Independent native rebuild / 独立原生重建

Full beginner GUI route / 完整初学者界面路线

The numbered lessons now give every click, node, field and check. Begin with the P0 benchmark in M1. The compact tables below are a reference once you know the interface.

编号课程现已列出每个点击、节点、字段与核对步骤。从 M1 的 P0 基准开始，熟悉界面后再把下面简表当速查。

GUI basics · Worked derivations

界面基础 · 完整推导

Use the saved native configuration to check every node, then start in an empty Model Wizard. Parameter/expression files make long EOS entries inspectable; importing expressions is allowed, but the geometry, equations, boundaries, solver and result choices remain your rebuild task.

用保存的原生配置核对每个节点，再从空白模型向导开始。参数／表达式文件便于检查长 EOS 项；可以导入表达式，但几何、方程、边界、求解器和结果选择仍由你重建。

Model A / 模型 A

Node / 节点	Exact choice and purpose / 设置与目的
Model Wizard	3D → Coefficient Form PDE + Global Equations → Time Dependent. One physics pair per component.
Global Definitions	Use parameter table, source cards and expression map, plus p(T,ρ), u(T,ρ), g(T,ρ) in pfpp/pfpu/pfpg_expression.txt. Arguments and function results are dimensionless; wrap physical units in variables.
Geometry	Water outer sphere Lout=0.9 mm. Single/3/5 components subtract source matching spheres; add a receiver cylinder radius 50 μm, height 5 μm centered at z=0.5 mm. Form Union retains receiver boundaries.
Selections and integrations	Source boxes: entitydim=2, bounds center±1.001b, condition Inside. Receiver: entitydim=3, x/y±1.001a_det, z=r_obs±0.501h_det. Source and receiver integrations must be nonempty. Far boundary is Complement of inside ±0.8Lout boundary box.
Coefficient Form PDE	χ=φτ/a². ea=1/c², da=0, c=1, f=0. Initial χ=χ̇=0. Source Flux/Source g=−τQ/(a²A_b), q=0. Radiation g=−χ̇/c, q=1/Lout. No extra spherical radiation term in the radius law.
Global Equations	Copy each dependent-variable/equation/initial-state array from the exact source script. Radius second derivative, phase EOS constraints, dissipated/radiated energy integrators and receiver filter are solved together. Gas uses r, ediss, erout, pdet; each PFP adds θ,qL,qV.
Discretization	Quadratic Lagrange potential. Global size 150 μm maximum, 5 μm minimum, growth 1.5, curvature .3; resolved source surfaces local maximum a_i/2. The gas geometry uses fixed b=110 μm and the mesh made at case0.
Study activation	Each study solves only its own component physics. Global tags use unique component suffixes. Aux gas list 0…6 with source interpolation cards; initialize each from the declared rest state.
Solver	Fully Coupled + Direct PARDISO. Create the replacement before removing generated Segregated. Manual scales=1, relative tolerance 2e−5, generalized-alpha ρ∞=.75. Baseline 25 ns; authoritative case0 study 6.25 ns internal / 12.5 ns output.
Uniform component	No holes. Partition a 200 μm population sphere. PDE distributed source = −100τQ/(a²V_pop). One representative phase state coupled to mean population χ̇. Authoritative uniform study uses 6.25 ns internal and relative tolerance 2e−6.
Results	Evaluate filtered gauge pdet·100 kPa, separate raw receiver mean, R/a, wall Mach, phase and total-energy residuals. Use authoritative studies for gas0 and population. Compare with supplied independent references and numerical controls.

Node / 节点

Exact choice and purpose / 设置与目的

Model Wizard

3D → Coefficient Form PDE + Global Equations → Time Dependent. One physics pair per component.

Global Definitions

Use parameter table, source cards and expression map, plus p(T,ρ), u(T,ρ), g(T,ρ) in pfpp/pfpu/pfpg_expression.txt. Arguments and function results are dimensionless; wrap physical units in variables.

Geometry

Water outer sphere Lout=0.9 mm. Single/3/5 components subtract source matching spheres; add a receiver cylinder radius 50 μm, height 5 μm centered at z=0.5 mm. Form Union retains receiver boundaries.

Selections and integrations

Source boxes: entitydim=2, bounds center±1.001b, condition Inside. Receiver: entitydim=3, x/y±1.001a_det, z=r_obs±0.501h_det. Source and receiver integrations must be nonempty. Far boundary is Complement of inside ±0.8Lout boundary box.

Coefficient Form PDE

χ=φτ/a². ea=1/c², da=0, c=1, f=0. Initial χ=χ̇=0. Source Flux/Source g=−τQ/(a²A_b), q=0. Radiation g=−χ̇/c, q=1/Lout. No extra spherical radiation term in the radius law.

Global Equations

Copy each dependent-variable/equation/initial-state array from the exact source script. Radius second derivative, phase EOS constraints, dissipated/radiated energy integrators and receiver filter are solved together. Gas uses r, ediss, erout, pdet; each PFP adds θ,qL,qV.

Discretization

Quadratic Lagrange potential. Global size 150 μm maximum, 5 μm minimum, growth 1.5, curvature .3; resolved source surfaces local maximum a_i/2. The gas geometry uses fixed b=110 μm and the mesh made at case0.

Study activation

Each study solves only its own component physics. Global tags use unique component suffixes. Aux gas list 0…6 with source interpolation cards; initialize each from the declared rest state.

Solver

Fully Coupled + Direct PARDISO. Create the replacement before removing generated Segregated. Manual scales=1, relative tolerance 2e−5, generalized-alpha ρ∞=.75. Baseline 25 ns; authoritative case0 study 6.25 ns internal / 12.5 ns output.

Uniform component

No holes. Partition a 200 μm population sphere. PDE distributed source = −100τQ/(a²V_pop). One representative phase state coupled to mean population χ̇. Authoritative uniform study uses 6.25 ns internal and relative tolerance 2e−6.

Results

Evaluate filtered gauge pdet·100 kPa, separate raw receiver mean, R/a, wall Mach, phase and total-energy residuals. Use authoritative studies for gas0 and population. Compare with supplied independent references and numerical controls.

Model B / 模型 B

Node / 节点	Exact choice and purpose / 设置与目的
Model Wizard	3D → Solid Mechanics + Global Equations → Stationary. Enable geometric nonlinearity.
Parameters	Full baseline parameters. a=300 nm radius, B=20, G=10 kPa, K/G=1000, p₀=100 kPa. σ_LL=.054 N/m is an assumed effective PFP/gel coefficient from a water/PFP reference, not a measured gel interface.
Geometry and selection	Difference: sphere radius B·a minus sphere radius a. Cavity boundary Box, Inside, entitydim=2, ±1.001a. All surrounding gel volume retained.
Solid Mechanics	Hyperelastic Material: Neo-Hookean, Nearly Incompressible, pressure mixed formulation, K/G user-defined; ρ=1006.5 kg/m³. Free outer traction. Rigid Motion Suppression only. No cavity prescribed displacement.
Surface energy	Follower Pressure p_l−p₀ on cavity. Boundary Weak Contribution in material quadrature: expanded variation, cofactor/surface variables. Constant σ_LL energy/stress assumption.
Deformed geometry measures	Material-frame integration ic on cavity. CN = cof(F)n₀, Js=\|CN\|. V=−ic[(X+u)CN1+(Y+v)CN2+(Z+w)CN3]/3. R=(3V/4π)^(1/3); A=ic(Js). This measures solved shape without imposing sphericity.
Finite PFP closure	p,u,g EOS functions; ρ_l=ρ_lref exp(qL), ρ_v=ρ_vref exp(qV), T=Tref θ. f_v=(V−m/ρ_l)/[m(1/ρ_v−1/ρ_l)]. Solve Laplace, Gibbs equality and retained-energy residual using the literal expression table and full builder.
Mesh	Free Tetrahedral; quadratic displacements; global hmax=B·a/4, hmin=a/6, growth1.45, curvature.25; cavity local hmax=a/5. Native fine study 77,481 solved DOFs.
Baseline initial guess	lam=1.8; initialize displacement [(r_ref³+a³(lam³−1))^(1/3)/r_ref−1]·(X,Y,Z). Global θ,qL,qV=(1.1655711769961976,−.11475258143625314,1.5473196370961337). This is an initial guess, not a constraint.
Study / solver	Aux Qdep list11.55,11.65,11.75,11.82,11.9 pJ, previous solution reuse. Fully Coupled + PARDISO, stationary relative tolerance1e−7. Manual displacement scale a, mixed pressure G, global variables1.
Sensitivity initialization	Before recomputing the separate sensitivity study: set lam=2.5441334743639912 and global θ,qL,qV=(1.1398107875349053,−.093845518489624163,1.3349352599864706). Cases0–3, Qdep=13 pJ; global G/σ functions given in GelSensitivityValid3D.java. Restore baseline guesses before baseline recomputation.
Results and acceptance	Export Qdep,R/a,max principal stretch,T,f_v, shell thickness, phase/mass/energy residuals. Check analytic pressure/work and independent mesh changes. The local λ_f=2.1 screen is a proxy, not measured cavity-fracture certification.

Node / 节点

Exact choice and purpose / 设置与目的

Model Wizard

3D → Solid Mechanics + Global Equations → Stationary. Enable geometric nonlinearity.

Parameters

Full baseline parameters. a=300 nm radius, B=20, G=10 kPa, K/G=1000, p₀=100 kPa. σ_LL=.054 N/m is an assumed effective PFP/gel coefficient from a water/PFP reference, not a measured gel interface.

Geometry and selection

Difference: sphere radius B·a minus sphere radius a. Cavity boundary Box, Inside, entitydim=2, ±1.001a. All surrounding gel volume retained.

Solid Mechanics

Hyperelastic Material: Neo-Hookean, Nearly Incompressible, pressure mixed formulation, K/G user-defined; ρ=1006.5 kg/m³. Free outer traction. Rigid Motion Suppression only. No cavity prescribed displacement.

Surface energy

Follower Pressure p_l−p₀ on cavity. Boundary Weak Contribution in material quadrature: expanded variation, cofactor/surface variables. Constant σ_LL energy/stress assumption.

Deformed geometry measures

Material-frame integration ic on cavity. CN = cof(F)n₀, Js=|CN|. V=−ic[(X+u)CN1+(Y+v)CN2+(Z+w)CN3]/3. R=(3V/4π)^(1/3); A=ic(Js). This measures solved shape without imposing sphericity.

Finite PFP closure

p,u,g EOS functions; ρ_l=ρ_lref exp(qL), ρ_v=ρ_vref exp(qV), T=Tref θ. f_v=(V−m/ρ_l)/[m(1/ρ_v−1/ρ_l)]. Solve Laplace, Gibbs equality and retained-energy residual using the literal expression table and full builder.

Mesh

Free Tetrahedral; quadratic displacements; global hmax=B·a/4, hmin=a/6, growth1.45, curvature.25; cavity local hmax=a/5. Native fine study 77,481 solved DOFs.

Baseline initial guess

lam=1.8; initialize displacement [(r_ref³+a³(lam³−1))^(1/3)/r_ref−1]·(X,Y,Z). Global θ,qL,qV=(1.1655711769961976,−.11475258143625314,1.5473196370961337). This is an initial guess, not a constraint.

Study / solver

Aux Qdep list11.55,11.65,11.75,11.82,11.9 pJ, previous solution reuse. Fully Coupled + PARDISO, stationary relative tolerance1e−7. Manual displacement scale a, mixed pressure G, global variables1.

Sensitivity initialization

Before recomputing the separate sensitivity study: set lam=2.5441334743639912 and global θ,qL,qV=(1.1398107875349053,−.093845518489624163,1.3349352599864706). Cases0–3, Qdep=13 pJ; global G/σ functions given in GelSensitivityValid3D.java. Restore baseline guesses before baseline recomputation.

Results and acceptance

Export Qdep,R/a,max principal stretch,T,f_v, shell thickness, phase/mass/energy residuals. Check analytic pressure/work and independent mesh changes. The local λ_f=2.1 screen is a proxy, not measured cavity-fracture certification.

The scripts below reproduce the files from scratch using COMSOL’s supported Java API. They are also an audit trail for the manual recipe. Native HTML reports can simplify displayed derivative expressions in their reporting context; use the exact builder and the expanded surface/post variable imports for deformation gradients, rather than copying displayed identity constants.

下列脚本通过 COMSOL 支持的 Java API 从头重现文件，也提供手动步骤的审计线索。原生 HTML 报告在其报告上下文可能简化显示导数表达式；变形梯度应采用精确构建脚本及展开的表面／后处理变量导入项，不能照抄所显示的单位阵常量。

Model A: PressureEnvelope3D → SmallGasRefinement3D → UniformPfpFinalRefinement3D → FinalModelA3D Checks: WorstGasTime3D, WorstGasMesh3D, FivePfpDomain3D Model B: FreeCoreShellFine3D → GelSensitivityValid3D → FinalModelB3D Checks: FreeCoreShellPfp3D, free-gel/thermal/EOS controls, independent reference scripts