Determinants of Dust Exposure in Tunnel Construction Work

ABSTRACT

In tunnel construction work, dust is generated from rock drilling, rock bolting, grinding, scaling, and transport operations. Other important dust-generating activities are blasting rock and spraying wet concrete on tunnel walls for strength and finishing work. The aim of this study was to identify determinants of dust exposure in tunnel construction work and to propose control measures.

Personal exposures to total dust, respirable dust, and ⍺-quartz were measured among 209 construction workers who were divided into 8 job groups performing similar tasks: drill and blast workers, shaft drilling workers, tunnel boring machine workers, shotcreting operators, support workers, concrete workers, outdoor concrete workers, and electricians. Information on determinants was obtained from interviewing the workers, observation by the industrial hygienist responsible for the sampling, and the job site superintendent. Multivariate regression models were used to identify determinants associated with the dust exposures within the job groups.

The geometric mean exposure to total dust, respirable dust, and ⍺-quartz for all tunnel workers was 3.5 mg/m3  (GSD = 2.6), 1.2 mg/m3 (GSD = 2.4), and 0.035 mg/m3  (GSD = 5.0), respectively. A total of 15 percent of the total dust measurements, 5 percent of the respirable dust, and 21 percent of the ⍺-quartz exceeded the Norwegian OELs of 10 mg/m3 , 5 mg/m3 , and 0.1 mg/m3, respectively. Job groups with highest geometric mean total dust exposure were shotcreting operators (6.8 mg/m3), tunnel boring machine workers (6.2 mg/m3), and shaft drilling workers (6.1 mg/m3). The lowest exposed groups to total dust were outdoor concrete workers (1.0 mg/m3), electricians (1.4 mg/m3), and support workers (1.9 mg/m3). Important determinants of exposure were job group, job site, certain tasks (e.g., drilling and scaling), the presence of a cab, and breakthrough of the tunnel. The use of ventilated, closed cabs appeared to be the single most important control measure for lowering exposures.

MATERIALS AND METHODS

Sampling Strategy and Job Groups

Between 1996 and 1999 16 different tunnel construction sites were visited, and measurements of dust exposure were performed. Prior to measurement, the tunnel workers had been divided into job groups in which the workers performed similar tasks. The groups consisted of workers excavating (drill and blast workers, shaft drilling workers, tunnel boring machine [TBM] workers); workers performing protection and securing work (shotcreting operators who apply wet concrete, support workers); and workers performing finishing work (concrete workers, electricians). Other concrete workers, who worked outside the tunnel, served as an internal reference group in the epidemiological study. (2) A random sample of workers from each job group was measured. Exposure to dust was determined by personal sampling and the aim was to measure exposure to two or more agents for each person for at least two days. Under the labor agreements of the workers the work shift was 10 hours with two breaks of 30 minutes each. The sampling time was limited to 5–8 hours because of the limited battery capacity of the sampling equipment. High dust concentrations further increased power consumption. However, the sampling time was considered representative for the whole work shift because the sampling periods were selected randomly within a shift and tasks were often repeated on the same day. A more detailed description of sampling strategy and job groups has been reported elsewhere.

Determinants

Information on potential determinants within each job group was obtained from three sources: 1) the workers themselves, 2) the industrial hygienist responsible for the sampling, and 3) the job site superintendent. The workers were interviewed after the sampling was completed about the type and the duration of the tasks they had performed during the sampling. In addition, they were asked for their perception of the exposure conditions. The industrial hygienist observed the workers throughout the sampling period and recorded information such as the type of operator cab. The job site superintendent provided general information about the construction site: the size of the tunnel, the type of explosives, and the type of equipment used. The superintendent also provided information on unusual occurrences during the sampling, such as the temporary shutdown of ventilation fans and special tasks.

Dust Exposure

Total dust was collected on acrylic copolymer membrane filters (Versapore 800, Gelman Sciences, Ann Arbor, MI), with a 0.8 µm pore size, fitted in 25 mm closed-face aerosol filter cassettes (Gelman Sciences) at a sampling flow rate of 2 L min-1. Respirable dust was collected on 37 mm cellulose acetate filters with a pore size of 0.8 µm using a cyclone separator (Casella T13026/2, London, U.K.) at a sampling flow rate of 2.2 L min-1.

The particle mass was measured with a microbalance (Sartorius AG, MC 210 p, Goettingen, Germany), with a detection limit of 0.06 mg (0.063 mg/m3 based on 8 hour sampling).

The ⍺-quartz content in the respirable dust sample was measured by X-ray diffraction using NIOSH Method 7500.

Data Analysis

Measured exposure values were used without further adjustment for the unsampled time because they were regarded as representative of the whole work shift. Using cumulative probability plots, the exposure data were best described by lognormal distributions. The exposure data were therefore ln-transformed before further statistical analyses. Standard measures of central tendency and distributions (arithmetic means [AM], geometric means [GM], medians, and geometric standard deviations [GSD]) were calculated. For statistical tests a significance level of 0.05 was chosen.

The GM is calculated from the arithmetic mean of the logtransformed exposures, AM logX , by exp (AM logX). The GSD is calculated similarly from the standard deviation of the logtransformed exposures, SD logX , by exp (SD logX). The GM has to be divided and multiplied with the GSD to obtain confidence limits; for example, the lower and upper 95 percent confidence limits are approximately GM/GSD2 and GM x GSD2, respectively.

Differences in exposure levels among the job groups were evaluated using Kruskal-Wallis test because the Levene’s test showed that variances were not homogeneous (p < 0.05). The analysis of the important determinants for each job group started with univariate models using t-tests and one-way analysis of variance (ANOVA) of the categorical variables suspected of influencing the personal exposure levels. The determinants evaluated were: season, work shift, job site, cross-section of the tunnel (<50, 50–100, >100 m2), equipment (no cab, open cab, closed cab), work height (ground level, >5 m above ground level), the type of explosive, the type of accelerator, before versus after breakthrough of the tunnel to the other side, and operation of TBM machine (yes/no). Only those determinants with sufficient measurements are reported. The means of different strata of the determinants were compared using Bonferroni’s post hoc tests. In case of heteroscedasticity the Kruskal-Wallis test was used instead of ANOVA and the Mann-Whitney test instead of t-tests and Bonferroni post hoc tests.

The duration of the tasks was described by the percentage of the total sampling time. Correlations between continuous predictor variables (i.e., task duration as the percentage of the total sampling time) were evaluated using Pearson’s correlation coefficient. No variables were excluded from the modeling because no correlation coefficient exceeded 0.6. Multivariate regression models were developed for each job group using a forward stepwise regression procedure. Job tasks occurring with a frequency of n <= 3 were not included in the analysis. The model was built in steps beginning with the variable with the lowest p-value and adding variables (p to enter <0.20) until further additions did not result in statistically significant p-values for the added variables (p to remove >0.10), earlier variables lost their significance, or the regression coefficients changed by more than 10 percent. Finally, plausible interactions between explanatory variables were added and kept in the model when a partial F test was significant (p < 0.05). All measurements were considered as independent observations in the analysis. Residuals were studied to assess the fit of the final model. All data analyses were performed using SYSTAT 9.0 and SPSS 10.0 (SPSS Inc., Chicago, IL).
RESULTS

In total 209 workers participated in the exposure study and most of the workers (77%) were monitored on more than one occasion. The geometric mean exposure of total dust, respirable dust, and ⍺-quartz for all tunnel workers was 3.5 mg/m3 (GSD = 2.6), 1.2 mg/m3 (GSD = 2.4), and 0.035 mg/m3 (GSD = 5.0), respectively (Table I). Comparison of job groups by the Kruskal-Wallis test showed statistical differences among the groups for each of the three agents (p < 0.01). The geometric mean exposure levels of total dust in the job groups varied from 1.0 (outdoor concrete workers) to 6.8 (shotcreting operators) mg/m3. The geometric mean exposure levels of respirable dust varied from 0.20 (outdoor concrete workers) to 2.8 (shaft drilling workers) mg/m3, and the geometric mean exposure levels of ⍺-quartz varied from 0.002 (outdoor concrete workers) to 0.39 mg/m3 (TBM workers).

Determinants

Two of the tunnel construction sites investigated were associated with power plants, four with railway installations, seven with road construction, one with a sports center, and two with cleaning/purification plants. The cross-section area of the tunnels varied from 13 m2 (in a shaft) to 340 m2 (a rock cavern). Information on the season, work shift, and job site was available for most job groups. Other determinants evaluated were, for example, the type of equipment used and tasks performed (Table II). The percentage of time spent on different tasks by the job groups during the air monitoring are given in Table III. On average, the workers carried out two to three primary tasks during the sampling.


Drill and Blast Workers

The type of drill rig was a major determinant of all three types of exposures when drilling was performed for >1.0 hours. Workers using a drill rig with no operator cab had the highest exposure to total dust (7.1 mg/m3), which was significantly different from the exposures resulting from using an open cab (2.0 mg/m3) or a closed cab (1.6 mg/m3), p < 0.05 (Table IV). Similar patterns were found for the respirable dust and ⍺-quartz exposure.

Statistical modeling of the determinants of exposure indicated that the tasks of mechanical scaling (removal of loose rock using a hydraulic jackhammer), shotcreting, drilling (with no cab), and repairing the ventilation duct increased the total dust exposure level, while assisting with the drilling operation (e.g., detaching the drill head when it was stuck in the drilling hole, etc.) was associated with a decreased exposure (Table V). Mechanical scaling and drilling with no cab also were associated with increased respirable dust exposure levels, while repair work (repairing equipment), drilling assistance, performing "miscellaneous tasks" (e.g.,tidying upwork area, organizing equipment, etc.), manual scaling (removal of loose rock using hand tools), and rock bolting were associated with decreased exposures. For ⍺-quartz exposure, drilling (with no cab), shotcreting, mucking (gathering of the rock using a shovel), transport of the rock out of the tunnel, and repairing the ventilation duct were associated with increased exposures, while rock bolting was associated with decreased exposures. These models for the drill and blast workers explained 27–38 percent of the variance of the three types of dust exposures.

Shotcreting Operators

The shotcreting operators essentially performed only one task. They sprayed wet concrete (shotcrete) onto the tunnel walls for rock support either during the excavation process to protect the workers from falling rock (before tunnel breakthrough) or after the excavation has been completed for permanent rock support (after breakthrough). Effects on exposure of the type of shotcreting rig used were studied before breakthrough, and was an important determinant of exposure (Table VI). The geometric mean exposure of the workers using shotcreting rigs with closed cabs was 85 percent (total dust) and 73 percent (respirable dust) lower than operators using no cabs (p < 0.05). Workers using open cabs were exposed to dust levels that were between these two types of shotcreting rigs.

Performing shotcreting before or after tunnel breakthrough when using an open cab was an important determinant of exposure (Table VI). The geometric mean exposure of the shotcreters after tunnel breakthrough was 78 percent lower for total dust compared to that before breakthrough (p < 0.01), but no significant difference was found for respirable dust. The type of accelerator had a small and nonsignificant effect on both total dust and respirable dust exposures. No variable had a significant effect on ⍺-quartz exposure.

Statistical modeling of the determinants of total dust exposure showed that the presence of a cab and whether the job was performed before or after tunnel breakthrough were determinants associated with decreased exposure (Table VII). For respirable dust exposures only the presence of a cab was associated with decreased exposures. These models explained 66 and 52 percent of the variance of the dust exposures (Table VII), respectively. No significant models for ⍺-quartz exposure were found.

TBM Workers

The TBM workers did not operate the tunnel boring machine every day due to repair work on the TBM machine and on the ventilation ducts. When the TBM was operated, the workers had significantly higher exposure levels for all three agents than when the machine was not operated (p < 0.001) (Table VIII). The geometric mean exposure of the workers when the TBM was not operated was 81 percent (total dust), 79 percent (respirable dust), and 90 percent (⍺-quartz) lower than when the TBM was operated. There were no significant differences in exposures between work shifts (when operating the TBM), but exposures were lower during the day compared to evening shifts. The respirable dust exposures while operating the TBM were 50–70 percent lower in the summer (p < 0.01) than during the winter and spring season, whereas no difference in total dust and ⍺-quartz exposure was found by season, although similar patterns were observed.

Statistical modeling of the measurement results showed that the "miscellaneous tasks" and surveillance of loading broken rock onto the conveyor belt were determinants associated with increased total dust, respirable dust, and ⍺-quartz exposure on days when the TBM machine was operated. These models explained 10–17 percent of the variability in exposure (Table IX).

Tunnel Concrete Workers

Tunnel concrete workers at site A (a railway installation) had significantly higher exposures compared to those same workers at other sites (cleaning/purification plants and a sport center) for the three measured agents (p < 0.001) (Table X). The geometric mean exposures of the workers at the other sites (B, C, D) were 48 percent (total dust), 55 percent (respirable dust), and 87 percent (⍺-quartz) lower than those of workers at job site A. The height that the work was being done had no significant influence on exposure. Measurements on both work shifts (day and evening) were only performed at site A. For this site, there was no significant difference in exposure between the work shifts for any of the measured agents.

Statistical modeling of the determinants of total dust exposures found that job site A and the task of welding were associated with increased exposures of the tunnel concrete workers. Working at job sites A and B increased the respirable dust exposures, while the tasks of concreting and demolition of wooden forms containing the concrete decreased exposures. An interaction between the task of demolition and job site B was found for respirable dust, indicating an increased exposure when performing this task at job site B compared to other job sites. For ⍺-quartz exposure the job sites A, B, and C, were associated with increased exposure levels, while the tasks of demolition, drilling, and outdoor work were associated with decreased exposures. These models explained 36–85 percent of the variance of the dust exposures (Table XI).

Outdoor Concrete Workers

The geometric mean exposure of the outdoor concrete workers was 71 percent (total dust), 78 percent (respirable dust), and 94 percent (⍺-quartz) lower than the tunnel concrete workers (Table I). The geometric mean exposures of total dust respirable dust, and ⍺-quartz at one job site (railway installation) were 0.88 mg/m3 (GSD = 1.7), 0.23 mg/m3 (GSD = 1.7), and 0.003 mg/m3 (GSD = 1.8), respectively, versus on the other site (railway installation) 1.2 mg/m3 (GSD = 1.8), 0.19 mg/m3 (GSD = 1.7), and 0.003 mg/m3 (GSD = 1.8), respectively (not shown). The differences between these exposures at these work sites were not significant. Statistical modeling of the determinants of respirable dust exposure found that iron work and demolition were associated with increased exposures, and iron work was associated with increased ⍺-quartz exposure. These models explained 15 percent and 12 percent of the variance of the dust exposures, respectively. No significant models for total dust exposure were found (Table XII).

Electricians, Shaft Drilling Workers, and Support Workers

The geometric mean total dust, respirable dust, and ⍺-quartz exposures for the electricians in the winter season were 1.5 mg/m3 (GSD = 1.8), 0.67 mg/m3 (GSD = 1.4), and 0.014 mg/m3 (GSD = 2.1), respectively (not shown). For the spring season the geometric mean total dust, respirable dust, and ⍺-quartz exposure was 1.1 mg/m3 (GSD = 1.7), 0.85 mg/m3 (GSD = 1.4), and 0.022 mg/m3 (GSD = 1.2), respectively (not shown). These differences were not significant.

Determinants of exposure in shaft drilling and support work were not evaluated due to too few measurements in these groups (n = 7 and n = 16, respectively).

Perception of Exposure

Only job groups with measurements taken across all three evaluations of the work conditions (i.e., drill and blast; TBM; shotcreting; support) were evaluated. Fifteen percent of the measurements were reported by the workers to have been taken under conditions that were worse than usual, and 7 percent of the time conditions were reported to have been better than usual. When the conditions were reported to be worse than usual the most frequent explanation was that the ventilation system was not functioning. The geometric mean total dust, respirable dust, and ⍺-quartz exposures when work conditions were reported to be "better than usual" were 2.7 mg/m3 (GSD = 2.8), 0.86 mg/ m (GSD = 2.3), and 0.025 mg/m3 (GSD = 6.2), respectively. The exposures when work conditions were reported to be "worse than usual" were 6.0 mg/m3 (GSD = 2.0), 2.0 mg/m3 (GSD = 1.7), and 0.10 mg/m3 (GSD = 3.8), respectively. The exposures when work conditions were reported to be "as usual" were 3.9 mg/m3 (GSD = 2.9), 1.4 mg/m3 (GSD = 2.6), and 0.035 mg/ m3 (GSD = 4.8), respectively. The workers who reported "worse than usual" had significantly higher exposures than those who reported "better than usual" for all measured agents (p < 0.05) (not shown). There was also a trend indicating an increase in exposure from "better than usual" to "worse than usual." When analyzing the data by the four job groups, the same trends were found (not shown).