CRITICALITY ISSUES WITH HIGHLY ENRICHED FUELS IN A REPOSITORY ENVIRONMENT
Larry L. Taylor
Lockheed Martin Idaho Technologies Co.
1955 Fremont Ave.
Idaho Falls, ID 83415-3114
Ph: 208-526-9175
email: larryt@inel.gov
Lawrence C. Sanchez and Jonathan S. Rath
Sandia National Laboratories
1515 Eubank SE, MS-1328
Albuquerque, NM 87185-1328
Ph: 505-848-0685
email: lcsanch@nwer.sandia.gov
ABSTRACT
Existing regulations (10 CFR 60) preclude the disposal of fissile material in a repository without the attendant double contingencies as they apply to operational facilities concerned with criticality safety1. While much of the spent nuclear fuels (SNF) identified for the repository consist of commercial fuels with relatively low enrichments, there will be a limited number of DOE-owned SNF packages that will contain highly enriched uranium (HEU) with enrichments in excess of 60%. Of the double contingencies available at the time of package emplacement, none of those contingencies will be monitored nor will they be remediable after repository closure. Without imposition of single parameter limits (mass, spacing, "poisons," concentration), there will be criticality risks associated with the disposal of any fissile material within the repository. Imposition of a single parameter limit may eliminate criticality as an issue for a single-package event but will result in excessive cost to dispose of such fuels.
The work presented in this paper focused on two attributes of criticality in a geologic repository for near-field and far-field scenarios: (1) scenario conditions necessary to have a criticality, and (2) consequences of a nuclear excursion that are components of risk. All criticality consequences are dependent upon eventual water intrusion into the repository and subsequent breach of the disposal package. Key criticality parameters necessary for a critical assembly are (1) adequate thermal fissile mass, (2) adequate concentration of fissile material, (3) separation of neutron poison from fissile materials, and (4) sufficient neutron moderation (expressed in units of moderator to fissile atom ratios). Key results from this study indicated that the total energies released during a single excursion are minimal (comparable to those released in previous solution accidents), and the maximum frequency of occurrence is bounded by the saturation and temperature recycle times.
INTRODUCTION
Regardless of the double contingency criteria established for any given spent nuclear fuels (SNF) package at loading (e.g., fissile limits, neutron "poisons," fuel spacing within a package, "dryness," etc., in a geologic time frame), one or more of any designed contingency can be expected to be compromised. Prior studies have addressed the potential failure of one or more SNF packages that contain fission products and actinides within the 10,000-year isolation stipulated for the repository2,3.
"Criticality" in a performance assessment of a geologic repository corresponds to the risks resulting from a nuclear criticality excursion (where risk is the combined effect of consequences and probability). Water infiltration scenarios resulting in a post-closure repository are predicted to (1) breach a waste package, (2) promote the necessary "reconfiguration" of package contents, and (3) provide the necessary moderation/reflection needed to allow neutron propagation. As a subset failure mode of an SNF package, at least one scenario involved a top-breach of the waste package with long-term water drip. The breaching of the package may result in a criticality in one of several scenarios: internal (in situ), near-field (within several tunnel diameters of the waste package), or far-field (a significant distance away from the package).
THEORY
Eventually, all issues related to any possible occurrence of a criticality in a post-closure time frame of the repository rely heavily on the presence of water. Fissile loads distributed throughout the various DOE SNF canisters containing enriched material have no identified, credible events that cause fuel reconfiguration leading to criticality inside or outside the package without the presence of water. Water entering a breached waste package results in the addition of both reflection and moderation of the waste package contents as well as the necessary mechanism to provide reconfiguration of the fissile material, or separation from any neutron absorber within the package.
Reactivity features of various fissile materials (e.g., 235U, 239Pu, and 233U) differ slightly in their nuclear resonances and fissioning for thermal neutrons. However, defense fuel canisters containing enriched 235U dominate the packages with identified risks of criticality. The 239Pu as an isotope predominates in the N-reactors fuel category, but its fissile atom density is too low to create a significant issue for criticality potential even in a densely packed SNF canister with fuels of low enrichment. 233U as a fissile material is limited to one fuel category that is still dominated by 235U atom densities.
Given the fact that any criticality experienced in the repository will require the presence of water, models that incorporate the physics associated with water-moderated critical systems should govern the characteristic behavior of the criticality. Doppler temperature coefficients, water density/saturation changes, and resonance behavior of the fissile materials with increasing temperatures all suggest a natural shutdown mechanism exists for any criticality that might occur. The rate of water influx, rate of neutron production, and the thermal properties of the repository would govern initiation of any criticality. System shutdown would similarly be governed by the rate of heat generation (negative temperature coefficient) and desaturation based on water being driven out of the system.
RESULTS
During scenario development for criticality in a repository, an infinite number of configurations can be identified that result a critical assembly. Using a systematic approach, near-field and far-field criticality scenarios were investigated. The saturation recycle time governs the frequency of criticality excursions even though the temperature recycle time is minimal.
Near-field criticalities are defined as those events that occur outside, but in proximity to the disposal package (within a few tunnel diameters of the package). Contributions to such a criticality event include actinide transport from a disposal package as soluble or colloidal species. Expectations of a totally degraded outer package would result in a significant accumulation of iron oxide on the floor of the drift. Intermittent groundwater flow across exposed fissile material in a breached package may create incremental fissile material additions to any rust or invert (concrete) material encountered on the drift floor. Alternatively, fissile accumulation may occur from a gross breach in the bottom of the package and dropping partially intact fuel shards out of the bottom of the package and onto the drift floor. In either case, the ability to maintain adequate water in proximity to effect a critical system is certainly problematic. Neither chemically bound water nor hydrated compounds can provide the atom density of hydrogen necessary to significantly affect neutron moderation in such an "open" system.
The mechanism(s) for evaluating the consequences of a criticality in a moderated system corresponded to a select set of conditions using an uncoupled far-field thermal-hydraulics (THX) conceptual model. This model used a set of assumed conditions comprised of a mixture thermal fissile material (TFM), water, and tuff. The uncoupled calculations were required to determine the time history of the water saturation given the groundwater transport through interconnected pores and fractures of the host rock. This flow is indicated by changes in fluid density and/or vaporizing the liquid phase because of heat generated in a criticality. BRAGFLO_T (X1.00), a multiphase fluid and energy simulator, predicted the uncoupled water saturation and (moderator) temperature changes resulting from a single critical event of an assumed system size (energy release) and power output. A total of 45 uncoupled calculations examined a range of initial saturation (65, 75, and 85 percent), and heat generating zone radii of 0.5, 1.0, and 1.5 m. The size of the modeled critical events ranged from 2.0 x 1018 to 1.0 x 1020 total fissions per event. From these calculations, both the average saturation change and average temperature history changes were used to determine the recycle (or reflux) times. These recycle times are simply the periods of time for the change in average saturation and temperatures to return to near initial conditions. Saturation recovery times generally ranged from 150 to 10,000 days, and temperature recovery times ranged from 12 to 110 days.
The dynamic system feedback for criticality is comprised of several major components consisting of prompt shutdown mechanisms. Because the temperature recycle time is shorter than the saturation recycle time, the prompt shutdown mechanism dominates the feedback processes.
In support of these criticality calculations, Sandia National Laboratories (SNL) developed a NARK (NucleAR Kinetics and dynamics) code that uses self-adaptive ordinary differential equation (ODE) solution algorithms to solve sets of "stiff" or "nonstiff" coupled ODEs common to nuclear kinetics and dynamics. These calculations were uncoupled from the THX calculations; hence, they were denoted as uncoupled nuclear dynamics (UDX) calculations. The subsequent calculations were performed for a range of thermal feedback coefficients. The goal of these calculations was to demonstrate that prompt negative coefficients dominate and control the shutdown mechanisms of a moderated system that might develop in near-field or far-field scenarios. The values used in these calculations are identified in Table I.
Table I. NARK UDX Sensitivity Input Parameters
Several uncoupled NARK calculations were run to calculate power excursions (rise and fall above initial power condition). The calculations yielded a range of total fissions (2.0 x 1018 to 1.5 x 1020) that are similar to documented criticality accidents in aqueous systems7. Preliminary static calculations using the MCNP code (via a pre-processor "RKeff ") provided results that revealed a critical state (i.e., Keff = 1.0) could be achieved at a concentration of 5.0 kg/m3 (i.e., concentration = TFM mass/volume of the mixture [TFM + rock + water]) for Topopah Springs welded tuff (saturation = 100%, porosity = 13.9%). Thus, the static criticality concentration was incorporated into the uncoupled dynamic code calculations by varying the TFM mass consistently with the concentration of 5.0 kg/m3. This constraint, using TFM mass values (25, 50, 100 kg) to bound the UDX calculations, resulted in radii of 1.0067, 1.3365, and 1.6839 m, respectively. The results of the uncoupled nuclear dynamic calculations indicated that excursions resulted in small quantities of release energy and additional fission yield products. To achieve a criticality in a far-field location, multiple HEU package failures (interspersed with commercial packages) and transport to a common accumulation point are necessary to accumulate a fissile mass needed to support a criticality. Similar criteria may exist for accumulated TFM both in situ (inside the breached package) and near-field. The latter situation is depicted in Figure 1.
Fig. 1. S-Curve for pure 235U (ideal cases).
In an attempt to understand the importance and sensitivity of the excursion input parameters, the net integrated fissions from a criticality were identified to be the result of a combination of initial reactivity, TFM mass, and the prompt feedback temperature coefficient. This relationship is depicted in Figure 2. The nuclear dynamics scaling parameter (romTFM/|aT|) versus the maximum number of integrated fission displays as a linear trend in log-log space.
Fig. 2. Integrated Energy
The output from the UDX calculations revealed how fast power excursions shut down because of prompt feedback mechanisms. UDX results yielded a range of integrated fissions from 1.30 x 1016 to 5.54 x 1019, and excursion durations ranged from 3.17 x 103 to 3.46 x 105 seconds. Selected time behavior results are presented in Figure 3.
Associated with any SNF package, including commercial fuels, will be the potential presence of neutron "poisons." This requirement may be implemented to assure subcriticality for any package in transport or after emplacement and prior to repository closure as part of the double contingencies associated with enriched fuels in operating facilities. The established concentration levels of any neutron absorber material will likely be predicated on the expected retention time within the 10,000-year life of the repository. Retention in this case is defined as the ability to provide some degree of assurance that adequate neutron absorber material will remain within proximity of any fissile material. Differential separation because of differing solubilities can result in movement/transport of one material away from another that might allow formation of a critical mass. Such a separation can occur both within a breached package and in the geologic setting. Materials under consideration for inclusion in any SNF package for the purpose of criticality control include boron within a borated stainless steel matrix, hafnium, cadmium (where it is already integral to the fuel assembly), and, perhaps, gadolinium in an amorphous metal-glass composite.
Fig. 3. Select UDX results.
CONCLUSIONS
Disposal of SNF packages containing significant fissile mass within a geologic repository must comply with current regulations relative to criticality safety during transportation and handling within operational facilities. However, once the repository is closed, the double contingency credits for criticality safety are subject to unremediable degradation, (e.g., water intrusion, continued presence of neutron absorbers in proximity to fissile material, and fissile material reconfiguration).
In situ criticalities pose the highest probability of a criticality for any enriched package because they represent the greatest TFM concentration for all time within the repository. In addition, the in situ event also provides the most significant opportunity for both moderation and reflection within a package. It is also possible to study various scenarios that provide the "optimal" H/U ratio for a given fuel type or enrichment.
Perhaps less than 100 HEU SNF packages containing HEU contribute to a significant risk in the near-field. In this case, the fuel assemblies inside the SNF package would last longer than the disposal package itself. Water and corrosion to completely destroy the waste package to the point where solid fuel "debris" accumulates on the drift floor. But water may not be present in adequate quantities to either reflect or moderate such a mass or both.
Far-field criticalities are even more problematic given that both sufficient water and contact time with the contents of a breached waste package are needed to transport TFM. Furthermore, all of the following assumptions may be required: (1) no significant TFM retention along the drift floor, (2) multiple HEU package failures that are themselves interspersed with commercial packages, (3) a single collection point with adequate volume and suitable geometry, (4) geochemistry favorable to TFM retention, and (5) retained water quantities adequate to promote a criticality. Mechanisms that promote TFM dilution for transport are unlikely to also favor reconcentration to the degree needed to create a critical system. Additionally, any package failure and transport of uranium with low enrichment to a common point drain will tend to denature any HEU at the common location.
Doppler temperature effects appear to dominate the shutdown mechanisms of a moderated criticality, with temperature recovery times measured in hours versus the void coefficient delays measured in days.
The presence (and retention) of neutron poisons during initial packaging may be necessary to assure safety of a fully flooded SNF canister with intact fuel assemblies for transportation and handling within operational facilities. However, any failure scenario must necessarily address the degraded package condition, where changes in the spatial atom densities of both the fissile material and neutron poison(s) must be analyzed because of chemical alteration by an aqueous system. Such an analysis would require examination of the separation of materials in a waste package (because of differences in robustness of the contents) and differential solubilities between the poisons and the TFM for transport outside the package.
REFERENCES