The last decade has seen the development and application of data-driven methods taking off in nuclear engineering research, aiming to improve the safety and reliability of nuclear power. This work focuses on developing a reinforcement learning-based control sequence optimization framework for advanced nuclear systems, which not only aims to enhance flexible operations, promoting the economics of advanced nuclear technology, but also prioritizing safety during normal operation. At its core, the framework allows the sequence of operational actions to be learned and optimized by an agent to facilitate smooth transitions between the modes of operations (i.e., load-following), while ensuring that all safety significant system parameters remain within their respective limits. To generate dynamic system responses, facilitate control strategy development, and demonstrate the effectiveness of the framework, a simulation environment of a pebble-bed high-temperature gas-cooled reactor was utilized. The soft actor-critic algorithm was adopted to train a reinforcement learning agent, which can generate control sequences to maneuver plant power output in the range between 100% and 50% of the nameplate power through sufficient training. It was shown in the performance validation that the agent successfully generated control actions that maintained electrical output within a tight tolerance of 0.5% from the demand while satisfying all safety constraints. During the mode transition, the agent can maintain the reactor outlet temperature within $\pm 1.5$ °C and steam pressure within 0.1 MPa of their setpoints, respectively, by dynamically adjusting control rod positions, control valve openings, and pump speeds. The results demonstrate the effectiveness of the optimization framework and the feasibility of reinforcement learning in designing control strategies for advanced reactor systems.

Best Estimate Plus Uncertainty (BEPU) approaches for nuclear reactor applications have been extensively developed in recent years. The challenge for BEPU approaches is to achieve multi-physics modeling with an acceptable computational cost while preserving a reasonable fidelity of the physics modeled. In this work, we present the core multi-physics computational framework developed for the efficient computation of uncertainties in Light Water Reactor (LWR) simulations. The subchannel thermal-hydraulic code CTF and the nodal expansion neutronic code PARCS are coupled for the multi-physics modeling (CTF-PARCS). The computational framework is discussed in detail from the Polaris lattice calculations up to the CTF-PARCS coupling approaches. Sampler is used to perturb the multi-group microscopic cross-sections, fission yields and manufacturing parameters, while Dakota is used to sample the CTF input parameters and the boundary conditions. Python scripts were developed to automatize and modularize both pre- and post-processing. The current state of the framework allows the consistent perturbation of inputs across neutronics and thermal-hydraulics modeling. Improvements to the standard thermal-hydraulics modeling for such coupling approaches have been implemented in CTF to allow the usage of 3D burnup distribution, calculation of the radial power and the burnup profile, and the usage of Santamarina effective Doppler temperature. The uncertainty quantification approach allows the treatment of both scalar and functional quantities and can estimate correlation between the multi-physics outputs of interest and up to the originally perturbed microscopic cross-sections and yields. The computational framework is applied to three exercises of the LWR Uncertainty Analysis in Modeling Phase III benchmark. The exercises cover steady-state, depletion and transient calculations. The results show that the maximum fuel centerline temperature across all exercises is $2474\mathrm{K}$ with $1.7\%$ uncertainty and that the most correlated inputs are the ²³⁸U inelastic and elastic cross-sections above 1 MeV.