Abstract Computational models and simulations (CM&S) are increasingly used in medical device development, yet their adoption in regulatory decision-making remains limited due to challenges in establishing model credibility. This study presents a structured and reproducible verification framework for a computational model of a simplified medical device involving conjugate heat transfer using OpenFOAM's chtMultiRegionFoam solver. Code verification is performed using analytical benchmarks for laminar pipe flow, radial heat conduction, and transient thermal diffusion yielding observed orders of convergence of p = 1.95, p = 2.08, and p = 2.15 (at 0.025 s) respectively. Solution verification evaluates discretization error through systematic mesh and time step refinement and regression-based extrapolation using the StREEQ method. Results show that spatial and temporal discretization errors in transient coupled simulations interact and cannot be assessed independently. Co-refinement is required to capture true convergence behavior, while independent refinement yields polluted convergence rates. Computational cost analysis demonstrates diminishing returns with refinement and highlights that runtime does not vary monotonically with discretization parameters due to solver convergence behavior. The use of a simplified geometry enables rigorous verification while retaining the dominant physics of the system. Grounded in a risk-informed framework consistent with ASME V&V 40 Standard, this study demonstrates that a level of verification rigor beyond minimal practice can be achieved through targeted model simplification and systematic analysis under realistic computational constraints
Abstract This study evaluates computational fluid dynamics (CFD) methods for urban atmospheric dispersion using high-resolution three-dimensional experimental data obtained through magnetic resonance imaging (MRI). Laboratory experiments were conducted in a controlled water channel featuring a 3D-printed urban array with dual contaminant injection sites. By applying magnetic resonance velocimetry (MRV) and magnetic resonance concentration (MRC), volumetric measurements of mean velocity and contaminant concentration were acquired at an isotropic resolution of 0.7 mm, yielding extensive datasets for direct, quantitative comparison with large eddy simulation (LES) results using SIERRA/Fuego. We applied established metrics-fractional mean bias, normalized mean square error, geometric mean, geometric variance, FAC2, and normalized absolute difference to assess model performance. Additionally, new metrics including local normalized difference (LND), norm of relative difference (RD), projection coefficient (PC), and a mutual information distance (D) were introduced to capture both bias and variance in the data. Sensitivity analysis using synthetic datasets informed the development of intrinsic, data-based acceptance criteria. Overall, excellent agreement between CFD predictions and MRI measurements is demonstrated, highlighting the potential of combining detailed experimental data with advanced quantitative metrics to improve the evaluation and fidelity of CFD models for urban dispersion scenarios.
Abstract Physics models of engineering systems often require geometric representations of complex physical domains. When smooth or detailed features are simplified for model construction or meshing, the resulting geometric defeaturing can introduce a form of model error that is distinct from numerical discretization error. This work investigates that effect for a two-dimensional composite unit cell containing a centered circular filler that is represented by regular polygons with 3–26 sides. The system response quantity is the dimensionless effective thermal conductivity, k*, evaluated over fill fractions from 1% to 78% and filler-to-matrix thermal-conductivity ratios from 0 to 100. Code verification is performed using the Method of Manufactured Solutions, and solution verification is performed using a global-deviation grid convergence index to estimate numerical uncertainty in k*. The geometric-fidelity effect is then quantified by comparing each polygonal representation with a 26-sided reference representation and by interpreting this difference relative to the estimated numerical uncertainty. The results show that geometric representation can materially affect k*, particularly for high fill fractions and large conductivity contrasts, while numerical uncertainty is generally small on the finest meshes. Across the full ensemble, 1981 of 2016 fine-mesh cases had relative numerical uncertainty Unum/k* below 1%, with a median value of 0.02%, while low-sided polygon cases often exceeded the 1% geometric-convergence threshold. The study demonstrates the importance of treating geometric fidelity as a separate uncertainty source in verification, validation, and uncertainty quantification workflows for composite and porous-media thermal models.
Abstract Engineering simulations based on high fidelity mathematical models support safety, performance, and mission critical decisions in nuclear energy, aerospace, and defense. The ASME V&V 20 standard quantifies a range that bounds true model error for a scalar quantity of interest at discrete validation setpoints where experimental data exist. Because decisions are made at application points where direct experimental data are unavailable, a method is needed to extend model error knowledge from validation setpoints to application points for margin analyses. This article develops and analyzes a regression based procedure extending ASME V&V 20 scalar validation metrics from validation setpoints to application points within a validation space. The approach performs weighted polynomial regression on the upper and lower bounds of the V&V 20 modeling error interval, quantifies uncertainty from residual scatter and coefficient estimation, and propagates uncertainty to a selected application point. Numerical uncertainty at the application point is incorporated consistent with V&V 20, yielding a modeling error interval suitable for safety and performance margin assessments.
Abstract Oblique shock reflections occur in a variety of instances when a supersonic flow is compressed by a concave contact surface. These reflections can vary significantly in complexity depending on the initial Mach number of the planar shock, the number of facets on the contact surface, and their corresponding angles. Standard analyses of oblique shocks, such as shock polar analysis, provide only limited information for shocks over multifaceted reflectors. Thus, it is necessary to explore and develop additional techniques to assess these types of problems. This exploratory study proposes and examines the utility of using the method of characteristics (MOC) as a physics-informed reference to test whether hydrodynamic code (hydrocode) solutions in smooth supersonic regions are consistent with a characteristic-based continuation of hydrocode-extracted upstream data. This technique will be applied to a single-wedge and double-wedge simulation to examine method-to-method discrepancies and refinement behavior. The resulting characteristic nets and summary metrics demonstrate that MOC can be useful for assessing consistency and refinement behavior in simulations of oblique shock reflections.
