This chapter, together with the appendices, describes the steps of the risk assessment, the data used in the analysis, and the models and methods used to quantify risk and uncertainties.

The models used in this risk assessment address the spread of tritium releases from the NTLF, contact with human populations, metabolism within the human body, and potential health impacts. The primary goal of these models is to clarify how tritium

The models’ parameters were verified by matching calculated tritium concentrations with the tritium concentrations measured in air and surface runoff in the vicinity of the NTLF and in Berkeley. Using site-specific data and the calibrated parameters, the models determine the tritium concentrations in air, surface water, soils, and plants adjacent to the NTLF. These contamination levels provided the basis for estimating the corresponding doses as well as the risks to people in the three zones.

A tritium atom has two neutrons, while the common hydrogen (protium) atom has none. Tritium’s atomic weight is 3. Deuterium, with one neutron, has an atomic weight of 2, and protium’s atomic weight is 1 (see Figure 4-1). Tritium decays to helium through beta decay—the emission of a beta particle from its nucleus:

Tritium’s radioactivity per unit mass makes it a good tracer and label. A count of nuclear decays per unit time from a known mass or volume allows specific activity (analog of concentration) to be calculated. Because nuclear disintegrations are easily monitored, any molecules that have incorporated tritium are easily differentiated from non-labeled forms. Therefore, tritium has great utility for experiments in which the transport or fate of particular chemicals must be identified. Tritium can be incorporated, via substitution for protium, into any molecule or compound that contains hydrogen.

The radioactive half life of tritium is 12.3 years. The radiological units commonly used to characterize exposure, dose, dose equivalent and radiation protection standards for tritium are listed in Table 4-1.

Table 4-1
Common Radiological Units

A small fraction of tritium is incorporated into organic molecules exposed to tritium gas and HTO. This organically bound tritium (OBT) has a different metabolism than HTO. Following exposure to HTO, tritium may be converted to OBT and retained in the body for a longer period [Murphy, 1984]. Table 4-2 gives the physical and chemical characteristics of tritium that are relevant for a dose and risk assessment.

Table 4-2
Physical and Chemical Characteristics of Tritium

The NTLF is designed and operated so that multiple layers of defense restrict the release of tritium to the environment.

Tritium comes to the NTLF facility chemically bound to a uranium bed contained in a sealed metal cylinder. At the NTLF the cylinders are attached to a closed system inside a ventilated glove box. The cylinders are heated to release the chemically-bound tritium from the uranium. All formation reactions take place in the closed system within the glove box.

Two main components of this system are designed to trap or recirculate tritium to prevent its release to the environment:

Tritium can enter the ambient air compartment from the NTLF via three major pathways (see Figure 4-2):

All postulated releases from the NTLF go to the air compartment as HTO. When tritium is released to the environment—whether into air, water, or soil—it spreads rapidly.

Figure 4-2. NTLF exhaust schematic.

The predominant influences on the relative distribution of HTO in the environment are [Murphy, 1993]:

The two influences suggest the importance of understanding the water budget of the release site. The environmental cycling of tritium follows quite closely that of natural hydrogen as it occurs in gas, water, and organic molecules. Perhaps because of hydrogen abundance, no sequestering processes (bioaccumulation) in biota have evolved for hydrogen (or tritium), and thus there is no mechanistic reason to expect biomagnification, which is a metabolically mediated buildup in biological tissues. However, because hydrogen can bind to organic molecules, we might expect some accumulation as a result of OBT.

A review of the field study literature relating to the fate and transport of HTO in the environment provided some insight into the distribution of HTO subsequent to a continuous atmospheric release, and proved helpful in developing a customized HTO fate and transport model. The sections that follow look at major components of this model. Appendix A contains details of the model structure and the rationale for the model components.

For Zones 1 and 2, we used a two-compartment model to estimate the steady-state distribution of HTO between air and soil as a result of a constant HTO emission to the atmosphere. This is an idealization, constructed to obtain an approximate HTO distribution that is consistent with HTO affinities, site-specific characteristics, and observed environmental parameters (e.g., precipitation, runoff, and infiltration). See Appendix C for an account of site-specific characteristics.

Using the principle of mass conservation for each compartment, the model provides an algorithm for predicting the steady-state HTO concentrations in the soil and air compartments. The overall model structure is illustrated in Figure 4-3. The premises of this modeling approach are as follows:

Figure 4-3.  Compartment model for assessing the environmental fate of tritiated water emissions from the NTLF.

The steady-state equations describing gains and losses in the two compartments were solved for the steady-state inventory in each compartment. Table 4-3, gives the gains and losses considered for each compartment. Equations 4-1 and 4-2 express gains and losses for the air and soil compartments, respectively.

and

Table 4-3
Summary of HTO Exchange and Loss Between Air and Soil

By solving Equations 4-1 and 4-2 simultaneously we obtain:

These solutions are used later to estimate tritium risk to human health. The rate constants used in the equations were selected to ensure that the model reproduces the available measured concentrations of tritium in the air and water from the NTLF site.

For Zone 3 we used the CAP88-PC microcomputer radionuclide dispersion and dose-assessment code supplied and approved by US/EPA. This model is able to calculate doses and risks to individuals at various distances from the tritium source. For the point of maximum off-site exposure, the dose represents the cumulative exposure from all significant exposure pathways (inhalation, ingestion, air immersion, and surface exposure). The methods and parameters used to calculate the dose are very conservative. In this risk assessment, however, we only use the tritium concentration in air calculated by CAP88 as input data in our model for risk calculations.

For releases of tritium from the NTLF, pathways leading to internal exposures of tritium include:

In constructing dose models one needs to define the characteristics of individuals in various age/sex categories and the characteristics of the microenvironments in which they live or from which they obtain air, water and food. This section defines the types of anatomical and activity data needed to carry out the dose assessment and provides representative values for these parameters. Appendix B describes how these data were obtained and evaluated. An arithmetic mean value and a coefficient of variation (CV)—the arithmetic standard deviation divided by the arithmetic mean—are defined for all exposure factors used in this assessment.

In the basic exposure model, the exposure duration (ED) is the amount of time, in years, that the exposed population is assumed to be in contact with a specified environmental contaminant. The averaging time (AT) is the period, in days, over which exposure is averaged. The estimates of risk given in section 4.7 are based on population exposures and the committed dose in a defined population on an annual basis. Thus, for this risk assessment ED and AT represent the same period of time and the ratio ED/AT in Equation 4-7 is [1 y]/[365 d].

The arithmetic-mean and CV of body weight and body surface area for three age groups—infant, child, and adult is given in Appendix B. The body weight for infants is needed for estimating exposures to tritium in breast milk. Appendix B also contains an evaluation of available data on population breathing rates, ingestion of water during recreational activities, ingestion of breast-milk by infants, and ingestion of home-grown fruits, vegetables and grains. The descriptions, parameter symbols, and values prepared for these inputs are summarized in Table 4-4.

Footnotes to Table 4-4:

(a) Listed are the arithmetic-mean value and (in parentheses) the estimated CV. Body weights are from Najjar and Roland [1987], breathing rates are from ICRP [1975], tap water intakes are from Yang and Nelson [1986] and Ershow and Cantor [1989], and food intakes are from Yang and Nelson [1986].

(b) The child category covers ages 0 to 15, the adult category covers ages 16 to 70, the combined category is used to represent lifetime equivalent exposure and is obtained by multiplying the child category by 15/70, the adult category by 55/70, and then summing these products.

(c) Breast milk intakes are from Butte et al. [1984] and Whitehead and Paul [1981].

Activity patterns define the frequency and duration that could expose individuals to tritium-contaminated media—air, water, soil. For occupational exposures the exposure time per day is typically 8 to 10 hours and the exposure frequency (EF) is 200 to 300 days per year (assuming 50 working weeks per year and from 4 to 6 working days per week). For non-occupational exposures, the U.S. EPA [1989a] reports the average time spent at home indoors as 8 to 16 h/d and the time spent at home outdoors as 0.3 h/d, with an estimated CV in both cases of 0.14.

For estimating possible doses from water recreation activities in creeks running down from the Berkeley Hills, we assume every adult in Zone 3 wades 2 hours per day for 30 days annually (the U.S. EPA [1992a] reports that for recreational swimming, the time spent swimming ranges from 0.5 h/d to 1 h/d and the frequency ranges from 5 d/y to 150 d/y, see Table 4-5).

Table 4-5
Values and Units of Exposure Parameters Used to Characterize Activity Patterns.

This section looks at how exposures to tritium are linked to potential health impacts within the exposed population. First, a review of the potential health effects associated with tritium establishes the relation between tissue concentration and radiological dose. To establish the appropriate relation between exposure and dose, we next look at metabolism and dosimetry, followed by regulatory standards that have been established to limit exposure and dose. A derivation of the risk factors used to estimate the incidence of health impacts within the exposed population concludes the section.

To quantify the likely magnitude of health risks associated with exposures to tritium, it is necessary to address tritium metabolism and dose in terms of radiation biology and radiation protection methods.

The unit for energy absorption in living tissues is the gray (Gy). One Gy represents one joule of energy deposited per kilogram (kg) of tissue.

Because few studies are available on tritium-induced health injury in humans, health-risk estimates for tritium must be based on the large number of experiments with animals and cell cultures [Straume, 1993]. These experiments show that exposure to tritiated water results in mutations and cell disruptions that can lead to cancer, heritable genetic defects, and reproductive and developmental effects [Straume, 1993]. These effects are qualitatively similar to the injuries caused by low linear-energy-transfer (LET) radiations produced by cesium-137 and cobalt-60, such as x-rays and gamma-rays, for which a significant amount of human data is available. The quantitative difference between tritium beta emissions and a standard low-LET radiation is expressed by the relative biological effectiveness (RBE). The RBE for tritium ranges from 1 to 3 [Straume, 1993]. This means that although the health effects of the beta doses from tritium are believed to be qualitatively similar to health effects of gamma rays, the beta rays are believed to be more potent per unit energy deposition than gamma rays. However, there have been no human experiments that demonstrate the validity of these conclusions for the weak beta energy of tritium decay. As has been noted by UNSCEAR [1993] there is some likelihood that no health effects would occur at low doses and low dose rates.

There is general consensus that the current understanding of the metabolism of HTO is sufficient for radiation protection purposes [Hill and Johnson, 1993]. When it enters the body, HTO quickly reaches equilibrium with hydrogen in the body and imparts a uniform dose to all soft tissues. The ICRP [1978] recommends the assumption that internalized HTO is completely absorbed and mixed rapidly with total body water so that the concentration in all body waters—blood, urine, sweat, sputum, insensible perspiration, and exhaled water vapor—is the same. However, following HTO exposures, a small fraction of tritium will eventually be incorporated into organic molecules as OBT and will have a different metabolic rate than total body water tritium. Because there is no preferential distribution of OBT to any specific body tissue, it can be assumed that both HTO and OBT are uniformly distributed throughout the body, once taken in through some exposure route.

Human studies indicate that, generally, tritium retention can be characterized by a three-component exponential function. These three components have biological half-lives of 6 to 12 days, 10 to 34 days, and 130 to 550 days, with respective median values of 9, 22, and 340 days [Hill and Johnson, 1993]. The first biological half-life is for the body water metabolism of hydrogen and reflects the turnover of water within the pool of body water. The second half-life appears to be related to tritium involved in carbon-tritium bonds, and the last is for tritium incorporated into organic molecules having very slow turnover rates [Hill and Johnson, 1993].

Tritium can also enter the body as OBT through food products. The fate of OBT after ingestion depends primarily on the human body’s hydrogen metabolism. Usually, direct uptake and metabolism of OBT has been ignored in models of tritium dosimetry. However, recent reports based on more advanced environmental tritium models indicate that ingestion doses from OBT in food could be higher than doses from equivalent quantities of tritium ingested or inhaled as HTO [Etnier et al., 1984]. To properly assess combined exposures to HTO and OBT, we describe the retention of tritium following exposure with a model that accounts for intake of both. Following the recommendation of Hill and Johnson [1993], we use the four-compartment model proposed by Etnier et al. [1984] to describe tritium retention. One compartment is used to represent the retention of tritium in body water and the other three compartments are used to represent the retention of tritium in different types of organic molecules. This model is illustrated in Figure 4-4.

Under steady-state conditions, these parameters can be used to assess the body burden associated with a continuous input of tritium as HTO or OBT by estimating the effective tritium half life, , for a given intake. When HTO is ingested, based on the ranges of compartment half-lives discussed above, the model in Figure 4-4 gives an estimated mean value of equal to 15 days with a CV of 0.2. When OBT is ingested into either the labile or fast fat compartment, the model in Figure 4-4 gives an estimated mean

From this steady state inventory and the body weight, BW (in kg), we calculate the total body burden (in Bq/kg) and the corresponding dose rate in Gy per year as

where dist means disintegration.

International, federal, and state agencies have all set standards for controlling human exposures to tritium in both occupational and nonoccupational settings.

Detrimental effects of toxic chemicals on humans, animals, and plants can be classified as stochastic or nonstochastic effects [ICRP, 1977]. Stochastic effects are those for which the probability of an effect occurring, rather than the severity of effect, is proportional to dose without threshold. Nonstochastic effects are those for which the severity of effect is a function of dose and for which a threshold may exist. For example, the human health effects of carcinogens and many types of genetic effects are assumed to be stochastic. In contrast, the effects of neurotoxins, such as lead and mercury, are assumed to be nonstochastic. The detrimental effects of body burdens of tritiated water are stochastic. This section identifies risk factors that can be used to estimate the likely occurrence of these effects in exposed populations.

Together with previous studies, a large number of radiobiological studies conducted over the past decade indicate that tritium in body water produces the same portfolio of radiogenic effects as that observed following whole-body exposure to penetrating ionizing radiations such as x-rays and gamma rays [Straume and Carsten, 1993], including cancer, heritable genetic defects, and reproductive and developmental effects. Nevertheless, it has been observed and reported that the beta rays produced by tritium have greater biological effectiveness (i.e., damage potential) than x-rays or gamma rays of the same energy [Straume and Carsten, 1993].

We have shown that an annual dose equivalent from tritium exposure can be calculated from tritium concentration in body water by converting this body-water concentration to an estimate of equivalent whole-body energy deposition in Gy. Based on the average dose in Gy for a population, we derive the risk factors to predict the potential incidence of detrimental effects. However, studies showing deleterious effects from tritium uptake were in the million-Bq-per-liter range. These ranges may not be applicable to the concentrations around the LBNL tritium facility.

Exposures to tritiated water appear to induce DNA changes, indicating that HTO might be an initiator of cancer. In addition, there is evidence that HTO can cause transformations in mutant stem cells and thus enhance later stages of the progression from normal to cancer cells, thus indicating that HTO might also be considered a promoter of cancer. Types of cancers associated with HTO dose include leukemia and nonleukemia (primarily soft-tissue carcinomas) [Straume, 1993].

Straume [1993] has developed a stochastic dose-response model of fatal cancer risk for HTO doses that is based on the dose-response data for human cancer and a modification for the biological effectiveness of HTO relative to gamma rays. This model expresses the probability of cancer per Gy of dose and is illustrated in Figure 4-5. In this model the risk factor for cancer incidence within a population has respective median and mean values of 0.080 and 0.089 fatal cancers per Gy of dose. According to Straume [1993], the full range of this distribution is from 0.018 to 0.580, and the CV is approximately 1.3. The International Commission on Radiological Protection [ICRP, 1990] recommends a risk factor of 0.05 per Sv for fatal cancers in populations that include adults (of all ages) and children. Thus, the risk factor of 0.089 per Gy used here has an implicit relative biological effectiveness (RBE) of 1.8. However, Straume’s [1993] data for RBE used to construct his risk factor suggests that the RBE he used was actually 2.3. This discrepancy suggests that Straume’s model differs somewhat from the ICRP [1990] approach.

The factor 0.089 is the risk factor for fatal cancer. In order to determine the total number or incidence of cancers, we multiply the number of fatal cancers by 1.2. This ratio is used by the ICRP in their stochastic dose-response model. This ratio implies that for every 5 fatal cancers predicted to result from radiation exposure, there will be on average a likelihood of 1 nonfatal cancer in addition to five fatal cancers.

To estimate genetic risk to humans from low-level HTO exposures, we use a model developed by Straume [1993], who has used ranges and distributions of gamma-ray-induced risks to construct a stochastic-dose response function for the number of heritable genetic defects expected per Gy of population dose. The distribution of defects per unit dose obtained from this model is shown in Figure 4-6. This distribution has respective median and mean values of 0.006 and 0.0066 defects per live-born Gy of population dose in the first generation. The range of this distribution is from 0.0014 to 0.033, and the CV is approximately 1.3.

To estimate reproductive and developmental risk in humans from low-level HTO exposure, we use the model of Straume [1993] who has used ranges and distributions of gamma-ray-induced risks to construct a range of likely values for the dose-response function for HTO doses. From these data we developed a stochastic-dose response function for the number of reproductive and developmental effects expected per Gy of dose to the live-born infant. The probability density function of this model is shown in Figure 4-7. This distribution has respective median and mean values of 0.15 and 0.17 defects per live-born Gy of population dose. The range of this distribution is from 0 to 0.4, and the CV is approximately 0.45.

In this section, we quantify the potential health risks to the populations in and near the NTLF facility. The purpose of this assessment is to determine how tritium doses and potential health effects might be distributed among the occupational and nonoccupational populations of LBNL, the University of California (UC), and the cities of Berkeley and Oakland.

Three geographic zones and associated populations are defined in order to carry out this analysis (see Figure 4-8). Zone 1 includes the natural bowl that surrounds the NTLF facility and its tritium release stack. The population potentially at risk within this zone is the working population and its off-spring. Zone 2 includes the entire LBNL site; the UC property occupied by the Lawrence Hall of Science, the Samuel Silver Space Sciences Laboratory, the Mathematical Sciences Research Institute, and the Botanical Garden; and some single-family residences along the ridge above the LBNL site. Zone 3 includes the main UC Berkeley campus and areas of the city of Berkeley beyond Zone 2 and within 2 km of the NTLF.

Landscape, climate and hydrology parameters used in the analysis are derived from site-specific information. These parameters are summarized in Table 4-6. The references and methods used to obtain these values are described in detail in Appendix

C. The model used to determine tritium concentrations in Zones 1 and 2 is a site-specific model, which is described in Appendix A. It is based on the CalTOX model, which is a standard model used by the California Department of Toxic Substances Control for assessing risks at toxic substances release sites [McKone, 1993]. In Zone 3, we use the EPA radionuclide air dispersion model CAP88 to determine tritium concentration in air. As is discussed in Chapter 1 of this report, the site-specific model and the CAP88 model give close dose and risk estimates in Zone 2. The CAP88 model is not appropriate in Zone 1, because of the proximity to the source and because of the complex terrain. For application in Zone 3, we note that the CAP88 model calculates ingestion doses by pooling all vegetation within 80 km of the tritium source and assuming this vegetation provides some fraction of the food consumed by any individual in this area. Since this algorithm is not realistic, based on what is known about food distribution patterns in this area, we decided to use only the calculated air concentration from CAP88 in combination with the dose models developed for Zone 2 to estimate the exposure, dose and risk in Zone 3.

Table 4-6
Landscape properties for the NTLF Environment.

The NTLF currently releases tritium at a rate of approximately 100 Ci/y (3.7 trillion Bq/y). This has been typical since 1990. This value is lower than during the period 1982 to 1990, but higher than current releases by about a factor of 2 or more. According to the 1990 LBNL Site Environmental report, essentially all of LBNL’s tritium emissions originate from the NTLF [LBNL Site Report, 1990]. The total inventory of tritium that is attached to the uranium beds (U-beds) is roughly 10,000 Ci—the U-beds are designed with a stoichiometric capacity for 25,000 Ci of tritium, and thus are always under-utilized. The release of the entire inventory, even in the case of an extreme event, is unlikely. It is even less likely that it would persist for an extended period (i.e., more than a few days). Note that a catastrophic accident would be a one-time release and certainly would not persist.

The NTLF stack is located on the slope west of Building 75 (Figure 4-9). Its surrounding area is in a natural bowl or amphitheater. We used this bowl to characterize the initial mixing volume of the stack release. To be conservative, we used 15m as the mixing height for Zone 1, which may lead to an overestimate of the tritium concentration in air.

Table 4-7
Tritium Turnover Rates for the Air and Soil Compartments in Zone 1

Table 4-8
Tritium Concentrations and Inventories in Zone 1,
Corresponding to a Release of 100 Ci/year

Table 4-9
Tritium Turnover Rates for the Air and Soil Compartments in Zone 2

In Zone 3, air concentrations are obtained using CAP88 (see Appendix F) with an

Table 4-10
Tritium Concentrations and Inventories in Zone 2, Assuming a Release of 100 Ci/Year

From measurements taken from air, rain water, soil, and surface water near the NTLF [Thorson, 1996], rain water appears to be washing tritium out of the air and into surface soil and the interstitial water of the soil . Some of the tritium in the soil water moves as runoff in the Chicken Creek Basin from which it is carried into Strawberry Creek and down toward the Berkeley Campus. This transfer by rainwater and runoff is used to assess tritium doses to those who come in contact with the lower portion of Strawberry Creek on the UC Berkeley Campus. This section illustrates how we estimated tritium concentrations in water and air associated with the segment of Strawberry Creek that flows through the Berkeley Campus.

Only a small fraction of the water flowing down Strawberry Creek (into Zone 3) originates from Zone 1 (bowl area). Assuming most of the stream flow stems from runoff, this fraction can be estimated as:

Based on the fraction derived in Equation 4-16, we estimate that the tritium concentration in Strawberry Creek is on the order of 1% of the tritium water concentration in Zone 1. Based on the mass-balance model for water evaporation (Equation 4-17) in the area of the UC Campus, we estimate that evaporation from Strawberry Creek could contribute no more than about 0.01% to the total water inventory in air above the campus.

This analysis was carried out as follows. We assumed that the flow path of Strawberry Creek through the UC Berkeley campus and city is on the order of 2000 m (1.3 mile) and that it has an average width of 1 meter and depth of 0.5 meter. For evaporation of water from fast-moving water bodies, Southworth [1979] has shown that the rate of water evaporation is limited by the rate of mass transfer of saturated vapor in the air above the water. Furthermore, he has shown that this evaporation rate depends on the wind speed over the water, the water flow velocity, and the difference between the saturated and average relative vapor concentrations in air. This evaporation rate is given as follows:

(4-17)

Table 4-11
Tritium Concentrations and Inventories in Zone 3, Assuming a Release of 100 Ci/Y

This section presents the mathematical expressions used to estimate the inhalation, dermal uptake, and ingestion exposures in each of the three zones.

Values used for the parameters in these expressions are discussed in Section 4.5 and Appendix B.

Values used for the parameters in this expression are discussed in Section 4.5 and Appendix B.

Values used for the parameters in this expression are discussed in Section 4.5 and Appendix B.

The total body burden in Bq/kg from all exposure routes for an average individual within the population is the sum of inhalation and ingestion intakes and dermal uptake in Bq/day multiplied by the removal time. The removal time is equal to the effective biological half-life, , divided by 0.693.

As discussed above, is 15 days for HTO and 48 days for OBT.

The body burden of tritium in a fetus is assumed to be the same as the mother.

The total dose in Gy associated with ingestion, inhalation, and dermal uptake is calculated from the product of body burden and a conversion factor (Equation 4-27) that translates from Bq/kg to Gy/y, as given by Equation 4-14.

The population risk for cancer represents the estimated increase of fatal cancers per year due to exposures including adults and infants. The magnitude of the population for cancer risk depends on the number of exposed adults and the number of infants born to exposed adults.

Individual Lifetime Risk (cancer) = Dose (adult) x ED x (4-29)

The population risk for genetic effects represents the estimated increased hereditary effects per year per live-born child within the exposed population. The magnitude of the population risk for heritable defects depends on the total number of live births and the exposure to the parents of a child.