Assessment of the Uncertainty Associated with the Use of Superposition in CANDU Safety Analysis

There are significant benefits related to Canada deuterium uranium (CANDU) reactors’ capability to fuel the reactor during normal operations including up to 30 months’ operation without an outage, ability to utilize natural uranium-based fuels, and application for use in flux shaping and reactivity management. However, such flexibility also involves rigorous attention to the time-dependent nature of the core, since fueling may introduce local variations in the power distribution that must be demonstrated to fit within the allowable operating limits, and the initial fueling state may affect the behavior of some accident transients. For CANDU safety analysis, this time dependence is sometimes addressed using the “superposition” principle, in which the relative impact of some perturbation to the reactor core is assumed to be the same across different initial core states, greatly simplifying the analysis process but introducing a new source of uncertainty resulting from this assumption. A two-stage study has been completed to examine the sensitivity of key figures of merit in safety simulations to both nuclear data and initial core burnup distribution. First, core follow studies were performed with the PARCS code and reference nuclear data using an emulator that mimicked the decisions of a fueling engineer. Then, using the emulator, studies were performed with 60 nuclear datasets (the aforementioned reference nuclear dataset and 59 randomly selected nuclear datasets from the uncertainty files provide with the SCALE code package). For each of the 60 cases, emulated fueling was performed for 1095 operating days and the important figures of merit were extracted for each day and for each case. The time-dependent figures of merit (liquid zone controller level, maximum channel power, maximum bundle power) were compared across all of the cases and show that the effect of the nuclear data on the distributions of the key parameters was small, mainly resulting from the fueling decisions, which alters the fueling pattern to accommodate any changes in nuclear cross sections. Finally, for each day within each nuclear dataset, a representative failure of a liquid zone controller was introduced and the observed changes in the channel powers and non-failed liquid zone controller responses were observed. These results are then used to assess the uncertainty in the ”superposition” principle utilized in some safety analyses of reactivity device-driven transients to simplify analyses of CANDU reactors. The results of this work quantify the potential uncertainty introduced to safety analysis as a result of the use of superposition on the channel overpower (i.e., the relative increase in each channel power resulting from the liquid zone controller drain) and maximum channel overpower (i.e., the largest relative increase in channel power anywhere in the core resulting from the zone controller drain).