UC Berkeley

Damaged fuel from core melting accidents like Three Mile Island and Fukushima Daiichi need to be retrieved and disposed. Retrieval of damaged fuel is considerably more complex than for standard fuel as fuel is relocated and deformed during the accident. Two major aspects need to be considered when planning damaged fuel retrieval: prevent to create conditions in which criticality could be achieved and protect workers and equipment from radiation. In the case of the Fukushima Daiichi Power Plant, damaged fuel cannot be covered with water, making retrieval operations more complex as water provides radiation shielding. This work provides an assessment of the radiation emissions from the Fukushima Daiichi damaged fuel necessary to determine shielding requirements. In an initial scoping study, the photon and neutron dose rate of the fuel debris in a canister was evaluated in several parameters including canister designs, source term, geometry of fuel debris. This study provided the basis for the development of methods for the evaluation of the dose rate and uncertainty. The fuel debris was modeled by using random sampling on the size and distribution of fuel debris in a canister. The model uses the Monte Carlo method and performs random sampling on the geometry of the fuel debris to estimate the average dose rate and its standard deviation. SCALE and MCNP codes were used for neutron and photon transport calculations, and a Python code was developed to generate samples and to manage calculations for the evaluation. The Python code efficiently manages calculations of dose rate and flux by applying multiprocessing and parallel computing. The output data was evaluated by basic statistics, and a method which converts the output data into a cumulative format is proposed for the data fitting and the regression analysis of nonlinear data. The model provided variability and uncertainty of the radiation dose rate and flux in detail. However, this method is time consuming and computationally expensive, therefore, an analytical method is developed to approximately evaluate the flux and uncertainty in relatively short time. The uncertainty of dose rate was evaluated as the standard deviation and the relative range of estimated dose rates. The uncertainty depends on the size of fuel debris, and uncertainty of a canister with larger debris is larger. The standard deviation can be larger than 30% of the average dose rate, and the relative range can be larger than 90% of the average dose rate depending on the condition of fuel debris. The uncertainty increases when canisters are contained in a transport cask which has several canisters in it. The average dose rate also depends on the size of debris, and a canister with larger debris has a smaller dose rate. The distribution of fuel debris in a canister also affects the dose rate and uncertainty. Variability of the dose rate and uncertainty by changes in the location of the detector was also evaluated. The maximum dose rate was estimated at the vertical center of a canister at 1 m from the surface of a canister which is loosely packed with fuel debris. In case of a canister with the close packed fuel debris, the maximum dose rate was estimated at the vertical center of the pile of debris which is not same as the vertical center of a canister. The uncertainty also depends on the vertical location of the detector, and the uncertainty can be minimized when the dose rate is estimated at the vertical center of the debris pile. The dose rate and uncertainty are variable when they are estimated near a canister, therefore, the dose rate and uncertainty of fuel debris can be characterized in more detail when they are estimated near a canister, and this can be applied for the characterization of the fuel debris. Modified designs which can reduce the uncertainty of dose rate were evaluated. The uncertainty can be reduced by dividing the inner space of a canister by using inner containers or partitions which restrict the distribution of fuel debris in the canister. In case of the horizontal division, size of fuel debris can be restricted depending on the number of partitions used for the design. More partitions restrict the size, and uncertainty can be reduced since uncertainty of smaller debris is smaller. Vertical division does not restrict the size and it only restricts the vertical distribution of fuel debris. Even if the horizontal division can minimize the uncertainty when many partitions are installed in a canister, the vertical division of the inner space is more efficient than horizontal division in reduction of the uncertainty. Evaluated results in this dissertation can help the safe retrieval of fuel debris in Fukushima Daiichi nuclear power plant. Models and methods developed in this dissertation can be used for the characterization of damaged nuclear fuel which can be made by an accident in the future. They also can be applied to develop methods to manage spent fuel of advanced nuclear reactors such as the fuel of pebble bed reactors.