Ann. occup. Hyg., Vol. 46, No. 1, pp. 103-112, 2002
© 2002 British Occupational Hygiene Society
Published by Oxford University Press
Article |
A Field Evaluation of the Impact of Transfer Efficiency on Worker Exposure During Spray Painting
Department of Environmental Sciences and Engineering, School of Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
Received 6 October 2000; in final form 11 April 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This paper presents a mathematical model to predict breathing-zone overspray concentrations produced during spray painting as a function of the overspray generation rate, ventilation and work practices. The overspray generation rate required an estimate of the spray gun transfer efficiency, which was provided by a previously developed mathematical model. These models were evaluated in the field under two different scenarios: first in a controlled environment that approximated the assumptions of models, and then under actual spray painting conditions. Results from the first test showed the model overestimated transfer efficiency, but the measured exposures and predicted exposures were not significantly different. During actual spray painting operations, all task exposures were within a factor of three of the model predictions, and there was no statistical difference between the measured and predicted values. The predicted average exposure of each worker was within the 95% confidence interval. The overall mean exposure was within one standard error of the model prediction. The current study expands on the original exposure model by including a transfer efficiency model to provide a better estimate of the overspray generation rate. The theoretical foundation between exposure and its primary determinants is established, and this knowledge can be applied to design and can evaluate optimal control interventions. Also, the general methodology presented here for developing an exposure model is applicable to operations other than spray painting.
Keywords: exposure modeling; spray painting; transfer efficiency; high volume–low pressure spray gun
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Mathematical modeling of worker exposure is a useful tool that serves different purposes, e.g. retrospective exposure assessment or evaluation of control devices. A fairly new concept is to define worker exposure as a function of its primary determinants, including process parameters, ventilation and work practices. The relationship between exposure and these factors can be established quantitatively by various means. For example, integral control-volume analysis can solve for the average room concentration as a function of contaminant generation rate and airflow rate. Computational fluid dynamics is a more flexible, but complex, method that creates numerical simulations of an industrial process. Statistical association or dimensional analysis are other important approaches that relate exposure to process parameters. Since the model can incorporate all aspects that determine exposure, unbiased and accurate predictions may be possible. Application of such a model could prevent implementation of expensive and sometimes impractical control interventions by evaluating the efficiency and feasibility of the devices a priori. The process designer can also use the model to make a quantitative comparison among alternative control techniques, and optimize the intervention selected.
Carlton and Flynn developed a model to identify the important parameters that characterize exposure during spray painting operations in a side-draft spray booth (Carlton and Flynn, 1997a). These included the overspray generation rate, air velocity field, worker size, and orientation to the airflow. Dimensional analysis and wind tunnel experiments were performed to determine the functional relationship among these factors. Positive field results demonstrated the ability of this model to predict mean breathing-zone concentrations (Carlton and Flynn, 1997b). This model was developed with a conventional spray gun, and subsequently generalized to a high volume–low pressure (HVLP) spray gun (Flynn et al., 1999).
All parameters in the exposure model, except the overspray generation rate, can be measured or estimated easily with minor uncertainties. The overspray generation rate is a function of the transfer efficiency of the spray gun; it increases as transfer efficiency decreases. Transfer efficiency is defined here as the fraction of sprayed paint that coats the surface. It is time dependent and varies with spray gun set-up and painting techniques, which are mostly based on worker experience. Transfer efficiency was determined under highly idealized conditions in previous studies (Carlton and Flynn, 1997a,b). Liquid was sprayed onto a flat plate at a fixed nozzle pressure and a gun-to-surface distance of 8''. However, depending on the workpiece geometry and work practices, these parameters are seldom constants. Heitbrink et al. found that transfer efficiency has a great effect on particle overspray concentrations and suggested that a physical model of overspray production would be a useful tool (Heitbrink et al., 1996). Kwok has studied the atomization and transport processes in spray painting, but his work did not link the spray gun transfer efficiency to worker exposure (Kwok, 1991). To provide better estimation of the overspray generation rate, Flynn et al. developed a model that relates transfer efficiency with spray gun settings, paint properties and gun-to-surface distances based on physical assumptions (Flynn et al., 1999). Supportive laboratory studies validated the model with non-volatile materials (Tan and Flynn, 2000). Further study extended the model to account for volatilization effects, which occur in real-life situations (Tan and Flynn, 2001). The current study is the first field application and evaluation of this transfer efficiency model.
The purpose of this study is to evaluate the exposure model and the transfer efficiency model. In the first part of this work, the modified exposure model and the transfer efficiency model (Flynn et al., 1999) were examined separately with non-volatile vacuum pump oil. The experiment was conducted in a controlled environment that best approximated the model assumptions. It was designed to examine the effects of work practice and worker painting experience, since the worker was the only difference from the laboratory study. The second part of the study was to validate the modified exposure model by following three workers during their routine painting work in a spray booth. The transfer efficiency model, adjusted for volatile material, was applied to determine the overspray generation rate in the exposure model. This is the primary improvement from the original exposure model (Carlton and Flynn, 1997a,b).
|
MODEL DESCRIPTION |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This section summarizes the contaminant exposure model and the spray gun transfer efficiency model. Detailed model development can be found in previous publications (Carlton and Flynn, 1997a,b; Flynn et al., 1999; Tan and Flynn, 2000, 2001).
Contaminant exposure model
The contaminant exposure model predicts a dimensionless breathing zone concentration of overspray paint mist while a worker spray-paints a flat surface in either of two orientations (90° or 180°) within a side-draft spray booth. Equation (1) shows that the dimensionless concentration is governed by the worker’s size and orientation to the booth airflow, air velocity in the booth, overspray generation rate, and air momentum fluxes from the spray gun and around the worker. The overspray generation rate is a function of the spray gun transfer efficiency, as shown in equation (2). The model that predicts transfer efficiency is presented in the next section.
where C is the total overspray mass concentration in the breathing-zone (mg/m3); U the average air velocity in the side-draft spray booth (m/s); H the height of the worker (m); D the breadth of the worker (m); mo the overspray generation rate (mg/s); Fg the air momentum flux from the spray gun (kg·m/s); Fm the air momentum flux through the projected area of the worker (kg·m/s); 
, 
, 
values that depend on worker orientation to the airflow in the spray booth; 
the transfer efficiency of an air spray gun based on total mass of paint sprayed; and mL the mass of paint sprayed (mg/s).
The predicted overspray mass concentration, C, is adjusted to determine the time-weighted average. Since the spray gun is only activated sporadically during the actual spray-painting task, the predicted C will be weighted for the time the spray gun is on, as shown in equation (3).
where C is the predicted time-weighted average (TWA) breathing-zone concentration (mg/m3); Cspray the predicted breathing-zone concentration when the spray gun is on (mg/m3); tspray the time in which the spray gun is on (min); Coff the breathing-zone concentration while the spray gun is off (mg/m3); and toff the time in which the spray gun is off (min).
The model assumes no exposure while the spray gun is off, hence, Coff in equation (3) is zero. Cspray is weighted by the time spent spraying while standing in two specific orientations. This model assumes only 90° or 180° orientation with respect to the predominant direction of the airflow in the booth. Generally, a worker orients the workpiece so the freestream flows either to the worker’s side (defined as the 90° orientation) or to the worker’s back (defined as the 180° orientation). In this study, the orientation of a worker standing between the spray gun and the exhaust fan is also classified as 180°. Equation (4) weights Cspray for the time the worker spends in either of the two orientations.
where C90 or C180 are the predicted breathing-zone concentrations while worker’s orientation is 90° or 180° to the airflow in the spray booth (mg/m3); and t90 or t180 are the times for which the worker is standing at 90° or 180° to the airflow in the spray booth (min).
Transfer efficiency model
The transfer efficiency model simulates the paint deposition process by assuming it acts like an impactor. It assumes paint droplets larger than d50, the impactor 50% collection efficiency cut size, impact the workpiece, while smaller droplets escape as overspray. The cumulative size distribution of droplets (
v) that are smaller than the cut size is estimated by a model developed by Kim and Marshall (1971), as shown in equation (5). The specific size ratio, d*, as shown in equation (6), is defined as the ratio of d50 to the mass median diameter of the paint droplets that reach the target surface. Equation (5) calculates the fraction of paint as overspray. Transfer efficiency is the fraction of paint that impacts, and can be calculated according to equation (7).
When non-volatile material, e.g. oil, is sprayed, it is assumed that an empirical equation derived by Kim and Marshall (1971) estimates the mass median diameter of the droplets:
where MMD is the mass median diameter of paint droplets (µm); 
L the paint density (g/cm3); n the air density at the nozzle exit (g/cm3); 
the paint surface tension (dynes/cm); v the relative velocity of air and paint (cm/s); µL the paint viscosity (poise); A the nozzle area for air flow (cm2); mL the mass of paint sprayed (g); and ma the mass of air sprayed (g).
However, equation (8) is not suitable for estimating particle size distributions when volatile paint is sprayed. Continuous solvent evaporation makes it difficult to predict the particle size distribution of droplets that reach the target surface. A subsequent study (Tan and Flynn, 2001) based on a mass balance extended the model to include volatile materials by predicting upper and lower bounds on transfer efficiency. The upper bound is still found using equation (8) to estimate the mass median diameter. The lower bound is found by adding a correction that accounts for solvent evaporation. The median of these bounds was taken as the transfer efficiency prediction in the paint study.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The field sampling took place in a spray booth at the Corrosion Control Section of Shaw Air Force Base, South Carolina. Aircraft ground equipment of various shapes and sizes are primed and painted in this booth. Workers use polyurethane primers and paints of assorted color and chemical composition. Only one painter is allowed in the booth during each day’s spray painting operation. Personal protective equipment, including a supply air respirator, disposable Tyvek suit and chemical resistant gloves, is required. The spray booth measures 4.3 m wide x 3.7 m high x 14.6 m deep (14' x 12' x 48'). Fresh air is pulled through a 4.3 m x 3.7 m (14' x 12') filter door. Exhaust fans, which are located opposite to the filter door, collect paint overspray mists on filters. Workers use a DeVilbiss HVLP spray gun, model JGHV-531–44F, in a pressure-feed cup configuration, or a DeVilbiss conventional gun, model JGA-510, in a siphon-feed cup configuration.
Non-volatile oil test
The first part of the study evaluated the exposure model and the transfer efficiency model in a controlled setup with a non-volatile material. Inland 99 neutral parrafinic vacuum pump oil was chosen as a non-flammable paint surrogate. The main advantage for evaluating the models with non-volatile oil was that accurate transfer efficiency measurements and total breathing-zone mass concentrations could be obtained. All 10 painters, including two supervisors, were called in to perform tests with the oil. These 10 workers have a wide range of painting experience. Half of the workers have less than two years experience; three workers have been painting for two to eight years; the other two workers have more than 15 years of experience. An experienced worker was expected to produce a higher transfer efficiency, since the worker has better control of the spray gun and adjusts their technique based on constant visual feedback. The worker always applies the paint to the surface from a constant distance at a controlled speed (Roobol, 1997). Workers were instructed to spray oil continuously onto a metal plate using the same technique as if they were spraying paint. They were also asked to spray only within the boundaries of the plate, so all impacted material could be collected. However, in the real-world situation, spray painting is an intermittent process. A spray painter usually starts and ends each stroke outside the workpiece to prevent excess paint running at the edges. Each worker sprayed for 1 min with a calibrated DeVilbiss HVLP spray gun, model JGHV-531–44F; this was repeated three times. Thirty pairs of transfer efficiency and exposure measurements were thus obtained.
The spray gun settings, the object being sprayed and the oil properties remained constant for all runs. A 1.02 m x 0.66 m flat metal plate was used for a simple workpiece; this allowed better experimental reproducibility among runs. The plate was placed parallel to the airflow in the spray booth; hence, the freestream flowed to the worker’s side. A trough was placed underneath the plate to collect oil overflow. The plate and the trough were covered with clean, heavy-duty aluminum foil before each test to ensure that there was no oil contamination from the previous run. The whole set was weighed before and after spraying to determine the total oil mass deposited on the surface. The spray pot was weighed before and after each 1 min test to determine the mass of oil sprayed. Transfer efficiency was calculated as the ratio of the oil mass on the plate to the oil mass sprayed. Workers stopped after spraying for approximately 1 min, and were requested to freeze their motion once stopped. The gun-to-surface distance, an important factor in the transfer efficiency model, was measured at this time. Since the worker was stopped in the middle of the action, the measured gun-to-surface distance should be a reasonable estimate of the distance a worker usually keeps when painting a flat object. Note that the gun-to-surface distance would be highly variable when a worker paints an irregularly shaped object. The measured transfer efficiencies were then compared to predicted values.
Total breathing-zone oil mass concentration was measured in accordance with the National Institute for Occupational Safety and Health (NIOSH) Method 0500. A 37 mm polyvinylchloride membrane (PVC) filter was mounted in a modified cassette with a SKC Aircheck pump (Model 224-PCXR8), operated at 2.0 l/min to collect the overspray oil droplets. The filter cassette with its inlet facing downward was attached to the worker’s lapel. The inlet of a closed-face cassette was expanded to 15 mm to mimic the IOM sampler. This technique should help to reduce the undersampling effects associated with both open- and closed-face cassette samplers (Carlton and Flynn, 1997b). Lidén and Kenny found that extremely large particles may enter the IOM sampler orifice, but are less likely to enter the 37 mm orifice (Lidén and Kenny, 1994). The measured exposures were then compared to model predictions, using the overspray generation rate (mo) predicted by the transfer efficiency model.
Volatile paint test
To validate the exposure model in a real world situation, a separate experiment was conducted by sampling three workers while they were priming and painting the regular aircraft ground equipment. The objects being painted included two different types of trailers, two five-gallon drums and several metal parts. Two workers painted the trailers on separate dates, while the other worker painted the drums and metal parts on two other days. On each sampling day, the entire operation was divided into three to six separate task samples. For example, priming the drum and four metal parts was considered one task, and spraying blue paint on the drum was another. Total breathing-zone paint concentration, which includes solid and solvent components, was measured in the worker’s breathing-zone. Paint solid overspray concentrations were measured as total dust concentrations using NIOSH Method 0500. The sampler setup was similar to the one that measures total oil mass concentration, as described above. The only difference is that two large charcoal tubes were placed in parallel to back up the filter cassette and capture solvents in the worker’s breathing-zone. The charcoal tubes also collected solvents that evaporated from the droplets on the PVC filter. An American Industrial Hygiene Association accredited laboratory analyzed the charcoal tubes using the NIOSH Method 1550 for the mass of total hydrocarbons, which determined the solvent overspray concentrations.
Parameters in the models were measured directly. Each worker’s height and breadth were taken with a tape measure. Air velocity in the spray booth was determined with a calibrated Alnor model 8565 thermo-anemometer with the pieces in the spray booth. Velocities ranged from 0.44 to 0.49 m/s during the 4 day sampling period. Physical properties of the vacuum pump oil, primer and paints were measured, as shown in Table 1. The viscosity was determined by a Zahn cup viscometer. A Fisher Surface Tensiomat Model 21 was used to measure surface tension. All tasks were recorded with a video camcorder to identify variables that were difficult to quantify. These variables included total spraying time and work practices. Information such as task time and worker activities other than spray painting was also recorded. Since only the worker was allowed in the booth, the gun-to-surface distance cannot be measured directly. Also, constant movement of the worker made it infeasible to estimate the distance from the videotape. Thus, workers were asked to spray non-volatile material onto a flat plate, so that the gun-to-surface distance can be determined by the method used in the oil test. Although this method was not a direct assessment, it provided an estimate of the average gun-to-surface distance a worker usually maintained.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The oil test examined the predictive ability of the transfer efficiency model. Average transfer efficiencies measured from all 10 workers ranged from 0.72 to 0.92. Figure 1 compares the average measured transfer efficiencies and the average predicted values. Error bars in the x direction are the uncertainty bounds on the transfer efficiency predictions. The primary quantifiable uncertainty in the model was the mass median diameter of particle size distribution, which was estimated by equation (7). Kim and Marshall reported an uncertainty in this estimate of ±24% at a 95% confidence interval (CI) level (Kim and Marshall, 1971). This was used in calculating the error bars. The letter next to each point identifies one worker. From Fig. 1, six of the 10 average transfer efficiency measurements agree with the model predictions within the uncertainty limits. The Wilcoxon signed-rank test was performed to compare the predicted and measured transfer efficiencies. This test is a non-parametric analogue to the paired t-test. The result indicates that the hypothesis of no difference between predicted and measured transfer efficiencies is rejected at the 0.05 level (P = 0.004). The model overestimated transfer efficiency in this study.
|
The exposure model was also evaluated with oil data. The measured breathing-zone mass concentration ranged from 110 to 295 mg/m3. Figure 2 shows the comparison between the average measured oil mass concentration and the average predicted values. The transfer efficiency model was used in conjunction with the exposure model to estimate the predicted mass concentrations of the overspray oil droplets. This estimate was assumed to be the major source of uncertainty in the exposure model. Therefore, error bars on the x-axis were derived from the corresponding transfer efficiency errors shown in Fig. 1. Figure 2 shows that six of the 10 oil mass concentrations are within the uncertainty limit of the model predictions. The Wilcoxon signed-rank test result shows that the hypothesis of no difference between predicted and measured exposures is not rejected at the 0.05 level (P = 0.32).
|
Figure 3 shows the relationship between each worker’s measured transfer efficiency and length of painting experience. The x-axis, which shows months of painting experience, is a logarithmic scale. As expected, transfer efficiency increases with painting experience. The increasing transfer efficiency then reaches a maximum value, which is limited by the design of the spray gun used. Beyond this value, transfer efficiency is not affected by worker’s experience. However, workers D, E and F did not follow the trend described above. Worker D had substantially lower transfer efficiency despite having 7 yr of painting experience. Workers E and F, on the other hand, had very high transfer efficiencies even though they have been painting for less than 2 yr.
|
For the second study using actual spray painting tasks, 20 paint samples were collected over the 4 day sampling period. The measured breathing zone mass concentration ranged from 13 to 275 mg/m3. Solid and solvent each contributed half of the total mass concentration for most samples, while some samples had about 10–30% solids. On the first and the fourth day, worker X primed and painted two five-gallon drums and eight pieces of metal parts of different shapes. Nine samples were collected from worker X during these 2 days. On the second and the third day, workers Y and Z primed and painted a trailer, with five and six samples collected from each worker respectively. Figure 4 illustrates the comparison between the measured total overspray mass concentration and the predicted levels for all 20 tasks. The uncertainty line represents experimental errors on the model prediction, which are based on the measurement imprecision of parameters in the model. One example is the variation of the air velocity measurement in the spray booth. Seven task exposures (35%) fall within the uncertainty lines. All 20 task exposures are within a factor of three of the model prediction. The Wilcoxon signed-rank test result suggests that the hypothesis of no difference between predicted and measured task exposures is not rejected at the 0.05 level (P = 0.1836).
|
Each worker’s measured and predicted mean overspray mass concentrations, as well as the overall group exposure from the 20 tasks, are shown in Fig. 5. Error bars represent one standard error of the mean. Two of the three predictions are within one standard error of the measurements. The overall exposure prediction is also within one standard error of the overall group exposure.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The oil test evaluated the transfer efficiency model and the exposure model in a controlled environment. Although it did not completely reflect the actual spray painting operation, it allowed for a better assessment on how work practices affect transfer efficiency and worker exposure. The statistical results showed that the model overestimated transfer efficiency, but that the exposure model agreed well with the field data. Worker D was considered an extreme case in the oil study, since he sprayed differently from all other workers. The workpiece on which the workers sprayed oil was 1.02 m in height. All workers, except worker D, were spraying from a squatting position or with their back bent to keep the spray gun perpendicular to the workpiece surface. Worker D stood straight and sprayed downward during the entire sampling time. The spray gun was pointed at the trough and caused many oil droplets to deposit on the ground instead of the workpiece. This resulted in inaccurate sampling and modeling. Worker D is thus excluded from the following discussion.
Six of the nine workers had measured mean transfer efficiencies that matched the predicted mean, as shown in Fig. 1. The model had a tendency to overestimate transfer efficiency. The most plausible reason was that oil mass was lost when workers started and ended each stroke outside the target. As explained earlier, this technique prevents excess paint from running at the edges. Although all painters were instructed to spray only within the boundaries of the plate, some still accidentally sprayed outside the surface. This problem caused the measured transfer efficiency to be less than predicted. Three workers, C, E and H, were beyond the model uncertainty limits in Fig. 1. Workers C and H each had more than 15 yr of painting experience. They tended to have more difficulty in changing their habits and sprayed more oil outside the workpiece. Normally, the amount of paint sprayed outside a flat-surface object should be small because the spray gun is usually off beyond the workpiece edges. These workers were instructed to trigger the spray gun at all times during sampling, and thus resulted in more oil loss if spraying outside the object. Worker E showed an inattentive attitude by moving the spray gun in a large arc at a rapid pace. The transfer efficiency model assumed all air streamlines are perpendicular to the target surface. When spraying in an arc, more oil droplets missed the workpiece than predicted. Although these three workers produced transfer efficiencies beyond the model uncertainty limits, the relative errors between their measured transfer efficiencies and the predicted values were <7%.
The exposure model was first evaluated with the oil data from 10 workers. Figure 2 shows that workers C, F and G had measured mean exposures larger than the predicted values by 29, 37 and 62% respectively. Interestingly, these errors decreased with the length of the workers’ painting experience. Worker F demonstrated the most controlled and stable spraying technique among all 10 workers. He kept the spray gun at a reasonably close distance and sprayed across the plate at a uniform speed. No reasonable hypothesis was found to explain why worker F’s mean exposure was higher than expected. Unlike the others, workers C and G were extremely concerned about their gun-to-surface distances being measured, because they believed that their performance was being evaluated. For this reason, it was possible that they moved the spray gun closer to the workpiece just before measuring the distance. If this occurred, transfer efficiency would be overestimated; hence, their exposures would be higher than expected. However, the statistical result showed no significant difference between the measured and predicted oil exposures.
Transfer efficiency is affected by spray gun and coating type, painter skill level, and geometry and size of the target. Snowden-Swan and Worner evaluated the impact of these factors, and found that the most consistent parameter that exerts influence on transfer efficiency is painter skill level (Snowden-Swan and Worner, 1993). They found that the differences in transfer efficiency due to skill level exceeded that due to spray gun and coating types. In Fig. 3, painters with more years of experience did display better spraying technique and achieved higher transfer efficiencies, with the exception of workers E and F. Workers E and F only had 12 and 16 months painting experience respectively, but they obtained very high transfer efficiencies (>85%). As explained above, worker F showed the best spraying technique and controlled the spray gun close to the target. Undoubtedly, he achieved the highest transfer efficiency. Although worker E did not control the spray gun well, he kept the spray gun very close to the target surface at all times. This demonstrated that gun-to-surface distance is a crucial element that determines transfer efficiency. Proper training and adequate experience are the two most significant factors in achieving optimal transfer efficiency of a spray gun.
Carlton and Flynn evaluated the exposure model in the field with the use of a conventional spray gun (Carlton and Flynn, 1997b). Samples were taken from eight workers for 55 separate tasks. Twenty-two tasks (40%) fell within the estimated error of the model prediction, while 39 tasks (71%) were within a factor of three of the prediction. Four of the eight workers had predicted average exposures within one standard error of the measured mean; the overall predicted mean exposure was also within one standard error of the measured mean. Flynn et al. revised the exposure model to include HVLP spray guns (Flynn et al., 1999). The encouraging field results from the previous (Carlton and Flynn, 1997b) and current studies demonstrate the model’s ability to predict exposure with different types of spray guns.
The most important improvement of the current study, compared to the previous field evaluation (Carlton and Flynn, 1997b), was the application of the transfer efficiency model. The exposure model incorporated the transfer efficiency model for a better estimate of the overspray generation rate. Hence, the model related exposure to many important parameters that were not accounted for in the original design. These parameters included spray gun settings and performance, liquid composition and properties, and gun-to-surface distance. The association provides a theoretical foundation for designing optimal control interventions to reduce exposure generated by spray painting operations. Examples of controls include changing a worker’s orientation in the spray booth, setting the spray gun’s nozzle pressure at different ranges and/or training workers to spray closer to the workpiece.
As shown in Fig. 5, two of the three workers had predicted exposures within one standard error of the measured means. The predicted mean of the other worker was within the 95% CI of the measured mean. The predicted overall group exposure, based on all 20 individual tasks, was within one standard error of the measured overall mean. The models were developed in a well-controlled wind tunnel with a mannequin spraying non-volatile oil onto a flat surface. They were not expected to work well in all situations in the real world, but their performance in predicting the average exposure level is encouraging. The predominant limitations that affect the predictive ability of the models were identified, as follows. An actual workpiece is seldom a flat surface as the models assume. It usually has a complex shape, sometimes with holes and other inaccessible areas. The spray gun has to be held at various distances and angles to coat these surfaces. Many assumptions in the transfer efficiency model are thus invalid. Movement of workers and obstruction of a large workpiece can disrupt the freestream flows in the spray booth. In both cases, actual overspray distribution may have differed from that produced in the laboratory. Worker orientation to the airflow in the spray booth is another important source of uncertainty. In this study, a worker never painted at a fixed orientation or position. He/she moved around the workpiece swiftly to apply a sufficient amount of paint onto the surface. Sometimes workers even stood between the spray gun and the exhaust fan. The exposure model was designed for only two orientations, 180° or 90° to the booth airflow. It was very difficult to distinguish worker orientation and classify this into two groups. The predictive ability of the exposure model was therefore compromised. Despite these limitations, the models, which were based on the functional relationship with their primary determinants, worked reasonably well. The positive field results presented support the future development of conceptual models for predicting exposure in the work place. Future modeling research for spray painting should continue to simulate the real world situations in order to overcome the limitations described above. For example, sophisticated methodology is needed to measure gun-to-surface distance as a function of time. Non-intrusive instruments, such as a phase Doppler anem-ometer, can be used to measure the paint aerosol size and its velocity. Such information would be valuable for improving the transfer efficiency model by quantifying its uncertainty.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This study presents a field evaluation of two quantitative models during spray painting operations. One model defined worker exposure as a function of the overspray generation rate, air velocity field, and worker dimensions and orientation to the airflow in the spray booth. The overspray generation rate is a function of the spray gun transfer efficiency, which is very difficult to measure directly. Thus, a separate model was used to predict the transfer efficiency based on spray gun settings and spraying conditions. The exposure model and the transfer efficiency model were examined first in a controlled environment that fitted most assumptions in both models. Six of the nine workers had mean transfer efficiencies within the model uncertainty limits, and the statistical results showed that the model overestimated transfer efficiency. Although the model overestimated three of the workers’ transfer efficiencies, the relative errors were <7%. Six of the nine worker exposures were not different from the predicted values within the uncertainty limits in this case. The statistical result indicated no significant difference between model predictions and measurements. The second part of the study validated the exposure model with data collected during actual spray painting tasks. All exposure measurements were within a factor of three of the model predictions, and there was no statistical difference between predicted and measured exposures. Two of the three workers’ predictions were within one standard error of the exposure measurements. The overall group prediction was within one standard error of the mean exposure.
The current study expanded the original exposure model to account for more parameters that influence exposure generated during spray painting operations. These new relationships provide a comprehensive basis to integrate engineering controls and work practices for reducing exposure. Future research should focus on better simulation of the real world situation. Although spray painting is the target application, the model development and evaluation procedures are also applicable to other airborne exposure situations. Positive results from this study encourage future interests in similar modeling techniques.
Acknowledgements–The authors would like to thank the staff of the Shaw Air Force Base Corrosion Control Shop, whose assistance was invaluable in the collection of samples, especially the shop supervisor Tsgt. Shelly Hardy Jr. The authors would also like to acknowledge Lt. Chet Bryant of the Shaw AFB Bioenvironmental Engineering office for technical support, and the US Air Force Armstrong Laboratory Analytical Division, Brooks AFB, for supporting and conducting sample analysis. This work was supported by a pilot project research grant number T42/CCT410423-06 from the US National Institute for Occupational Safety and Health (NIOSH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIOSH.
|
FOOTNOTES |
|---|
* Author to whom correspondence should be addressed: 881 Airport Road 12C, Chapel Hill, NC 27514, USA. Tel: +1-919-932-6141; fax: +1-919-966-7911; e-mail:ceciliat{at}mail.com
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMODEL DESCRIPTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Carlton GN, Flynn MR. (1997a) A model to estimate worker exposure to spay paint mists. Appl Occup Environ Hyg; 12: 375–82.
Carlton GN, Flynn MR. (1997b) Field evaluation of an empirical-conceptual exposure model. Appl Occup Environ Hyg; 12: 555–61.
Flynn MR, Gatano BL, McKernan JL, Dunn KH, Blazicko BA, Carlton GN. (1999) Modeling breathing-zone concentrations of airborne contaminants generated during compressed air spray painting. Ann Occup Hyg; 43: 67–76.
Heitbrink WA, Verb RH, Fischbach TJ, Wallace ME. (1996) A comparison of conventional and high volume–low pressure spray-painting guns. Am Ind Hyg Assoc J; 57: 304–10.
Kim KY, Marshall WR. (1971) Drop-size distributions from pneumatic atomizers. AIChE J; 17: 575–84.
Kwok K. (1991) A fundamental study of air spray painting. Ph.D. dissertation, University of Minnesota, Minneapolis, MN.
Lidén G, Kenny LC. (1994) Errors in inhalable dust sampling for particles exceeding 100 micrometers. Ann Occup Hyg; 38: 374–84.
Roobol NR. (1997) Industrial painting: principles and practices. Cincinnati, OH: Hanser Gardner.
Snowden-Swan L, Worner P. (1993) Determining transfer efficiency and VOC emissions. Metal Finish; 91: 73–8.
Tan Y, Flynn MR. (2000) Experimental evaluation of a mathematical model for predicting transfer efficiency of a high volume–low pressure air spray gun. Appl Occup Environ Hyg; 15: 785–93.[Medline]
Tan Y, Flynn MR. (2001) Methods for estimating the transfer efficiency of a compressed air spray gun. Appl Occup Environ Hyg; in press.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. R. FLYNN A Stochastic Differential Equation for Exposure Yields a Beta Distribution Ann. Hyg., April 1, 2004; 48(5): 491 - 497. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||