Exposure to diesel exhaust emissions on board locomotives

ABSTRACT

Measurements of diesel exhaust emissions (DEEs) were taken in the cabs of leading and trailing locomotives on 48 runs, under winter and summer conditions, on 9 different routes. The cab windows were kept open during the summer runs and closed during the winter runs. The average measurement duration was 9.5 hours. There was virtually no exposure to DEEs in the lead locomotives during winter or summer and very little in the trailing locomotives during winter. The average elemental carbon (EC) concentration in the trailing units of the summer trials was greater than or equal to the proposed American Conference of Governmental Industrial Hygienists' threshold limit value/time-weighted average of 20 µg/m3 on 26% of the runs, and was greater than or equal to 10 µg/m3 on 63%. The concentrations of the gaseous components (nitric oxide, nitrogen dioxide, and carbon monoxide) were from 10 to 20 times below their respective threshold limit values. Mean EC concentration was 2.9 µg/m3 (detection limit 2 µg/m3) during the winter runs and 17.1 µg/m3 during summer. DEEs appeared to be fairly uniformly distributed in the trailing cabs. Configuration of the locomotives had a major impact on EC concentration, with the mean concentration being nearly three times higher in the forward–backward mode than in the forward–forward mode. Descriptive statistics such as means, medians, standard deviations, and so forth, are provided. Various types of statistical comparisons are reported. Recommendations for controlling exposure are made.


MEASUREMENT PROTOCOL

National Institute for Occupational Safety and Health (NIOSH) method 6014 was used to measure NO and NO2. In this method, air at a nominal flow rate of 25 cc per minute is drawn through a sorbent tube that collects NO2. The air then passes through a second tube, which oxidizes any NO that may be present in the samples to NO2). The oxidized NO2 is then collected in the third tube in the train. The first and last tubes are analyzed for NO2. The concentration of NO is calculated from the NO2 collected in the third tube.

CO was measured using Gastec 1DL, carbon monoxide long-term, colorimetric dositubes (Gastec Corp., Japan), One end of the tube was broken open to allow air to enter by diffusion. If CO was present, the reagent in the tube changed color. The tube was calibrated to read ppm-hours, depending on the length of the color stain. Knowing the time duration for which the tube was left open, CO concentration in parts per million can be readily calculated.

EC measurements were made following the protocol in NIOSH method 5040 It is a thermo-optical method, and speciation of organic, carbonate, and EC is accomplished through temperature and atmosphere control. Air at a nominal flow rateof 2 L/min is passed through 37-mm diameter, quartz fiber filters
in open-faced cassettes for about 8 hours. Normally, a 1.54 cm2 rectangular punch of the filter is analyzed, and OC and EC are reported as micrograms of carbon per square centimeter. Total EC
and OC are calculated by multiplying the reported values by the sample deposit area, As mentioned previously, the NIOSH 5040 protocol requires air samples to be taken with open-faced cassettes, which means it is not size selective. This issue has been examined in detail by Birch and Carey, leading to the conclusion that “EC results obtained when collecting diesel aerosol samples with a variety of samplers were not statistically different". The authors of that paper go on to say (1) that the collection of a respirable dust fraction should be considered only when other dust is present at levels that could overload the filter, and (2) when sampling in coal mines an impactor with a submicrometer cutpoint would be necessary to minimize collection of coal-source LC.

In addition to the EC analysis, the mass concentration of DPM was measured gravimetrically and reported as total dust (TD). The same filters were used for the EC analysis.

 
Smoking on board locomotives has been prohibited since the enactment of Canadian federal legislation in 1989. Every attempt was made during the present survey to ensure that there was no smoking in either the lead or the trailing locomotives. Observers were present in the leas locomotive, and only the survey personnel were present in the trailing locomotive.

Measurement Locations

Measurements were taken in the leading and in the first trailing locomotive. Three samplers for DPM and three CO dositubes were located in each of the units, one on the engineer’s console, one on the conductors console, and one on the firewall at the back of the cab. Two samplers for NOx, were located in each of the units, beside the DPM samplers on the consoles. No personal samples for diesel emissions were taken for the following reasons. The locomotive crew was in the lead locomotive, where there was very little likelihood of exposure to diesel emissions. As mentioned earlier, noise exposure of the crew was also measured during this survey. It was considered more important to take personal samples of noise by placing noise dosimeters on the crew. Personal samples for diesel emissions were not taken so that the crew would not be overburdened with instruments. In addition, because the crew members rarely left their stations it was felt that area samples would be quite representative of their personal exposures. In the trailing locomotive only the survey staff was present, and it was considered that area samples would adequately represent the exposure of the deadheading crew.

The instruments were switched on just as the train was getting under way, All the samples except for NOx were taken continuously for the full duration of the trip, roughly 8 hours. The NOx sorbent tubes were changed after 4 hours of operation as required by the NIOSH method. At the end of each run the instruments were collected and the flow rate of the pumps verified (Matheson Rotameter, model 763011-603; Montgomeryville, Pa.). Samples were rejected if the flow rate had varied by more than +-5% of the initial flow rate. In addition to the instruments that gave long term measurements, a Gastec pump was used with the appropriate tubes to take instantaneous measurements of NO, NO2, and CO.

RESULTS


A total of 278 EC, 181 NO,, and 280 CO valid samples were taken in the leading and trailing cabs of locomotives on 48 trips (24 in the winter and 24 in the summer) on 9 routes. The average measurement duration was 9.5 hours. Descriptive statistics for EC, TD, and NO measurements under winter and summer conditions (as well as overall) are shown in Table I. Only the measurements in the trailing locomotive are shown, because there was virtually no exposure to DEFs in the lead locomotive. Statistics related to NO2, and CO are not included, because 90% of the NO, samples and 75% of CO samples were below the detection limits of the respective methods (0.1 ppm for NO, and 0.2 ppm
for CO). The maximum NO, concentration was 0.3 ppm and the maximum CO concentration was 4.5 ppm. Their respective TLVs are 3 ppm and 25 ppm.

All the statistics have been calculated from concentrations averaged over the three measurement locations in the trailing locomotive: the engineer’s console, the conductor’s console, and the
back wall. The mean EC concentration was 2.9 ug/m3 during the winter runs and 17.1 ug/m3 during the summer runs. The corresponding TD concentrations were 0.16 and 0.19 mg/m3, and
the NO concentrations were 1.20 and 1.06 ppm, respectively, during winter and summer.

Figure 1 shows the cumulative normal probability plot of the log transform of the average EC concentration in the trailing locomotive, The equation of the regression line for all samples (sum-
mer and winter) is:

In(Ave EC conc) = 1.638 + 1.22*z

where z is the score of the cumulative normal probability distribution for a given probability (r2=0.97). The various percentiles of the EC distribution can be calculated using this equation. Because nearly 35% of the winter samples were below the detection limit and the maximum concentration encountered was 8.7 ug/m3 (less than half the proposed TLY), it is more interesting to
examine the data from the summer runs separately. The corresponding regression equation is:

In(Ave EC conc) = 2.538 + 0.72*z (r2 = 0.98)
