Znd. Eng. Chem. Res. 1987,26, 1645-1653 Superscript * = denotes best-fit criterion
Literature Cited Carter, J. W.; Barrett, D. J. Trans. Znst. Chem. Eng. 1973,51,75-81. Cassidy, R. T.; Holmes, E. S. AZChE Symp. Ser. 1984, 80(233), 68-75. Chihara, K.; Suzuki, M. J . Chem. Eng. Jpn. 1983, 16, 53-60. Garg, D. R.; Yon, C. M. Chem. Eng. Prog. 1986, 82(2), 54-60. Kaguei, S.; Yu, Q.; Wakao, N. Chem. Eng. Sci. 1985, 1069-1076. Kayser, J. C.; Knaebel, K. S. Chem. Eng. Sci. 1986,41, 2931-2938. Kayser, J. C.; Knaebel, K. S. Chem. Eng. Sci. 1987, in press. Knaebel, K. S.; Hill, F. B. Chem. Eng, Sci. 1985, 40, 2351-2360.
1645
Kowler, D. E.; Kadlec, R. H. AZChE J . 1972, 18, 1207-1218. Matz, M. J. M.S. Thesis, The Ohio State University, Columbus, 1986. Miller, G. W.; Knaebel, K. S.; Ikels, K. G. AZChE J . 1986, in press. Pan, C. Y.; Basmadjian, D. Chem. Eng. Sci. 1970, 25, 1653-1664. Sircar, S.; Kumar, R.; Anselmo, K. J. Znd. Eng. Chem. Process Des. Deu. 1983, 22, 10-15. Yang, R. T.; Cen, P. L. Ind. Eng. Chem. Process Des. Deu. 1986,25, 54-59. Yoshida, H.; Ruthven, D. M. Chem. Eng. Sci. 1983, 38, 877-884.
Received for review July 29, 1986 Revised manuscript received March 23, 1987 Accepted April 25, 1987
Dominant Mechanisms for Color Differences in the Mechanical and the Electrostatic Spraying of Metallic Paints Stuart L. Inkpen and James R. Melcher* Laboratory for Electromagnetic and Electronic Systems, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Color differences in metallic paints sprayed by traditional mechanical sprayers and newer electrostatic sprayers have motivated the detailed investigation of these devices. The dominant mechanism for this color difference has been identified. T h e amount and size distribution of the metallic flake deposited by the electrostatic sprayer is significantly different than that deposited by the mechanical process. For the devices investigated, the mass percent of flake and average flake size deposited by the mechanical sprayer are respectively twice and more than twice what they are for the electrostatic sprayer. This is a consequence of the low efficiency of the mechanical sprayer. These results imply that for color control with low-efficiency sprayers (mechanical), cognizance is required of the disparity between input and workpiece flake content and size distribution and that for highly efficient processes (electrostatic), a new approach to color control is required.
Background The painting industry is responding to environmental concerns by using electrostatic sprayers to increase the deposition efficiency. In the automotive industry, the efficiency is between 10% and 40% for a mechanically applied paint coat but between 60% and 90% when applied electrostatically. By use of these two different processes, most paints can be matched by varying standard painting parameters. However, electrostatic deposition of metallic paints tends to result in a darker appearance. This is often ascribed to alignment of the conducting flake by the electric field. Although this mechanism is not eliminated as a contributing factor, it is shown here that this is not the dominant mechanism, at least for the specific equipment studied here. Rather, it is found that the selective deposition of drops by the mechanical sprayer (that is, minimized in the electrostatic sprayer) is the dominant mechanism. Possible Mechanisms for Electrically Induced Color Variation The basic configuration of a high-speed turbobell electrostatic painter is shown in Figure 1, where four process stages are distinguished (a) formation of drops, (b) flight to the workpiece, (c) impact with the workpiece, and (d) postimpact. Whether the atomization and delivery are purely mechanical or purely electrostatic, or any combination, these stages are potential contributers to the coloration of the final product. The spray devices chosen for this project were a Binks hand-held sprayer and a RansOSSS-5S85/S7/2626-1645$01.50/0
burg Turbobell. The Binks spray gun uses mechanical atomization and delivery, while the Ransburg Turbobell uses both electrostatic and mechanical atomization and delivery. There are four mechanisms proposed that may contribute to the electrically induced color variation: (1)flake orientation (formation, flight, impact, and postimpact), (2) electrophoresis of the flake or other pigments (postimpact), (3) differential evaporation of solvents (flight), and (4) variation in the amount and size distribution of deposited flake (flight). The final angle of the flake has a large effect on the color (Wojtkowiak, 1983). I t is possible that this orientation could be modified in any of the paintingstages by electrical forces, although the earlier the stage, the less likely the persistence of the effect. Electrically induced flake orientation is briefly considered in the next section, where it is found unlikely that it is the dominant mechanism for electrically induced color variations. Electrophoresis of the flake as well as the other pigments would cause their spatial variation and a resultant change in color. This requires an electric field in the paint and a substantial double layer surrounding the particles. Experiments show that for the paints used, electrophoresis requires field strengths on the order of lo4 V/m for noticeable motion of the flake or pigments in times of interest. Fields in the paint are typically estimated to be 2 orders of magnitude less. If there were a mechanism for imposing a high electric field in the paint, such as a substantial corona current in the air, electrophoresis would 0 1987 American Chemical Society
1646 Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987 Table I. Properties of Unmixed Paint / ~
- - - - ---I
I
freq, Hz 100 1 X lo3 1 X lo4 strain rate, s-l
0
1 2
Point
Flight F o r m a t ion
Figure 1. Stages of paint spraying.
be a viable mechanism. However, corona currents are not commonly found at the workpiece in properly operated painting processes. The higher efficiency of the electrostatic paint process does give a wetter (richer in solvent) paint layer. Experimentally it has been shown (Inkpen, 1986) that the paint used has a total loss in mass during the application and curing process of 60%. During mechanical application of the paint, the loss is 34%; electrostatic application results in a loss of 26%. The mechanical process loses more solvent during flight than the electrostatic process. The possibility exists that this difference in evaporation causes a color variation. However, the difference, approximately lo%, is thought unlikely to cause the color variation observed. One of the benchmarks established in electrostatic painting is a report by Bell and Hochberg (1981). Through private communication, they have proposed a mechanism that attributes the color variation to an increase in the fraction of nonmetallic solids ending up on the workpiece. This has the effect of creating a lower metallic flake content in electrostatically applied paint. Their postulate is based on the following reasoning: Given that small drops are entrained in the air flow and generally constitute the majority of the overspray (waste) and that the electrostatic process deposits more small drops than a standard mechanical sprayer, and if it is true that large drops carry more metallic flake per unit mass than small drops, then the electrostatically applied paint has a lower content of flake. Most of the following gives this postulate quantitative verification, documenting not only the electrical effect on the total flake content, but on the distribution of flake as well. A typical sprayer has numerous operating parameters to adjust. Each parameter can have an effect on each stage of the spraying process, and variations in these inputs are likely to be interrelated. The fluid mechanics and paint atomization are strong influences in the steRs of this postulate. The efficiency of the process is as much a function of the drop distribution and fluid flow as the application of electric forces. For example, the efficiency of a typical mechanical sprayer can be easily changed by varying the pressure of the shaping air and that applied to the atomizing nozzle. As the shaping air velocity and the atomization pressure is increased, the efficiency typically decreases because the number of small drops increases and the higher air velocity causes more turbulent mixing. The effects of varying these two parameters are interrelated, as both cause finer atomization and change the velocity profile in the spray as well as the degree of turbulence. Most parameters are similarly related and affect the atomization, evaporation, and transport. The operator of a spray device can also adjust the contents of
5 10
flake paint no-flake paint u , S~/ m +/toa u , S/m ~ 7.28 1.02 X lo4 8.88 1.17 X lo4 6.52 1.03 X lo4 7.96 1.18 X lo4 6.38 1.05 X lo4 7.44 1.20 X lo4 shear stress/strain rate, N.s/m2
tr/tOa
0.67 0.51 0.39 0.3 mass density, kg/m3 928
Dielectric constant.
0.38 0.32 0.27 0.25 916
Conductivity.
the paint, usually by the addition of solvents, again affecting most processes in the system. A more complete discussion of the physical processes at work in mechanical and electrostatic paint sprayers than can be given here is available (Inkpen, 1986). Color Variation Using Flaked P a i n t s The amount of color variation between the two processes, mechanical and electrostatic, is very dependent on the specific type of device and the particular paint. There are, however, general trends that are obvious. Electrostatic paint finishes tend to be darker and exhibit more of the color of the base paint, i.e., the paint material with the flake removed. One of the paint colors that exhibits the largest variation is light blue. Differences in the two colors can be quantified by using the color vector (MacAdam, 1985). The coordinates in color space.(CIE Lab) measure the normal perception of whitelblack, L*, redlgreen, a*, and blue/yellow, b*. Typical results (Du Pont nonaqueous automotive enamel K-383Y- BR055) are as follows: electrostatic sample, L* = 41.6, a* = -3.8, b* = -28.8; mechanical sample, L* = 52.3, a* = -3.4, b* = -19.4. The L* for the electrostatic sample is lower than the mechanical sample; the sample is darker. The coordinate b* has greater magnitude, indicating a bluer color. The total difference, E = 15.8, is the distance in color space between the two samples (the average person can detect a E N 0.9). Orientation of Flake Because orientation requires a finite time, it is most likely to contribute to the color in the stages of greatest duration, flight and postimpact. The characteristic time for electrical orientation of a flake in a Newtonian fluid, ignoring inertial effects, is the electroviscous time (Melcher, 19811, T,, = VICE2 ( V is the absolute viscosity, e is the electrical permittivity, E is the electric field). The time for the relaxation of charge is 7, = e / u (e is the permittivity and is the electrical conductivity), less than 0.1 ms in paint. (Properties of paint used in these studies are summarized in Table I.) Even though a drop in flight is subject to a large dc field, the paint charge relaxation time is so short that there is a field inside for only a small fraction of the flight time. Thus, flake or no flake, there is little torque on a spherical drop due to the dc field. If a flake is large enough to distort the drop to an oblong shape, so that it forms a "wetted flake", the dc field can induce a torque. Indeed, the electroviscous time based on the dc field and the viscosity of air is short compared to a flight time of about 0.1 s. However, the results of experiments discussed later show the number of wetted flakes is likely to be insignificant.
Ind. Eng. Chem. Res., Vol. 26,No. 8,1987 1647
. .
I
Figure 3. Analysis of electrostaticsprayer showing conclusions in bold boxes with quantitativeand qualitative corroborationsindicated by heavy and broken lines, respectively.
Figure 2. Cmsa &ion of paint layers (a. top) mechanically sprayed and (b, bottom) electrostatically sprayed.
A spherical drop enveloping one or more flakes could experience a time-average torque because of the ripple component of the electric field. Estimates of the electroviscous time, based on the viscosity of the air and the fraction of the field penetrating the drop at 120 Hz of the measured 10% ripple, are about equal to the flight time (Inkpen, 1986). However, because the electrical torque is generated by flakes inside the drop, T~ should he multiplied by the cube of the ratio of the drop radius to a dimension typical of the flake size t o estimate the actual orientation time. Even more, any in-flight orientation would have to persist through impact to have an effect. Thus, i t seems unlikely that in-flight orientation has a significant effect on coloration. Postimpact orientation is governed by the viscosity of the deposited paint and the field in the paint. There is some dc field in the layer resulting from the current due the charge carried by the paint drops. This current c a w an electric field that depends only on the current density and the conductivity of the paint. These quantities are reliably measured or estimated and lead to an electroviscous time on the order of tens of hours, much longer than any residence time in the spray zone. [However, an imposed corona current has been used to obtain substantial field in the paint and vertical flake alignment (Inkpen et al., 1987).] Time-varying fields do penetrate the deposited paint. By use of the power supply ripple, the same calculation as done previously for a spherical drop shows that there is a relatively small ac field in the deposited layer. Even more, the flake can no longer transmit ita torque to the paint/air interface but must rotate relative to the paint. The viscosity of paint is 4 orden of magnitude higher than that of air and the resulting electroviscous time is larger than any reasonable residence time of the sample in the field. An ac postdeposition flake alignment has been demonstrated, but only by raising the frequency and field strength and reducing the viscosity of the paint (Inkpen et al., 1987). Cured samples of both mechanically and electrostatically applied paint layers were sliced to expose the cross section. After this surface was polished to 0.1 am. the flakes could be oberved in the paint layer. T w o sample cross sections are shown in Figure 1. The angles between the flakes and the substrate, for flakes larger than 2 pm in length, were measured. The mean angle for the mechanically sprayed sample is 17.8 pm with a standard error of 2.4 pm. while the mean angle for the electrostatically sprayed sample is
Figure 4. Analysis of mechanical sprayer showing conclusions in bold boxes with quantitative and qualitative corroborations indicated by heavy and broken lines. respectively.
15.8 r m with a standard error of 1.8 pm. There is no evidence of vertical flake alignment in the electrostatic sample. Note, however, that the average flake angle is small; the flakes are almost parallel to the painted surface. This indicates that the impact dynamics may well dominate flake orientation in both painting processes. From Figure 2. it should also be noted that the electrostatic sample appears to have less flake present. This could result in two ways. First, the flakes could be aligned. Thus, in a single slice. less flake would be present (however, the occasional flake standing on end would he expected and is generally not observed). Second, there is simply less flake in the layer. This is pursued in the following sections.
Flake Selection: The Dominant Mechanism for Discoloration Investigation of flake selection as a mechanism requires data on the toiA amount of flake actually making it to the surface. Details of this mechanism are obtained from knowledge of the drop size and the amount of flake carried by differing drop sizes. For the electrostatic sprayer, where almost all of the spray reaches the large planar workpiece used, this information will be found as a function of position on the workpiece. For the mechanical sprayer, where much of the spray never reaches the workpiece and what does is confined to a relatively small area, one spaceaverage sample is made at the workpiece and the overspray is sampled at three locations. Outlines of the investigation presented in the next sections are given by Figure 3 and Figure 4 for the elec trmtatic and the mechanical sprayers, respectively. Of the three experimental approaches listed in the left columns
1648 Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987
of boxes, the first (sampling of paint cross sections and microscopic estimates of flake angle and density) has already been described. Because the electrical sampling resolves the spatial distribution of drops and flake on the workpiece while the mechanical sampling places more emphasis on the overspray, the second and third approaches to the electrostatic sprayer (the second and third boxes in the first column in Figure 3) differ from those for the mechanical sprayer (Figure 4). Heavy solid and broken lines, respectively, indicate quantitative and qualitative corroborating evidence.
Analysis of Drop and Flake Distributions for an Electrostatic Sprayer Once dropg have been sampled in either the electrostatic or mechanical sprayer, the drops are then cured before the data are taken. At the expense of ignoring the evaporation data for each drop size, this avoids problems of evaporation during processing. Electrostatic Paint Sprayer. The Ransburg Turbobell I1 (Figure 1) was used with a 70351-01 bell. It was driven at a nominal 28 000 rpm, resulting in an edge speed of 108 m/s with a bell voltage of -90000 V. In this device, paint is forced at a controlled rate onto the inner surface of the bell where the centrifugal force causes the formation of a thin sheet that flows to the serrated edge of the bell. At the bell edge, a combination of electrical and fluid forces causes the sheet to break into drops. These drops are then carried to the surface by the air flowing around the outer edge of the bell (shaping air) and by the electrical forces. The high angular velocity at the bell results in large radial momentum that causes the drops’ initial trajectory to have a large radial component. Because of the short charge relaxation time, induction charging of the drops dominates with any corona emission effectively eliminated once steady-state spraying is achieved. The grounded workpiece was 0.3 m from the bell tip, the paint flow rate was typically 100 mL/min, and the shaping air pressure was maintained at 20 psi. This pressure resulted in a maximum axial air velocity a t the bell tip of approximately 20 m/s. The deposited spray pattern was approximately 35 cm in radius, and this area was “coveredn in roughly 45 s. Sampling Drops at the Paint Surface. To sample drops impinging at a given location on the workpiece, an electrically shielded sampler was developed. To prevent drop deposition, a sampling plate is insulated from the rest of the grounded target and taken to a potential, V. The same polarity is used as for the bell. This creates a field near the surface that prevents the charged drops from impacting the sampling plate. This plate is then momentarily grounded, collapsing the electrical shield, to sample the spray cloud. Thus, during the sampling period, field and flow are not modified by the device. Although restricted to the study of charged drops, the electric shutter has the advantage of being switched off essentially instantaneously, of not interfering with the gas flow during the critical time just after it is switched off, and of leaving the test surface essentially as it is on the actual workpiece while the sample is being taken. The field shape around the high-potential electrode, assuming it is long compared to its width, was calculated numerically by using the method of charge simulation and is shown in Figure 5. This figure illustrates the strong field at the edges of the plate in comparison to the weaker field at the center. Moving outward along this center line perpendicular to the electrode, the field decreases, eventually reaching zero, at a critical point, x = no in Figure 5, where the field due to the electrode balances that due to the spray equipment.
Figure 5. Electrical shutter: electric field plot.
1
0.5
> \
-+
w
0
I
%/.e (X1-t)
-
2
3
Figure 6. Electrical shutter: x-directed electric field
The details of the drop trajectories depend on the specific gas velocity and the drop conditions in the vicinity of the electrode. However, for design purposes, it is along this center line that the “electric umbrella” provided by raising the potential of the sampling plate is most likely to “leak”. If inertia is ignored, and in the absence of particle drag, the drops will follow the field lines exactly. Therefore, ensuring the critical point exists provides a lower bound on the shutter potential, V, required to effectively shield the drops. Figure 6 shows a plot of the normalized x-directed electric field strength due to the sampling electrode as a function of the distance from the center of the electrode. For a critical point to exist, the applied field from the painter, E,, must be less than the maximum field in Figure 6, in normalized units approximately 0.6. Therefore, V E, < 0.6- or _v > 1.67EP 1 1 where 1 is half the narrowest width of the electrode. Because the fluid flow at the surface is zero, the effects of the flow will not change the validity of this result. Inertia, however, will have an effect on the larger drops. An estimate taking into account the effects of inertia on the larger drops as they find themselves in the decelerating flow near the workpiece (Inkpen, 1986) shows that for practical shielding, E,l/V should be about 0.2 so that roll is about 1.4 (as illustrated in Figure 6). Practically, the shutter potential is limited by the insulation strength of the gap between the test area and the rest of the workpiece. Corona emission and the possibility of arcing must be avoided. The plates used were at 6.5 kV and consisted of conducting glass microscope slides. The effectiveness of the shutter was tested by exposing two test
Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987 1649
-,,
number distributions and statistics on subsets of these distributions. The drop number distribution function, D, is a function of the drop size, diameter, w , and the total flake area in a drop, fd. 'I
- ' a b -
D
= D(W,/d)
(2)
Therefore, the number of drops, Pd, with diameters between w and w + A is
Similarly. the flake number distribution, F, is also a function of the drop size and the flake area, f.
Figure 7. Bark-illuminated drops.
plates to the spray cloud for approximately 20 s and grounding one plate for 0.25 8. The density of splattered paint on the samples was compared. The ratio of the area density for the nongrounded sample to the grounded sample was 0.18. In the experiment, the samples were subjected to the spray cloud for 10 s, indicating there should be less than 10% error. Actually, the error was substantially less because the data were taken from the edge of the sample where the shield was much more efficient. The design approximations were found to give good estimates of the shielding performance. Data Processing Technique. The test plates are glass slides coated with tin oxide to make one surface conducting. This allows the drops to be examined under a microscope using both front and back illumination. A photograph of a back-illuminated sample at a magnification of 200 is shown in Figure 7. The flake appears opaque in these pictures, making it easily identified. Yet, at the same time, the drop outline is clear. Information can be obtained on both the flake size and the drop size as well as the amount of flake in a given drop. To assemble these data for statistical analysis, the drops and associated flakes were digitized by using an electromagnetic tablet. The drop and flake area were measured by estimating a diameter that was judged to best represent the area. This greatly aided in the speed of processing the data and suggests how best to automate the method by using computerized pattern recognition equipment. The raw data are in terms of the splattered drop area and the flake area. However, the masses of the drop and the flake are of most interest. Given the dried density of the paint, this can be obtained from the volume of each drop and flake. The flake, as seen in the cross section of Figure 7, is approximately uniform in thickness with an average thickness of 0.75 Nm. Therefore, flake volume is approximately equal to flake area multiplied by 0.75 Nm. Drop volume is more difficult to calculate as the drop profile depends on the splattered drop radius. Drop profiles were measured for a range of drop sizes by using a Dektak surface analyzer. The measurement tip of this device has a radius close to the drop sizes measured, requiring that the profiles be corrected to account for the finite size of the tip. The average height was fitted to a fourth-order curve in drop radius. The data were processed by computer t o obtain the drop size distribution, the flake size distribution, the mass percent of flake in a drop interval and for the entire data set, the percent of total flake, the average flake area, and the total mass. Also, basic statistics on the entire sample such as averages, standard deviations, and an estimate of their error were calculated. The data represent discrete points from two
F = F(w,f) (4) Therefore, the number of flakes, Iy0 within drops having diameters between w and w + A is (5) All data can be written in terms of these two distribution functions, and appropriate scaling functions. For example, mass in the drop diameter interval w to w + A as a percent of the total mass in a sample is
where C ( w ) is the drop height as a function of diameter, to is the thickness of the flake, pp is the density of dried paint, and pAI is the density of aluminum. Results of Electrical S h u t t e r Experiment. Samples were taken a t the paint surface, 0.3 m from the bell, a t seven radial positions. A sample was taken every 10 cm, starting at the center of the target and moving horizontally to the right. The samples were numbered consecutively, starting at the outer edge of the pattern and moving into the center. Therefore, the sample at the center was no. 5. The other two samples were taken at 40 cm to the left of center and 20 em up from the center, providing a check on the symmetry of the spray pattern. The experiment was carried out using paint with flake and the same paint with the flake removed. Mass as a percentage of the total on all samples is plotted in Figure 8a for flake paint and for no-flake paint. The mass percentage of aluminum flake is plotted in Figure 8b as a function of radius. In practice, an object is usually painted by passing it through the spray cloud. Therefore, it experiences a mass average of the spray cloud over radius. This average, from the sampled data, gives a mass percent of aluminum flake equal to 1.82%. Figure 8c is a plot of average drop size vs. radius for flake paint and no-flake paint. These curves have the same trend as Figure 8a, mass vs. radius, increasing toward the center and then decreasing slightly a t the center. Note the very different average drop sizes between flake and no-flake paint. Figure 8d is a plot of the average flake area vs. radius. The average flake area has approximately the opposite trend as the mass and drop size, increasing toward the edge of the spray pattern and then decreasing at the
1650 Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987
30m
Table 111. Drop D a t a f o r Electrostatic Spraying of No-Flake P a i n t
3 20 (a)
Drops std dev = 18.6 pm av diameter = 33.4 pm total no. of drops = 1282 with probability of 95% error in the mean = 1.02 std dev between 18.14 and 19.16
Mass Data total mass: drops = 7.81 119 data for drop intervals no. of drops 70 of total mass % of total max droD diameter 2.57 6.71 7.41 9.98 10.76 9.91 10.53 9.91 7.51 5.23 19.42
x no-flake
10
B
0 flake
t
I
I
I
I
T
0.00 0.05 0.25 0.90 1.96 3.15 5.42 7.64 8.13 7.69 64.81
5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 m
30tI zotI I Figure 8. Results of electrical shutter experiment as a function of radial position on the paint target: (a) total mass, (b) mass percent of aluminum, (c) average drop diameter, (d) average flake area. Table 11. Drop a n d Flake D a t a for Electrostatic Spraying of Flake P a i n t Drops std dev = 13.4 pm av diameter = 12.9 pm total no. of drops = 3860 with probability of 95% error in the mean = 0.42 std dev between 13.21 and 13.64
IO
20 30 40 50 Droo Diameter Cum1
IO
20 30 40 50 Drop D i o m e t e r C p m l
h
Flakes std dev = 71.7 pm2 total no. of flakes = 1234
av area = 33.1 pmz with probability of 95% error in the mean = 4.00 std dev between 69.78 and 73.78
Mass Data flake = 8.28 total mass: drops = 4.55 pg % flake bv mass = 1.82 data for drop intervalsn 1
2
3
4
5
37.07 23.89 9.53 8.06 5.78 4.43 3.68 2.23 1.66 1.22 2.46
O.OO*b
0.17 0.75 1.61 3.63 5.27 7.07 9.63 8.79 9.28 9.26 44.54
0.00 0.01 0.20 0.75 1.60 4.37 6.07 7.03 11.23 8.19 60.25
0.00 0.03 0.22 0.37 0.55 1.12 1.15 1.52 2.20 1.61 2.46
0.08* 0.97 2.76 4.54 9.64 11.18 11.83 12.07 9.08 37.84
pg
X
6 4.5 6.7 9.0 11.7 15.0 18.0 20.6 30.8 29.9 52.8
~li 20
10
7 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 m
“Key: 1 = number of drops as a percent of the total; 2 = number of flakes as a percent of the total; 3 = mass as a percent of the total; 4 = flake mass as a percent of the total flake mass; 5 = mass percent of flake; 6 = average flake area; 7 = maximum drop diameter for this interval. *In this and other tables, the asterisk indicates the intervals where the number of flakes or drops is less than 10.
extreme edge. It should be noted that the standard deviation of the average flake area has this same trend. The
50
100
150
200
250
Flake A r e a Cpm21
Figure 9. Electrostatic sprayer: deposited distribution of paint drop diameter for (a) flake, (b) no flake, and (c) deposited distribution of flake area.
plot of mass percent of aluminum, Figure 8b, exhibits a similar trend as Figure 8d for flake area, increasing toward the outer edge of the spray cloud. The error bars on the graphs represent the error in the mean with a probability of 95 7 0 * The statistics for the sum of the five test positions, representing the mass average, are shown in Table I1 for flake paint and in Table I11 for no-flake paint. Care should be taken in interpreting the data, as some of the intervals have an insignificant number of data points. The intervals where the number of flakes or the number of drops is less than 10 are marked with an asterisk. Table I1 for flake paint shows that in drops smaller than 15 pm, the flake content is extremely small, almost nonexistent. These data
Ind. Eng. Chem. Res., Vol. 26, NO. 8, 1987 1651 are used to plot the drop distributions for flake paint, Figure 9a, and for no-flake paint, Figure 9b. Not only are the average drop sizes much different but the drop distribution for no-flake paint is almost flat in comparison to the distribution for flake paint. The flake area distribution is shown in Figure 9c; note the high percentage (72%) of flakes with an area less than 50 pm2.
Analysis of Drop and Flake Distributions of a Mechanical Sprayer Because the drops are not charged, sampling in the case of the mechanical sprayer is actually more difficult than for the electrostatic sprayer. The trajectory of charged drops is easily modified with the application of an electric field. The specific geometry of the spray gun and the mechanisms used in atomization and delivery of the paint greatly effect the sampling technique. Once again, a description of the spray gun is appropriate before the sampling technique is described. Description of Mechanical Spray Gun. The mechanical spray gun is a Binks Model 98-1081. The gun atomizes the paint by forcing it through a cylindrical nozzle into a region where the paint is impacted by high-velocity air. Approximately 1-cm downstream from the atomizing region, the spray is shaped by several jets of air. With the target approximately 10 in. from the nozzle, the resulting spray pattern on impact is approximately elliptical with a major axis of 14 cm and a minor axis of 4 cm. Although this pattern represents the dense part of the spray, paint is not confined solely to this region. The waste paint outside the main spray constitutes the overspray and accounts for the low efficiency of the process. Delivery of the paint to the target is due to drop entrainment in the air flow. This air flow is highly turbulent and has a high average velocity, 30 m/s, 7.5 cm from the nozzle. The basic operating parameters are paint pressure, air pressure, nozzle opening, air valve opening, and the distance to the paint target. The paint pressure and air pressure are easily controlled by conventional methods (they were usually set to 10 and 30 psi, respectively). During spraying, the spacing between the target and the gun were kept in the range 8-10 in. Drop Sampling. To obtain a picture of the spray cloud, including the now significant overspray, it was sampled a t four positions, three a t various positions between the spray gun and the paint target, and one in the main spray cloud at the paint target. The sample a t the paint target was an average of the main spray pattern. It was obtained by passing the gun over the target fast enough so that a coating of discrete drops was obtained. Small electrostatic precipitators were used to sample the cloud between the gun and workpiece. These devices are shown in Figure 10. The exit of the precipitator is connected to a variable pump so that the entrance velocity can be approximately matched to the velocity in the absence of the precipitator. The three positions between the spray gun and the paint target, also shown in Figure 10, are numbered in order of increasing distance from the spray gun, 4 being on the paint target. In order to obtain paint that would normally constitute the overspray, the precipitators are outside of the main spray stream. The flow is predominantly radial at these three positions, directly into the precipitators. Results. As in the electrostatic sprayer, samples were taken using both flake and no-flake paint. The spray gun parameters were adjusted by painting several test plates before the experiment. Because the overspray was precipitated while the gun was in a fixed position, no feedback on the quality of the painting was possible. Also, for the
2 cm
1
lOcm
Conducting Glass Slide
Figure 10. Details and positions of electrostatic precipitators.
sample at the paint surface, a single pass of the spray gun made it impossible for the operator to ensure appropriate gun settings. The quality of the spray was totally dependent on the previous spraying of the test samples. The same processing technique as described for the electrostatic sprayer was used. Each position in the overspray can be treated as discrete information. However, the sample at the paint surface is an average of the spray making it to the target and should represent the statistics of a normal paint finish. The average drop diameters for samples 1-3 are similar, regardless of the presence of flake, with an overall average of 6.4 pm. Sample 4,at the surface of the paint target, has an average drop diameter of 36 pm with flake and 38 pm without flake. The standard error for both of these latter samples is approximately4 pm, indicating the average drop diameters are not significantly different. The aluminum mass percent is 0.99% for sample 1,0.32% for sample 2, 0.80% for sample 3, and 1.94% for sample 4. The drops in the overspray (collected as samples 1-3 and found to be small) contain very little flake, while the drops deposited on the paint target (collected as sample 4 and found to be large) carry a high flake content. Data from sample 4 are shown in Table IV for flake paint and in Table V for no-flake paint. The drop distribution for sample 4 is shown in Figure l l a for flake paint and in Figure I l b for no-flake paint. The flake area distribution a t the workpiece is shown in Figure l l c .
Average Flake Deposition and Mass Balance Atomic absorption measurements were used to provide a check on the average (overall drop sizes a t a given location) percent of flake (the middle row in Figures 3 and 4). These data, augmented by information on the total mass loss during the painting and curing process, provide a check on the mass balance found by averaging over the data taken from the drop-flake studies. Atomic Absorption Measurements. In preparation for atomic absorption measurements, samples were painted on stainless steel slides and then cured so that results could be expressed as a mass percent of dried solids. The paint (flake included) was then scraped off the substrate, and dissolved. Aluminum content of the solution and hence of the solids was then determined by atomic absorption. Samples were painted by using both the electrostatic and mechanical sprayers. The samples painted electrostatically produced consistent values of 1.86% f 0.02%. The mechanical sprayer, however, provided samples with a wide range of aluminum content, between 2.02% and 3.31%.
1652 Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987 Table IV. Drop and Flake Data for Mechanical Spraying of Flake Paint Drops av diameter = 36.0 pm with probability of 95% error in the mean = 3.76 std dev between 27.28 and 31.06
:
std dev = 29.0 pm total no. of drops = 229
total mass: % flake by 1
1.75* 8.30 13.10 13.97 12.66 8.73 7.86 2.18* 3.49 4.80 23.14
i
10
10
20
30
40
50
;im€E Drop Diameter C p m l
Flakes av area = 74.2 pm2 with probability of 9570 error in the mean = 15.77 std dev between 142.16 and 157.97
I
std dev = 149.6 pm2 total no. of flakes = 346
Mass Data flake = 5.20 X drops = 2.68 pg mass = 1.94 data for drop intervals" 3 4 5 6 2 o.oo* 0.00 0.00 0.00 o.oo* 0.04 0.00 0.00 o.oo* 0.21 0.00 0.00 0.66 0.86 2.55 37.0 1.73* 1.24 0.52 0.82 14.9 2.60* 1.50 1.28 1.66 33.0 2.89 2.03 2.00 1.91 51.4 2.89 0.93 0.64 1.34 41.3 1.16* 2.03 0.90 0.85 20.9 3.18 3.60 1.35 0.73 13.8 7.23 78.32 87.77 92.44 2.04 87.5
I
s
IO
IO
20
30
40
50
pg Drop Diameter C p m l
n
7 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 =
"Key: 1 = number of drops as a percent of the total; 2 = number of flakes as a percent of the total; 3 = mass as a percent of the total; 4 = flake mass as a percent of the total flake mass; 5 = mass percent of flake; 6 = average flake area; 7 = maximum drop diameter for this interval.
Table V. Drop Data for Mechanical Spraying of No-Flake Paint Drops std dev = 37.1 pm av diameter = 38.0 pm total no. of drops = 187 with probability of 9570 error in the mean = 5.32 std dev between 34.60 and 39.95 Mass Data total mass: drops = 3.25 p g data for droD intervals no. of drops % of total mass L70 of total max drop diameter 1.60* 0.00 5.0 11.76 0.04 10.0 13.90 0.19 15.0 15.51 0.42 20.0 11.23 0.69 25.0 4.81* 0.53 30.0 0.90 35.0 4.81* 0.85 40.0 3.21* 1.45 3.74* 45.0 1.90 3.74* 50.0 cn 25.67 93.40
This reflected the difficulty experienced in obtaining consistent quality finishes. As a general rule, the higher the aluminum content, the better the quality of the sample. A target similar to the electrical shutter was used so that a comparison of the aluminum content for each of the samples taken, using the electrical shutter, was made. A measurement of the aluminum content of the paint before spraying yielded 0.73% aluminum. The amount of mass lost in the painting and curing process was 60%. This yields a final value for the flake content deposited, assuming the process is 100% efficient, of 1.83% aluminum. A plot of the mass percent of aluminum vs. radius using the atomic absorption data and the electrical shutter data is shown in Figure 8b. These results are quite close to
30tI 20
50
100 150 200 Flake A r e a Cpm21
250
Figure 11. Mechanical sprayer: deposited (sample 4) distribution of paint drop diameter for (a) flake, (b) no flake, and (c) deposited distribution of flake area.
those obtained from the detailed drop analysis. An exception is the point farthest from the center where the number of drops is quite small, increasing the error in the electrical shutter results. Similarly, because the mass is small, there is a high error in the atomic absorption measurements for this point. The mass average of the electrical shutter results across the spray pattern radius should reflect the mass percent of aluminum in a electrostatic paint sample. Again this result, 1.82%,compares quite closely to the atomic absorption result of 1.86%. The mechanical sprayer results do not, however, coincide as well. The drop analysis results at the workpiece, sample 4,give a mass percent of aluminum of 1.94%. This is just slightly outside the range of 2.02% to 3.31%, measured by using the atomic absorption technique. Recall that the volume of the paint solids is calculated by using an empirical relationship between drop height and drop radius. The difference between the atomic absorption data and the drop analysis results can be accounted for by noting that the curve for drop height is based on drops mainly of a smaller size than some of the large drops in this sample. Mass Balance. The aluminum content is totally accounted for in the electrostatic sprayer. The mass percent for a 100% efficient process (1.83%) is almost identical with the electrostatic sprayer result (1.82% or 1.86%). This is consistent with the assumption that the electrostatic sprayer is 100% efficient; all the aluminum flake and paint solids are deposited on the target. However, the results for the mechanical sprayer, 1.94-3.31%, reflect the low efficiency of the mechanical process. In fact, in order to obtain high-quality paint finishes, i.e., higher flake content, a lower efficiency process is required. Flake content is increased by increasing the overspray. Because the overspray has low flake content, this increases the flake content deposited on the paint target. This is most often accomplished by adjusting the
Ind. Eng. Chem. Res., Vol. 26, No. 8, 1987 1653 Table VI. Color vs. Mass Percent Aluminum: Mechanical Sprayern % A1
color
E
L*
a*
b*
before spraying
after cure
ref 2.4 5.3 9.0 16.0 21.1
52.3 50.9 48.0 45.3 40.3 29.3
-3.4 -3.4 -3.0 -2.5 -0.3 4.9
-19.4 -21.5 -22.6 -25.0 -29.6 -35.6
0.70 0.59 0.49 0.37 0.21 0.00
3.27 2.78 2.24 1.73 1.03 0.00
15.8
41.6
-3.8
Electrostatic Sample -28.8
0.70
1.85
Linear Interpolation on L to Match Electrostatic Sample 12.9
41.6
-0.9
-28.4
"L* = brightness, a* = green
-
0.25
red, b* = blue
1.21 -+
yellow.
atomizing orifice size and the air velocity to form a finer spray literally blowing the fine low flake content drops away in the overspray. Since all the overspray is not examined, a total mass balance is not possible. Note, however, that the numbers for mass percent of aluminum in the overspray are less than the 100% figure, indicating that these numbers could be representative of an overall mass balance.
Effects of Aluminum Content on Color The effects of the aluminum content on the color were analyzed by changing the aluminum content in the paint used in the mechanical sprayer. A series of various paint mixtures was used to create a table of color vs. aluminum content, commonly called a color vector. This was done both in terms of the mass percent of aluminum in the paint before spraying and finally deposited. These results as well as the color of an electrostatic sample are illustrated in Table VI. Interpolating the mechanical sprayer data on the color coordinate L* to match the electrostatic sample results in an aluminum content of 1.21%. This result is also illustrated in Table VI. Note, the b* color coordinate matches almost exactly, yet the a* coordinate is quite different. Although changing the aluminum content does decrease the color variation, the lower flake content than actually observed in the electrostatic sample and the mismatch in color space indicate that another mechanism plays an important role. This mechanism is most likely the deposited flake distribution. Conclusions First, the flake content of the paint deposited electrostatically and mechanically is quite different. For a metallic paint finish of high quality, the mechanical sprayer deposits an aluminum mass percent greater than 3.0%, while the electrostatic process deposits approximately 1.85%. The electrostatic process has a larger spray pattern within which the flake content varies from 1.40% flake in the middle to more than 3.30% at the edge. Only a sample which passes through the entire cloud ends up with essentially a flake content consistent with depositing all of the paint on the workpiece (1.85%). This spatial variation is most likely to be of practical importance in painting nonuniform areas, where the same time-average exposure to the sprayer cannot be ensured a t each point on the workpiece. Second, the flake size distributions deposited by the two processes are quite different. The mechanical sprayer
deposits flakes that, on the average, have twice the area per flake as the flakes deposited by the electrostatic sprayer. Again there is the complication that the distribution of flake area (size distribution) deposited by the electrostatic sprayer also varies as a function of radius. The various methods, corroborations, and conclusions are outlined in Figures 3 and 4. The parameters of the mechanical sprayer can be adjusted to vary the transport efficiency. Therefore, the device can be tuned to provide different flake contents and flake size distributions. Of course, the price paid is that of the lost paint and of the attendant air pollution. However, if the efficiency of the electrostatic sprayer is to be exploited, there is no flexibility for tuning color by varying the flake content or flake size distribution. As this is a direct result of the high efficiency, all other highly efficient deposition processes are similarly restricted. A more general approach to the painting system must be adopted where the paint, as well as the device, is tuned to obtain the required color. As the electrostatic device is restricted in its ability to tune color, more emphasis should be placed on the tuning of the flake content and the flake size distribution in the paint.
Acknowledgment This work was coordinated by Dr. Carlton Speck of the General Motors Research Laboratories' Electrical Engineering Department, with advice on painting technique from Dr. J. Komjathy and other members of the GM Advanced Engineering Staff's Advanced Coatings Laboratory and Dr. H. Kuo of the GM Research Laboratories' Polymer Department. Advice was given by Dr. J. Hochberg of E. I. du Pont de Nemours & Co., Inc. E. P. Warren, P. Jeganathan, and W. B. Westphal assisted in the experiments. The work formed part of a Ph.D. Thesis submitted to the Department of Electrical Engineering and Computer Science a t MIT (Inkpen, 1986). This work was supported by the General Motors Research Laboratory under contracts with the departments headed by Dr. P. D. Agarwal and Dr. T. C. Wang, with the research under the supervision of Dr. Carlton Speck. The electrostatic paint spraying equipment was supplied by Ransburg Electrostatic Equipment, a division of Ransburg Corp. Paint was supplied by E. I. du Pont de Nemours & Co., Inc. Support was also provided to S.L.I. by a Canadian Scholarship from the Natural Science and Engineering Research Council. Registry No. Al, 7429-90-5.
Literature Cited Bell, G. C.; Hochberg, J. "Mechanisms of Electrostatic Atomization, Transport, and Deposition of Coatings", Presented at the 7th International Conference in Organic Science and Technology, Athens, Greece, 1981. Inkpen, S. L. PhD Thesis, MIT, Cambridge, MA, 1986. Inkpen, S. L.; Jeganathan, P.; Melcher, J. R., "Corona Induced Color Patterns in Metallic Paints", unpublished results, 1987. MacAdam, D. L. Color Measurement; Springer-Verlag: Berlin, 1985; p 219. Melcher. J. R. Continuum Electromechanics: M.I.T. Press: Cambridge, MA, 1981. Woitkowiak. J. J. "Aluminum Flake Orientation of a Metallic Tow coat Exhibiting Telegraphing", Research Memorandum, 1983; General Motors Laboratories, Polymers Department, Warren, MI. Received for reuiew November 17, 1986 Accepted April 17, 1987
