Second part of the OpenFOAM 14 series: a 3D duct from eight vertices, the hex block, cell counts, simple and symmetric grading, boundary patches, blockMesh and checkMesh with expected values.
First part of the OpenFOAM 14 case-construction series: system, constant and time directories, the FoamFile header, dictionary syntax, dimensions and units, field files, macros and foamDictionary.
Three inlet velocities, same mesh and solver: maximum velocity and pressure gradient scale as Poiseuille predicts. How to isolate the effect of a boundary condition.
The same case in serial and with four MPI processes: SCOTCH decomposition, mpirun, reconstructPar, a 3.5× speed-up and numerical equivalence checked point by point.
A Python script using the Gmsh API builds the channel from command-line parameters: same logic, different geometries and meshes, validated in OpenFOAM 14 against Poiseuille.
The laminar channel rebuilt with a structured Gmsh mesh imported through gmshToFoam: patches, changeDictionary and a comparison with blockMesh giving identical results.
The same laminar channel on three meshes (2,000, 8,000, 32,000 cells): the error against Poiseuille scales with order about 2. How to set up and read a mesh convergence study in OpenFOAM.
First case of the OpenFOAM 14 path: a laminar parallel-plate channel built from scratch with blockMesh, solved with foamRun and validated against the Poiseuille profile (0.15% error).
Starting from the characteristic curves published by General Electric, I digitized and analyzed the 2D21 thyratron grid current to build a reproducible compact mathematical model. In this article I describe the full process, from replicated digitizations and uncertainty estimation to family normalization, identification of common functions, and quantitative validation of the 2D21 Level-1 v0.1 model.
SPICE model of the 6BX7-GT derived from GE datasheet curves, with digitization, Koren fitting, independent validation and ngspice testing.