By

Douglas J. Crawford-Brown PhD.
Institute for Environmental Studies
University of North Carolina at Chapel Hill

April, 1996

I

Introduction

This paper provides a preliminary comparative risk assessment of two different options for
managing and treating the U.S. chemical agent stockpile at several facilities:

1) direct incineration, with separation of chemical agents from explosives taking place
approximately at the time of incineration; and

2) immediate separation of chemical agents from explosives, followed directly by
neutralization of the chemical agent and storage.

The facilities at which chemical agents will be treated and/or managed include the Anniston Army
Depot; Tooele Army Depot; Pine Bluff Arsenal; Umatilla Depot Activity; Aberdeen Proving
Ground; Pueblo Depot Activity; Newport Army Ammunition Plant; and Blue Grass Army Depot.
The present report uses the Anniston Army Depot as a representative case for the comparative risk
assessment.

The chemical agents to be treated and managed are:

• the blister agents H; HD; HT (all based on sulfur mustard) and Lewisite (an arsenical); and

• the organophosphate nerve agents GA; GB; and VX.

These agents all are stored currently as bombs; cartridges; mines; projectiles; rockets; spray tanks;
and ton containers (Army, 1995). Associated with these munitions are explosive charges which
would release the agents into the environment if activated by accidents such as forklift damage,
fires or aircraft accidents at the site.

The U.S. Department of Defense (DOD) is in the process of demilitarizing these weapons, and has
selected the treatment and management option of incineration. Before proceeding, DOD is required
to perform a risk assessment through the United States Environmental Protection Agency (EPA)
Resource Conservation and Recovery Act (RCRA). A screening risk assessment has been
published in response to this requirement (Army, 1995). In it, DOD describes the four primary
steps for demilitarization in their proposed treatment and management option:

1) transport and handling of munitions from the existing storage area;

2) dis-assembly of the munitions, separating explosives from chemical agents;

3) incineration of the chemical agents and the bulk containers; and

4) management of waste materials remaining after incineration.

The waste materials include both atmospheric releases and residual solid waste. Four furnace
systems/incinerators are designed to carry out these activities (described below).

The need for a formal risk assessment arises from the fact that the incineration process will release
known hazardous materials into the environment; and the storage of the nerve agents within
munitions presents the possibility that charges associated with these munitions may explode before
dis-assembly, releasing the original chemical agent into the environment. The agents may be
released into the environment due to external accidents, such as aircraft crashes onto the storage
site, and internal accidents such as forklifts striking the munitions and causing explosions.

The particular materials of concern in all of these cases (i.e., the substances of potential concern
[SOPCs]) include:

• a variety of dioxins and furans, some of which have been identified to date and some of which have yet to be identified;

• 48 other organic substances including the original chemical agents and packing material, called principal organic hazardous constituents (POHCs); and chemicals produced from incomplete combustion of the incinerated materials, called products of incomplete combustion (PICs):

° 20 metals;
° 2 acid gases (hydrogen chloride and hydrogen fluoride);
° chlorine;
° particulates (Army, 1995).

A complete listing of these materials is provided in Table 1 (See p. 9) , reproduced from the DOD
screening risk assessment (Army, 1995) The materials considered quantitaively in the DOD
assessment are:

° dioxin-like compounds, particularly 2,3,7,8 TCDD-TEQ (toxic equivalents);
° Diethyl Phthalate;
° Di-n-Butyl Phthalate;
° Di-n-Octyl Phthalate;
° bis-(2-Ethylhexyl) Phthalate;
° 4-Methylphenol;
° Napthalene;
° GB (a nerve agent);
° HD/HT (nerve agents);
° VX (a nerve agent);
° Total PCBs;
° 2,4-Dinitrotoluene;
° 2,6-Dinitrotoluene;
° Diphenylamine;
° RDX;
° Antimony;
° Arsenic;
° Barium;
° Beryllium;
° Cadmium;
° Chromium;
° Lead;
° Manganese;
° Mercury;
° Nickel;
° Selenium;
° Silver;
° Thallium;
° Vanadium.

In addition to these compounds, the present assessment includes consideration of hydrogen
chloride, since this acid gas is an important component of the non-cancer health effects produced
by incineration (Research Triangle Institute [RTI], 1995).

For each of these materials, a complete risk assessment includes consideration of risks from
inhalation of the materials released into the air, called the direct exposure pathway; as well as
ingestion and dermal absorption of these materials once they deposit onto the ground and travel
through the different media (water, soil, food, etc), called the indirect exposure pathways .
This complete assessment considers human health risks, including both cancer and non-cancer
effects, and ecosystem risks. The present report contains a comparative risk assessment for only
the human health risks. The comparative merits of the alternative treatment and management
options considered here would be unchanged by including ecosystem risks.

Exposures to the public typically are calculated by considering both the average health risks in a
population, as well as the health risk to some maximally exposed individual (or MEI). In the DOD
assessment, and in the present comparative risk assessment, the health risks to a maximally
exposed individual is used as the measure of desireability for each treatment and management
option. This individual is a subsistence farmer who consumes homegrown vegetables, beef and
milk, and who ingests fish from the East Aboga Fish Hatchery. The calculated risk includes both
the direct exposure pathway (inhalation) and the indirect exposure pathway (ingestion of the food
mentioned above), and the calculations follow the guidelines of the EPA methodology for
hazardous waste combustion facilities (EPA, Guidance 1994) Pathways of exposure are shown in
Figure 1. The location of the maximally exposed individual is shown in Figure 2, taken from the
DOD screening risk assessment (Army, 1995). The measure of risk is the probability of adverse
effect (either cancer or non-cancer) over a lifetime.

In all calculations performed in this report, the methodology adopted in the DOD assessment is
used to provide a baseline case, with corrections noted in the text. No attempt is made to reproduce
these calculations independently here, as this was beyond the scope of the project; instead, the
calculations in the DOD report were reproduced in the initial stages of this study simply to ensure
they were correct given the assumptions made by the DOD risk assessors.

The risk assessment considers two components of risk: incineration and storage. The risks of
incineration are developed from the methodology and information in the DOD screening risk
assessment (Army, 1995). In addition to performing a risk assessment for incineration, DOD has
performed a preliminary risk assessment for accidents taking place during storage (Army, 1988).
These values have been adopted here for the comparative risk assessment, using the information
specific to the Anniston site.

The DOD baseline case for incineration, after suitable adjustments described below, indicates a risk
to human health of potential regulatory concern. Due to this, it is necessary to consider alternative
treatment and management options for the stockpiles. In the DOD procedures, incineration takes
place after separation of the chemical agents from the explosives. Prior to that time, there is a
possibility of explosion, resulting in releases of the original chemical agent into the environment.
Even after separation, there is a possibility of release due to external events, although the time
between separation and incineration generally will be short.

Given the risk from incineration and the risk from acute releases due to accidents, a second
treatment and management option is considered in the present report. This has been chosen to be
significantly different from the proposed DOD procedure, so as to provide a comparative risk
assessment across the range of options. In this second option, the explosives and nerve
agents are separated immediately; the nerve agent is neutralized using techniques
described by the National Research Council (National Research Council [NRC], 1993);
and the neutralized material stored until a process other than incineration is
selected and shown to be technologically feasible. Both the DOD and this alternative
strategy present risks to the public and to workers at the site.

Since the methodology selected here is preliminary and screening in nature, the results should be
used only to indicate whether or not a more complete comparative risk assessment should be
performed prior to any final decisions on disposal method.

II

Methodology

For the risk from incineration, the methodology for risk estimation uses information on:

• operational characteristics of the facility;

• emissions source terms, or the amount of different materials released into the air;

• meteorological data, needed to estimate the direction of air movement and dispersion into the atmosphere, as well as deposition onto the ground;

• topographical/terrain data, needed to estimate dispersion into the atmosphere, as well as deposition onto the ground;

• hydrogeological data, needed to estimate where local bodies of water are located; the average annual rainfall; the average annual irrigation; the average annual surface water runoff; the average annual evapotranspiration; and flow characteristics of the watershed emptying into the lake considered here;

• demographic data, needed to determine population density; and

• biological data, needed to estimate rates of ingestion for food and water, and rates of inhalation of air.

Similar information is needed to estimate the risks from accidents during storage, but the source
term is related to the kind and severity of the accident rather than the operational characteristics of
the incinerator.

As in the DOD assessment (Army, 1995) all emissions estimates are obtained from the Johnston
Atoll Chemical Agent Disposal System (JACADS). The components of the incineration facility of
interest here are a Liquid Incinerator System (LIC) for treating the liquid chemical agent; a
Deactivation Furnace System (DFS) processing the drained rockets, energetics and any residual
agent; a Metal Parts Furnace (MPF) for processing drained metal ton containers, bombs and
projectiles, and any residual agent; and the Dunnage Incinerator (DUN) for processing dunnage.
Risks from accidents are based primarily on the major source of an accident, which is the rate of
airplane crashes into sites, and on accidents involving forklifts. These rates, and the resulting
release, are based on generic values for aircraft and industrial accidents in the United States,
coupled with estimates of the stockpile to be managed and treated at the Anniston site.

The process of risk estimation then includes a series of calculations:

• Estimation of the rate of release of chemicals and metals following incineration, or the total release of chemical agent following an accident;

• Use of the dispersion models to estimate the concentration of the various materials in the atmosphere at locations surrounding the facility;

• Estimation of the rate of deposition of the various materials onto land and water surfaces;

• Estimation of the rate of movement of the various materials into the important lake from the land and from tributaries;

• Estimation of the concentration of the various materials in the water of the lake and in other water supplies;

• Estimation of the rate of uptake of the various materials into fish (for the lake), animals used for food, and crops (due to uptake from the soil and due to direct deposition onto leaf surfaces);

• Estimation of the rate at which humans inhale the contaminated air or ingest the contaminated food; this is summarized as an Average Daily Rate of Intake (ADRI) in units of grams of material per kilogram of body mass per day; the body weight used here is 70 kg, as this is the weight that corresponds to the rates of ingestion and inhalation employed in the risk assessment; and

• Estimation of the risk of cancer and/or non-cancer health effects due to this rate of intake; for cancer, the probability of effect is obtained by multiplying the ADRI by a slope factor (in units of probability of effect per gram per kilogram per day); for non- cancer, the regulatory goal is to prevent the ADRI from rising above 25% of a prescribed limit at any point in time; this is summarized as a hazard index (HI), which is the ratio of the ADRI for the scenario of interest divided by the ADRI corresponding to the regulatory limit; all values of the HI for different substances are summed to produce a composite HI.

A. INCINERATOR OPERATIONS

Results from the DOD study for incinerator operations are presented in Table 18 of the screening
risk assessment (Army, 1995). The total lifetime cancer risk for the subsistence farmer is
estimated by DOD to be approximately 6 x 10-6 (six chances of death per million). This is due in
large measure to the risks from chromium (Cr) and the various forms of dioxins (PCDDs) and
furans (PCDFs). Approximately one-third (2.3 x 10-6 ) is due to the PCDDs and PCDFs, and the
remainder (3.7 x 10-6 ) to Cr.

It is necessary to correct these values specific to the PCDDs and PCDFs due to inappropriate
assumptions used in the DOD assessment. These corrections are as follows:

• The DOD study assumed that the subsistence farmer obtains 40-45% of his/her beef and milk from the farm. A better conservative assumption is that 100% is obtained from the farm. This raises the PCDD/PCDF cancer risk to 5.4 x 10-6, or the total cancer risk to 9.1 x 10-6.

• The DOD study uses an upset factor of 2.8 to account for emissions during upset conditions. This does not account for emergency conditions, in which the release rate is 100 times that of normal operating conditions. A more reasonable assumption is that emergency conditions would occur 1% of the operating time, with an overlap of poor combustion conditions (upset) and emergency conditions occurring 20% of the time emergency conditions occur (20% of 1%=0.2x0.01=.002 or 0.2%). On this assumption, that normal operation occurs 79.2% of the time, poor operating condition upsets occur 19.8% of the time, emergency conditions 0.8% of the time, and overlapping emergency and upset conditions 0.2% of the time, the emissions would be 5.57 times greater than normal. Thus, the total upset and emergency value would be about twice the 2.8 value used in the Anniston assessment (Army, 1995). This raises the PCDD/PCDF cancer risk from incineration to 10.8 x 10-6, or the total cancer risk to 14.5 x 10-6.

• The DOD study uses only the slope factors for identified PCDDs and PCDFs. This ignores the fact that between 80% and 90% of the dioxin-like compounds in the emissions
  have not been identified to date. The DOD study makes the assumption that these unidentified compounds are not carcinogenic. A more reasonable conservative assumption, as recommended in EPA guidance (EPA, 1994), is that the unidentified compounds are equally carcinogenic as the identified compounds. In the present report, the slope factor is multiplied by a factor of 5 to account for this feature of the DOD risk assessment. This raises the PCDD/PCDF cancer risk to 54 x 10-6, or the total cancer risk to 57.7 x 10-6.

• The DOD study fails to display the full extent of the uncertainty in the cancer slope factor. In the EPA 1994 Dioxin Health Assessment Report (EPA, Assessment 1994), which presents both human epidemiological data and animal data, the slope factor based on human data is higher than the value adopted by DOD by a factor of 20. If this uncertainty is reflected in the analysis, this raises the PCDD/PCDF cancer risk to between 54 x 10-6 and 1080 x 10-6, or the total cancer risk to between 57.7 x 10-6 and 1084 x 10-6 .

These values clearly all are in excess of guidelines established at 10 x 10-6, regardless of the "end"
of the confidence interval upon which one focuses (i.e. regardless of whether one looks at the
lower end of 57.7 x 10-6 or the upper end of 1084 x 10-6).

As to non-cancer risks, the results of the DOD study are displayed in Table 22 of the screening risk
assessment (Army, 1995). The estimate of the hazard index (HI) for a resident is 0.1. If the
correction for emergency conditions (a factor of 2) is applied, this HI rises to 0.2.

This estimate does not, however, account for the non-cancer effects of dioxin-like compounds.
The DOD risk assessors chose to ignore these effects, although the ADRI was estimated to be
approximately 0.02 picograms per kilogram per day. If the correction factors for dioxin-like
compounds mentioned above are applied (with the exception of the correction for slope factor,
which applies only to cancer effects), the ADRI for the dioxin-like compounds is:

ADRI = 0.02 x 2 x 2 x 5 = 0.4 picograms per kilogram per day

This may be compared with the Agency for Toxic Substances and Disease Registry (ATSDR)
suggested minimal risk level (MRL) of 1 picogram per kilogram per day (ATSDR, 1989). The
Anniston incineration facility would, therefore, produce an HI of 0.4/1.0 or 0.4. This must be
added to the value of 0.2 for the other compounds considered in the DOD report to obtain a total HI
of 0.6.

This HI is above the threshold regulatory value of 0.25. The DOD assessment does not, however,
consider hydrogen chloride (at least that acid gas is not highlighted in Table 2 of the DOD report,
and highlighting indicates a compound was considered in the assessment). In a recent assessment
of the health risks from facilities burning fuel, however, hydrogen chloride dominated calculations
of HI, representing more than 50% of the non-cancer health impact (RTI, 1995). Even if the non-
conservative assumption that hydrogen chloride non-cancer risk is equal to that of the compounds
identified in the DOD assessment is adopted, the HI for incineration would be 1.2, which is above
a target value of 0.25.

B. ACCIDENT RISKS

The risks from accidents are described in the DOD programmatic environmental impact statement
(Army, 1988). The primary causes of accidental releases are airplane crashes onto the site and
damage from forklifts resulting in detonations. As a result, the final estimate of risk depends
critically on whether air traffic is controlled or uncontrolled in the airspace over the site. The DOD
assessment does not incorporate control of air traffic, but mentions this management strategy as a
possibility in the future. Both scenarios will be considered here.

If air traffic is not controlled, the results of the DOD assessment for storage risks are displayed in
Figure 4.3.2 and Table 4.5.4 of the 1988 report (Army, 1988). The expected fatalities (see Table
4.3.2) are between 10-3 and 10-2 . The person-years at risk (Table 4.3.2) is between 105 and 106 .
If the midpoints of these two intervals is used, the expected annual rate of fatality (EARF) is:

EARF = 5 x 10-3 / 5 x 105 = 10-8 fatalities per person per year.

For 25 years of storage (also assumed in the DOD analysis), the lifetime probability of fatality will
be:

Lifetime Probability of Fatality from Accidents = 25 x 10-8 = 0.25 x 10-6

This value is an average over the exposed population, which includes effects out to 100 kilometers.
As displayed in Figure 4.2.1 of the 1988 DOD report (Army, 1988), most individuals who are
exposed are in the dosage range producing a rate of fatalities of between 0% and 1%. The
subsistence farmer in this analysis is located at a distance of approximately 4 kilometers from the
facility. At this distance, the risk from an accident is between 50 and 100 times that of the average
for the exposed population. The subsistence farmer will, therefore, have a lifetime risk of between
13 x 10-6 and 26 x 10-6 .

If air traffic over the site is restricted, the probability of a release from aircraft accidents is reduced
greatly. As a lower limit, it may be assumed that all such accidents are avoided. As shown in Table
4.5.4 of the DOD analysis (Army, 1988), the risk from aircraft accidents constitutes 90% of the
total risk from storage. Removing the aircraft accidents would, therefore, lower the risks calculated
above by a factor of 10. With complete restriction of air traffic, therefore, the value of EARF from
above would be reduced to:

EARF = 0.1 x 10-8 = 10-9 fatalities per person per year.

For 25 years of storage (also assumed in the DOD analysis), the lifetime probability of fatality will
be:

Lifetime Probability of Fatality from Accidents = 25 x 10-9 = 0.025 x 10-6

Again, this value applies to the average exposed individual. The risk to the subsistence farmer
would be between 50 and 100 times this value, or between 1.25 x 10-6 and 2.5 x 10-6 . The above
values for the risks from accidents do not reflect the impact of neutralization of chemical agents.
The alternative management and treatment strategy, as mentioned previously, is to separate the
chemical agent from explosives, neutralize the agent, then store the resulting material. This
removes the risk from incineration, but presents a risk from continued storage.

The degree of lowering of the risk from storage depends on the degree of neutralization. It also
depends on the relative contribution from aircraft accidents and forklift accidents, since the effect of
a forklift accident is greatly reduced through separation of the chemical agent from the explosives.
Since aircraft accidents dominate the risk in the DOD analysis, however, it is assumed here that the
residual risk from storage after neutralization is directly proportional to the residual chemical agent
after neutralization.

Three degrees of neutralization are considered here: 90% neutralization; 95% neturalization; and
99% neturalization. For each of these, the risk from storage noted above (between 13 x 10-6 and
26 x 10-6 if aircraft are not restricted above the site; between 1.25 x 10-6 and 2.5 x 10-6 if aircraft
are restricted above the site) is multiplied by the fraction (100% - %neutralization / 100%) to obtain
the residual risk from storage after neutralization. This produces the following six ranges of
estimates:

• No restriction of aircraft; 90% neutralization: The residual risk is between 13 x 10-6 x (100 - 90 / 100) = 1.3 x 10-6 and 26 x 10-6 x (100 - 90 / 100) = 2.6 x 10-6

• No restriction of aircraft; 95% neutralization: The residual risk is between 13 x 10-6 x (100 - 95 / 100) = 0.7 x 10-6 and 26 x 10-6 x (100 - 95 / 100) = 1.3 x 10-6

• No restriction of aircraft; 99% neutralization: The residual risk is between 13 x 10-6 x (100 - 99 / 100) = 0.13 x 10-6 and 26 x 10-6 x (100 - 99 / 100) = 0.26 x 10-6

• Restriction of aircraft; 90% neutralization: The residual risk is between 1.3 x 10-6 x (100 - 90 / 100) = 0.13 x 10-6 and 2.5 x 10-6 x (100 - 90 / 100) = 0.25 x 10-6

• Restriction of aircraft; 95% neutralization: The residual risk is between 1.3 x 10-6 x (100 - 95 / 100) = 0.07 x 10-6 and 2.5 x 10-6 x (100 - 95 / 100) = 0.13 x 10-6

• Restriction of aircraft; 99% neutralization: The residual risk is between 1.3 x 10-6 x (100 - 99 / 100) = 0.013 x 10-6 and 2.5 x 10-6 x (100 - 99 / 100) = 0.025 x 10-6

III

Summary and Implications

The results of this study are summarized in Table 2 (See p 9.), which compares the range of
predicted risks for the two treatment and management options (the DOD proposal of incineration
and the alternative of separation and storage). From this table, it may be seen that the risks vary
across several orders of magnitude between the different options, from a low of between 0.013
and 0.025 per million (the Alternative treatment and management option with restricted aircraft over
the site and 99% neutralization) to a high of between 57.7 and 1084 per million (DOD option with
adjustments noted in the text of this report). Only the DOD option, with corrections noted in this
paper, exceeds the limiting risk of 10-5 .

This preliminary screening comparative risk assessment indicates that there is reason to believe that
alternatives to incineration may be preferable from the standpoint of public health risk. A complete
comparative risk assessment for alternative treatment methods should, therefore, be conducted
prior to final decisions.

REFERENCES

Agency for Toxic Substances and Disease Registry. (1989, June). Toxicological profile for
2,3,7,8-tetrachlorodibenzo-p-dioxin. ATSDR/TP-88/23.

US Department of the Army. (1988, January). Chemical Stockpile Disposal Program: Final
Programmatic Environmental Impact Statement.Vol. 1, Sections 1-8. Aberdeen Proving
Ground, MD: Program Manager for Chemical Demilitarization.

US Department of the Army. (1995, March 27). Final Screening Risk Assessment, RCRA Part B,
Risk Assessment No. 39-26-1399-95, Anniston Chemical Demilitarization Facility,
Anniston Army Depot, Anniston, Alabama. Aberdeen Proving Ground, MD:
Commander US Army Chemical Demilitarization and Remediation Activity.

US Environmental Protection Agency. (1994, April 15). Risk assessment implementation
guidance for performing screening level risk analyses at combustion facilities burning
hazardous wastes. Revised Draft. Washington, DC: Office of Emergency and Remedial
Response and Office of Solid Waste.

US Environmental Protection Agency. (1994, August). Health assessment document for 2,3,7,8-
tetrachlorodibenzo-dioxin (TCDD) and related compounds.Vol. III. EPA/600/BP-
92/001C.

US National Research Council. (1993). Alternative technologies for the destruction of chemical
agents and munitions. Washington, DC: National Academy Press.

Research Triangle Institute. (1955). Preliminary uncertainty analysis for the characterization of
human health risks from direct inhalation exposures to electric utility HAP emissions.RTI
Project No. 94U-5846-105.

Table 2. Comparison of the lifetime risks from the two treatment and management options
considered in this report. The risk is to the maximally exposed individual, taken to be a subsistence
farmer. A range is reported for each based on the uncertainties mentioned in the text. The options
are ranked from low (the lowest risk) to high (the highest risk). The DOD option refers to
separation at or near the time of incineration, followed by incineration. Risks are dominated by
incineration, rather than storage. The Alternative option is separation of the explosives and nerve
agent immediately, followed by neutralization at the indicated efficiency, followed by storage for
25 years.

Option and Conditions Lifetime Risk (chances per million)

Alternative; Aircraft restriction; 99% neutralization 0.013 to 0.025

Alternative; Aircraft restriction; 95% neutralization 0.07 to 0.13

Alternative; Aircraft restriction; 90% neutralization 0.13 to 0.25

Alternative; No aircraft restriction; 99% neutralization 0.13 to 0.26

Alternative; No aircraft restriction; 95% neutralization 0.7 to 1.3

Alternative; No aircraft restriction; 90% neutralization 1.3 to 2.6

DOD; Original Assessment 6

DOD; With corrections; Aircraft restriction 57.7 to 1080

DOD; With Corrections; No aircraft restriction 57.7 to 1080

Table 1. Substances of potential concern (SOPCs) as identified in the DOD analysis and as used in
the present analysis.

Figure 1. The pathways for exposure to the substances of potential concern resulting from the
treatment and management facility.

Figure 2. The locations of various subpopulations used in the DOD analysis. Only the subsistence
farmer was used in the present analysis. It may be noted that the exposed subpopulations are
located approximately 5 to 10 kilometers from the proposed site of the disposal facility.

CWWG Home Page

Contact us: Chemical Weapons Working Group Kentucky Environmental Foundation P.O. Box 467 Berea, KY 40403 phone: 859-986-7565 fax: 859-986-2695 For comments about this WWW page contact Lois Kleffman.