Validation of the Griffin application for TREAT transient modeling and simulation

The Transient Reactor Test Facility (TREAT) at Idaho National Laboratory is a graphite-moderated, air-cooled facility that was restarted in 2018 and is specifically built to conduct transient reactor tests. The Department of Energy focused modeling and simulation efforts between 2015 and 2019 on obtaining multiphysics solutions using the Griffin reactor physics application. This research has: (1) developed a consistent analysis method that automatically maps the neutronics data generated with Serpent onto the Griffin model, (2) improved the predictive capability of the traditional MCNP and Point Kinetics models that are commonly used in TREAT modeling and the preparation of engineering calculations, (3) determined the average recoverable energy in TREAT fuel per fission event, and (4) calculated a dynamic energy coupling factor that demonstrates the dependency of the energy deposition in the experiments during transients. Three experiments conducted in 2018 form the basis for the validation of Griffin for the transient modeling of TREAT. The first two experiments are 1.5% and 2.6% Δk/k transient prescription experiments with flux wires placed in the M8CAL test rig. The third experiment is a 0.6% Δk/k with the Minimal Activation Retrievable Capsule Holder (MARCH) test rig and the Separate Effect Test Holder (SETH) capsule, a light-water reactor rodlet in dry conditions. The core behavior predicted by Griffin for the transient prescription transients agrees well with the measured core power, peak power, and reactor feedback. The exception is the total energy deposition, which is underpredicted. The core behavior for the MARCH-SETH transient is significantly influenced by delayed neutrons since the reactivity insertion is delayed critical (i.e. ρ<β_eff). The Griffin model experiences a stronger temperature feedback, leading to a lower peak power and total energy deposition. Increasing the number of temperature tabulation points significantly improves the solution for the MARCH-SETH transient. In addition, updating the 1.5% Δk/k transient model to ENDF/B-VIII.r0 data resolves the discrepancy in the prediction of the power tail and the total energy deposition. This is attributed to the new graphite S(α,β) data, which leads to lower values of the temperature coefficient of reactivity. The predicted energy coupling factors in both wire and rodlet experiments is the paramount parameter for TREAT modeling and simulation. In both cases, the Griffin predictions compare very well with measured values and exhibit consistent behavior, which is not observed in MCNP predictions. This work shows that Griffin is able to resolve the time-dependent effects from the control rod movement and the core heat-up in both the core and experiments.