2002-01-2749 The Deformation of Single Droplets Impacting onto a Flat Surface Ana Sofia Moita and António Luís N. Moreira Instituto Superior Técnico Copyright © 2001 Society of Automotive Engineers, Inc. ABSTRACT This paper presents an experimental study of the deformation of spherical liquid droplets impinging onto dry and flat surfaces making use of a CCD-camera with a high spatial resolution. The experiments consider different liquids (water and Diesel oil) and the effects of droplet velocity and diameter at the impact in a range of Weber numbers up to 1100 and Reynolds numbers up to 77400. Emphasis is put on the nature of the surface target. To consider this effect, two surface materials were used (Perspex and aluminium) with surface roughness varying from less than 5µm (considered as a smooth surface) up to Ra=66,6µm. For the range of droplet diameters considered in the experiments, the corresponding dimensionless values of Ra/Rdroplet vary from 1,5x10-5 and 2,5x10-2. In a first step, the experiments focus on the spread of the liquid film and analysis of the results suggest that, provided that the Reynolds number of the droplet at the impact is large (Re>2000), the energy dissipated at the wall is not affected by the nature of the surface. The effect of surface roughness appears to be important for low Reynolds numbers, typically Re<1000. Depending upon the physical parameters at the impact, the droplet may splash at the first contact with the surface (prompt splash) or the liquid film at the wall may break-up on secondary droplets during spread. In a second step, the experiments emphasize the effect of surface roughness on the onset of splash for different liquids. The experimental results are analysed in terms of the critical Weber number for which splash occurs, and compared with correlations reported in the literature accounting for the effects of surface roughness and droplet liquid and suggest the likely influence of, not only the nature of the surface (e.g. surface profile and material), but also of the liquid. INTRODUCTION The dynamic behaviour of an individual droplet impacting onto a solid surface includes several individual phenomena, such as deformation, fingering, splashing and rebound. The onset and physical description of those phenomena are usually characterized based on dynamic similarity arguments making use of dimensionless numbers characterizing the relative magnitude of the forces acting upon the surface of the droplet (Reynolds, Weber and Ohnesorge numbers). As an example, Chandra and Avedisian [1] considered the impact of n-heptane droplets onto a polished stainless steel surface, using a flash photographic method. The authors considered the influence of the surface temperature, but they establish different regimes based, mainly on Reynolds and Weber numbers. Marmanis and Thoroddsen [2] scaled the fingering pattern in water droplets impacting onto solid non-heated surfaces. Later, Thoroddsen and Sakakibara [3] considered the time evolution of the fingering pattern in water droplets impacting onto non-heated glass surfaces. By adjusting the release height of the droplets, the authors achieved Weber and Reynolds numbers at the impact as high as 1000 and 15000, respectively. Estimative of these dimensionless numbers were, in turn, obtained by scaling the forces (surface tension, shear and gravity) with physical properties of the liquid (density, viscosity and surface tension) and considering all lengths proportional to the diameter and velocity at the instant of impact, respectively. However, the physical properties of the impacting surface such as temperature, roughness or inclination alter the boundaries of the problem under study and similarity arguments cannot be applied. Some work has already been developed in this context: Stow and Hadfield [4] established a splash parameter for the occurrence of splash, which depends on the surface roughness. Later, Range and Feuillebois [5] added the influence of liquid properties and surface material to the correlations purposed by Stow and Hadfield [4]. Riboo et al. [6] carried out a qualitative analysis of the various outcomes of a drop impacting on solid horizontal surfaces with various roughness and wettabilities; Šikalo et al. [7] studied the influence of the physical properties of the liquids on the droplet impact upon horizontal surfaces, but they also included the influence of surface roughness. It seems that the effect of the inclination angle of the impact surface hasn’t been much studied. From the reviewed literature one may report Karl et al. [8], who emphasized the deformation of water and ethanol drops, with diameters ranging from 100 to 200µm, upon inclined surfaces, heated well above the Leidenfrost temperature of the droplet liquid. Kang et al. [9] studied the dynamic behaviour of water droplets impacting upon heated surfaces, inclined with an angle between 0º and 60º. between 0.6ms-1 and 5ms-1. The time history of a droplet is recorded by a high-speed camera triggered by the passage of the droplet through a laser beam hitting onto a photodiode. A function generator creates an adjustable delay in the signal that triggers the camera. The camera is connected to a personal computer to store the recorded frames and to a TV set to allow observation of the droplet deformation during recording. Figure 1 shows the experimental set-up and Figure 2 shows the trigger mechanism used for the high-speed camera. The fact that the properties of the impact surfaces influence droplet dynamic behaviour and don’t allow to apply similarity arguments is the reason why it appears, from the reviewed literature that it is not possible to correlate the several phenomena occurring during droplet deformation with proper dimensionless numbers. The present study is part of a research study aimed at to account for the complexity introduced by the influence of non-scaled parameters on the description of droplet impact. The present paper reports an experimental analysis of droplet deformation on dry flat surfaces of different materials and with different surface profiles. The experiments were carried out for different regimes commonly characterized by non-dimensionless numbers. The analysis considers different fluid properties, namely water and Diesel oil, droplet velocity and droplet diameter at impact to vary the Weber and Reynolds numbers and the results are compared with a simplified theoretical model as presented by Pasandideh-Fard et al. [10]. Also, the onset of splash is described by a critical Weber number and the results are analyzed in terms of the influence of liquid properties, surface material and profile. Figure 1: Experimental Set-Up. EXPERIMENTAL Experimental Set-up Droplets are generated at the tip of a hypodermic needle and fall by action of gravity onto a flat and dry surface. The flow rate into the needle is controlled by a syringe pump. Using different fluids (Diesel oil and water in the present case) as well as needles with different diameters, it is possible to vary the initial diameter of droplets (diameter at impact). Also, the release height of the droplets is adjustable in order to change the impact velocity of the drops. In this way the present set-up allows to cover a large range of droplet diameters and impact velocities, the only limitation being the nonsphericity, which may occur when the diameter or/and velocity of droplets is too large. The experiments reported here consider droplets with initial diameters ranging from 2.6mm to 3.5mm and with impact velocities Figure 2: Triggering mechanism for the high-speed camera. Experimental Method All quantities necessary to characterize the deformation of the droplet on the surface, such as the instantaneous droplet diameter and the height of liquid above the surface, contact angles and number of fingers are measured in pixels, directly from the recorded images. The impact velocity is obtained from the distance, in pixels, traveled by the drop between the two last frames before impact. The uncertainty of the measures from the pictures varied between 1 and 3 pixels. The uncertainty of the measurements is mainly due to the difficulty to establish droplets contour. Another source of inaccuracy is the the fact that, for hight values of drop diameter and release height, droplets aren’t perfectly spherical, due to the effect of the gravitational force.Therefore, both horizontal and vertical diameters were measured. The equivalent droplet diameter was considered, as in Šikalo et al. [7], to be D = ( D h Dv ) 2 1/ 3 values agree very well with those obtained using equation (1), which gives 2.8mm and 3.5mm for the minimum and maximum diameters used, respectively. During the experiments, each measurement was repeated five times in order to confirm the reproducibility of the phenomena and to obtain average values. , where Dh is the Profile A horizontal diameter and Dv is the vertical diameter. However for the ranges presented in this paper the nonsphericity effect is not very important since the difference between Dh and Dv is about 9% for the worst case. Profile B Profile C Profile D The instant of impact (t = 0 s) is taken from the recorded sequence of images. This procedure provides a velocity and a length scales for dimensionless analysis. The initial diameter of the droplet is also estimated considering the equilibrium between gravity forces and surface tension forces at the tip of the needle, so that: d =3 6σd needle ρ liq g (1) where: σ - is the surface tension of the liquid; dneedle – is the diameter of the needle; ρ - is the density of the liquid; g – is the specific gravity. Calibration of the image acquisition system allows establish a correspondence between the measured number of pixels and the real dimension. According to this calibration, the smallest droplet considered in the present report has a diameter of 2.6mm ± 0.1 and the biggest droplet has a diameter of 3.5mm± 0.17. These 250 200 Intensity in the grayscale Measurements were taken using image intensity profiles. Each image is converted into a matrix of values, according to a grayscale, where the minimum value corresponds to black and the maximum to white. Intensity profiles along a Diesel oil droplet are shown in Figure 3. The intensity profiles are in fact vectors of values according to the grayscale. For example, Figure 3 (a) shows four paths along the image of the droplet, namely paths A, B, C and D. Each path is represented by a vector of values in the grayscale and correspond to a curve in the plot in Figure 3 (b). The regions in each curve with larger gradient correspond to the edge of the droplet along that path. Repeating this procedure for several paths (Figure 3 only considers four to illustrate the method) one may be able to determine the shape of the droplets. (a) Profile A Profile C Profile B Profile D 150 100 50 0 0 5 10 15 20 25 Pixel (b) Figure 3: (a) Paths of grayscale values along a Diesel oil drop with a diameter of 2.9mm, released from a height of 83cm. Frame rate: 5000fps. (b) Plot of the curves corresponding to each path represented in figure (a). Characterization of Surface Roughness In order to investigate the influence of the nature of the surface target on droplet deformation and splash, this study considers the use of aluminium and perspex plates with different surface properties. The surface is characterized by the mean roughness Ra, to allow comparison with other results reported in the literature. Surface roughness, Ra, is measured with a profilometer that measures roughness in a range between 5µm and 250µm. The surfaces of the plates considered here are not perfectly homogeneous but, in the region where droplet impact and deformation occurs, deviations from the mean roughness are smaller than 10%. However, some materials showed regular surface profiles while others showed random profiles.Table 1 summarizes the characteristics of the surfaces used in the present study. Table 1: Characteristics of the surfaces used in this study. Surface Ra µm ± 10% Smooth Perspex <5 Rough Perspex with regular profile 66.6 Aluminium with random profile 18.84 Aluminium with random profile 15.2 Aluminium with regular profile 22.2 Aluminium with regular profile 6.7 Droplet spreading The sequence in Figure 4 shows the typical deformation pattern of a water droplet with the occurrence of perturbations and later, the development of fingers, without crown formation, as reported by Levin and Hobbs [11] and Thoroddsen and Sakakibara [3]. 1.8ms 3.3ms 9.8ms 39.8ms 0.8ms 2.3ms Droplet spreading is characterized by two dimensions: the diameter of the wetted area, d, and the droplet height above the surface, h (See Figure 4). Normalizing these quantities by the initial droplet diameter, D0, yields the spread factor, β(t) and the dimensionless height, ξ(t), given by: β (t ) ≡ ξ (t ) ≡ RESULTS 0.3ms Figure 4: Water drop impacting on a horizontal, dry aluminium surface. We=486, Re = 11965. Frame rate: 2000fps. 1.3ms d (t ) D0 h(t ) D0 (2) (3) Figure 5 shows the variation of d/Do and h/Do with time for a water and for a Diesel oil droplet with the same Weber number, We=487, impacting onto a smooth perspex surface (Ra<5µm). The water droplet has a high Reynolds number, of about 11400, while the Diesel oil droplet has a lower Reynolds number, of about 2550. The results in the Figure 5 show the periodic deformation of the water droplet, expanding and recoiling. After the impact, the drop spreads as a thin liquid film on the surface, until it reaches the maximum spreading diameter (spread factor β(t)=4.8), at t=4.7ms.The dimensionless height reaches a minimum value (ξ(t) = 0.17) at this point and should increase afterwards. However, the plot in the Figure shows that the dimensionless height remains constant while the droplet continues its deformation. This is due to the concave form of the droplet surface, which does not allow visualization of the center of the droplet from a side view. Similar observations can be inferred from the results reported by B. S. Kang and D. H. Lee [9], for a droplet of water impinging onto a surface heated at the temperature of nucleate boiling regime. However, the phenomena is more noticeable in the present case, which must be associated with the nature of the surface and velocity at the impact. 2.8ms 4.8ms 7.3ms 19.8ms 29.8ms 49.8ms 71.8ms After spreading the water drop recoils, only until about t=6ms, when the dimensionless height reaches a peak (ξ(t) =0.23). This first recoiling is not very clear in the spread factor curve. Then the liquid film starts spreading again, until it reaches a new maximum spread at about t=15ms. Afterwards the liquid film recoils until it reaches its final state. During spreading, some liquid breaks through the lamella, due to the action of inertial forces, which overcome surface tension forces. This is why the spread factor is larger for water than for the Diesel oil droplet. However, if the spread factor doesn’t account with the liquid breaking through the lamella, Diesel oil drop has a larger spread. Later, during the recoiling, this liquid film is separated from the liquid that will remain deposited in the central region, leading to the formation of two small secondary drops, as it is shown in Figure 6. The perturbations observed in the lamella (See Figure 6) are fingers, which are formed during the spreading process. Despite the formation of large fingers, splash does not occur, probably because the impact surface is very smooth. Therefore the surface properties, which are not accounted by the dimensionless numbers considered here, may have influenced the establishment of two different regimes (deposition/splash), delaying the occurrence of splash. 6 5 d/Do 4 Diesel oil droplet Re=2558 3 2 Water droplet Re=11383 1 0 0 0,01 0,02 0,03 0,04 0,05 0,06 t [s] 0,6 Diesel oil droplet Re=2558 0,5 h/Do 0,4 Water droplet Re=11383 0,3 0,2 0,1 0 0 0,012 0,024 0,036 0,048 0,06 t [s] Figure 5: Time variation of the spread factor and the dimensionless height, for a water droplet and a Diesel oil droplet, both with We=487. The periodic behaviour of the deformation process of the water droplet cannot be observed for the Diesel oil droplet. Maximum spread diameter occurs at about t=5ms, and has a value of about ξ(t) = 3.4. Moreover, no significant recoil is observed for the Diesel oil droplet, which is confirmed both by the spread diameter and dimensionless height curves that remain practically constant after reaching the maximum spread. Naturally, if the Diesel oil drop has a larger spread than the water drop, its dimensionless height is smaller. The differences observed in the behaviour of the two droplets may be explained, in part, by the lower surface tension of the Diesel oil, which may not be large enough to bring the liquid back to the central region of the droplet. Also, it seems that the system Diesel oil/perspex surface has a higher wettable liquid/solid interaction, so the liquid film progression is easier. The influence of viscosity does not seem to be important during the first stage of spreading, since inertial forces are dominant. However, the higher dissipation due to shear forces may explain the very small (almost inexistent) recoiling stage of the Diesel oil droplet, which has a lower Reynolds number. Figure 7 shows the variation of d/Do and h/Do with the time for a water and for a Diesel oil droplet with the same Reynolds number, Re=2558, impacting onto a smooth perspex surface (Ra<5µm). For this Reynolds number the water droplet has a very low Weber number and, in this case, differences between the two droplets are even clearer. The water droplet has a periodic behaviour, contracting and expanding about 16 times as shown in the images of Figure 8, in a way that can be described as a spring-mass system, as by Haitao Xu et al. [11]. For this case, the water drop has a very low impact velocity, and, therefore, has a small spread diameter. Maximum spread occurs at t=4.5ms and is about 2. Although this drop has a smaller impact velocity, the maximum spread occurs nearly at the same time as for the drop of Figure 5. For the initial stages of spreading, the inertial forces are dominant, as expected since droplet diameter has a contribution ∝ d3, which is bigger than the contribution of droplet velocity. After reaching the maximum spread, and consequently the minimum dimensionless height, the liquid film begins to recoil and reaches a height peak of about 0.8, at instant t=11ms. At this instant, a phenomena close to partial rebound, is clearly observed, as it is shown in Figure 8. The following peaks of dimensionless height are successively smaller, due to dissipation of energy caused by shear forces with the wall. It is also interesting to investigate the influence of surface roughness in droplet spreading. Figure 9 shows the time evolution of the spread factor and dimensionless height of Diesel oil drops impacting onto a rough aluminium surface (Ra =18.84µm ±10%) and on a smooth perspex surface (Ra<5µm). The droplets have nearly the same diameter and the same impact velocity, so changes in the behavior may be attributed to surface properties only. The results show similar qualitative trends, in the sense that both cases show the periodic deformation behaviour previously reported. Quantitatively, the effects of surface roughness seems to be negligibly small in the initial stage (t* < 0,5), which is in accordance with the fact that the initial rate of change depends more on the kinetic energy of the droplet at impact (e. g., Fukai et al., [12]). Further, the maximum spread is larger on the smooth surface as also observed by other authors (e. g., Fukai et al., [12]; Sikalo et. al., [7]) and the minimum nondimensional height is larger on the smooth surface, 17.25ms 19.75ms 22.25ms 24.75ms 27.25ms 29.75ms Figure 6: Recoiling of a water droplet upon a perspex surface, with We=487. The droplet, with an initial estimated diameter of 3.6mm, was released from a height of 495mm. Frame rate:2000fps. 4 1,2 3,5 1 2,5 Diesel oil droplet We=487 2 1,5 Water droplet We=47 1 Diesel oil droplet We=487 0,8 h/Do d/Do 3 0,6 Water droplet We=47 0,4 0,2 0,5 0 0 0 0,01 0,02 0,03 0,04 0,05 0,06 t [s] 0 0,012 0,024 0,036 0,048 0,06 t [s] Figure 7: Time variation of the spread factor and the dimensionless height, for a water droplet and a Diesel oil droplet, both for Re=2558. in accordance with the requirement of mass conservation. However, other authors observed that the surface only affects the height of the droplet on the recoil phase. The difference may stem from the much smaller value of the Reynolds number considered in this case (about 400 compared to 3264). It may be so, since the increased dissipation due to roughness takes more time to act when the kinetic energy at impact is larger. Also, a larger dissipation of the initial energy at the surface will decrease the period of the deformation, as also observed in Figure 9. Therefore, the results presented here suggest that, as the kinetic energy at impact increases, the effect of surface roughness may be delayed for later stages of the deformation process. before impact 1.0ms (after impact) 2.0ms 4.0ms 12.5ms 5.0ms 7.5ms 10.0ms 14.0ms 17.5ms 22.5ms 27.5ms 30.0ms 25.0ms 32.5ms Figure 8: Water droplet impacting onto a perspex smooth surface. We=47, Re=2558. Initial periods of deformation. Frame rate:2000fps. 3,5 1.0 2,5 0,8 2.0 0,6 ξ(t) β(t) 3.0 1,5 0,4 1.0 0,5 0,2 0.0 0 5 10 15 t* Aluminium rough surface Ra =18.84µm 0.0 0 5 10 15 t* Perspex smooth surface Ra<5µm (b) (a) Figure 9: Variation of β(t) and ξ(t) with the dimensionless time for a Diesel oil droplet, impacting onto an aluminium rough surface (We=18; Re=464, Ra=18.84µm±10%) and on a smooth perspex surface (We=13; Re=394, Ra<5µm). It is worth discussing the physical meaning of these results based on a model proposed by Pasandideh-Fard et al.[10]. In this model, it is assumed that the film of liquid expands radially on the surface, with velocity VR, as shown in Figure 10. This model considers that the droplet spreads into a cylindrical disk of diameter D(t) and height h. The velocity VR is given by conservation of mass: VR d (t ) 2 = (4) U o 4 D(t )h 4 h= πDo3 (5) 6 or h= 2 Do3 2 3Dmax (6) Combining eqs. (5) and (6) gives: 2 3 U o Dmax VR = d (t ) 2 2 8 D(t ) Do (7) Uo Do d(t) h D(t ) 3 Uo = t Dmax 8 Do (8) Maximum diameter occurs when: t* = t where U0 and D0 are, respectively, the droplet velocity and diameter before impact. When the disk (liquid film) reaches its maximum diameter, Dmax, the liquid volume within the disk equals the volume of the initial droplet, i.e.: πD 2max Assuming, as Pasandideh-Fard et al.[10], an average value d(t)~D0/2 and knowing that VR=δD(t)/2δt, the evolution of splat diameter (D(t)) is given by integration of eq. (7): Uo 8 ≈ Do 3 (9) Therefore, the dimensionless time (i.e. the time normalized by the impact velocity and by the initial droplet diameter), t*, required for the droplet to reach its maximum extent, is a constant. Back to the Diesel oil and water drops considered in Figures 5, 7 and 9, the values for t* may be determined and compared with those obtained by Pasandideh-Fard et al. [10], as shown in table 2. The results in Table 2 show that the experimental values of t* differ from theoretical values of Pasandideh-Fard et al. [10], being systematically larger than the theoretical value of 8/3 for Reynolds numbers above 2558 but smaller for Reynolds numbers below 464, regardless the liquid or the surface roughness. This analysis, firstly suggest that the behaviour of droplet deformation is determined by the balance between inertial forces and viscous forces acting against the spread of the liquid film over the wall. The maximum spread diameter may also be predicted by applying the energy conservation condition to the model of Pasandideh-Fard et al. [10]: droplet kinetic energy, KED, plus the energy required to keep the spherical shape of the droplet, SE1, must equal the final surface energy, SE2, plus the work done in deforming the droplet against viscosity, W, i. e: KED + SE1 = SE2+W VR The authors found a relation for W, based on appropriate length scales, and achieved an equation for the maximum spread: D(t) Figure 10: Model of droplet spreading, according to Pasandideh-Fard et al. [10]. D max We + 12 = Do 3(1 − cosθ a ) + 4 We ( Re ) (10) where We is the Weber number, Re is the Reynolds number and θa is the contact angle between the liquid and the impact surface. We=487, Re=11383 (Water drop onto a smooth perspex surface, Ra<5µm) We=47, Re=2558 (Water drop onto a smooth perspex surface, Ra<5µm) We=487, Re=2558 (Diesel oil drop onto a smooth perspex surface, Ra<5µm) We=4, Re=198 (Diesel oil droplet onto smooth perspex surface, Ra<5µm ) We=13, Re=390 (Diesel oil droplet onto smooth perspex surface, Ra<5µm ) We=18, Re=464 (Diesel oil onto rough aluminium surface, Ra=18.84µm±10%) t* (present work) t* (PasandidehFard et al.) [%] 4.0 8/3 32.5 4.25 8/3 36.5 4.25 8/3 36.5 ∆ Table 2 Comparison between experimental values of t* and the value obtained by Pasandideh-Fard et al. [10] The scatter observed in Figure11 for low Reynolds numbers is attributed to the effect of the surface. It is expected that the effect of surface roughness may be that of altering friction forces at the wall, which may be accounted in Reynolds number with a modified viscosity. 6 5 1.3 1.61 8/3 8/3 - 51.8 - 39.3 Dmax/Do Experimental Conditions 4 3 2 1 0 1.66 8/3 - 38.5 Following the analysis of the values in Table 1, it is expected that the maximum diameter attained by the liquid film during spread may be determined mainly the ratio between inertial forces and viscous forces at the wall, I e, by the Reynolds number. This is shown in Figure 11, where the two lines corresponds to curves of Dmax/Do obtained from equation (10) with limiting values of the Weber number of 40 and 400 an limiting values of the contact angles of 5o and 60o as measured by Mundo et al. [14]. The symbols in Figure 11 represent the experimental results obtained in the present work. The results show a good agreement with the model of Pasandideh-Fard et al. [10] for Reynolds numbers above 2000, but the scatter is larger for small values of Reynolds number, for which the assumptions for the dissipative term may not be appropriate, as also mentioned by Chandra and Avedisan [1] and by Mundo et al. [14] who used similar assumptions. Furthermore, these results suggest that, for high values of Re, Reynolds number alone may qualitatively describe the steady state parameters characterizing droplet deformation, regardless liquid properties and the nature of the surface, at least provided that splash does not occur. It is worth mentioning at this point that, if surface roughness is not much smaller than the liquid film thickness, it is expected the film to detach from the surface. In this case, splash occurs and the physics of phenomena are not described in the same way. However, for the experimental conditions considered here, the ratio of surface roughness to film thickness attained a maximum value of about 6x10-2, which is not large enough to promote secondary atomisation (e. g., Mundo et al. [14]). 0 2000 4000 6000 8000 10000 12000 Reynolds number Figure 11 - Comparison between experimental results of (Dmax/D0) and the model of PasandidehFard et al. [10] It is worth mentioning that some difficulties were found in measuring the spread diameter, due to the illumination conditions, which may have contributed for some differences in Figure 12. Also, the value for the contact angle was estimated by consulting the literature, in particular the experimental results of Fukai et al.[12] and Pasandideh-Fard et al. [10] and. For the Diesel oil droplet, the value of the contact angle was estimated, based on the behaviour of water droplets with sodium dodecyl sulphate, (surface tension of 50x10-3 N/m). Referring to the accuracy of these results, it is important to note that they depend strongly on the choice of the operational parameters of the CCD camera, and it is always necessary to optimise the time resolution with the spatial resolution. When the deformation process is slower it can be used a lower frame rate, that allows a better spatial resolution, although at the expense of the time resolution. Finally it is interesting to observe with more detail the periodic behaviour of water droplets. Figure 12 depicts the evolution of the number of periods in a water droplet as a function of the release height. Number of periodes 18 16 14 12 10 8 6 4 2 0 0 200 400 600 800 1000 Droplet release height [mm] Figure 12: Evolution of the number of periods as a function of the droplet release height. The graphic shows that the number of periods reduces as the release height increases. This may suggest that with an increase of the impact velocity, as there is a larger spread and a bigger change of height above the surface, there is a higher dissipation of surface energy. For lower release heights, there is a lower spread and the surface energy is high enough to produce the phenomena close to partial rebound, observed in the water drop, and allows the periodic behaviour also observed in this droplet. Influence of surface roughness in the splashing limits When the Weber and Reynolds are high enough, the droplet may splash as it impinges at the wall or as a result of hydrodynamic perturbations developed during spread, as shown in Figure 13 for a Diesel oil droplet. In each case the liquid does not deposit on the surface. number at which splash occurs was determined for water and Diesel drops impacting onto various dry plates with different surface characteristics (see Table 1). The results are plotted in Figure 14 as function of log(Ra/Ro), where Ro is the droplet radius at the impact, together with measurements obtained reported in the literature considering other surfaces. The curves in Figure 14 represent groups of results obtained by (o) Stow and Hadfield [4] with water drops onto aluminium and by Range and Feuillebois [5] with (•) water droplets on glass, (×) water droplets impacting onto transparencies produced by 3M, (+) water droplets on aluminium and (∗) water droplets on Plexiglas. From their experimental results Stow and Hadfield [4] developed a correlation for the splashing limits, according to the empirical formula: R0 u 10.69 = S T ( Ra ) (11) where the critical value of a droplet to splash, ST, depends on the arithmetic roughness Ra. The authors also rewrote equation (11) in terms of a critical Weber and a critical Reynolds number (Wec and Rec, respectively), so that: Re c0.31 Wec0.69 = ξ ( Ra ) (12) where ξ is a dimensionless critical number for splashing, function of Ra. The curve fitting of their results follow this empirical correlation (circles in Figure 14). Figure 13: Splash of a Diesel oil droplet upon a perspex surface, with We=630. Droplet has an initial diameter of 2.8mm. Frame rate:2000fps. Although it is known that the onset of splash is associated with high velocities of impact when smooth surfaces are considered, the experiments in the present work showed that surface roughness promotes splash, as also observed by Stow and Hadfield [4], Range and Feuillebois [5], Mundo et al. [14] or Cossali et al. [15]. In order to study in detail the influence of surface roughness in the splashing limits, the critical Weber Figure 14: Critical Weber for splashing as a function of dimensionless roughness Ra/Ro. Adapted from Range and Feuillebois [5]. Our results: Water drops impacting onto an aluminium surface, Ra=6.7µm±10%( ), an aluminium surface, Ra=15.2µm±10%( ), an aluminium surface, Ra=18.84µm±1%( ), an aluminium surface, Ra=22.2µm±10%( ) and a perspex surface, Ra=66.6µm±10% ( ). On the other hand, Range and Feuillebois [5] fitted their results making use of the correlation of Wu [16]. This author stated that for small Ohnesorge numbers, (Ohnesorge number, Oh = ν ρ / 2 R0σ , where ν is the liquid kinematic viscosity, ρ is the liquid density, σ is the liquid surface tension and Ro is the initial droplet radius) the influence of viscosity can be neglected and the correlation of Stow and Hadfield may be improved to: R Wec = a log b 0 Ra (13) Wu obtained the values of a=6.47 and b=1.87 by fitting his formula to the experimental results of Stow and Hadfield with a least-squares method. For each plate material Range and Feuillebois [5] adjusted the values for the coefficients a and b, in order to fit their experimental results with the correlation of Wu [16]. characteristics of the surface. It was noticed in the measured roughness profiles, that the frequency of the surface perturbations in the surfaces with random roughness was very high. So, although Ra is large, maybe the high frequency makes the surface to behave as a smooth surface. For the surfaces with a regular profile, it was observed that the splash occurs in a later stage of the spreading. Possibly, the regular profiles promote the flow of the film over the surface and only later it becomes unstable, generating fingers. Then, some of the fingers detach from the film and generate secondary droplets. From all the plotted results it seems that the results do not always follow the suggested correlations, particularly the correlation by Stow and Hadfield [4]. As for the correlation proposed by Wu [16], it seems that there is a dependence of the coefficients of a and b, not only with the surface material but also with the liquid, since it is necessary to adjust the values of these coefficients whenever the surface material or the liquid of the droplet are changed. Further investigation may allow to determine in detail the dependence of coefficients a and b with those parameters. In Figure 15 the results obtained with Diesel oil droplets impacting onto aluminium and perspex surfaces are compared with results reported by Range and Feuillebois [5] with ethanol and water in order to analyse the influence of fluid properties. The experimental results of the present work plotted in Figure 14 were obtained with water droplets impacting, mainly onto aluminium surfaces (only one of the surfaces is made from perspex). It was observed that the critical Weber limiting splashing/deposition regimes is not a well defined number, but instead it varies with the experience. The range of variation was considered as the accuracy due to the reproducibility of the splashing phenomena and was used as the vertical dimension of the symbols in the Figure. In a similar way, the horizontal dimension of the symbols was taken as the accuracy in the measurements of Ra (10%). Observing our results in Figure 14 it is noticeable that the data obtained from the impact of water droplets on aluminium surfaces follow the curve for 3M transparencies obtained by Range and Feuillebois [5]. As for the perspex plate, the point is also near this curve, but there is not enough data to establish the distribution of the results using perspex plates. It is also clear that our results do not follow the correlation proposed by Stow and Hadfield [4]. Although it follows a similar behaviour, the values of the critical Weber are higher, even when higher values of Ra are considered. Even though our experimental conditions are slightly different from those used by Stow and Hadfield [4], it does not explain the differences observed. One possible explanation may lay, not in the value of Ra, but in other Figure 15: Critical Weber for splashing of droplets as a function of dimensionless roughness Ra/Ro, for different liquids. Adapted from Range and Feuillebois [5]. Our results: Water drops impacting onto an aluminium surface, Ra=6.7µm±10%( ), an aluminium surface, Ra=15.2µm±10%( ), an aluminium surface, Ra=18.84µm±1%( ), an aluminium surface, Ra=22.2µm±10%( ) and a perspex surface, Ra=66.6µm±10% ( ). Diesel oil droplets impacting onto an aluminium surface, Ra=18.84µm±10%( ) and a Perspex surface, Ra=18.84µm±10%( ). characterized by the dimensionless roughness Ra/Ro. However, this parameter alone does not describe completely the nature of the surface and other characteristics describing the surface profile may also be considered. Therefore, existing correlations do not apply to all cases considered in the present work. However, more systematic measurements are still necessary in order to develop appropriate correlations. ACKNOWLEDGMENTS It is observed that Diesel oil droplets have higher critical Weber numbers, both for aluminium and perspex plates. From the point of view of the influence of the surface tension, Diesel oil has the lowest surface tension, so Diesel oil droplets should splash easier. The fact that Diesel oil drops were slightly smaller than the water drops may explain the higher splash velocities, but does not necessary explains the higher value for the critical We, since the Weber number accounts for the influence of both diameter and velocity. However, liquid viscosity, which is smaller for water, is not taken into account. For example comparing the Diesel oil and water droplets impacting onto the aluminium random rough surface, Reynolds number for Diesel oil is smaller than for water (ReDiesel oil drop = 3590< Rewatercdrop = 8960). It is therefore suggested that the effect of viscosity can not be neglected for the Diesel droplets. CONCLUSIONS This work reports an experimental study of the deformation and splashing of water and Diesel oil droplets impinging on rough surfaces. The results show that an accurate characterization of the physical processes occurring during deformation and splash must take into account the forces involved (surface tension, inertial and shear forces), as well as the liquid properties and the nature of the impact surface. The analysis based on dimensionless numbers only, neglects the effects of the properties of the surface, which may influence the phenomena and the establishment of different regimes. In general, the models reported in the literature to describe droplet deformation do not take into account surface properties, although these do not alter the qualitative behavior of liquid film spreading. If splash does not occur, the effect of surface roughness is expected to alter friction forces at the wall and, therefore, the energy dissipated during spread. It is found that, provided the Reynolds number of the droplet at the impact is large (Re>2000), the energy dissipated at the wall is not greatly affected by the nature of the surface, which appears to be important for low Reynolds numbers, typically Re<1000. However, the nature of the surface significantly alters the onset of splashing. Correlations for splashing found in the literature take into account surface roughness The authors acknowledge the contribution of the National Foundation of Science and Technology of the Ministry for Science and Technology by supporting this study through the project POCTI/ 1999/ EME/ 32960 and by supporting A. S. H. Moita with a Research Grant. REFERENCES 1. Chandra, S., Avedisian, C. T., “On the collision of a droplet with a solid surface”, Proc. R. Soc. London, Ser. A 432, 13-41, 1991. 2. Marmanis, H., Thoroddsen, S. T., “Scaling of the fingering pattern of an impacting drop”, Physics of Fluids, vo.8, nº6, 1344-1346, 1996. 3. Thoroddsen, S. T., Sakakibara, J., “Evolution of the fingering pattern of an impacting drop”, Physics of Fluids, vol.10, nº6, 1359-1374, 1998. 4. Stow, C. D., Hadfield, M. G., “An experimental investigation of fluid flow resulting from the impact of a water drop with an unyielding dry surface”, Proc. R. Soc. of London, Ser. A 373, 1981. 5. Range, Kai, Feuillebois, François, “Influence of Surface Roughness on Liquid Drop Impact”, Journal of Colloid and Interface Science, 203, 1630, 1998. 6. Riboo, R., Marengo. M., Tropea, C., “Outcomes from a drop impact on solid surfaces”, ILASSEurope’99, 1998. 7. Šikalo, Š, Marengo, M., Tropea, C., Ganic, “Analysis of impact of droplets on horizontal surfaces”, Experimental Thermal and Fluid Science, nº25, 503-510, 2001. 8. Karl, Alexander, Anders, Klaus, Rieber, Martin, Frohn, Arnold, “Deformation of liquid droplets during collisions with hot walls: experimental and numerical results”, Part. Syst. Charact. vol.13, 18619, 1996. 9. Kang, B. S., Lee, D. H., “On the dynamic behaviour of a liquid droplet impacting upon an inclined heated surface”, Experiments in Fluids, vol.29, 380387, 2000. 10. Pasandideh-Fard, M., Qiao, Y. M., Chandra, S.,Mostaghimi, J., “Capillary effects during droplet impact on a solid surface”, Physics of Fluids, vol.8, nº3, 650-658, 1996. 11. Levin,Z., Hobbs, P. V., “Splashing of water drops on solid and wetted surfaces: Hydrodynamics and charge separation”, Phils. Trans. R soc. of London, Ser. A 269, 1971. 12. Haitao, Xu, Yongchang Liu, Ping He, Haiqing Wang, “The TAR Model for calculation of droplet/wall impingement”, Journal of Fluids Engineering, 593-597, 1998. 13. Fukai, J., Shiiba, Y., Yamamoto, T., Miyatake, O., Poulikakos, D., Megaridis, C. M., Zhao, Z., “Wetting effects on the spreading of a liquid droplet colliding with a flat surface: experiment and modeling”, Physics of Fluids, vol.7, nº2, 236-247, 1995. 14. Mundo, CHR., Sommerfeld, M., tropea, C., “Dropletwall Collisions: Experimental studies of the deformation and break-up processes”, Int. J. Multiphase Flow, vol21, N02, 151-173,1995. 15. Cossali, G. E., Coghe, A., Marengo, M., “The impact of a single drop on a wetted solid surface”, Experiments in Fluids, 22, 463-472, 1997. 16. Wu, Z. N., “Modélisation et calcul implicite multidomaine d’écoulements diphasiques gasgouttelettes”, Ph.D. Thesis, Université Pierre et Marie Curie, Paris, France, 1992.
© Copyright 2025 Paperzz