Publications

Simplified expressions for the attenuation of radionuclide releases by sprays and by water pools are devised. These expressions are obtained by correlation of the 10th, 50th and 90th percentiles of uncertainty distributions for the water pool decontamination factor and the spray decontamination coefficient. These uncertainty distributions were obtained by Monte Carlo uncertainty analyses using detailed, mechanistic models of the pools and sprays. Uncertainties considered in the analyses include uncertainties in the phenomena and uncertainties in the initial and boundary conditions during dictated by the progression of the severe accidents. Final results are graphically displayed in terms of the decontamination factor achieved at selected levels of conversatism versus pool depth and water subcooling or, in the case of sprays, versus time.

The WETCOR-1 test of simultaneous interactions of a high-temperature melt with water and a limestone/common-sand concrete is described. The test used a 34.1-kg melt of 76.8 w/o Al{sub 2}O{sub 3}, 16.9 w/o CaO, and 4.0 w/o SiO{sub 2} heated by induction using tungsten susceptors. Once quasi-steady attack on concrete by the melt was established, an attempt was made to quench the melt at 1850 K with 295 K water flowing at 57 liters per minute. Net power into the melt at the time of water addition was 0.61 {plus_minus} 0.19 W/cm{sup 3}. The test configuration used in the WETCOR-1 test was designed to delay melt freezing to the walls of the test fixture. This was done to test hypotheses concerning the inherent stability of crust formation when high-temperature melts are exposed to water. No instability in crust formation was observed. The flux of heat through the crust to the water pool maintained over the melt in the test was found to be 0.52 {plus_minus} 0.13 MW/m{sup 2}. Solidified crusts were found to attenuate aerosol emissions during the melt concrete interactions by factors of 1.3 to 3.5. The combination of a solidified crust and a 30-cm deep subcooled water pool was found to attenuate aerosol emissions by factors of 3 to 15.

This report is a followup to the work presented in NUREG/CR-5423 addressing early failure of a BWR Mark I containment by melt attack of the liner, and it constitutes a part of the implementation of the Risk-Oriented Accident Analysis Methodology (ROAAM) employed therein. In particular, it expands the quantification to include four independent evaluations carried out at Rensselaer Polytechnic Institute, Argonne National Laboratories, Sandia National Laboratories and ANATECH, Inc. on the various portions of the phenomenology involved. These independent evaluations are included here as Parts II through V. The results, and their integration in Part I, demonstrate the substantial synergism and convergence necessary to recognize that the issue has been resolved.

Mechanistic models of aerosol decontamination by an overlying water pool during core debris/concrete interactions and spray removal of aerosols from a Mark I drywell atmosphere are developed. Eighteen uncertain features of the pool decontamination model and 19 uncertain features of the model for the rate coefficient of spray removal of aerosols are identified. Ranges for values of parameters that characterize these uncertain features of the models are established. Probability density functions for values within these ranges are assigned according to a set of rules. A Monte Carlo uncertainty analysis of the decontamination factor produced by water pools 30 and 50 cm deep and subcooled 0--70 K is performed. An uncertainty analysis for the rate constant of spray removal of aerosols is done for water fluxes of 0.25, 0.01, and 0.001 cm{sup 3} H{sub 2}O/cm{sup 2}-s and decontamination factors of 1.1, 2, 3.3, 10, 100, and 1000.

Spray systems in nuclear reactor containments are described. The scrubbing of aerosols from containment atmospheres by spray droplets is discussed. Uncertainties are identified in the prediction of spray performance when the sprays are used as a means for decontaminating containment atmospheres. A mechanistic model based on current knowledge of the physical phenomena involved in spray performance is developed. With this model, a quantitative uncertainty analysis of spray performance is conducted using a Monte Carlo method to sample 20 uncertain quantities related to phenomena of spray droplet behavior as well as the initial and boundary conditions expected to be associated with severe reactor accidents. Results of the uncertainty analysis are used to construct simplified expressions for spray decontamination coefficients. Two variables that affect aerosol capture by water droplets are not treated as uncertain; they are (1) {open_quote}Q{close_quote}, spray water flux into the containment, and (2) {open_quote}H{close_quote}, the total fall distance of spray droplets. The choice of values of these variables is left to the user since they are plant and accident specific. Also, they can usually be ascertained with some degree of certainty. The spray decontamination coefficients are found to be sufficiently dependent on the extent of decontamination that the fraction of the initial aerosol remaining in the atmosphere, m{sub f}, is explicitly treated in the simplified expressions. The simplified expressions for the spray decontamination coefficient are given. Parametric values for these expressions are found for median, 10 percentile, and 90 percentile values in the uncertainty distribution for the spray decontamination coefficient. Examples are given to illustrate the utility of the simplified expressions to predict spray decontamination of an aerosol-laden atmosphere.

This paper discusses a nonideal solution model of the metallic phases of reactor core debris. The metal phase model is based on the Kohler equation for a 37 component system. The binary subsystems are assumed to have subregular interactions. The model is parameterized by comparison to available data and by estimating subregular interactions using the methods developed by Miedama et al. The model is shown to predict phase separation in the metallic phase of core debris. The model also predicts reduced chemical activities of zirconium and tellurium in the metal phase. A model of the oxide phase of core debris is described briefly. The model treats the oxide phase as an associated solution. The chemical activities of solution components are determined by the existence and interactions of species formed from the components.

A classic model of aerosol scrubbing from bubbles rising through water is applied to the decontamination of gases produced during core debris interactions with concrete. The model, originally developed by Fuchs, describes aerosol capture by diffusion, sedimentation, and inertial impaction. This original model for spherical bubbles is modified to account for ellipsoidal distortion of the bubbles. Eighteen uncertain variables are identified in the application of the model to the decontamination of aerosols produced during core debris interactions with concrete by a water pool of specified depth and subcooling. These uncertain variables include properties of the aerosols, the bubbles, the water and the ambient pressure. Ranges for the values of the uncertain variables are defined based on the literature and experience. Probability density functions for values of these uncertain variables are hypothesized. The model of decontamination is applied in a Monte Carlo sampling of the decontamination by pools of specified depth and subcooling. Results are analyzed using a nonparametric, order statistical analysis that allows quantitative differentiation of stochastic and phenomenological uncertainty. The sampled values of the decontamination factors are used to construct estimated probability density functions for the decontamination factor at confidence levels of 50%, 90% and 95%. The decontamination factors for pools 30, 50, 100, 200, 300, and 500 cm deep and subcooling levels of 0, 2, 5, 10, 20, 30, 50, and 70{degree}C are correlated by simple polynomial regression. These polynomial equations can be used to estimate decontamination factors at prescribed confidence levels.

Physical and chemical phenomena that will affect the behavior of radionuclides released from fuel in the Annular Core Research Reactor during a hypothetical, core disruptive accident are described. The phenomena include boiling of water on heated clad, metal-water reactions, vapor nucleation to form aerosol particles, coagulation of aerosol particles, aerosol deposition within bubbles rising through the shield pool, vapor dissolution in the shield pool, and revaporization of radionuclides from the shield pool. A model of these phenomena is developed and applied to predict the release of radionuclides to the confinement building of the Annular Core Research Reactor. It is found that the shield pool provides overall decontamination factors for particulate of about 2.8 {times} 10{sup 5} and decontamination factors for noble gases of about 2.5--3.7. These results are found to be sensitive to the predicted clad temperature and bubble behavior in the shield pool. Slow revalorization of krypton, xenon and iodine from the shield pool is shown to create a prolonged, low-intensity source term of radioactive material to the confinement atmosphere.

Analyses have shown that early rupture of the Mark-I boiling water reactor containment-by the direct action of core debris depends strongly on the time that core debris is superheated above its liquidus. The analyses of the duration of superheat in the core debris are compared to predictions obtained with the CORCON-MOD 3 computer code. The predicitons of this computer code as functions of the core debris mass, composition, and initial superheat are used to create a polynomial response surface. This response surface is used in a Monte Carlo analysis to produce probability distributions for the duration of superheat in core debris in the drywell of a Mark-I containment. It is concluded that to a high level of confidence (>90%) the duration of superheating predicted with the CORCON-MOD 3 code is less than what has been used in the analyses of the threats to the Mark-I containment liner. Based on these results, to the extent superheat duration dictates the threat to the liner, analyses in NUREG/CR-5423 would appear to overestimate the threat to the liner in comparison to threats estimated using the predictions of the duration of superheating obtained with CORCON-MOD 3.

Severe nuclear reactor accidents - accidents involving the melting of the reactor core - dominate the residual risk associated with the use of nuclear power. The uninterrupted progression of a severe reactor accident is expected to lead to the expulsion of core debris into the reactor containment. Many safety-significant phenomena may be hypothesized to occur when core debris is expelled from the reactor coolant system. The exact nature of these events depends on whether or not the coolant system is pressurized at the time of melt expulsion and whether or not expulsion is into water. Regardless of what transient events are associated with the initial expulsion of core debris from the reactor coolant system, a protracted period of core debris interactions with the structural concrete of the reactor is expected in most analyses of severe reactor accidents.