IV. CONCLUSIONS AND RECOMMENDATIONS
During most of the past 40 years, thousands of kilograms of uranium have been discharged by FMPC each year into the Great Miami River as liquids or released as dust and gases into the air. Although FMPC no longer produces uranium metal, the site continues to store radioactive materials that were used at Fernald and other DOE sites. There are also four silos, large concrete underground storage tanks in a waste disposal area, that are a source of 222Rn.
The primary assessment domain includes residents within an 8-km (5 mi) radius around the FMPC production area. RAC estimates that more than half of the particulate material released from FMPC to the atmosphere was deposited in this area. Concentrations of uranium deposited at the outer boundary of the assessment domain would be expected to be only about one-hundredth those at the site boundary. This information is appropriate for identifying the most highly exposed population.
Atmospheric pathways are expected to have dominated the total dose from FMPC releases, and the transport of uranium particles is believed to be strongly influenced by the particle-size distribution. Other sources of dose from FMPC releases considered include the radioactive progeny of 222Rn produced by the decay of 226Ra, a component of waste materials stored in the K-65 silos, and direct γ-radiation exposures from the silos. Short-term, “episodic” releases (which increase the composite uranium release rate by a factor of at least 10 for a period of fewer than 10 days) are treated with special dose assessment procedures. The time increment over which the average release rate is calculated should be specified (hourly, daily, weekly).
Particle-size distributions are based on an analysis of measurements made in 1985, the only time such measurements were made. The authors have fit a polynomial function to the observed data for this purpose. Actually, the data would seem to be fit adequately by a lognormal distribution. Such a fit provides parameters that can readily be used in
standard lung-deposition models, greatly simplifying the process of calculating dose. It is strongly recommended that these data be analyzed by someone familiar with risk-related experimental particle-size analysis before a commitment is made to a technique that will require special dose-calculation methods.
No site-specific meteorologic data were available for the assessment. Although the authors are clearly reluctant to use climate data from either the Cincinnati or the Dayton airports, they have presented a compelling analysis of why the Cincinnati data, adjusted for local patterns, should be used. The Monte Carlo sampling method for using the Cincinnati data is clever; a better method probably does not exist.
The near-field deposition of particles and building-wake effects should continue to be reviewed in the context of near-field dispersion and runoff because these mechanisms could have considerable effects on the pattern of the introduction of uranium and other materials into the environment. The endpoint objectives are appropriate, and the structure of the report summary followed by detailed appendixes is a good one. Although the report assesses effects out to 50 mi, this appears unnecessary in view of the clear significance of near-field transport and deposition.
Episodic releases should be considered separately for dispersion modeling as well as for other evaluations. There were major episodic releases in the early years that did not come from the roof vents. The actual mode of each major episodic release should be considered and a determination should be made of whether a puff model, a wake model, or perhaps a plume model is the most appropriate for a particular release.
Measurements are compared with modeled estimates, and the results generally were found to be within a factor of 2. The model tends to overestimate observed concentrations in air, primarily because of assumptions about the removal of large particles by deposition. Uncertainties in meteorologic data, the source term, and air measurements themselves also pose problems, but the overall results of the comparison are a reasonable validation of the model predictions.
The estimate of the deposition of uranium in the environment is 2 × 105 kg to 9 × 105 kg. It is not clear whether this range refers to the total period of operation nor is it clear how well the calculations correlate with soil measurements. The question of onsite source depletion of coarse particles from low-elevation puff releases remains. Such releases could represent the largest release fraction during episodic events.
There is considerable uncertainty in determining dispersion-input parameters. A consistent procedure should be outlined for propagating uncertainties and, in particular, for taking into account the substantial systematic errors in much of the early sampling and analytic data. There is also appreciable uncertainty associated with early episodic releases, and it will be important to conduct a sensitivity analysis to eliminate minor, but highly uncertain, contributions to the total dose.
There is considerable uncertainty in the assessment of radon releases. A major complication for the study is the variation in the radon background concentration in individual homes. The uranium isotopes are the most important contributors to the dose resulting from atmospheric releases, and inhalation is the most important pathway. The most important radionuclides released in surface water are found to be 228Ra, 226Ra, 99Tc, and uranium radioelements. No screening exercise was carried out for the groundwater releases.
The authors of the report conclude that they have provided “estimates of the number of individuals that may have been exposed to environmental releases from the FMPC between 1952 and 1988.” They actually estimate the number of residents at various points in time. To derive “the number of individuals that may have been exposed,” they also would need to add population turnover rates to the model. Their conclusion should be changed so they do not claim to have done more than they did.
The purpose of the demographic modeling process was to provide a matrix of population size by calendar time, which is a fundamental building block in estimating the
person-dose distribution and the person-years at risk for possible epidemiologic studies. The demographic estimates provided are credible and basically adequate for the intended purpose. The authors were careful to check for internal consistency and validity. It is recommended, however, that more detailed age categories be provided. Finally, where there are major uncertainties or problems in handling some of the scenarios, these should be presented as such, the proposed solution indicated, and the reasons for choice of the solution made clear.
These conclusions have addressed the methods and modeling processes used by the RAC in the dose reconstruction project under review. To facilitate the public understanding of the implications of the dose reconstruction, it would be helpful if the final dose estimates for the surrounding population were evaluated in terms of the increases over background radiation levels. Such perspective should be included in the final summary for this project.
