Finite strain materials

using MuesliMaterialsusing MuesliMaterials: check, updateCurrentState, storedEnergy, deformationGradient,    firstPiolaKirchhoffStress!, firstPiolaKirchhoffStressNumerical!,    secondPiolaKirchhoffStress!, KirchhoffStress!, CauchyStress!,    convectedTangent!, materialTangent!, spatialTangent!,    setProperty!, MaterialProperties

Finite strain models take the deformation gradient $F$ rather than a strain tensor, and offer several stress measures. This page uses a neo-Hookean material.

Building the material

Hyperelastic materials are constructed from a property map:

props = MaterialProperties()setProperty!(props, "young", 210000.0)setProperty!(props, "poisson", 0.3)material = NeoHookeMaterial(props)check(material)
true
mp = NeoHookeMP(material)

The undeformed state

At $F = I$ a hyperelastic material must be stress free and store no energy:

I3 = [1.0 0 0; 0 1 0; 0 0 1]updateCurrentState(mp, 1.0, Itensor(I3))P = Itensor()firstPiolaKirchhoffStress!(mp, P)P
MuesliMaterials.ItensorAllocated with 3x3 entries
0.0 0.0 0.0
0.0 0.0 0.0
0.0 0.0 0.0
storedEnergy(mp)
0.0

A deformed state

updateCurrentState takes the deformation gradient as an Itensor — note that this is a general tensor, not a symmetric one:

F = [1.05 0.02 0.0     0.0 0.98 0.01     0.0 0.0 1.01]updateCurrentState(mp, 1.0, Itensor(F))deformationGradient(mp)
MuesliMaterials.ItensorAllocated with 3x3 entries
1.05 0.02 0.0
0.0 0.98 0.01
0.0 0.0 1.01
storedEnergy(mp)
349.4144047009163

The stress measures

MUESLI exposes the usual family, each filled into a preallocated tensor. Note which are symmetric: $S$, $\tau$ and $\sigma$ are, $P$ is not.

P = Itensor();   firstPiolaKirchhoffStress!(mp, P)S = Istensor();  secondPiolaKirchhoffStress!(mp, S)τ = Istensor();  KirchhoffStress!(mp, τ)σ = Istensor();  CauchyStress!(mp, σ)

They are related by $P = FS$, $\tau = PF^{T}$ and $\tau = J\sigma$. Those identities are the quickest way to confirm you are reading the right quantity:

Pm = convert(Matrix{Float64}, P)Sm = convert(Matrix{Float64}, S)τm = convert(Matrix{Float64}, τ)σm = convert(Matrix{Float64}, σ)J = F[1, 1] * (F[2, 2] * F[3, 3] - F[2, 3] * F[3, 2]) -    F[1, 2] * (F[2, 1] * F[3, 3] - F[2, 3] * F[3, 1]) +    F[1, 3] * (F[2, 1] * F[3, 2] - F[2, 2] * F[3, 1])(PeqFS = Pm ≈ F * Sm, τeqPFt = τm ≈ Pm * F', τeqJσ = τm ≈ J .* σm)
(PeqFS = true, τeqPFt = true, τeqJσ = true)

Checking an implementation

MUESLI can differentiate the stored energy numerically, which is the check to run when you are unsure whether a model is wired up correctly:

Pnum = Itensor()firstPiolaKirchhoffStressNumerical!(mp, Pnum)isapprox(convert(Matrix{Float64}, P), convert(Matrix{Float64}, Pnum); rtol = 1e-5, atol = 1e-6)
true

Frame indifference

Rotating the deformation must leave the material response unchanged: $S$ is untouched, $\sigma$ rotates with the body, and the stored energy is the same.

θ = 0.4Q = [cos(θ) -sin(θ) 0.0; sin(θ) cos(θ) 0.0; 0.0 0.0 1.0]rotated = NeoHookeMP(material)updateCurrentState(rotated, 1.0, Itensor(Q * F))S2 = Istensor(); secondPiolaKirchhoffStress!(rotated, S2)σ2 = Istensor(); CauchyStress!(rotated, σ2)(    S_unchanged = convert(Matrix{Float64}, S2) ≈ Sm,    σ_rotates = convert(Matrix{Float64}, σ2) ≈ Q * σm * Q',    W_unchanged = storedEnergy(rotated) ≈ storedEnergy(mp),)
(S_unchanged = true, σ_rotates = true, W_unchanged = true)

Tangents

Three tangents are available, in the configuration their names suggest:

c = Itensor4()convectedTangent!(mp, c)c[1, 1, 1, 1]
225076.65050228298
cm = Itensor4(); materialTangent!(mp, cm)cs = Itensor4(); spatialTangent!(mp, cs)(material = cm[1, 1, 1, 1], spatial = cs[1, 1, 1, 1])
(material = 259684.492193581, spatial = 263020.2400550819)

Other finite strain models

SVKMaterial, YeohMaterial, MooneyMaterial and ArrudaBoyceMaterial follow the same pattern, as does FplasticMaterial for finite plasticity. Saint Venant–Kirchhoff, for instance:

svk = SVKMaterial(props)svkmp = SVKMP(svk)updateCurrentState(svkmp, 1.0, Itensor(F))Psvk = Itensor()firstPiolaKirchhoffStress!(svkmp, Psvk)Psvk
MuesliMaterials.ItensorAllocated with 3x3 entries
14037.793269230768 1818.8019230769237 15.83076923076923
1662.2307692307697 1862.0942307692162 842.6048076923076
2.220446049250313e-16 799.4538461538461 6756.608653846149