A Survey of Exposure to Diesel Engine Exhaust Emissions in the Workplace

ABSTRACT

Forty sites were visited during a survey of exposures to diesel engine exhaust emissions. Personal and background exposure to gaseous components, respirable dust, elemental carbon, organic carbon and total carbon were measured and details of control systems were recorded. The results show a wide spread in exposure patterns reflecting the different work practices, job categories of employees and the control methods used. However, sites where fork-lift trucks were in use consistently produced the highest exposures. The survey results suggest that the measurement of elemental carbon could be used as an indicator of exposure to diesel engine exhaust emissions. 


METHODS

Survey strategy

Sites were selected in the UK to provide a cross-section of typical occupational exposures to diesel exhaust emissions. After preliminary enquiries, 40 premises were selected for inclusion in the survey and were classified, on the basis of similarity of operation, into the groups shown in Table 1. Groups 16 were reasonably well defined categories while group 7 comprised a miscellaneous collection of operations which did not fit neatly into the other six groups. Each site was visited to gather the following information: descriptions of the workplaces and the types of controls in operation (for example local exhaust, respiratory protective equipment); details of relevant work activities including range and timing of employee's work, and estimates of exposure to diesel exhaust mainly by personal sampling but also supported by background samples where appropriate.


Sampling and measurement


At the time of this survey, it was not clear which measurement would best represent exposure to diesel exhaust. Although there has been a trend towards the measurement of carbon particles, we did not want to exclude measurement of gases. The emphasis of this work was the measurement of personal exposure and as the measurement of exposure to carbon requires a pump for sample collection we needed to measure the gases diffusively to avoid overburdening operators with additional pumps. Consequently, long-term diffusive colorimetric gas detector tubes were used although we recognized that there was the potential for some interference from the gaseous components of the exhaust and that there may be detection limit problems. Carbon monoxide was measured with Draeger 50/a-D or Gastec Dosi No. 1 DL tubes (detection limit about 6 p.p.m. over 8 h); carbon dioxide was measured with Draeger 500/a-D or Gastec Dosi No. 2 D tubes (detection limit about 65 p.p.m. over 8 h for Draeger tube, 250 p.p.m. for Gastec tube); nitrogen dioxide was measured with Gastec Dosi No. 9 D tubes (detection limit about 0.1 p.p.m. over 8 h).

The respirable fraction of diesel exhaust particulate was collected onto pre-weighed 25 mm quartz filters using a cyclone sampling at 1.91. min"1 (the quartz filters were cleaned in a furnace at 550°C for 4.5 h before use to remove residual carbon). After sampling, the filters were re-weighed (appropriately blank corrected), treated with dilute hydrochloric acid (200 ul of 1.5% w/w) to remove carbonates, then analysed for elemental and organic carbon using a Strohlein Coulomat. The Coulomat uses a two-stage analysis. In the first stage, the quartz filter is placed in an i.r. furnace and heated to about 800 C under a stream of nitrogen. The organics adsorbed onto the carbon particles are vaporized and catalytically converted to carbon dioxide. 

The carbon dioxide is adsorbed into barium perchlorate and quantified using a coulometer where the results appear as 'counts' each of which is equivalent to 0.02 ug carbon. The result is converted to the mass of carbon in the vaporized organic substances and is referred to as 'organic carbon' (OC). This represents the organic carbon adsorbed by the carbon particles in the exhaust without discriminating between the many substances that may be present. After the first stage of the analysis is complete, there should remain only carbon particles on the quartz filter. In the second stage, the furnace is again heated to 800°C but this time under oxygen. Under these conditions, carbon is converted to carbon dioxide and again quantified in the coulometer. This result is referred to as 'elemental carbon' (EC) and represents the carbon particles in the diesel exhaust. The OC and EC results can be added to provide a 'total carbon' (TC) figure. Field blank filters were analysed with each batch of samples and the sample results corrected appropriately. Pyrolytic formation of EC from diesel exhaust particulate during the OC measurement stage has not been detected using the conditions of analysis reported here.

The measurement of aldehydes (acrolein, formaldehyde and acetaldehyde) was attempted at some background locations using glass fibre filters treated with 2,4-dinitrophenylhydrazine (DNPH) sampling at 200 ml min~1. However, it became apparent that not only is DNPH unsuitable for the measurement of acrolein because of the formation of multiple reaction products but the DNPH itself was affected at some locations by other components of diesel exhaust. Consequently, this measurement was abandoned.

Personal samples were collected in the breathing zones of workers potentially exposed to diesel exhaust during their working shift. Static samples were placed in locations where personnel worked and occasionally in those areas where it was impractical to collect personal samples. Wherever possible, sampling times were at least 6 h duration to ensure collection of sufficient carbon for analysis but some processes inevitably resulted in shorter sampling periods. All results reported here are representative of 8 h time weighted exposures.

In summary, each set of personal and background samples provided results on: respirable particulate, elemental carbon, organic carbon, total carbon, carbon dioxide, carbon monoxide and nitrogen oxides.

RESULTS

A total of 691 respirable dust samples were collected. One personal sample was omitted from analysis because of pump failure. A further 5 personals and 29 background samples were declared invalid because of filter damage (producing negative weight gains) leaving a total of 656 respirable dust samples split between 215 personals and 441 background samples. Close examination of the respirable dust results required the elimination of a further 65 samples (26 personal, 39 background) because the sum of the elemental carbon and organic carbon analyses exceeded the respirable particulate on the filter. This probably reflects the difficulties in handling quartz filters which are prone to shed fibres during handling. All respirable dust masses in the survey were corrected by weighing blank filters in the usual way. A total of 189 personal and 402 background respirable dust results were accepted as valid.

Estimates of the limit of detection for weighing quartz filters (based on three times the standard deviation of the blanks) ranged during the survey from about 21 up to 120 ug for a random selection of 10 batches of samples. The median value of 45 ug was adopted as representative of the overall limit of detection. The median air volume sampled was 714 litres. Combining this with the limit of detection for weighing gives 63 ng m-3 as the detection limit for airborne concentration of respirable dust. Any values reported below this concentration were assigned a value equal to half the detection limit as is the usual convention. The estimated detection limit for both EC and OC was 10 ug. Using the median volume of air given above, this equates to an airborne detection limit of 1 4 u g m-3 for both EC and OC. Values reported below this concentration were allocated a value of half the detection limit. Total carbon is simply the sum of elemental and organic carbon.

Of the respirable dust samples collected, 220 personals and 244 backgrounds were analysed for EC and OC (these included some samples which produced invalid respirable dust results but which were nevertheless analysed for carbon content). During analysis, nine EC and one OC analysis failed. Consequently, results are reported for 217 personal and 237 background EC analyses and 220 personal and 243 background OC analyses.

Respirable dust

Table 2 shows that the geometric mean (GM) for 189 personal exposures to respirable dust was 208 ug m-3. Ambulance depots recorded the lowest exposures (118ugm~3) but exposures where forklift trucks were in use were nearly twice those of the other groups (369 ug m~3). The fork-lift truck category also recorded the maximum respirable dust value. Except for the ambulance category, the exposures for background samples are all lower than the corresponding personal samples. The geometric standard deviations (GSD) for the personal samples are around two for each workgroup except for the ro-ro ferries with a value of 2.7 which perhaps reflects the non-repetitive nature of the work. Two background samples at site 34 (group 7) produced respirable masses of about 8 mg m~3 which are clearly unusually high.

Elemental carbon

The GM concentration for 217 personal exposures to elemental carbon was 25 ug m^3 (Table 3). Fork-lift trucks provided the highest GM exposure (66 ug m~3), ro-ro ferries were ranked second (37 ug m~3) and the lowest exposures were at vehicle testing stations ( l l u g m ~ 3 ) . Background exposures to elemental carbon within the work categories were very similar to the personal exposures with an overall background exposure of 24 ug m~3. Within the groups, fork-lift trucks were an exception since the geometric mean background was about 30% higher than the corresponding personal exposure. Ro-ro ferries were the only category where personal exposures exceeded background exposures to any degree. The close equivalence of personal and background exposures suggests that the elemental carbon is relatively homogeneously dispersed in the measured atmospheres. This might indicate that those people in the same work area but not directly involved in the process generating diesel exhaust might also be significantly exposed.

Organic carbon

The pattern of exposure changed for organic carbon (Table 4). Fork-lift trucks again gave the highest GM personal exposure (99 ug m~3) but not markedly different from railway repair (91 |a.gm~3) and bus garage/repair (90 ug m~3). This suggests that the latter two job groups are exposed to a higher proportion of organics adsorbed onto the carbon particles than the fork-lift trucks category. Whereas the overall means for personal and background exposure to elemental carbon were similar, they are different for organic carbon at 70 ug m~3 for personals and 43 ug m~3 for backgrounds (the ratio of these two results for organic carbon is similar to the ratio of the corresponding results for respirable dust). This suggests that those closest to the source of organic carbon receive a higher exposure to organic carbon than those more remote from the source.
Total carbon

In terms of total carbon (elemental + organic) shown in Table 5, fork-lift trucks provided the highest GM personal exposures (188 ug m~3) followed by bus/garage repair (124ugm~3) and railway repair (113ugm~3). The overall mean of personals (102 ug m~3) was greater than for the backgrounds (65 ug m~3). It is interesting to note that total carbon accounts for 49% of the overall mean personal respirable concentration in mass terms for the personal exposures. The corresponding percentage for background samples is very similar at 46%. In both cases, the remaining 50% or so of the respirable mass will be made up of the elements burned off from adsorbed organics carbon during the measurement of organic carbon together with combustible and non-combustible general workplace dust.

Personals vs backgrounds

Generally, personal exposures will be expected to be higher than background exposures since the operator will usually be closer to the source of the pollutant. This is the case with respirable dust in this survey (208ugm~3 personal, 140ugm~3 background), organic carbon (70 ug m~3 personal, 43 ug m~3 background) and total carbon (102ugm~3 personal, 65ugm~3 background). The ratio of personal to background concentration is 1.5 for respirable dust, 1.6 for organic carbon and 1.6 for total carbon. However, for reasons not yet understood, elemental carbon deviates from this trend and the personal and background overall geometric means are essentially the same (25 ug m~3 personal, 24 ug m~3 background).

When the four sets of results (Tables 2-5) are ranked on the basis of highest personal exposure, fork-lift trucks rank first with bus/garage repair and railway repair consistently occupying second or third place. The ranking for ambulance depots on this basis is dubious as it is based on only three personal results.

Percentiles

Personal exposure percentiles are given in Table 6. Whereas the 95th percentile for most job groups is less than about 110 ugm-3 EC, the fork-lift truck group has a corresponding percentile of 390 ug m~3 indicating significantly higher exposures.

Relationship between elemental and organic carbon

Figure 1 shows the relationship between elemental carbon and organic carbon for personal samples. There is clearly no straightforward relationship between these two measurements. Other work in this laboratory (HSE, 1999c) has shown that the relative proportion of elemental and organic carbon in diesel exhaust varies depending on the load on the engine. For example, when idling at low revs, one particular engine produced elemental carbon which was about 30% of the total carbon emitted (i.e. organic carbon predominated). At higher revs and higher engine loads, the proportion of elemental carbon in the total carbon increased to as much as 80% of the total carbon. This information can help to interpret the results in Fig. 1 on the basis of the workloads on the engines. The percentage of EC was calculated as a proportion of TC for the GM personal concentrations at each site where the concentrations of EC were greater than the detection limit. The results were ranked in ascending order of percentage EC.

Sites 3-5, 29-31 and 33 had percentages of EC above 40% (range 40-69%). These samples should originate mainly from engines that are working under some load and generating higher proportions of EC. Sites 3-5 are ro-ro ferries where there would be expected to be heavy loads on the engines during start-up and manoeuvring of vehicles on deck. Sites 29-31 are in the fork-lift truck category where there is likely to be much activity requiring frequent acceleration of the engine under load. Site 33 is a toll booth where vehicles will have to pull away from a standing start, again putting a heavy load on the engine.

At the other end of the scale, sites 6-9, 11, 14, 32, 36 and 39 have percentages of EC up to about 20% (range 5-21%). These samples would be expected to derive from lightly loaded engines. Sites 6-9 and 14 are railway repair and maintenance depots where it is likely that the engines are left idling for some time while adjustments are made and tests carried out. Site 11 is a railway station where engines may be left running while trains are standing at platforms. Site 32 is in the fork-lift truck category and stands apart from the three other fork-lift truck sites. However, this may not be surprising since the work activity included supplying an assembly line as well as warehouse duties. Site 36 is an airport coach set-down area and site 39 is in an airport baggage handling area. Between the two arbitrary percentages of 20% EC and 40% EC, there is a range of activities which would be
expected to be a mixture of heavy and lightly loaded engine duties.

The relevance of these findings depends on the risk to health associated with EC and OC since some operations will generate proportionately more of one than the other. Those operations that place heavy loads on the engine might be expected to present more problems if EC is regarded as a health risk. Lighter engine loadings, particularly with frequent idling, will be mainly associated with OC as the dominant feature.

Because the fork-lift truck category produced higher exposures than the other job categories, the four sites which make up this group were examined in more detail. Table 7 summarizes the geometric mean personal exposures at each site for the four measurements shown. All four measurements rank site 29 as producing significantly higher exposures than the other sites. The GM respirable dust at these four sites was significantly greater than the GM for all results (Table 2). The result for elemental carbon for sites 29-31 was also higher than for the mean of all results (Table 3) but site 32 was lower. A similar pattern was repeated for organic and total carbon where site 32 produced results similar to the overall means. Thus, amongst the fork-lift truck category, sites 29-31 confirm the significantly higher exposures than in the other work categories but site 32 produced results which were more in line with the overall mean. It is interesting to note that the ratios of elemental carbon to organic carbon change significantly between the sites in Table 7. At site 29, elemental carbon predominates which could indicate that the engines are being worked under some load. At sites 30 and 31, organic carbon is present at a slightly higher proportion than elemental carbon, but at site 32, organic carbon predominates markedly perhaps indicating engine usage under light load.

Relationship between respirable dust and elemental/organic carbon

Figures 2 and 3 show the relationship between respirable dust and elemental carbon and organic carbon, respectively, for personal samples. There appears to be no direct relationship between respirable dust and elemental carbon (Fig. 2) so respirable dust could not be used as a surrogate for elemental carbon. Some structure may be present in Fig. 3, probably because organic carbon accounts for the bulk of the DEEE mass in most samples but even in this graph there is considerable scatter.
Gas detector tubes

Measurements with the gas detector tubes produced results of questionable value because the exposures were low and, for most samples, less than the detection limit of the tubes. To calculate the various statistics, gas detector tube results recorded as less than the limit of detection were assigned values equal to half the limit of detection as follows: carbon monoxide 3 p.p.m.; carbon dioxide 125 p.p.m.; nitrogen dioxide 0.1 p.p.m.

Of the 202 detector tubes used to measure personal exposure to carbon monoxide, 113 produced no response; 118 of the 216 background tubes deployed also produced no response. The result was an overall personal GM exposure of < 6 p.p.m. Only the ambulance depot category produced measurable exposures (38 p.p.m.) but this was based on only two results above the detection limit so they may not be representative.

Amongst the 202 personal measurements for carbon dioxide, 49 produced a response less than the detection limit. A total of 116 background results (out of 216) were less than the limit of detection. The highest GM personal exposure was in the forklift truck category (1200 p.p.m.) and the two results from the ambulance depot category produced an exposure of similar magnitude lending some weight to the carbon monoxide results for this job group. Contrary to findings using the other measurements, vehicle testing ranks third when using carbon dioxide as the index of exposure (900 p.p.m.). The overall background exposure (270 p.p.m.) is only just above the detection limit but is less than the corresponding personal exposure value (650 p.p.m.). 500 400^ CO E possibly suggesting a job-related exposure. Atmospheric carbon dioxide will of course contribute to these results.

Of the 202 gas detector tubes used to measure personal exposure to nitrogen dioxide, 176 recorded values less than the limit of detection; 157 of the 216 background tubes produced responses less than the limit of detection. However, the measurements did again confirm the fork-lift truck category as providing the highest GM exposure (0.2 p.p.m.) supporting to some extent the conclusions drawn from other measurements.

The data suggest that time weighted average gas detector tubes are unlikely to have sufficient sensitivity to measure exposures to carbon monoxide, carbon dioxide or nitrogen dioxide resulting from diesel exhaust except where exposures are relatively high.

