Silica Dust Exposures During Selected Construction Activities

ABSTRACT

This study characterized exposure for dust-producing construction tasks. Eight common construction tasks were evaluated for quartz and respirable dust exposure by collecting 113 personal task period samples for cleanup; demolition with handheld tools; concrete cutting; concrete mixing; tuck-point grinding; surface grinding; sacking and patching concrete; and concrete floor sanding using both time-integrating filter samples and direct-reading respirable dust monitors. The geometric mean quartz concentration was 0.10 mg/m3 (geometric standard deviation [GSD]=4.88) for all run time samples, with 71% exceeding the threshold limit value. Activities with the highest exposures were surface grinding, tuck-point grinding, and concrete demolition (GM[GSD] of 0.63[4.12], 0.22[1.94], and 0.10[2.60], respectively). Factors recorded each minute were task, tool, work area, respiratory protection and controls used, estimated cross draft, and whether anyone nearby was making dust. Factors important to exposure included tool used, work area configuration, controls employed, cross draft, and in some cases nearby dust. More protective respirators were employed as quartz concentration increased, although respiratory protection was found to be inadequate for 42% of exposures. Controls were employed for only 12% of samples. Exposures were reduced with three controls: box fan for surface grinding and floor sanding, and vacuum/shroud for surface grinding, with reductions of 57, 50, and 71%, respectively. Exposures were higher for sweeping compound, box fan for cleanup, ducted fan dilution, and wetted substrate. Construction masons and laborers are frequently overexposed to silica. The usual protection method, respirators, was not always adequate, and engineering control use was infrequent and often ineffective.

METHODS

After consultation with a group of construction contractor safety directors, a list of eight activities common to many large construction projects was selected, based on frequency of occurrence and expected level of dust produced (see Appendix A). These activities were cleanup; demolition using hand tools; concrete cutting with handheld or table mount saws; concrete and mortar mixing; tuck-point grinding; surface grinding; sacking and patching concrete; and concrete floor preparation with a sandpaper disk (floor sanding).

Sampling occurred from August 2000 through January 2001 for 42 on-site days at nine large construction sites representing six contractors. Projects included five cast-in-place concrete office buildings ranging in size from three to five stories, two concrete block two story structures, one concrete tilt-up one story office building, and a major renovation of a university library.

Site days were selected primarily at the convenience of researchers, usually without advance knowledge of that day’s scheduled construction activities. Volunteer subjects were recruited at the start of the shift by a referral from the foreman or during the shift as workers were observed conducting activities of interest. Activity is defined as the period of sampling, including ancillary functions performed in support of the activity. Task is used to define only the dust-producing portion of the activity.

Sample and Data Collection

Sampling was conducted for the entire activity period a worker was engaged in the target task or doing other tasks to support additional target task work. Occasionally there were large time gaps between target task occurrences when setup, cleanup, or other tasks were completed in support of the target task.

Two types of sampling devices were used. A 10-mm Dorr-Oliver nylon cyclone with polyvinyl chloride (PVC) filter calibrated at 1.7 L/min was used to measure average concentrations during full work activities. Personal DataRam (pDR) light-scattering photometers (models 1000 and 1200) fitted with BGI cyclone preselectors and PVC filters and calibrated to 2.2 L/min were used to assess task-specific exposures, peak levels, and run-time averages. The pDRs were positioned with the pump in a small backpack with 12-inch silicon tubing extending from the cyclone inlet to the subject’s shoulder.

Subjects were fitted with either a nylon cyclone or a pDR sampling device on each sampling day. Subjects monitored with a nylon cyclone were asked to record their tasks on a task card that delineated task, tool, respiratory protection used, work area (enclosed, inside, partially enclosed, or outside), and whether anyone nearby was making dust (Y/N). Subjects monitored with a pDR were observed by a researcher who recorded the following variables for each minute: task, tool, work area, respiratory protection used, controls employed, estimated cross draft, and whether anyone nearby was making dust.

Work area was categorized as outside, partially enclosed (not all walls and windows in place), inside, or enclosed. Examples of enclosed areas are stairwells and confined plastic enclosures. Respirators encountered included dust masks and half-face cartridge respirators. At some sites respirator protection was mandated by management during dusty operations, whereas it was a matter of worker choice at other sites. The control strategies employed varied among sites, with some sites having a much greater emphasis on controls for dust reduction than other sites. Cross draft was estimated (none, low, medium, high) using prior researcher experience and observation of visible dust as a guide. Two researchers conducted the observations and classified work area and cross draft for modeling. Although between-researcher agreement was not quantified, the researchers worked together for the first several site days and made joint decisions on classification to assure reasonable concordance on how these variables were classified.

To calibrate the three sampling devices (nylon cyclone, pDR 1000, and pDR 1200) to each other, side-by-side area samples were collected. The three devices were placed together in a basket on a tripod with sampler inlets located within 2 inches of each other. Twenty sample sets were collected, with duration ranging from 18 to 59 min. Sampling was conducted at construction sites with samplers positioned close to operations producing moderate to high concrete dust concentrations. Each pDR instrument response was paired to its respective cyclone filter result, giving a total of 31 sample pairs. Regression analysis was conducted using the nylon cyclone sample as the dependent variable, and the resulting regression line was used to calculate adjusted respirable dust (and silica) concentrations for either run-time averages or 1-min concentrations from the pDRs.

Respirable dust samples were analyzed following National Institute for Occupational Safety and Health (NIOSH) method 600. Filters were equilibrated in an environmental chamber (relative humidity 32%) for at least 2 hours before weighing on a Mettler MT-5 analytical balance with a resolution of 0.001 mg. The laboratory’s limit of detection was 5.0μg. Quartz analysis followed NIOSH method 7602, (19) using a Fourier transformed infrared spectrophotometer. The quartz limit of detection was 5.5μg.

Analyses were conducted on run-time average dust and quartz concentrations from nylon cyclone and pDR (adjusted) results and also on the 1-min adjusted dust concentrations from the pDR.

For samples below the analytical detection limit, the detection limit divided by the square root of 2 was used as the value for all data analysis. All data were lognormally transformed before analysis, because the data were generally lognormally distributed. Geometric means, geometric standard deviations, and parametric exceedance fractions (27) were used for summary statistics in run time average and 1-min data sets. The parametric exceedance fraction was selected over the actual exceedance fraction, because the actual exceedance fraction is extremely unstable for a small sample size, and the intent was to exploit the material in this data set to the maximum extent.

The determinants of exposure concentrations were assessed using multiple linear regression modeling for 1-min concentrations for the three activities that had at least six sampling sessions: surface grinding, hand demolition, and cleanup. Factors were added to the model stepwise and included if the coefficients were significant (p<.05).
RESULTS

For the side-by-side area samples (Figure 1), no difference was observed between the two pDR results, and a clear linear relationship (on the log scale) was observed between the nylon cyclone and pDR results. The R2 was .67, the standard error was 0.61, and the observed regression line was

ln (nylon cyclone) = 0.1714 + 0.6932 X ln(pDR)

This relationship was used to adjust the pDR-derived concentrations to be comparable with the standard nylon cyclone values. Part of the residual variation may derive from particle distribution in various tasks. Thorpe and Walsh found in a controlled study that the monitor to cyclone ratio varied somewhat with stone particle size with ratios of 0.91–0.97 at 4μm and 1.121.22 for 6.4μm. The present calibration study, which occurred under field conditions, is valid because it was done under conditions observed in the study, despite the increased variability that may occur.

Respirable dust and quartz run-time average concentrations by activity are presented in Table II. There were 113 samples collected, representing eight activities and one "mixed" activity category. Sample duration averaged 202 (standard deviation =97) min. The highest exposures were during surface grinding and tuck-point grinding, and the lowest exposures were for cleanup and sacking/patching concrete. Mixed samples may be elevated due to amount of surface grinding represented in these samples. Although geometric means for respirable dust were substantially less than the American Conference of Governmental Industrial Hygienists threshold limit value (TLV) of 3.0 mg/m3, 34% of the sample distribution would be expected to exceed the respirable dust TLV of 3 mg/m3.

The geometric mean quartz concentration was 0.11 (geometric standard deviation [GSD]=5.21) mg/m3 with 75% of the runtime average quartz concentration distribution exceeding the quartz TLV of 0.05 mg/m3. For five of the eight activities more than half the sample distribution exceeded the quartz TLV. Quartz, as a percentage of the respirable dust samples, ranged from 2.2 to 21.0% depending on activity.

Respiratory protection use was assessed with quartz data because respirator decisions would normally be based on quartz exposure. As quartz exposure increased, greater respiratory protection was employed, with geometric means (GMs) of 0.03, 0.12, and 0.20 mg/m3, respectively, for no respirator, dust mask, and cartridge respirator (Table III). When no respirator was used, the TLV was exceeded for 46% of samples. When dust mask and cartridge respirators were used (respirator protection factor of 5 for dust mask and 10 for cartridge), concentrations exceeded the respirator’s protection concentration for 43% dust mask samples and 38% of cartridge respirator samples. For higher exposure activities (tuck-point grinding and surface grinding) respirators were always used, with a preference for cartridge respirators. Laborers and brick masons tended to favor no respirators or dust masks, whereas cartridge respirators were more common among cement finishers.

Dust control methods were employed rather infrequently, with only 12% of samples using some form of control, including water, area fans, ducted fan exhaust, and sweeping compound.

Surface grinding was the activity with the highest exposures (Table II). Grinding samples included work with 4.5 and 7-inch grinding wheels, and both abrasive grinding (for finer finishing work) and diamond wheels (for more aggressive rough grinding). Quartz exposures when the 4.5-inch grinders were used were 33% less than exposures for 7-inch grinders, and exposures when the abrasive wheel was used were 60% less than exposures when the diamond wheel was employed.

The 1-min average data set offers the opportunity to look at exposures during only the target task or during the full activity sample period, including nontarget tasks. Table IV summarizes adjusted respirable dust exposure for each activity by task and tool (when appropriate). There were 39 sampling sessions conducted to collect 6365 min of data with an overall GM of 0.66 mg/m3 (GSD=3.07). The highest GMs were for tuck-point grinding and surface grinding, the lowest for sacking/patching and floor sanding. For some activities (tuck-point grinding, cleanup, demolition) the subject worked at the target task for most of the sampling session; other activities (concrete mixing, concrete cutting, floor sanding) involved much more time on support tasks than the target task. Some activities showed little difference in exposure between target and nontarget tasks (cleanup, sacking/patching, and floor sanding), and others showed clearly elevated exposure during the target task (grinding, demolition, cutting, and mixing). For cleanup, one tool, the backpack blower, generated higher exposures than other hand tools. For demolition the rivet buster and sledgehammer produced the highest exposures, and for concrete cutting the handheld and table saws were higher than the slab saw with water control.

Several measures to control dust levels were observed during sampling for the 1-min data set, including the following.

- A sweeping compound was used to control dust during sweeping. This compound, composed of dyed sawdust, sand, and mineral oil, is broadcast over the floor before sweeping. With the sweeping action, the dry dust tends to adhere to the oil-impregnated particles rather than become airborne. The amount of compound used and the uniformity of distribution are likely important factors in its effectiveness.

- A box fan was positioned in a work area with the intent of creating a cross draft to carry dust-laden air from the area. The position of the fan relative to dust generating activities varied greatly.

- An interior space was ventilated by a large exhaust fan and 12inch diameter duct, exhausting outside the building. Sometimes the space was isolated from adjacent spaces using plastic sheeting, whereas at other times there was no isolation. The duct was sometimes located close the dust generating source.

- A surface grinder with a shroud surrounding the disk was connected to a vacuum with high efficiency particulate air filter.

- Grout (substrate) to be ground was wetted with a hose prior to tuck-point grinding.

A task without controls is compared with the same task with controls in the same type of work area (inside, outside, etc.) in Table V. GMs were reduced for surface grinding or floor sanding inside when a box fan was in use (2.71 versus 6.27 mg/m3, and 0.21 versus 0.42 mg/m3, respectively), and for surface grinding outside when local exhaust ventilation was in use (1.42 versus 4.87 mg/m3). Sweeping compound, box fan for cleanup, ducted fan dilution, and wetted substrate produced higher exposures than the comparable task without dust control. It may be that control strategies were employed only when high exposures were anticipated, so increased exposures while using controls does not necessarily suggest that exposures increased with the use of these controls.

The three activities for which data were collected during at least six sampling sessions were modeled to identify which factors had most effect on exposure. For the surface grinding task, the tool, work area, cross draft, and controls were all highly significant (p< .001) and had an  2 of .61 (Table VI). For concrete demolition the model produced an R2 of .35 (Table VII), with task, tool, work area, and cross draft being highly significant (p< .001). For cleanup the R 2 was .14 (Table VIII) with tool, work area, cross draft, and nearby dust generation being highly significant (p< .001).
