Paul K. Chan

Variation of Burnable Neutron Absorbers in a Heavy Water–Moderated Fuel Lattice: A Potential to Improve CANDU Reactor Operating Margins

Abstract A CANDU lattice cell has been modeled using the Los Alamos National Laboratory’s MCNP 6 code and Atomic Energy of Canada Limited’s WIMS-AECL 3.1. Models for the CANDU 37-element fuel bundle have included a CANLUB coating, as a carrier for the neutron absorbers. The objective is to improve CANDU reactor operating margins by adding small amounts (~1 g) of neutron absorbers to each fuel element. For CANDU natural uranium fuel bundle design, the results indicate that (a) the fueling transient (due to the xenon-free effect) could be significantly reduced using gadolinium oxide (Gd2O3), with no significant impact on fuel burnup, and (b) the reactivity peak (due to plutonium production) could be reduced using europium oxide (Eu2O3), with minimal impact on fuel burnup. An appropriate mixture of Gd2O3 and Eu2O3 that will improve operation and safety margins while having a minimal impact on fuel burnup is determined. Reactivity and power calculations for various mixtures of Gd2O3 and Eu2O3 are reported here. It is concluded that ~180 mg Gd2O3 and ~1000 mg Eu2O3 (~4.9 ×10−3 wt% per bundle) are sufficient to suppress the refueling transient and lower the axial plutonium peak, with a 0.27% burnup penalty (which is a small impact). Fuel safety and performance are always important topics for a nuclear utility. This approach of a relatively simple application of burnable poisons to existing CANDU natural uranium fuel design offers the benefits of improving fuel utilization and safety margins.

Neutron Absorbers in CANDU Natural Uranium Fuel Bundles to Improve Operating Margins

Safety margins are particularly tight in natural uranium-fuelled CANDU reactors which are refueled on-power. During on-power refueling, the insertion of xenon-free fresh fuel bundles into the reactor core affects the reactor’s excess reactivity in such a way that this could lead to temporary power derating. It is desirable from a fuel management perspective, and to maintain safety margins to eliminate this xenon-free effect and any other power ripples such as the subsequent plutonium reactivity peak. A redesign of the CANDU NU fuel bundle with an appropriate combination of elements, with some including neutron-absorbers, could well address the issue of the xenon-free initial portion of the bundle’s irradiation and also lower the plutonium-peak that occurs shortly thereafter. This may improve the fuel utilization (by further optimizing the fuelling strategy) and provide improved safety margins (by lowering the maximum channel and bundle powers).The use of neutron-absorbers in fuel design and manufacturing has been a regular practice in Light Water Reactor fuels for more than three decades. In CANDU applications, neutron absorbers have also been considered for the conceptual Advanced CANDU Reactor and the Low Void Reactivity fuel designs, for which the fissile content is made of low enriched uranium (LEU) or MOX fuels. The application to CANDU natural uranium (NU) fuel, however, especially as burnable poisons, is a relative novel approach. The reason for this is that the neutron economy in natural uranium-fuelled CANDU reactors is a prime concern, thus the addition of extra neutron absorbers is generally shunned. In our proposed application of burnable poisons to existing CANDU NU fuel design, because of low excess reactivity for NU fuel, the amount of neutron-absorber is expected to be restricted to small quantities and in a manner whereby the poison effect is restricted to the initial period of excess reactivity of a newly inserted fuel bundle. This implies that the impact on neutron economy would be relatively minimal, but the fuel performance would be significantly improved.Small amounts and appropriate mixtures of neutron absorbers were selected (approximately 500 mg of absorbers in a CANDU fuel bundle having a nominal weight of 24 kg). Preliminary results indicate that the fuelling transient and the subsequent reactivity peak can be lowered to improve the reactor’s operating margins. A parametric study using the Los Alamos National Laboratories’ MCNP 5 and Atomic Energy of Canada Limited’s WIMS-AECL 3.1 codes is presented in this paper. Details of this project and future work are also to be discussed.Copyright

Fueling Study of a CANDU Reactor Using Fuels Containing Burnable Neutron Absorbers

Abstract Trace amounts of burnable neutron absorbers (BNAs) were used to tailor the reactivity of the 37-element, natural uranium (NU) fuel bundle used in CANDU reactors. The BNAs of interest included Gd2O3 and Eu2O3, which were added to the fuel in variable quantities and combinations. The fuel lattice was modeled using the WIMS–AECL 3.1 code, and core simulations were conducted using the Reactor Fuelling Simulation Program (RFSP). The fuel model assumes an equivalent and uniform distribution of BNAs in the CANLUB layer of each fuel element. The incorporation of BNAs is designed to improve CANDU reactor operating margins during on-power refueling by eliminating the fueling transient (FT) and reducing the magnitude of the plutonium peak (PP) that is characteristic of NU fuels. By adding an optimal combination of “fast-burning” and “slow-burning” BNAs, the FT and PP can be selectively reduced, and a significantly flatter trend in the burnup-dependent evolution of fuel reactivity can be achieved. The results of the study indicate that by adding ~150 mg [~8 parts per million (ppm)] of Gd2O3 and ~300 mg (~15 ppm) of Eu2O3 per fuel bundle, the best gain in the operating margins of a 2650-MW(thermal) (480-channel) model CANDU reactor can be achieved. Based on the simulation of refueling events, it was shown that the magnitude of average postrefueling channel power ripples can be reduced by an average of 100 kW and a maximum of 220 kW for powers observed immediately after refueling. This reduction in postrefueling powers was also shown to allow the average liquid zone controller level to decrease from ~48% to 10%. This decrease implies a potential relief on overpower protection (an operating margin).

A Three-Dimensional Analysis of the Local Stresses and Strains at the Pellet Ridges in a Horizontal Nuclear Fuel Element

A three-dimensional finite element model is being developed for a quarter fuel element, which is equivalent to a full fuel element using symmetry. The model uses the Multiphysics Object-Oriented Simulation Environment (MOOSE) framework developed at Idaho National Laboratory. The model facilitates an in-depth investigation into a variety of deformation phenomena for a horizontal nuclear fuel element including bowing, sagging, and stresses and strains.This paper presents a preliminary analysis of the local stresses and strains of the sheath (clad) at the pellet-to-pellet interfaces for low, normal and high linear powers. During irradiation the fuel pellets thermally expand and take on an hourglass shape. The hourglassing behaviour leads to higher local stresses and strains in the sheath at the locations of the pellet-to-pellet interfaces. The purpose of this work is to quantify these stresses and strains for varying linear powers, and to illustrate the effect that the material model chosen for the cladding has on the results. Preliminary results are presented for two sheath types: elastic, and elastic including diffusional creep. These models are benchmarked against a validated industry code called ELESTRES. The results indicate that the predicted sheath hoop strain is about half of what is determined by ELESTRES in both the elastic and elastic-creep cases. This highlights the requirement of a pellet cracking model in three-dimensional simulations. The elastic-creep model predicts less stress within the sheath than the elastic model as expected.© 2014 ASME

An Examination of CANDU Fuel Performance Margins Derived from a Statistical Assessment of Industrial Manufacturing Data

Abstract This study employs a novel approach to the prediction of CANDU [Canada deuterium uranium (reactor)] fuel reliability. Probability distributions are fitted to actual fuel manufacturing data sets provided by Cameco Fuel Manufacturing. They are used to form input for two industry-standard fuel performance codes: ELESTRES for the steady-state case and ELOCA for the transient case—a hypothesized 80% reactor outlet header break loss-of-coolant accident. Using a Monte Carlo technique for input generation, 105 independent trials are conducted, and probability distributions are fitted to key model output quantities. Comparing model output against recognized industrial acceptance criteria, no fuel failures are predicted for either case. Output distributions are well removed from failure limit values, implying that margin exists in current fuel manufacturing and design. To validate the results and attempt to reduce the simulation burden of the methodology, two dimensional reduction methods are assessed. Using just 36 trials, both methods are able to produce output distributions that agree strongly with those obtained via the brute-force Monte Carlo method, often to a relative discrepancy of <0.3% when predicting the first statistical moment and to a relative discrepancy of <5% when predicting the second statistical moment. In terms of global sensitivity, pellet density proves to have the greatest impact on fuel performance, with an average sensitivity index of 48.93% on key output quantities. Pellet grain size and dish depth are also significant contributors, at 31.53% and 13.46%, respectively. A traditional “limit of operating envelope” case is also evaluated. This case produces output values that exceed the maximum values observed during the 105 Monte Carlo trials for all output quantities of interest. In many cases the difference between the predictions of the statistical methods and the limit method is very prominent, and the highly conservative nature of the deterministic approach is demonstrated.