コンテンツにスキップ

graetz2d

英語 | 日本語

« Home · « 移流拡散・反応・輻射

orheo-graetz2d -- the Graetz problem as a callable tier for MF-6. --------------------------------------------------------------------------- A fully developed Poiseuille flow enters a heated section of length L whose walls are held at a constant scalar value C_w. The physics and its gates live in ctest graetz2d (tests/test_graetz2d.cpp); THIS binary exists so the multi-fidelity driver can ask one question -- "what mean Nusselt number does a finite section of this length at this Peclet number actually deliver?" -- and get a number back with its convention attached.

Geometry

graetz2d geometry

Animation

Fields

この計算が最後の出力ステップで書き出した場をそのまま描いたもの。3D ケースは中央断面。

graetz2d C

Plots

この計算が書き出した QoI の時系列・分布。参照値がある場合は水平線で重ねている。

graetz2d nu_x

Build & run

cmake -S native -B native/build && cmake --build native/build -j
./native/build/orheo-graetz2d

graetz2d <config-file> (KEY = VALUE lines; NOT KEY=VALUE argv)

このページの図を作った計算のコスト(Miyabi-G Grace CPU (72 cores, -stdpar=multicore)): 3.8 s wall。

Validation

Tier T1 — 自動ゲートあり
Gate ✅ 直接ゲート

Source & run details

View source

Item Value
Target orheo-graetz2d
Source native/benchmarks/graetz2d.cpp
Lattice D2Q9
GPU twin no
Modules advection_diffusion.hpp, channel2d.hpp
Gallery run 2026-07-28 @ c9b0572c
Args ``
Wall time 3.81 s
CPU time 223.17 s
Device Miyabi-G Grace CPU (72 cores, -stdpar=multicore)

Computation log

この計算の実際の出力(先頭と末尾)。

=== graetz2d: Ny=40 L=1200  Pe_Dh=100.0 (Dh=2H=80)  U=0.0400  D=0.032000  tau_g=0.5960 ===
velocity: settled Poiseuille, U_measured=0.039993 (target 0.040000)
  step    200  max|dC|/step=3.397e-03
  step    400  max|dC|/step=8.185e-04
  step    600  max|dC|/step=4.852e-04
  step    800  max|dC|/step=3.454e-04
  step   1000  max|dC|/step=2.688e-04
  step   1200  max|dC|/step=2.209e-04
  step   1400  max|dC|/step=1.894e-04
  step   1600  max|dC|/step=1.689e-04
  step   1800  max|dC|/step=1.563e-04
  step   2000  max|dC|/step=1.499e-04
  step   2200  max|dC|/step=1.482e-04
  step   2400  max|dC|/step=1.471e-04
...


C_m: inlet 0.005953 -> outlet 0.990005   theta_out=1.005494e-02
Nu_lm (LOG-MEAN, the coupling convention) = 7.666151
nu_mean_arith (plain average of Nu_x)     = 7.724729   [VERIFY which mean Shah & London tabulate before comparing]
Nu_x at the outlet                        = 7.584018
energy closure: whole domain 6.884e-02 (includes the singular inlet column and the first-order outflow clamp)
                interior [120,1080] 4.456e-03  <- the wall-stencil figure; this is the one a coupled gate means
wrote ./graetz2d.csv
RESULT Ny=40 L=1200 Pe_Dh=100.0 Dh=80 U=3.999300000e-02 D=3.200000000e-02 theta_out=1.005494492131e-02 nu_mean=7.666151222523e+00 nu_mean_arith=7.724728916780e+00 wall_flux=1.682576504e+00 advected=1.574206959e+00 closure=6.884e-02 closure_interior=4.456e-03 steps=30800