glass (t = 0, 1, 2, 3 ms). Fig. 15. Splash (a) in all direction (water droplet on dry rough glass, We = 391, a = 45, Wen = 196) and (b) in the forward direction (isopropanol droplet on smooth glass, We = 544, a = 45, Wen = 272). 104 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 x P bac im oin k pa t of ct Fig. 16. Partial rebound of a glycerin droplet on dry smooth glass: (a) ti = 0 (reference time of the image) and (b) ti = 7.67 ms (We = 93, a = 10, Dt = 2 ms). la m el la x x t ac nt o c meters, including film thickness. It is interesting to note that a droplet rebounds from a liquid film at larger angles than from the dry surface under the same conditions. The deformation of a droplet increases with an increase of the impact angle, and the contact time of the droplet with the surface decreases. For example, on dry glass at 8 the contact time of the glycerin droplet is about 14.5 ms. On a wetted surface at 30 (limit rebound/deposition) the contact time is about 10 ms. The contact time of an isopropanol droplet at 8 (D = 3.3 mm, We = 179, dry surface) is 29.68 ms, and the contact time of the isopropanol droplet at the impact of 15 (wetted surface) is about 22.99 ms. (The contact time is measured as the time between first contact of the droplet with the surface and the final lift-off of the droplet.) At larger impact angles the deformation is high, and the droplet elongation during its slipping along the target surface is larger. The elongation of a water droplet and an isopropanol droplet parallel to the surface in comparison with a glycerin droplet is also larger, due to lower viscosity. Higher deformations of an isopropanol droplet, due to its low surface tension and low viscosity, are observed at higher impact angles, so its shape differs from prolate or oblate one as it lifts off, as Fig. 18 shows. Fig. 19 illustrates possible outcomes of a water droplet impact onto a wall liquid film near a critical impact angle; complete coalescence (deposition) or partial coalescence. These outcomes are also observed for the other investigated liquids. At the coalescence of a droplet and the liquid film in an earlier recoiling phase, the droplet deposits (Fig. 19a). At the coalescence in the later recoiling phase, just before lifts off, the droplet disjoins and partially rebounds (Fig. 19b). (a) 4 We = 544, = 45º break up 3 x/D Front, contact 2 Back, contact Front, lamella 1 0 0.01 (b) 0.1 1 10 tu/D Fig. 17. Spread factors of the lamella and the contact line (a) typical view and (b) time evolution (isopropanol droplet, We = 544 and a = 45). distance from the point of impact to the contact line in the two directions, Fig. 17b. 3.1.2. Wetted surface In the case of the target surface covered with a liquid film, the outcomes of the impact may be similar to that for the dry surface: deposition (coalescence), rebound, partial deposition or splash, depending on the impact para- Fig. 18. Rebound of an isopropanol droplet from a wetted wall (We = 179, a = 15, Wen = 12, d 40 lm). Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 105 5 x=d/2, 90º x/D 4 We = 50, smooth glass Front, 45º 3 Front, 10º 2 Back, 45º Back, 10º 1 0 The images in Fig. 19 were taken for exactly the same impact conditions. The rather larger differences – rebound, partial rebound, deposition – indicate how sensitive the outcome is at this critical impact angle to non-controllable small perturbations. At this condition the repeatability is not nearly as high as at all other conditions. The deformation is affected by the ratio between normal inertial forces and surface forces. Therefore, the normal Weber number of the droplet appears as a parameter of importance to predict the degree of deformation and the droplet rebound limit. 3.2. Quantitative results 3.2.1. Droplet spreading on inclined surface The spread factor (x/D) in both forward and backward directions from the point of impact and the apex height (y/D) of a droplet (see Fig. 13b) as a function of nondimensional time after impact, t* = tu/D will be discussed. Fig. 20a and b show the effect of the impact angle on the spread factor of a 2.7 mm water droplet; the impact velocity was 1.17 m/s with an associated droplet Weber number of We = 50 and a Reynolds number of Re = 3260. Note that positive slope of xback with time (t*) refer to a film expanding away from the point of impact, whereas negative slope of xback indicates a slipping down the surface. Spread factors of a water droplet on the smooth glass (static contact angle h = 6–10) and on the smooth wax (h = 95–105) are shown in Fig. 20. If the spread factors shown in Fig. 20a and b are directly compared with one another, little influence of the surface is detected up to a dimensionless time of about t* = 0.7. This time decreases with a decrease of the impact angle. For example, it is about t* = 0.36 for a = 10. In the following stage of spreading, the effect of the surface is more important. For the wax surface with high static contact angles, the droplet begins to slide once the maximum spread factor in the back direction has been reached, as Fig. 20b shows (see Fig. 14b). No sliding is observed for the glass surface, as Fig. 20a shows (see also Fig. 14a). The lower impact -1 0.01 0.1 (a) 1 10 1 10 t* 4 2 x/D Fig. 19. Impact of a water droplet onto wetted surface (liquid film, d 0.1 mm): (a) deposition (coalesce in an earlier recoiling phase, ti = 0.99 ms) and (b) partial rebound (coalesce in a later recoiling phase, ti = 8.354 ms) (We = 39, a = 22, Wen = 5.5, t = 0, 5, 9 ms, time of the exposure). We = 50, wax x=d/2, 90º Front, 45º Front, 10º Back, 45º Back, 10º 0 (b) -2 0.01 0.1 t* Fig. 20. Spread factors of a water droplet on (a) smooth glass and (b) wax (a = 10–90, We = 50). angle leads to a larger droplet elongation (distance between leading and trailing edges). For the Weber number of 50 and the impact angle of 45 the forward spread factor becomes larger then for 90 at the dimensionless time about t* = 0.159, while at the Weber number of 391 this occurred at t* = 0.286. At the impact angle of 10 and the Weber number of 50 a water droplet slips on smooth glass at the first instant of the impact and the data of back spread factor are negative and exhibits larger scatter, as shown in Fig. 20a. No slipping of a water droplet was observed on smooth glass at larger Weber numbers. At the impact onto the wax the back edge of the contact line begins to slide at about t* = 0.80 (x/D = 0.348) for the Weber number of 50, and the slide begins at about t* = 1.623 (x/D = 0.618) for the Weber number of 391. The lower Weber number leads to a faster sliding of the back edge of the contact line. The normal impact velocity (u Æ sin a) scales the apex height well, as Fig. 21 shows. The partial or complete rebound of the droplet away from the surface was not observed for the inclined wax surface. The sliding of the water droplet on an inclined surface inhibits recoil of the droplet, as is observed for normal impacts. However, a water droplet on inclined wax surface does not rebound in the way observed for normal impacts. The sequences of the impact process of a water droplet on the smooth glass, wax and rough glass at 45 are shown 106 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 1 We = 50, smooth glass 0.8 90º 45º 20º 10º y/D 0.6 0.4 0.2 0 0.01 0.1 1 10 t* sinα Fig. 21. Effect of impact angle on the apex height of a water droplet in terms of t* Æ sin a (a = 10–90, We = 50). in Figs. 22–24, respectively. In the early phase of the impact (t up about 0.7 ms), the water droplet spreads in all direction almost symmetrically. In the later phase of spread the Fig. 22. A water droplet spreading on smooth glass (We = 391, a = 45). Fig. 23. Sequences of spreading of a water droplet on smooth wax (We = 391, a = 45). Fig. 24. Sequences of impact of a water droplet on rough glass (We = 391, a = 45). droplet elongates. The main part of water moves down, and in the back remains a trace of water (liquid film). The liquid film breaks up under capillary forces into two or more parts, and the liquid tends to reach an equilibrium state. The effect of contact angle of the wax is observed in the early phase of spread, as Fig. 23 shows. A raised rim is formed at the lamella periphery on the wax (t = 2.07 ms) due to capillary forces that limit the spread and decelerate the liquid flow close to the contact line. At t = 6.49 ms the droplet reaches its maximum elongation. After that, the back (trailing) edge of the contact line moves (slides) down faster than the front (leading) edge of the contact line. The droplet recoils due to large surface energy (large contact angle). The contact area (water–wax) decreases and the height of liquid increases. The liquid/solid contact shrinks and at the same time moves downward, contrary to the horizontal recoil, where the liquid/solid contact shrinks and moves inward until the rim reaches the center and rebounds. Therefore, the droplet does not rebound in this case. At t = 15.14 ms a small portion of the droplet pinches off from its front and forms a secondary droplet (t = 17.42 ms). The main body of liquid slides downward faster than the secondary droplet, they can rejoin, and the droplet approaches an equilibrium shape. The secondary droplets can also detach from the back. Fig. 24 shows the prompt splash of a water droplet impacting at a Weber number of 391 and impact angle of 45 onto the rough glass. The splash occurred in all directions from the lamella periphery. Very small droplets eject in early phase of impact (t = 0.19 ms) in all directions. Larger droplets eject later only downward as seen at t = 1.124 ms. When a droplet comes across a peak of roughness during the slipping on smooth glass it spreads and deposits on the surface rather than rebounding from the surface, even at very small (smaller than critical deposition/ rebound) angles, as Fig. 25 shows. There is a sharp point of roughness on the smooth glass, visible in an enlarged image (Fig. 25b) of the fourth exposure in Fig. 25a, at which the droplet adheres to the surface (pinning). Therefore, as result of (even low) roughness (Ra = 0.3 lm) of the wax the droplet spreads rather than rebounding at an impact angle of 2. Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 1.000 107 K=3 K=8 K=19 K=57.7 Glycerin Dry wall Wet wall Dry 2.45 mm Deposition 0.100 Oh Wet Isopropanol 0.010 Splash 1.8 mm 3.3 mm Rebound Water 2.7 mm 0.001 1 Fig. 25. Observations of slip and spread of a water droplet impacting onto a smooth glass surface with a sharp point above the exposure four (a) and the pinning (b), enlarged exposure 4 from the image (a) (We = 391, a = 2, time between exposures is Dt = 1 ms). 3.2.2. Droplet rebound Fig. 26 presents measurements of the critical impact angle for a droplet to rebound (the limit droplet rebound/ deposition or partial deposition) on dry glass. When the critical normal Weber number, Wencr = We Æ sin2a, is plotted against impact angle a, the data exhibit constant values of Wencr, as shown in Fig. 26. The limit between the rebound and deposition regimes based on the Sommerfeld and Tropea [27] parameter K = Oh Æ Re1.25 is shown in Fig. 27. The Ohnesorge number is defined by the relation Oh = We1/2/Re. Sommerfeld and Tropea [27] propose a splash limit of K = 57.7 for a prewetted surface using water droplets. The results are compared with the limits between the rebound/deposition regimes for dry and wet walls for K = 3 and K = 8, which were used in [25] to classify the regimes of their experimental results for monosized ethanol droplet–wall interactions. The parameter K does not predict well the experimental results of rebound/deposition limit in this study. At the limit for dry smooth glass, the average values of K are 1.48 for glycerin, 4.77 for water, 5.91 for isopropanol of 1.8 mm and 7.63 for isopropanol of 3.3 mm droplets. At the limit for wetted surfaces the average values of K are 6.4 for glycerin, 12.95 for water, 16.34 for isopropanol of 10 100 Ren 1000 10000 Fig. 27. Rebound/deposition limit based on the parameter K = Oh Re1.25. 1.8 mm and 19 (21 maximal values) for isopropanol of 3.3 mm droplets. For the impact on a dry surface, roughness affects the rebound/deposition limit, while on a wetted surface waviness, depth and velocity of liquid film also affect the rebound/deposition limit. 3.2.3. Partially rebound A highly viscous (glycerin) droplet, upon impact onto a smooth surface, is found to partially rebound at a wide range of impact angles (up to 45), as Fig. 16 suggests. The size of the rebounded droplet decreases with increasing impact angle. The fraction of the droplet that rebounds may change from 100% (complete rebound) to zero (deposition) over the range of the parameters considered in the experiment [28]. The rebound ratio, defined as the ratio of the diameters of droplets that are disjoined from the primary droplet, expressed as Ds/D, are presented in Fig. 28, where Ds is the diameter of the secondary droplet that disjoins after impact and D is the primary droplet diameter. An increase of the droplet Weber number decreases the rebound ratio for the same impact angle. The measured rebound ratio as a function of the normal capillary number Can = Wen/ Ren = lusin a/r coincides for all experimental conditions, as is seen in Fig. 28. 1.2 4 1 0.8 2 Ds /D Wencr 3 Glycerin D=2.45 Water 1 We 0.6 0.4 D=2.7 Isoprop. D=3.3 51 163 402 93 280 571 0.2 Isoprop. D=1.8 0 0 5 10 Impact angle, α (deg) 15 20 Fig. 26. Normal Weber number as a function of critical impact angle for dry smooth glass. 0 0.0001 0.001 0.01 0.1 Wen/Ren=Can 1 10 Fig. 28. Rebound ratios in terms of the normal capillary number for glycerin droplets on dry, smooth glass. 108 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 4.1. Results and discussion time step was from 107 s to 105 s, depending on the spread velocity and cell size. The adhesion boundary condition that describes the free surface in contact with the solid surface is given by a static contact angle. A liquid/solid combination gives this angle. The flow considered in this study was laminar, so the wall boundary conditions could be applied directly. A qualitative comparison of a water droplet shape obtained by simulation and a CCD camera at the same times after the impact instant is shown in Fig. 29. The static contact angles (advancing 105 and receding 95) were prescribed as a boundary condition. After contact with solid surface, the bottom part of the droplet compresses and deforms. Then a liquid film (lamella) ejects close to the contact line and spreads over the solid surface. At the beginning, the thickness of the lamella is smaller than 6 lm. The thickness increases with time during the spreading. Therefore, the simulation of water droplet impact requires grids with very fine cells close the wall to capture all details of the process. Liquid accumulates at the front (rim) of the lamella and its thickness increases with time, during the spreading and receding phases, as Fig. 29a and b illustrate. At the maximum spread, the major driving forces become capillary forces. The droplet begins to recoil in order to obtain the shape with minimum surface energy. Fig. 29b shows the simulated shapes of the droplet during The spread of a single droplet is studied for dry surfaces. Two liquids, water and 85% glycerin were used as test liquids. The difference of the dynamic viscosities of glycerin (0.116 Pa s) and water (0.001 Pa s) is high, while the difference of their surface tensions (0.063 and 0.073 N/m) is not so important. The static contact angle for both fluids on the wax is high (90–97 and 95–105) and it is low on the glass (13–17 and 6–10, respectively). Since a glycerin droplet at the same Weber number as a water droplet has a Reynolds number two orders of magnitude lower it spreads slowly. Therefore, this selection includes sufficient number of different variables for a numerical analysis of droplet spreading. The flow can be considered axisymmetric. Therefore, the numerical simulation was performed in a cylindrical coordinate system, where all changes in the azimuthal direction are assumed zero. The computational space domain in this study had the shape of a cylinder segment dimensions of r = 12 mm, z = 12 mm, with a segment angle h = 2. A fixed, non-uniform grid consisting of the square control volumes in the r– z plane and only one control volume in the azimuthal direction were created and refined in the region where the liquid droplet moves. The smallest cells were near the wall, in the region where the lamella is created and expands. The cells near the wall had a size between 12.5 · 6.25 lm and 0.78 · 0.78 lm depending on the case studied. The finest grid used in this study had in total about 50,000 cells. The time step was taken on the basis that the actual local Courant number must be less or equal to 0.2. Thus, the Fig. 29. Time sequence of water droplet impact onto a wax surface (We = 90). Comparison of the simulation (right) and experiment (left). (a) The spreading phase, (b) the receding phase. 4. Numerical simulation of droplet impact Within the last decade, various studies have examined numerically droplet impact. Fukai et al. [8] used a finite-element-based technique to model the droplet spreading process. Bertagnolli et al. [29] presented an adaptive-grid, finite element method to examine the impact of molten ceramic droplets in the context of thermal spraying, neglecting wetting effects. Pasandideh-Fard et al. [30] studied fluid flow, heat transfer and solidification during molten metal droplet impact. Rieber and Frohn [31] have developed 3D models to simulate droplet impact onto a liquid surface. Bussmann et al. [11] developed a fixed-grid Eulerian model, employing a volume-tracking algorithm. The normal impact of a water and glycerin droplets on smooth glass and wax were numerically simulated in [28]. The characteristic cases for which reliable experimental data exist [28] are analyzed. The results are presented for two droplet Weber numbers, one for a low impact Weber number (We = 90) and one for larger Weber number (We = 800). The simulations were performed using the commercial code Comet [32] based on a finite-volume numerical method. Good agreement with experimental results was obtained. Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 the receding phase. Excellent agreement is obtained between the experiment and the simulation. Unlike the former case, the simulated higher viscous droplet (glycerin) impacting onto the wax surface spreads slowly. The high viscous dissipation suppresses the droplet spreading. The spreading lamella is thick and the maximum spread is low. During the receding phase viscous forces again dominate capillary forces and no rebound is observed. The droplet approaches its final shape only very slowly. Fig. 30 illustrates the free surface contours obtained by simulation for a glycerin droplet in both the spreading and the recoiling phases. The numerical and experimental values of the spread factor and apex height in time for a water droplet are com- Fig. 30. Free surface contours (a) spreading, t = 0–1.5 ms (Dt = 0.05 ms), (b) recoiling, t = 1.5–88.5 ms (Dt = 1 ms) (glycerin droplet/wax, We = 800). 6 Experiment Calculation d/D 4 2 (a) 0 0.01 0.1 1 t* 10 100 1.5 Experiment Calculation 5. Concluding remarks The physics of droplet wall interaction is a key to many spray coating and heat transfer processes. Many fundamental questions of droplet impact have not yet been cleared up. For example, the mechanism causing the onset of splash, and other instability processes remains to be solved. In order to predict the outcomes of droplet impact a better knowledge of processes occurring during the final phase of the approach of a droplet toward a surface is necessary. In particular, entrainment and collapse of a thin air layer between an approaching droplet and a surface have not been solved so far. The first contact between the two surfaces occurs on molecular scales. Interfacial phenomena related to the physics of surfaces must be involved to model this process. Unfortunately, entrained-air-layer theories have not been advanced to the point where direct comparisons with experimental data can be made. The partial rebound ratio of a high viscous droplet impacting onto inclined smooth surface deserves further experimental and numerical investigations, since this outcome has not been studied and reported. From a practical point of view, the splash ratio, as a quantitative measure of secondary droplets formed at the impact, is more important than to fix the splash/deposition threshold only. The dynamic contact angle is required as a boundary condition for numerical calculation in capillary hydrodynamics. However, there is no general correlation of the dynamic contact angle as a function of the surface characteristics, fluid droplet diameter and impact velocity. Finally, non-Newtonian fluids are widespread in practice and need detailed experimental and numerical studies of their droplet interaction with walls. Because of the incomplete state of scientific study, most droplet/wall interaction problems have to be approached experimentally. Analyzing these phenomena and development of related models remains a challenge in wetting hydrodynamics. Therefore, the problem obviously deserves further theoretical and experimental studies. y/D E.N. Ganić acknowledges the support received from the Office of Naval Research under grant N00014-04-1-0389. 0.5 (b) pared in Fig. 31. The spread factor and apex height were determined very precisely with an uncertainty of less than 1% in any of the simulations in this study. Acknowledgment 1 0 0.01 109 References 0.1 1 t* 10 100 Fig. 31. Numerical calculations of the temporal evolution of (a) spread factor and (b) apex height, in comparison with the experimental data (water droplet on wax, We = 763). [1] G. Trapaga, J. Szekely, Mathematical modeling of the isothermal impingement of liquid droplets in spraying processes, Metall. Trans. 22B (1991) 901–914. [2] M. Rein, Phenomena of liquid drop impact on solid and liquid surfaces, Fluid Dyn. Res. 12 (1993) 61–93. [3] S. Chandra, C.T. Avedisian, On the collision of a droplet with a solid surface, Proc. R. Soc. Lond. A 432 (1991) 13–41. [4] A. Frohn, N. Roth, Dynamics of Droplets, Springer, Berlin, 2000. 110 Š. Šikalo, E.N. Ganić / Experimental Thermal and Fluid Science 31 (2006) 97–110 [5] I.V. Roisman, R. Rioboo, C. Tropea, Normal impact of a liquid drop on a dry surface: model for spreading and receding, Proc. R. Soc. Lond. A 458 (2002) 1411–1430. [6] I.V. Roisman, B. Prunet-Foch, C. Tropea, M. Vignes-Adler, Multiple drop impact onto a dry substrate, J. Coll. Interface Sci. 256 (2002) 396–410. [7] J. Fukai, Z. Zhao, D. Poulikakos, C.M. Megaridis, O. Mitayake, Modeling of the deformation of a liquid droplet impinging upon a flat surface, Phys. Fluids A 5 (1993) 2588–2599. [8] J. Fukai, Y. Shiiba, T. Yamamoto, O. Mitayake, D. Poulikakos, C.M. Megaridis, Z. Zhao, Wetting effects on the spreading of a liquid droplet colliding with a flat surface: experiment and modeling, Phys. Fluids 7 (1995) 236–247. [9] T. Mao, D.C.S. Kuhn, H. Tran, Spread and rebound of liquid droplets upon impact on flat surfaces, AIChE J. 43 (1997) 2169–2179. [10] M. Bussmann, J. Mostaghimi, S. Chandra, On a three-dimensional volume tracking model of droplet impact, Phys. Fluids 11 (1999) 1406–1417. [11] M. Bussmann, S. Chandra, J. Mostaghimi, Modeling the splash of a droplet impacting a solid surface, Phys. Fluids 12 (2000) 3121–3132. [12] Š. Šikalo, H.-D. Wilhelm, I.V. Roisman, S. Jakirlić, C. Tropea, Dynamic contact angle of spreading droplets: experiments and simulations, Phys. Fluids 17 (2005) 062103. [13] Š. Šikalo, M. Marengo, C. Tropea, E.N. Ganić, Analysis of droplet impact on horizontal surfaces, Exp. Thermal Fluid Sci. 25 (2002) 503– 510. [14] H. Marmanis, S.T. Thoroddsen, Scaling of the fingering pattern of an impacting drop, Phys. Fluids 8 (1996) 1344–1346. [15] S.T. Thoroddsen, J. Sakakibara, Evolution of the fingering pattern of an impacting drop, Phys. Fluids 10 (1998) 1359–1374. [16] R. Rioboo, M. Marengo, C. Tropea, Outcomes from a drop impact on solid surfaces, Atomiz. Sprays 11 (2001) 155–166. [17] M.M. Lee, T.J. Hanratty, The inhibition of droplet deposition by the presence of a liquid wall film, Int. J. Multiphase Flow 14 (1988) 129– 140. [18] Š. Šikalo, N. Delalić, E.N. Ganić, Hydrodynamics and heat transfer investigation of air–water dispersed flow, Exp. Thermal Fluid Sci. 25 (2002) 511–521. [19] R.W. Gent, N.P. Dart, J.T. Cansdale, Aircraft icing, Phil. Trans. R. Soc. Lond. A 358 (2000) 2873–2911. [20] G.S. Hartley, R.T. Brunskill, Reflection of water drops from surfaces, in: Surface Phenomena in Chemistry and Biology, vol. 57, Pergamon Press, 1958, pp. 214–223. [21] R.M. Schotland, Experimental results relating to the coalescence of water drops with water surfaces, Disc. Faraday Soc. 30 (1960) 72–77. [22] O.W. Jayaratne, B.J. Mason, The coalescence and bouncing of water drops at an air/water interface, Proc. R. Soc. Lond. A 280 (1964) 545– 565. [23] B. Ching, M.W. Golay, T.J. Johnson, Droplet impact upon liquid surfaces, Science 226 (1984) 535–537. [24] S.L. Zhbankova, A.V. Kolpakov, Collision of water drops with a plane water surface, Fluid Dyn. 25 (1990) 470–473. [25] G. Lavergne, B. Platet, Droplet impingement on cold and wet wall, in: C. Tropea, K. Heukelbach (Eds.), Proceedings of the 16th International Conference on Liquid Atom and Spray Systems – ILASS-Europe 2000, Darmstadt, Germany, 11–13 September 2000, paper VII.12. [26] Š Šikalo, C. Tropea, E.N. Ganić, Impact of droplets onto inclined surfaces, J. Coll. Interface Sci. 286 (2) (2005) 661–669. [27] M. Sommerfeld, C. Tropea, Experimental studies of deposition and splashing of small liquid droplets impinging on a flat surface, ICLAS94, Rouen, 1994, paper I-18. [28] Š. Šikalo, Analysis of droplet impact onto horizontal and inclined surfaces, Dissertation, TU Darmstadt, 2002. [29] M. Bertagnolli, M. Marchese, G. Jacucci, I.S. Doltsinis, S. Noelting, Thermomechanical simulations of the splashing of ceramic droplets on a rigid substrate, J. Comput. Phys. 133 (1989) 205–221. [30] M. Pasandideh-Fard, R. Bhola, S. Chandra, J. Mostaghimi, Deposition of tin droplets on a steel plate: simulations and experiments, Int. J. Heat Mass Transfer 41 (1998) 2929–2945. [31] M. Rieber, A. Frohn, A numerical study on the mechanism of splashing, Int. J. Heat Fluid Flow 20 (1999) 455–461. [32] Comet (ICCM, Institute of Computational Continuum Mechanics GmbH, Hamburg, Germany), 1998.