Fluxcore approach and capabilities

Fluxcore provides GPU-accelerated flow and conjugate heat-transfer calculations using the meshes and physical settings accepted by the central engineering service. Its purpose is to make repeated engineering analysis practical while preserving the evidence needed to assess a result.

The public simulation types are steady RANS, time-accurate URANS and CHT. Turbulence options include SST and Spalart–Allmaras, with supported DES/DDES controls for hybrid workflows. Laminar flow is currently exposed through the CHT contract. See the model compatibility table.

Speed with an engineering purpose

The five primitives and residuals guide explains density, three velocity components and pressure, and how the mean-flow equations appear as continuity, three momentum and energy residuals.

Efficient compute lets you compare designs, mesh resolutions and operating points. It does not remove the need for appropriate geometry, physical models, boundary conditions or convergence checks. Compare total time to a qualified engineering answer, including preparation and review, rather than iteration speed alone.

Mesh distribution and solver settings work together. Resolve the surfaces, layers and wakes that control the output, choose a compatible wall treatment, and retain enough solution data to inspect the result. Avoid using numerical stabilization as a substitute for missing spatial or temporal resolution.

The supported interface

Use architecture="fluxcore" or the service default. Discover advanced controls through get_solver_controls, and validate the complete request before submission. Features visible in source development are not necessarily available through the deployed API.

Saved requests, input artifact identities, convergence histories and output artifacts make runs traceable. A successful job state means the service completed its lifecycle; scientific convergence and publication of field data must still be assessed separately.