Evaluation of Diesel Exhaust Controls

ABSTRACT

The National Institute for Occupational Safety and Health (NIOSH) received a request from a fire department to conduct a health hazard evaluation (HHE). The department was planning to install diesel exhaust filtration systems on some of its fire apparatus and wanted NIOSH to evaluate the effectiveness of the control.

In response to the request, NIOSH worked with the department to select stations within the department that would be most appropriate for the study. NIOSH then conducted pre- and post-control evaluations of diesel exhaust in two fire stations. Site visits were made to document levels of the gas-phase and
particulate-phase (soot) constituents of the diesel exhaust, and to evaluate the control’s effectiveness in reducing soot exposures at the stations. Airborne concentrations were obtained for elemental carbon (Ce), sulfur dioxide (SO2), nitric oxide (NO), nitrogen dioxide (NO2), and volatile organic compounds (VOCs). The  first site visit was conducted before the ceramic filters were installed on the engines, and a post-control evaluation was conducted four months late. 

METHODS 
Station Selection

The fire department was composed of six fire stations, of which two were selected for this study. NIOSH representatives believed, based on recent diesel exhaust sampling efforts at other fire stations, that to make a determination as to the effectiveness of the engineering controls, stations with the highest poential exposures should be selected for the study. To determine which stations
would be expected to have the greatest potential diesel exhaust exposures, NIOSH looked at the following variables: age of the diesel-powered apparatus operating at each of the stations, the number of diesel-powered apparatus operating at each station, and the level of activity (number of calls) at each station. Based on these variables for the six different stations, NIOSH selected two stations in which the potential for diesel exhaust exposures was greatest.

Environmental Sampling

Pre- and post-control air sampling was conducted for diesel exhaust constituents in two firre stations. Samples were collected to characterize exposures occurring at the fire stations, and to
provide data for determining the effectiveness of the ceramic filters. During the pre-control evaluation, personal breathing zone (PBZ) and area samples were collected for Ce, a surrogate mea-
sure for whole diesel exhaust. A minimum of four PBZ samples were collected at each station during each sampled tour of duty. Area samples for Ce were collected in the apparatus bay and in the living quarters. Additionally, area samples were collected in the apparatus bay for oxides of nitrogen and VOCs, and grab samples were obtained for SO2. Three tours were sampled during the three-day sampling effort. In order to evaluate only the exposure that occurred at the fire stations, rather than that which occurred while riding in the emergency vehicles, NIOSH personnel collected sampling pumps from the employees as they departed the station. These pumps were paused until the employees returned to
the station, at which time they were promptly restarted and given back to the employees.

Post-control sampling was conducted on four tours over a four-day sampling period. This sampling was limited to area Ce samples collected in the apparatus bay. This decision was made because PBZ sampling for Ce and area samples for the gas-phase constituents yielded very low (or none detected) concentrations in the pre-control evaluation.

Air samples for Ce were collected and analyzed in accordance with NIOSH Method 5040. The samples were
collected on quartz-fiber filters in 37-millimeter(mm) diameter cassettes, connected via Tygon tubing to battery-powered air sampling pumps, which were operated at a low rate of three liters per minute (Lpm). PBZ and general area samples were collected over the three days of pre-control sampling. Additionally, one background sample was collected each day, away from sources of diesel exhaust emissions. Samples obtained after the installation of the ceramic filter were collected using the
same methods, but on 25-mm filters. The smaller filters provide increased sensitivity, due to the reduced volume of solvent needed to extract the smaller filter. 

Grab samples for SO2 were obtained when a diesel-powered apparatus entered or departed the station. These measurements were obtained using Sensidyne 5LC colorimetric detector tubes (Sensidyne, Inc., Clearwater, FL). These colorimetric tubes measure SO2 in a concentration range from 0.1 to 25 parts per million (ppm).

Area air samples for oxides of nitrogen were collected and analyzed in accordance with NIOSH method 6014. This method utilizes two triethanolamine (TEA)-treated molecular sieve sorbent tubes in series, separated by a chromate oxidizer tube, attached via Tygon tubing to a battery-powered sampling pump.
NO2 is collected on the first TEA sorbent tube, and is thereby separated from NO, which is oxidized by the chromate oxidizer tube, and is then collected on the second TEA sorbent tube. Samples to
assess the time-weighted average exposure to oxides of nitrogen were collected at a flow rate of 25 milliliters per minute (ml/min) in the apparatus bay.

To screen for VOCs, area air samples were collected using thermal desorption tubes in accordance with NIOSH method 2459. Thermal desorption tubes contain three sorbent beds in consecutive layers from front to back (Carbopack Y, Carbopack B, and Carboxen 1003), which are used to capture organic compounds over a wide range of volatility. Substances such as acetone, toluene, pentane, and hexane will be trapped with this sorbent tube. This method is an extremely sensitive and specic screening
technique; it will identify the compounds present on the sample in the parts per billion range. Samples were collected in the apparatus bay, beginning when the vehicles departed the station in response to an emergency dispatch, and the pumps were allowed to run for about two hours. The thermal desorption tubes were connected via Tygon tubing to battery- powered sampling pumps that were operating at a calibrated flow rate of 50 ml/min. Samples were analyzed using an automatic thermal desorption system interfaced directly with a gas chromatograph and mass selective detector (GC-
TD-MSD). Stock solutions in methanol containing known amounts of several compounds present in vehicle exhaust were used to prepare spikes to estimate the concentrations of solvents collected on the air samples.

To quantify compounds identified during the analysis of thermal desorption samples, samples were collected on charcoal tubes side-by-side with the thermal desorption tubes. The charcoal tubes
were placed in plastic holders connected via Tygon tubing to battery-powered sampling pumps that were operating at a flow rate of 200 ml/min. Sampling times matched those of the thermal desorption
tubes. Based on the results of the analysis of the thermal desorption tubes, the charcoal tubes were quantitatively analyzed for benzene, toluene, and xylene, using NIOSH method 1501.

Ventilation Assessment

A qualitative ventilation assessment was conducted at each fire station to determine the pressure differentials between the apparatus bay and the living quarters. The assessment included an overview of the heating, ventilating, and air conditioning (HVAC) system’s modes of operation, and a determination of the operating mode’s effects on relative pressures between the living quarters
and the bay. Smoke tubes were used to observe relative pressures through doorways separating the apparatus bay from the living quarters.

Statistical Methods


Preliminary determinations of the number of area samples required to detect a reduction of 50 percent or more in concentrations of Ce were made using levels found in a previous HHE. Power calculations indicated that a minimum of 26 Ce samples (13 pre-control and 13 post-control) from each fire station
would be needed to detect the 50 percent reduction in levels of elemental carbon with 90 percent power, at a level of significance of a = 0.050. Subsequent power calculations using measured pre-
control data indicated that 12 samples from each station would be sufficient. Concentrations in the pre-control data were found to follow a lognormal distribution; therefore, logarithms of concentrations were used for all calculations and statistical tests.

Several sample concentrations were below the method limit of detection (LOD). To be included in statistical analyses, these samples were assigned values equal to LOD divided by the square root of two. All transformations and calculations were done using the Statistical Analysis System (SAS)
v8.0, and plots were prepared using S-Plus v4.0.

Plots of concentrations of Ce by numbers of calls per day were made for each station, pre- and post-intervention. These are identified in Figures 1 and 2.


RESULTS

Pre-Control

Station 3. 

One 1989 E-One medic engine and one 1989 Quint were housed in station 3 during this study. The medic engine was powered by an eight-cylinder series 92-T Detroit Diesel engine. This engine was installed in 1998. The Quint was powered by an eight-cylinder series 92 Detroit Diesel engine, which was installed in 1994. All diesel engines used No. 2 Diesel Fuel.

During the sample period on day 1, the Quint made two runs, and the medic engine made four runs. On day 2, the Quint made three runs, and the medic engine made five runs. On day 3, the Quint and the medic engine each made two runs.

Elemental carbon 

A trace Ce concentration was detected on one of 12 (8%) PBZ samples collected during the three-day sampling campaign. No Ce was detected on the other 11 PBZ samples. Area samples collected in the apparatus bay ranged from none detected to 23.5 ug/m3. These 12 samples had a geometric mean (GM) of 6.1 ug/m3 (ge-ometric standard deviation [GSD] 2.3). A trace concentration of Ce was detected on seven of 30 (23%) samples collected in the living quarters of the fire station.

Sulfur dioxide 

Sulfur dioxide was not detected in any grab samples collected in the apparatus bay. This indicates that the concentration of SO2 in the bay did not exceed 0.1 ppm as the apparatus entered and departed the station.

Nitrogen oxides 

Oxides of nitrogen samples collected on day 1 were not valid and, therefore, were not reported. Trace concentrations of nitrogen dioxide were measured in the bay on days 2 and 3. The measurements indicated that concentrations in the bay were less than0.041 ppm as an 8-hour TWA. These concentrations were well below current exposure criteria. Nitrogen dioxide measurements were obtained during four-hour sampling periods. These sampling periods were too long to permit direct comparison against the short-term exposure criteria that are established for NO2. The concentrations do, however, offer a general idea as to the levels that may have existed during apparatus de- partures, if assumptions are made about the time-concentration pattern. Concentrations during the four-hour sampling periods ranged from none detected to trace, indicating that NO2 was detected in the apparatus bay at an average concentration of less than 0.27 ppm. These concentrations are averaged over four-hour sampling periods, and suggest that the 1 ppm TLV could have been exceeded when apparatuses departed the station.

Volatile organic compounds 

Low concentrations of several organic chemicals were identified in the analysis of the thermal desorption tubes. Identified chemicals included methyl t-butyl ether, C4-C7 alkanes, benzene, toluene, and xylenes. Toluene was also identified on the field blanks. Toluene and xylenes were present in the greatest abundance. Accordingly, the charcoal tube samples were analyzed for toluene and xylenes, as well as for benzene, due to its toxicity. Analysis of charcoal tubes recovered undetectable to trace quantities of benzene, corresponding to airborne concentrations of less than 0.003 ppm. The concentration of xylenes in the apparatus bay ranged from undetectable to
0.004 ppm. Field blanks were contaminated with toluene, which precluded accurate quantitation of toluene concentrations in the bay. The measured concentrations of benzene and xylenes were well below current exposure criteria.

Ventilation assessment 

The HVAC system in Station 3 supplies air to the living quarters of the fire station. Air is not
mechanically supplied to the apparatus bay. The evaluation was made with the HVAC manually turned to the “fan on” position, and again with the fans in the “off” position. Air consistently moved from the living quarters into the bay when the fans were operating. This condition is ideal for keeping diesel exhaust from migrating into the living quarters. When the fans were off, there was slight air movement from the bay into the living quarters.

Station 5. 

A 1997 E-One medic engine and a 1983 Crown Maxum truck were housed at Station 5. The E-One was powered by a 1997 series 60 six-cylinder Detroit Diesel engine. The truck was powered by a six-cylinder 92 non-computerized Detroit Diesel engine.

During the sample period on day 1, the truck made four runs, and the medic engine made five runs. On day 2, the truck made one run, and the medic engine made four runs. On day 3, the truck made one run, and the medic engine made two runs.

Elemental carbon 

Trace concentrations of Ce were detected on 4 of 15 PBZ samples collected during the three-day sampling campaign, indicating exposures in the range of 3 to 24 ug/m3. Area samples collected in the apparatus bay ranged from none detected to 22.6 ug/m3. The GM concentration for the 12 samples collected in the apparatus bay was 15.6 ug/m3 (GSD 1.3). Trace concentrations of Ce were detected on 10 of 28 samples collected in the living quarters of the fire station. No Ce was detected on the other 18 samples.

Nitrogen oxides

Nitric oxide was not detected in three of six samples collected over the three-day period. Three samples had trace quantities, indicating that NO concentrations in the bay were less than 0.41 ppm as an 8-hour TWA. The measured concentrations were well below current exposure criteria. Nitrogen dioxide measurements were obtained during four-hour sampling periods. Concentrations during these periods ranged from none detected to trace, which equate to concentrations of less than 0.09 ppm to 0.27 ppm in the bay.

Volatile organic compounds 

Low concentrations of the organic chemicals identified in Station 3 were also identified in samples collected in Station 5. Toluene was also identified on the field blanks. The charcoal tube samples were analyzed for toluene, xylenes, and benzene. Analysis of charcoal tubes recovered none detected to trace quantities of benzene, corresponding to airborne concentrations of less than 0.003 ppm. The concentration of xylenes in the bay ranged from 0.004 to 0.009 ppm. Field blanks were contaminated with toluene, which precluded accurate quantitation of toluene concentrations in the bay. Concentrations of benzene and xylene in the bay were well below current exposure criteria.

Sulfur dioxide 

Sulfur dioxide was not detected in any grab samples collected in the apparatus bay. This indicates that the concentration of SO2 in the bay did not exceed 0.1 ppm as the apparatus entered and departed the station.

Ventilation assessment 

As with Station 3, the HVAC system in Station 5 supplies air to the living quarters of the fire station, but does not supply air to the apparatus bay. The evaluation was made with the HVAC manually turned to the “fan on” position, and again with the fans in the “off” position. Air consistently moved from the bay into the living quarters when the fans were turned off. When the fans were on, there was slight air movement from the bay into the living quarters. This is opposite to the more desirable condition noted at Station 3.

Post-Control

Elemental carbon.

Station 3

During the sample period on day 1, the Quint and the medic engine each made four runs. On day 2, the Quint made one run, and the medic engine made five runs. On day 3, the Quint made no runs, and the medic engine made six runs. The Quint made no runs on day 4, while the medic engine made five runs. See Figure 1 for an illustration of the number of runs against the Ce concentrations, before and after the controls were installed.

Trace concentrations of Ce were detected on 4 of 16 area samples collected in the bay during the four-day sampling campaign. Four area samples collected on day 1 had trace amounts of Ce, indicating that Ce concentrations in the bay ranged from 1.3 to 5.1 ug/m3. The 16 samples had a GM Ce concentration of 1.5 ug/m3. Geometric mean Ce concentrations were reduced by 76 percent from pre-control levels.

Station 5

During the sample period on day 1, the truck and medic engine each made three runs. On day 2, the truck made three runs, and the medic engine made four runs. On day 3, the truck made no runs, and the medic engine made three runs. The truck made no runs on day 4, while the medic engine made four runs.
Trace Ce was detected on five of 16 area samples collected during the four-day sampling campaign. Ce was detected in two area samples collected on day 2, and in three area samples collected on day 4, indicating that Ce concentrations for these samples ranged from 1.3 t 5.1 ug/m3. The 16 samples had a GM Ce concentration of 1.4 ug/m3. Geometric mean Ce concentrations were reduced by 91 percent from pre-control levels.
