コンテンツにスキップ

pnp-full2d

英語 | 日本語

« Home · « 電気化学・マイクロ流れ

Full dynamic Poisson-Nernst-Planck (PNP) electrochemistry -- the Gouy-Chapman electric-double-layer (EDL) verification case. This extends the STATIC Debye-Hueckel electroosmosis of electroosmosis2d.cpp (which linearized the Poisson-Boltzmann potential and froze an electroosmotic body force) to FULL dynamic multi-ion transport: every ion species is time-marched by its own Nernst-Planck lattice-Boltzmann equation, and the electric potential is re-solved from the instantaneous space-charge density each step. The Boltzmann distribution is therefore an EMERGENT steady state of the transport, not an assumption -- so at convergence the potential satisfies the full NONLINEAR Poisson-Boltzmann equation and is checked against the exact Gouy-Chapman solution (the tanh profile), the nonlinear generalization of Debye-Hueckel.

概要: 完全動的 Poisson-Nernst-Planck。各イオン種を独立に時間発展させ、電位を毎ステップ空間電荷から解き直す。ボルツマン分布を仮定せず輸送の結果として創発させ、非線形 Gouy-Chapman 解と比較する。

Geometry

pnp-full2d geometry

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。

pnp-full2d phi

Build & run

cmake -S native -B native/build && cmake --build native/build -j
./native/build/orheo-pnp-full2d

pnp_full2d [lambda_D_cells] [n_lambda] [K_phi] [max_outer] [out_prefix]

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)): 43.9 s wall。

Validation

Tier T2 — 単点比較
QoI edl_profile
Reference Gouy-Chapman EDL profile
Gate 🧩 モジュールゲート (electrokinetics)

Source & run details

View source

Item Value
Target orheo-pnp-full2d
Source native/benchmarks/pnp_full2d.cpp
Lattice D2Q9
GPU twin no
Gallery run 2026-07-27 @ 8da31257
Args ``
Wall time 43.90 s
CPU time 3148.44 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)。

pnp_full2d  Ny=72 (Nx=4)  lambda_D=8.0 cells (9 lambda_D domain)  eps=128.0  c0=1.0  K_phi=10  Ma_target=0.15
Validation: full dynamic PNP steady state vs analytic Gouy-Chapman EDL Phi(y)=4 artanh[tanh(Phi_w/4) exp(-kappa y)]

=== Phi_wall = 1.0  (Debye-Hueckel would give Phi~1.000 e^-ky at wall; GC tanh-saturates)  D_ion=0.5000 (tau=2.000) ===
    outer 0: c-(wall)=1.00000  rel=1.00e+30
    outer 2000: c-(wall)=2.12378  rel=5.29e-01
    outer 4000: c-(wall)=2.42532  rel=1.24e-01
    outer 6000: c-(wall)=2.49910  rel=2.95e-02
    outer 8000: c-(wall)=2.52675  rel=1.09e-02
    outer 10000: c-(wall)=2.54125  rel=5.71e-03
    outer 12000: c-(wall)=2.54997  rel=3.42e-03
    outer 14000: c-(wall)=2.55542  rel=2.13e-03
    outer 16000: c-(wall)=2.55886  rel=1.35e-03
    outer 18000: c-(wall)=2.56105  rel=8.52e-04
...

    outer 168000: c-(wall)=16.59708  rel=1.60e-05
    outer 170000: c-(wall)=16.59732  rel=1.44e-05
    outer 172000: c-(wall)=16.59754  rel=1.31e-05
    outer 174000: c-(wall)=16.59773  rel=1.18e-05
    outer 176000: c-(wall)=16.59791  rel=1.07e-05
    outer 178000: c-(wall)=16.59807  rel=9.66e-06
    converged at outer 178000
  L2(phi vs Gouy-Chapman)=7.394e-03  L2(c+)=1.315e-03  L2(c-)=4.120e-02  L2(ions vs Boltzmann[phi_lbm])=4.900e-02
  peak drift u=0.0758 (Ma=0.131)  outer steps=178000  [CHECK]