Professor Turinsky’s interests include computational reactor physics and uncertainty quantification for simulations of physical phenomena. In the area of computational reactor physics, his students and he have mainly focused on the utilization of mathematical optimization capabilities to assist in making decisions associated with nuclear fuel management. This work has addressed the optimization of the core loading pattern (LP), i.e., where to place fresh and burnt fuel assemblies within the core, for pressurized water reactors (PWR) and boiling water reactors (BWR). The work on BWRs includes simultaneously optimizing the LP and control rod program and core flow rate as a function of cycle exposure. For multicycle applications, which involve decisions regarding numbers and compositions of fresh assemblies to procure each cycle over multicycles, mathematical optimization capability has been developed for PWRs.

Recent interest has focused on closed nuclear fuel cycles, where spent fuel is reprocessing to recapture uranium, plutonium and selected minor actinides, which are then recycled in a mixture of thermal and fast spectrum reactors. Decision variables for this application include when to start construction on new fuel cycle facilities, i.e., fuel fabrication, thermal and fast reactors, and separation, and of what sizes so as to meet the nuclear power portion of the nation’s electrical energy needs such that cost, fuel resource utilization, non-proliferation and environmental impact factors are addressed. The stochastic optimization methods employed include simulated annealing and genetic algorithms applicable to single and multi-objective problems. Because stochastic optimization is computationally intensive, highly efficient core simulators for PWRs and BWRs have been developed utilizing generalized perturbation theory and nonlinear solvers to address multiphysics introduced coupling.

Professor Turinsky’s interest in uncertainty analysis has led his students and him in several different directions. For fast spectrum reactors, they have evaluated the uncertainty in core attributes of safety significance due to cross-section uncertainty. For closed fuel cycles, they have quantified the uncertainty on radio toxicity and heat loads of material to be disposed of in a repository. To minimize the uncertainties of core and waste form attributes, they have evaluated the benefits of employing zero power fast critical facilities, optimizing their composition and instrumentation. In addition, adjustment of cross-sections to improve the agreement between measured and predicted instrumentation responses for BWR cores has shown great potential in improving core simulator fidelity. The application of this work has been broadened to the thermal-hydraulic performance of reactor systems, where now uncertainties in temperatures and critical heat flux margin due to uncertainties of thermalhydraulic parameters, e.g., heat transfer coefficients and pump head curve, are evaluated for accidents. Results are utilized to optimize instrumentation deployment in support of improving prediction accuracy and plant control.

More recently an interest in determining the overall uncertainty in predictions, which includes uncertainty originating due to parameters, e.g., code input, numerical error and modeling error, has led to work on adaptive model refinement and software verification and validation. The specific focus has been on the uncertainty in knowing the modeling error, whose determination needs to be completed during software validation. An outgrowth of this has been work on developing an Adaptive Model Refinement (AMoR) capability, such that model refinement would occur automatically to produce predicted values of the desired accuracy. In this manner computer resources are utilized at the appropriate level for the decisions to be made using the predicted values.