Permissible Exposure Levels for Selected Military Fuel Vapors (1996)

The U.S. Navy is in the final stages of designing a strategic sealift ship. The ship's mission is to transport fueled vehicles to avoid fueling at docking so that the vehicles can be deployed onto the field as soon as they are unloaded from the ship. To accomplish this mission, fueled vehicles must be stored in the ship's cargo holds, transported, and taken out of the cargo holds. The cargo holds will store fueled armored tanks, tanker trucks, other trucks of various sizes, trailors, jeeps, and helicopters. All the vehicles are designed to use jet-propulsion (JP) fuels JP-5 or JP-8. Diesel fuel marine (DFM) is used to operate the ship. The storage and operation of the fueled vehicles in the ship's cargo holds can be hazardous to service personnel because of exposure to fuel vapors and exhaust. The anticipated exposure scenario involves exposure to fuel vapors during fueling operations or while working in the vicinity of fueled vehicles.

Naval personnel might be exposed to fuels and fuel components in a variety of situations. For example, refueling of aircraft or ships can cause exposure to airborne fuel vapors displaced from fuel tanks. Personnel who repair and maintain fuel systems of aircraft and ships have the potential for sporadic high exposures. If precautions are not taken, vapors can reach relatively high concentrations because the work often is performed in confined spaces. For instance, during aircraft fuel-cell-maintenance operations, dilution ventilation and subsequent monitoring of the lower explosive limit are required to lower the fuel-vapor concentration before personnel can enter the cell (Bishop, 1982). The pilots and ships' crews also might be exposed. Crew members might be exposed in the bilge, because bilge water frequently contains marine fuel, either floating on the surface or emulsified. Also, high exposures have been reported in aircraft due to fuel-system leaks (Davies, 1964).

Thirty-three TWA personal-exposure measurements of JP-5 vapor were obtained from NAVOSHNET, the Navy occupational safety and health network, from November 1984 to February 1993. The TWAs ranged from <0.48 to 153 mg/m³. The geometric mean was 4.4 mg/m³, with a geometric standard deviation of 4.8. The 95th percentile was estimated to be 58.8 mg/m³. Table 4-1 shows these results broken down by task performed during monitoring.

Evaluating the toxicity of military fuels, setting PELs, and monitoring air to ensure that the PELs are not exceeded present special problems that result from the physical properties of volatile mixtures. The composition of vapor in equilibrium with a multicomponent liquid when the total pressure is not greatly different

from 1 atmosphere is described by the following equation:

where y_i and x_i are the mole fractions of component i in the vapor phase and the liquid phase, respectively; P_VPi is the vapor pressure (VP) of pure component i and P is the total pressure; and γ_i is the activity coefficient (Reid et al., 1987). The activity coefficient accounts for nonideal behavior of component i in the mixture. When γ_i equals 1, Eq. 1 becomes Raoult's law. Unfortunately, γ_i differs from 1 when intermolecular forces of component i, which hold the liquid together, differ from those of the liquid bulk. γ_i also depends on temperature and, to a lesser extent, on pressure. The overall effect of Eq. 1 is that the equilibrium vapor composition differs from the liquid composition—the vapor being enriched with more-volatile components.

When a multicomponent liquid mixture evaporates over time (for example, the vapors are carried away from the liquid by air movement), the liquid composition and, therefore, the vapor composition change. The most-volatile components are depleted most rapidly from the liquid phase, increasing the liquid mole fraction of the less-volatile materials and thus their vaporization rates. Therefore, it is not possible to assume that the composition of vapor to which personnel might be exposed is constant, much less to assume that the composition is the same as that of the liquid fuel from which the vapors emanate.

In addition to the effect of volatilization phenomena described in this section, fuel composition also depends on the source of crude oil and the refining processes employed in production (Bishop, 1982).

Two approaches have been used to develop limits of acceptable exposure to airborne gas and vapor mixtures. One method is to compute a composite limit on the basis of the toxicity and relative concentrations of individual components (Ball, 1959; Elkins, 1962). For example, the threshold-limit-value (TLV) formula for the mixture requires that the sum of the ratios of the component concentrations equal to the corresponding TLV less than one (ACGIH, 1992). Although well suited for some mixtures, this approach is impractical for fuels because they contain several hundred chemicals, of which only a small number have been studied toxicologically.

The alternative approach is to establish an exposure limit on the basis of the toxicity of the mixture and to state the limit in terms of the sum of all the component concentrations—for example, the TLVs for gasoline, VM&,P naphtha, and Stoddard solvent (ACGIH, 1992). Although more practical than considering each component individually, this approach ignores changes in vapor composition. The toxicity of the vapor mixture is a result of its

composition; however, the relative composition of airborne fuel vapors varies. The composition of a fuel vapor depends not only on the composition of original liquid fuel and the physical properties of its components but also on its evaporative history.

When test animals are exposed directly to liquid fuels (e.g., via gavage, instillation, and patch testing), the test material has a much higher percentage of low-volatile, high-molecular-weight components than do the vapors to which the animals would be exposed in an inhalation study. Thus, studies other than inhalation studies have questionable relevance for setting PELs for airborne fuel vapors.

Vapor mixtures for inhalation toxicity testing generated by partial vaporization of a liquid mixture might not have the same composition as the mixtures to which personnel are exposed. If these data are to be used for setting PELs, vapor compositions in the test chamber and in the occupational setting must be compared.

Airborne fuel vapors can be monitored by portable instruments or by collection on adsorbent material and subsequent laboratory analysis. Both approaches have advantages and limitations. Personal monitoring of breathing-zone exposure to organic vapors is generally performed by using either an activated carbon tube sampler and personal pump or a diffusive sampler. The Occupational, Safety, and Health Administration (OSHA) Method 48, which

was developed for sampling petroleum distillate fraction (PDF), is recommended for monitoring JP-4 and JP-5 fuel vapors (U.S. Navy Environmental Health Center, 1993). Samples are collected by drawing a stream of air through glass tubes containing activated charcoal. Then, the analytes are desorbed with CS₂ and analyzed by gas chromatography with a flame ionization detector (OSHA, 1990). Standards are prepared by adding CS₂ to known weights of liquid PDF, preferably taken from the source of the vapors. The sum of the constituent peak areas of the standards is regressed on the corresponding standard concentrations. The areas of the sample chromatogram peaks with the same retention times as the standard chromatogram peaks are summed to yield the sample response, and the concentration is calculated from the regression equation of the standards. Thus, this method yields a concentration estimate of total vapor weight per volume in air.

In diffusive samplers, the sampling rate is controlled by diffusion of the analyte through a quiescent air layer between the ambient air and the sorbent material. The uptake rate is determined by the ambient concentration and diffusion coefficient of the analyte in air. Diffusive samplers have been used for fuel-vapor sampling (Bishop, 1982). Two suppliers (3M, 1982; Pro-Tek, 1983) give sampling rates for a similar multicomponent hydrocarbon mixture, VM&P naphtha. However, each component of the vapor mixture has a different diffusion coefficient and, thus, a different sampling rate. Because a range of fuel-vapor compositions in breathing-zone air is expected, the sampling rate for each component must be estimated individually.

When rapid on-site measurements of fuel-vapor concentrations are needed (for instance, in area surveys or in checking confined spaces for explosive mixtures), portable instruments based on either continuous flame ionization detection (FID) or heat-combustion detection are used (ACGIH, 1989). Commercial portable FIDs produce ions in a sampled gas by burning organic compounds in a hydrogen-air flame. The conductivity of the gas is then measured. Instrument response is approximately propor

tional to the concentration of carbon atoms in the sample but is depressed somewhat for compounds containing electronegative elements, such as oxygen, sulfur, and halogens. Portable FID instruments generally have a wide linear range and lower limits of detection from 0.01 to 5 ppm for compounds such as methane (ACGIH, 1989).

Heat of combustion detectors oxidizes organic compounds in the sampled gas stream releasing heat, and the resulting increase in temperature is measured. Instrument response is displayed either in concentration units or in percentage of the lower explosive limit. The lower limit of detection of the most sensitive instruments is from 1 to 3 ppm, and the upper limit of the usable range varies considerably among the available instruments (ACGIH, 1989).

Most portable FID and combustible gas instruments are nonspecific, responding to nearly all organic compounds. Vapor mixtures producing equal instrument responses could have different compositions and, thus, different toxicities. Also, these instruments typically are calibrated against a test gas containing a single organic compound, such as methane. Fuel vapors at 1 ppm will likely produce a different response from the calibration gas at 1 ppm due to differences in the number of carbon atoms per molecule and in the response per carbon atom. Finally, instrument readings in parts per million must be converted to milligrams per cubic meter for comparison with permissible limits. Therefore, the average molecular weight of the vapor mixture must be estimated. The average molecular weight of the liquid fuel is expected to be considerably higher than the average molecular weight of fuel vapors. Thus, use of the average molecular weight of the liquid in unit conversion will result in overestimation of vapor concentration in milligrams per cubic meter.

In considering the limitations of the available methods, nonspecific methods are recommended only for rapid surveys and testing for explosion hazards, because the instrument response varies with vapor composition. Solid sorbent collection using either

pumped sorbent tubes or diffusive samplers accurately determines the total weight-per-volume concentration of fuel components collected. With diffusive sampling, conversion of the weight collected to airborne concentration is complicated because the sampling rates for each component differ. Thus, active sampling on sorbent tubes is generally the most satisfactory method for measuring personal exposure to fuel vapors.