Visualization and Collection of Overspray from Airless Spray Painting

SAVE OFFLINE

8 INTERNATIONAL SYMPOSIUM ON FLOW VISUALIZATION (1998)

VISUALIZATION AND COLLECTION OF OVERSPRAY FROM AIRLESS SPRAY PAINTING Gary S. Settles, James D. Miller, Thomas J. Hartranft, and Aaron D. Brandt Keywords: spray painting, liquid atomization and sprays, non-Newtonian fluids, environmental engineering, pollution control, two-phase flows Abstract Spray painting generates serious airborne pollution through “overspray”, the fraction (typically 50% or more) of the paint which fails to reach its target. This paper examines the phenomenology of “airless” spray painting by flow visualization, and shows how the knowledge thus gained can be used to capture the overspray efficiently near its source. Results show that the airless spray painting process combines the fluid-dynamic phenomena of wavy-sheet breakup, spray entrainment, spray impingement, inertial droplet impaction, and the turbulent wall jet. In view of this, the paper considers how the overspraycontaining wall jet might be induced to separate, roll up, and be drawn away in a particular direction where it can be filtered to remove the overspray particles. The design of an aerodynamically-correct overspray collector results from these considerations. 1 Introduction Millions of gallons of paint are sprayed every year worldwide. However, of the many coating operations with the potential to generate airborne pollution, none is more serious than spray painting. For example, some marine paints used for the protective coating of ships contain poisons and volatile organic compounds which are Contact author: G. S. Settles Gas Dynamics Laboratory, Mechanical Engineering Department Penn State University 301D Reber Bldg., University Park, PA 16802 USA

regulated as hazardous air pollutants. On average, perhaps only half of the sprayed paint actually forms a coating on its target. The other half, which escapes to the environment, is termed “overspray.” This issue thus has very important environmental implications. In fluid dynamic terms there are three ways to spray paint: airblast atomization, pressure atomization, or some combination of the two. The first is termed “conventional airspray” in the industry, while the second is known as “airless” paint spraying. Hybrid approaches make up the third category. Transfer efficiency in spray painting is the percentage of the total paint sprayed which eventually adheres to the target. Conventional airspray has a transfer efficiency typically in the range of only 20-30%, which has become environmentally unacceptable. Airless spray, on the other hand, has a transfer efficiency often above 50% – better, but with considerable room for improvement. Studies of the fluid dynamics of spray painting are few. Significant contributions were made, for example, by Kwok and Liu [1] and Hicks and Senser [2] for conventional airspray. Airless spray painting was considered by Settles [3]. Strategies to collect or eliminate the overspray, however, are essentially lacking except for the case of enclosed spray booths. This paper reviews the overall phenomenology of airless spray painting through flow visualization [3], then applies this knowledge to point the way toward aerodynamically-effective collection of the overspray near its source.

1-1

GARY S. SETTLES, JAMES D. MILLER, THOMAS J. HARTRANFT, AARON D. BRANDT.

2 Experimental methods 2.1 Spray painting equipment Graco™ commercial airless spray equipment was used in these experiments, including a 45:1 pressure intensifier, hose, and spraygun. Fed by compressed air at 0.7 MPa (100 psi), the equipment produced a paint pressure of 31 MPa (4500 psi). A 0.5 mm (0.021 inch) elliptical nozzle orifice was used. The nozzle standoff distance from the target was 30.5 cm (12 inches) and the spraygun was aimed perpendicular to the target, which was a flat plywood board. 2.2 Paints Sherwin-Williams™ A100 white latex paint was sprayed in the visualizations shown here. It is an innocuous water-borne paint with characteristics broadly typical of automotive or industrial as well as residential paints. Additional results not shown here were obtained with a 2-part anticorrosive epoxy primer produced for the US Navy, which showed reasonably similar atomization behavior but contained significant amounts of hazardous volatile organic compounds. Shear viscometry tests revealed that the latex paint was shear-thinning (as expected in a heavily-loaded paint containing polymer thickeners), but the epoxy primer was surprisingly Newtonian over the available measuring range. The extensional viscosity behavior of these paints and two other marine coatings used by the Navy is now being studied, insofar as it is expected to play a key role in the pressure atomization process 2.3 Flow visualization 2.3.1 Flash imaging of atomization A xenon flash of about 1 µsec duration was trained upon the near-field atomization process within a few cm of the airless spraygun nozzle orifice. Results were recorded on S-VHS videotape at the standard 30 Hz frame rate, to which the repeating flash unit was synchronized. No attempt at high framing rates was made. 2.3.2 Planar laser scattering The second approach used here involved a 2.75W CW Argon-ion laser beam, which was spread into a sheet by a simple glass rod. This Planar Laser 1-2

Scattering (PLS) technique is well-known and well-suited to examine complex particle-laden flows [4]. Here it was applied in three mutuallyperpendicular planes, two (x-z and y-z) parallel to and one (x-y) transverse to the z-axis along which the spray is directed (see Figure 1). Note that the exaggerated elliptical spray produced by the airless spraygun lies centered in the vertical y-z plane. Once again, all results were recorded using a 1/250 sec shutter speed on S-VHS videotape, from which the still images to be shown later were extracted.

Figure 1. Coordinate Frame for Spray Painting. 2.4 2-D experimental model In developing a concept for an overspray collector based on present results, a simple apparatus was constructed to represent a 2-D “slice” of the paint spray perpendicular to its y-axis. This model was a 3/4 scale replica of the actual process, had a thickness in the y-direction of 5 cm, and was made of plexiglass for purposes of flow visualization. The PLS techniques was used here as well as in the spray experiments, but no actual paint was sprayed, since this would have coated the plexiglass sidewalls. More will be said about this model in following sections.

VISUALIZATION AND COLLECTION OF OVERSPRAY FROM AIRLESS SPRAY PAINTING

3 Results and discussion 3.1 Paint atomization A microsecond-flash-illuminated close-up of latex paint atomization in the y-z plane is given in Figure 2. It shows, by comparison with classical results given, e.g., by LeFebvre [5], a phenomenon known as “wavy-sheet breakup.” The elliptical nozzle orifice causes the ejected paint to form a coherent, spreading sheet in the y-z plane. At least for Newtonian liquids, an instability at a certain natural frequency is known to cause such a sheet to develop waviness or a “flapping” motion. These waves, clearly visible in Figure 2, pinch off into filaments which then disintegrate into droplets. Increasing the paint viscosity impairs this process, while an increase in the relative speed between the liquid and the surrounding air promotes it.

3.2 Paint spray impingement on the target PLS images of airless spray painting with the light sheet in the y-z plane show turbulent structures formed by the atomized particles on their way toward the target (Figure 3). When the light sheet is oriented in the x-z plane (Figure 4), however, one observes, upon impingement of the spray, that a strong current of fine particles moves outward along the wall in the ±x-direction.

Figure 3. PLS image of spray in the y-z plane.

Figure 2. Microsecond y-z-plane image of latex paint sheet atomization by an airless spraygun. What happens when non-Newtonian fluids are sheet-atomized, on the other hand, is less understood. Researchers [6-8] agree that nonNewtonian effects impair atomization by causing streamwise ligaments, which lead to larger mean droplet diameters, but the details are sketchy. Viscous Newtonian fluids have enough extensional viscosity to form ligaments as well, but this becomes highly accentuated when long-chain polymers are present. For example, Figure 2 reveals significant streamwise elongation which is not seen when water alone is sprayed (e.g. [5]).

Figure 4. PLS image of spray in the x-z plane.

1-3

GARY S. SETTLES, JAMES D. MILLER, THOMAS J. HARTRANFT, AARON D. BRANDT.

This observation is confirmed by x-y-plane PLS images (e.g. Figure 5), in which the view is toward the target and the laser sheet is oriented parallel to the target and a few cm in front of it. Upon starting the spray, a toroidal vortex moves away from the spray centerline along the target surface, appearing as a ring. During steady-state spraying the majority of the overspray is seen to migrate rapidly away from the spray impingement zone in a direction perpendicular to the long axis of the spray ellipse, i.e. in the ±x-direction, which is horizontal in Figure 5.

Figure 5. PLS image of spray impingement in the x-y plane, with starting vortex. 3.3 The generation of overspray These flow visualization results reveal that the airless spray process and the generation of overspray involve a combination of several fluid dynamic “building-block” phenomena, about each of which something is known from previous work. The first phenomenon, wavy-sheet breakup, was mentioned earlier. The others are discussed in this section. The PLS images shown above reveal that the principal plane of observation of the overspray physics is the x-z plane. They further show that the flow pattern is quasi-two-dimensional (2-D) in this plane, with a comparatively small component of overspray in the y-direction. Tests with epoxy primer at a higher shear viscosity confirm the trend toward 2-D flow which is illustrated here for water-borne latex paint. Quasi-2-D flow is a powerful simplification. With it, the physics of the process can be summa1-4

rized in planar form. The several physical phenomena which combine to make up airless spray painting and overspray generation are thus best described in terms of Figure 6, a “flowfield model” in the x-z plane corresponding to the PLS image of Figure 4.

Figure 6. Flowfield model of airless spray painting in the x-z plane. 3.3.1 Spray entrainment Following wavy-sheet atomization, a dense quasielliptical spray is propelled toward the target. This spray strongly entrains the surrounding atmosphere, since each paint droplet induces an aerodynamic wake which follows it. The combined effect of these droplet wakes is a significant co-flowing airstream. Thus the attempt of the airless spray painting concept to avoid the collateral airstream produced by conventional airspray is not entirely successful. Spray entrainment has been discussed by MacGregor [9], who shows it to be dominated by the larger particles in the spray. 3.3.2 Spray impingement Upon reaching the target, the two-phase spray jet impinges thereupon. Similar impingements occur in spray cooling and diesel spray combustion phenomena. While the co-flowing airstream must

VISUALIZATION AND COLLECTION OF OVERSPRAY FROM AIRLESS SPRAY PAINTING

abruptly turn parallel to the target, the largest paint droplets have sufficient inertia to cross the aerodynamic streamlines and strike the target. Here, spray impingement works much like inertial impaction, a principle used to remove particles above a certain size from an airstream in environmental sampling. However, the smallest paint particles – by virtue of their low inertia – are able to follow the aerodynamic streamlines and turn parallel to the target, thus never impinging upon it. This discrimination of the behavior of large vs. small particles is characterized by a Stokes number [10], which compares the aerodynamic response time of the particle to a time scale associated with the structure of the flow. Thus the spray impingement zone separates the paint particle distribution into a large-particle fraction which strikes the target and a smallparticle fraction which does not. The latter is destined to become overspray. This mechanism of overspray generation was previously identified [1, 2] for conventional airspray painting and, more recently, for airless spray painting [3]. 3.3.3 Wall jet formation The quasi-2-D outflow of entrained air and overspray in the ±x direction along the surface of the target is known as a “wall-jet” flow [11]. This classical fluid-dynamic phenomenon has much in common with a boundary layer, but has no freestream velocity. Turbulent motion in the wall jet may lead to some additional wall contact and deposition of the remaining paint droplets, though most end up as overspray. The planar wall jet is known to grow linearly in thickness with distance from the impingement point, and eventually to separate from the wall [12]. In this case the separation of the wall jet (not directly observed in present experiments) spreads the overspray generally into the surrounding atmosphere. 3.4 Overspray collection Since the overspray takes the form of a wall jet, one may consider how to manage this wall jet for the purpose of preventing the general dispersal of overspray to the environment. By analogy with a boundary layer, the wall jet must first be forced to

separate from the target surface, forming an overspray-bearing free shear layer. This stream of overspray-laden air must then be collected efficiently and ducted to a remote location, where filters, etc. can be used to remove the paint particles and return the cleaned air to the atmosphere. By such an approach the overspray is captured near its source, thus avoiding the release of any significant quantity of pollution to the atmosphere. However, referring to Figure 6, the static pressure at the spray impingement point O is expected to be greater than atmospheric due to the stagnation of the entrained airstream. This causes a decline in static pressure on the target surface in the ±x-direction, a favorable pressure gradient zone in which the wall jet is accelerated away from impingement, and in which it cannot separate from the target surface. External means must therefore be used to impose sufficient adverse pressure gradient in the ±x-direction to separate the wall jet from the target as described above. Assuming that this can be done, the separated wall jet would then be redirected from the x- to the z-direction, i.e. back in the direction from which the paint spray was generated. The overspray particles and collateral airstream would execute, in that case, a 180° turn which generates significant concentrated vorticity about its center of curvature (roughly point C in Figure 6). Alternatively, one may induce such a flow pattern by introducing a vortex core at this location. Since the entire flow is approximately 2-D, this must be a line-vortex oriented parallel to the y-axis, thus inducing rotation in the x-z plane. 3.4.1 Computational fluid dynamics The initial attempt to generate such a flowfield was made with the potential-flow code 2DFlowPlus. A reflection plane represented the target, sources were used to generate and “separate” the overspray-laden flow, a sink served to collect it, and a vortex singularity induced the required circulation. Crude though this is, the streamline results (Figure 7) provide a compelling image of a flowfield dominated by a strong vortical flow. Moreover, the boundary of the vortical region (blue-red interface in Figure 7) 1-5

GARY S. SETTLES, JAMES D. MILLER, THOMAS J. HARTRANFT, AARON D. BRANDT.

suggested a bell-shaped contour for a “shroud” with which to contain the overspray and force the appropriate pressure distribution upon the target surface.

Figure 7. Streamlines of simple potential-flow model of overspray collector in the x-z plane.

Figure 8. Flowfield model of overspray collector in the x-z plane. Based on this, an overspray collector concept was developed as shown in Figure 8. It comprises a shroud which surrounds and moves with the spraygun, maintaining a gap between itself and the target surface. Behind the spraygun and opposite the target, the collector shroud termi1-6

nates in a suction outlet through which oversprayladen air exits and makeup-air slots through which atmospheric air enters in sufficient quantity to minimize residual airflow through the aforementioned gap. The overspray-laden wall jet is intercepted by the collector shroud, forced to separate from the target, and directed to the suction outlet by the shape of the collector shroud. Next a more elaborate 2-D computation was carried out using the Fluent CFD code to solve the Navier-Stokes equations, with appropriate boundary conditions extracted from the flow observations described previously. The solutions thus obtained include viscous effects and turbulence via a turbulence model. However, neither this code nor any known CFD approach can solve the problem of spray atomization of paint. Thus, in specifying boundary conditions, atomization was avoided altogether by providing an input to the code downstream of its completion. Even there, the detailed distribution of particles after atomization is presently unknown, and the calculation of two-phase flows with interacting air and paint particles is well beyond the present scope. It was thus decided that an acceptable level of approximation could be obtained by ignoring the particulate phase altogether, but realistically specifying the entrained airflow which results therefrom. Streamlines from this solution, shown in Figure 9, showed significant promise that this overspray collector concept might function effectively. In particular, the computed static pressure distribution on the target surface shows the expected above-atmospheric value at spray impingement (point O in Figure 9) and atmospheric pressure at the gap location G. A pronounced minimum in static pressure at point M, however, explains how the device actually functions. From O to M in Figure 9 a favorable pressure gradient accelerates the wall jet in the +x-direction. But, due to the pressure minimum M on the target opposite the vortex core C, a strong adverse pressure gradient must exist from M to G. This causes the separation of the overspray-laden wall jet at point S, shortly before gap G, whence the separated flow now follows the shroud contour to the suction outlet. A slight inflow at the gap G

VISUALIZATION AND COLLECTION OF OVERSPRAY FROM AIRLESS SPRAY PAINTING

assures that no overspray will escape to the environment.

Figure 11. Visualized overspray collector 2-D model flowfield in the x-z plane.

Figure 9. Streamlines of Navier-Stokes solution of overspray collector flowfield in the x-z plane. 3.4.2 2-D experimental model results As described earlier, a 2-D “slice” of this device was then constructed and tested using an airjet containing zinc stearate powder to represent the airless paint spray. An image of this 2-D model is given in Figure 10. An appropriate suction level at the suction outlet was provided by a throttle-modulated hose to a vacuum cleaner,

Figure 10. 2-D overspray collector model.

and the PLS technique described earlier was used for flow visualization. Note the use of a splitter plate to physically separate the suction outlet from the makeup air inlet. The 2-D model input airflow flow rate at the spraygun location (and hence the time scale for jet impingement and wall-jet formation) was varied by adjusting the airjet supply pressure at the nozzle. The model was tested over a range of supply pressures while videotaping the flow pattern. Frame-by-frame examination of the visualized flow then showed which supply pressure corresponded to a flow time scale similar to that of the actual spray painting flow visualizations discussed earlier (e.g. Figure 4). All subsequent model tests were then made at that supply pressure setting. As in the case of the Fluent™ calculation, suction outflow was then set so as to eliminate any outflow from the wall gap, and to ensure a slight inflow for positive overspray containment purposes. The key variable in these tests was the gap width between the collector and the target surface, which is shown as 2.5 cm at location G in Figure 9. This was a starting point, but a larger gap is clearly desirable in practice to account for surface protrusions on the target, etc. In fact, gap widths up to 15 cm were found to be workable, with an optimum value determined to be about 67 cm. The wall jet was observed to separate from 1-7

GARY S. SETTLES, JAMES D. MILLER, THOMAS J. HARTRANFT, AARON D. BRANDT.

the target and attach to the collector shroud via a free shear layer that showed notable stability over a broad range of gap widths, thus lending robustness to the overspray collector concept. A representative flow visualization image from these test is shown in Figure 11. A striking similarity is seen between the computed flow pattern of Figure 9 and the experimental flow pattern of Figure 11. This is taken as a general verification of the CFD results, and of the overall concept of the present overspray collector. Note, however, that the gap width in Figure 11 is 7.5 cm vs. 2.5 cm in Figure 9. The free shear layer which “jumps” the gap is clearly visible in the upper left corner of the Figure 11. A search of the relevant literature showed no similar approach to overspray collection. However, an “air curtain” approach did appear [13], in which a collector housing was surrounded by an airstream aimed toward the target. 3.4.3 A practical overspray collector Fig. 12 is an isometric drawing of a practical embodiment of the present overspray collector. For purposes of illustration, two sprayguns (2A and 2B) are shown, so arranged as to coat a band of dimension in the y-direction roughly twice that which is possible with a single spraygun. Since the flowfield within the collector, as demonstrated earlier, is approximately twodimensional in the x-z plane, the collector is simply terminated at top and bottom by the end caps (20) shown in Figure 12. The right half of the top end cap in Figure 12 is colored green, corresponding to the +x half-plane view of the collector flowfield shown above in Figures 7-11. Outlets (5A) and (5B), for the removal of overspray from the left-hand side of the collector, are shown unobscured in Figure 12 for clarity. The corresponding right-hand outlets have suction manifolds (21A) and (21B) connected to them, which then connect to suction hoses (22A) and (22B) for conveying the collected overspray away to a filtration device (not shown).

1-8

Figure 12. Isometric view of a practical embodiment of the present overspray collector. Adjustable louvers, vanes or dampers (23) modulate the flow of makeup air into the collector. This is required in order to regulate makeup air entrainment and to control the static pressure inside the collector, thus to optimize its operation. Structural member (24) supports and provides a hard connection point for the overspray collector. Actuating arm (25) connects to this structural member and effects uniform motion of the overspray collector over the target surface being coated. In contrast to spray-booth-type overspray treatments, the present collector is able to move with the spraying device, e.g. in the spray painting of large flat surfaces by way of a roboticallycontrolled traversing mechanism.

VISUALIZATION AND COLLECTION OF OVERSPRAY FROM AIRLESS SPRAY PAINTING

4 Conclusions The phenomenology of “airless” spray painting of a water-borne latex paint has been examined in this paper by flow visualization. Results show that the process combines the known fluiddynamic phenomena of wavy-sheet breakup, spray entrainment, spray impingement, inertial droplet impaction, and the turbulent wall jet. A quasi-two-dimensional flowfield is found, which lends considerable simplification to the problem. In view of this, the question was considered of how the overspray-containing wall jet might be induced to separate, roll up, and be drawn away in a particular direction where it can eventually be filtered to remove the overspray particles. A simple potential-flow model suggested the use of an aerodynamic shroud for this purpose. A moreaccurate Navier-Stokes solution then revealed the true function of such a shroud: the generation of a columnar vortex which induces the separation of the overspray-laden wall jet by way of an imposed adverse pressure gradient. After separation from the target surface, the shroud then conducts the overspray-laden airflow to suction ducting whence it is eventually treated. The design principles of an aerodynamicallycorrect overspray collector for airless spray painting have resulted from this research. Results shown here are now being used to design a fullscale overspray collector prototype. 5 Acknowledgments This research was supported by the US Navy through the Penn State Applied Research Laboratory.

Atomization and Spray Systems, Ottawa, pp. 145-149. May 18-21, 1997. 4. Settles GS. "An Overview of Planar Laser Scattering for the Visualization of High-Speed Flows." in Flow Visualization VI, ed. Y. Tanida et al., Springer-Verlag, pp. 628-633. 1992. 5. Lefebvre AH. Atomization and Sprays. Hemisphere Pub. Corp., NY, ch. 2, 1989. 6. Dexter RW. “Measurement of Extensional Viscosity of Polymer Solutions and its Effects on Atomization from a Spray Nozzle.” Atomization and Sprays, Vol. 6, pp. 167-191, 1996. 7. Mansour A and Chigier N. “Air-Blast Atomization of non-Newtonian Liquids.” Non-Newtonian Fluid Mechanics, 58, pp. 161-194. 1995. 8. Matta JE and Tytus RP. “Viscoelastic Breakup in a High Velocity Airstream.” J. Applied Polymer Science, 27, 397-405. 1982. 9. MacGregor SA. “Air Entrainment in Spray Jets.” Intl. J. Heat and Fluid Flow, Vol. 12, No. 3, pp. 279-283. Sept. 1991. 10. Wen F, Kamalu N, Chung JN, Crowe CT and Troutt TR. “Particle Dispersion by Vortex Structures in Plane Mixing Layers.” J. Fluids Eng., Vol. 114, pp. 657666. Dec. 1992. 11. Glauert MB, “The Wall Jet.” J. Fluid Mechanics, Vol. 1, Part 5, 1956, pp. 625-643. 12. Daws LF. “Movement of Airstreams Indoors.” J. Inst. Heating & Ventilation Engrs., Vol. 37, pp. 241-253, 1970. 13. Hockett WB. “Enhanced Recovery System.” US Patent 5,489,234. Feb. 6, 1996.

6 References 1. Kwok KC and Liu BYH. “How Atomization Affects Transfer Efficiency.” Industrial Finishing, pp. 28-32, May 1992. 2. Hicks PG and Senser DW. “Simulation of Paint Transfer in an Air Spray Process.” ASME FED-Vol. 178/HTD-Vol. 270, pp. 145-154, 1993. 3. Settles GS. “A Flow Visualization Study of Airless Spray Painting.” Proc. 10th Annual Conf. on Liquid 1-9