No pages match that search.
Tag
Every article tagged “Uncertainty quantification”.
17 articles
How transonic CFD simulation predicts wing drag, shock position and buffet onset, what the open AIAA Drag Prediction Workshops show about its scatter, how to set up mesh convergence, turbulence models and drag decomposition, and where a surrogate of the aerodynamic database must not extrapolate.
How CFD predicts downforce, drag and aero balance for race and high-performance road cars: ground effect, rolling-road and wheel boundary conditions, RANS against DES, correlation with the tunnel on open DrivAer data, and aero maps with surrogates that state their range.
How MRI radiofrequency fields heat implants and leads, why length, route and landmark matter, how ASTM F2182 and ISO/TS 10974 frame the evidence, how EMC coupling into powered devices is simulated, and where surrogates help without replacing the physics.
How stent finite element analysis follows a stent through crimping, release into the vessel and pulsatile loading, why nitinol needs a strain-based fatigue view, what the model cannot see, and how a surrogate-assisted workflow explores the design space without replacing the durability test.
How acoustic liner design links fan duct modes, liner impedance, grazing flow and certification noise, how it is simulated with CFD and finite element duct acoustics, and where AI surrogates help and where they do not.
How vibration fatigue analysis for eVTOL and VTOL structures works: rotor harmonics, stress power spectral density, Dirlik and rainflow, damping, ground vibration testing and FE model updating with MAC gates, and where AI helps.
Power electronics thermal design for SiC and IGBT inverters: electro-thermal coupling, thermal impedance and RC models, cooler and spreading, thermomechanical FEA, and how mission-profile temperature swings become a lifetime estimate with AI surrogates and honest uncertainty.
Neural operators, PINNs, reduced-order models and Gaussian processes compared, then the validation machinery that makes any of them usable: design of experiments, hold-out by region, residual checks, conformal intervals and a governed loop with the solver.
Have a model that will not converge?
Send the geometry, the physics and the decision it has to support. We answer with an approach, not a sales call.
