Skip to content
Snippets Groups Projects
Commit f1cdcabb authored by Patrick Jaap's avatar Patrick Jaap
Browse files

Fix in Mooney-Rivlin density: Use proper 4th term for Ciarlet

In Ciarlet the last factor is left unspecified. I could not follow the
choice used in this implementation. I guess it was intended to minimize
the energy under a hydrostatic stress F = t*I at the unit matrix, i.e.,
for t=1. A simple calculation shows that (independent on the dimension)

d = 2a + 4b + 2c

has to hold in this case. Values in the test were adjusted.
parent a97f2a69
Branches
No related tags found
1 merge request!69Fix in Mooney-Rivlin density: Use proper 4th term for Ciarlet
...@@ -64,13 +64,14 @@ public: ...@@ -64,13 +64,14 @@ public:
/* The Mooney-Rivlin-Density is given as a function of the eigenvalues of the right Cauchy-Green-Deformation tensor C = F^TF /* The Mooney-Rivlin-Density is given as a function of the eigenvalues of the right Cauchy-Green-Deformation tensor C = F^TF
or the right Cauchy-Green-Deformation tensor B = FF^T. or the right Cauchy-Green-Deformation tensor B = FF^T.
C = F^TF and B = FF^T have the same eigenvalues - we can use either one of them. C = F^TF and B = FF^T have the same eigenvalues - we can use either one of them.
There are three Mooney-Rivlin-Variants: There are three Mooney-Rivlin-Variants:
ciarlet: W(F) = mooneyrivlin_a*(normF)^2 + mooneyrivlin_b*(normFinv)^2*detF^2 + mooneyrivlin_c*(detF)^2 - ciarlet: W(F) = mooneyrivlin_a*(normF)^2 + mooneyrivlin_b*(normFinv)^2*detF^2 + mooneyrivlin_c*(detF)^2 -
((dim-1)*mooneyrivlin_a + mooneyrivlin_b + 2*mooneyrivlin_c)*ln(detF) (2*mooneyrivlin_a + 4*mooneyrivlin_b + 2*mooneyrivlin_c)*ln(detF),
where the last term is chosen s.t. W( t*I ) is minimal for t=1
log: W(F) = \sum_{i,j=0}^{i+j<=n} mooneyrivlin_ij * (I1 - 3)^i * (I2 - 3)^j + mooneyrivlin_k * ln(det(F))^2 log: W(F) = \sum_{i,j=0}^{i+j<=n} mooneyrivlin_ij * (I1 - 3)^i * (I2 - 3)^j + mooneyrivlin_k * ln(det(F))^2
square: W(F) = \sum_{i,j=0}^{i+j<=n} mooneyrivlin_ij * (I1 - 3)^i * (I2 - 3)^j + mooneyrivlin_k * 0.5 * (det(F) - 1)^2 square: W(F) = \sum_{i,j=0}^{i+j<=n} mooneyrivlin_ij * (I1 - 3)^i * (I2 - 3)^j + mooneyrivlin_k * 0.5 * (det(F) - 1)^2
For log and square: I1 and I2 are the first two invariants of C (or B), multiplied with a factor depending on det(F): For log and square: I1 and I2 are the first two invariants of C (or B), multiplied with a factor depending on det(F):
I1 = (det(F)^(-2/dim)) * [ first invariant of C ] I1 = (det(F)^(-2/dim)) * [ first invariant of C ]
= (det(F)^(-2/dim)) * (sum of all eigenvalues of C) = (det(F)^(-2/dim)) * (sum of all eigenvalues of C)
...@@ -99,7 +100,8 @@ public: ...@@ -99,7 +100,8 @@ public:
gradientInverse.invert(); gradientInverse.invert();
field_type frobeinusNormFInverseSquared = gradientInverse.frobenius_norm2(); field_type frobeinusNormFInverseSquared = gradientInverse.frobenius_norm2();
using std::log; using std::log;
return mooneyrivlin_a*frobeniusNormFsquared + mooneyrivlin_b*frobeinusNormFInverseSquared*detF*detF + mooneyrivlin_c*detF*detF - ((dim-1)*mooneyrivlin_a + mooneyrivlin_b + 2*mooneyrivlin_c)*log(detF); return mooneyrivlin_a*frobeniusNormFsquared + mooneyrivlin_b*frobeinusNormFInverseSquared*detF*detF + mooneyrivlin_c*detF*detF
- (2.0*mooneyrivlin_a + 4.0*mooneyrivlin_b + 2.0*mooneyrivlin_c)*log(detF);
} }
else else
{ {
......
...@@ -167,10 +167,10 @@ int main (int argc, char *argv[]) ...@@ -167,10 +167,10 @@ int main (int argc, char *argv[])
parameters["mooneyrivlin_a"] = "2.42e+6"; parameters["mooneyrivlin_a"] = "2.42e+6";
parameters["mooneyrivlin_b"] = "6.52e+6"; parameters["mooneyrivlin_b"] = "6.52e+6";
parameters["mooneyrivlin_c"] = "-7.34e+6"; parameters["mooneyrivlin_c"] = "-7.34e+6";
expectedEnergy = 68753239.6; expectedEnergy = 76302830.4;
expectedGradientTwoNorm = 31244643.1; expectedGradientTwoNorm = 40670527.3;
expectedGradientInfinityNorm = 9673572.39; expectedGradientInfinityNorm = 11116511.8;
expectedMatrixFrobeniusNorm = 1.67660965e+09; expectedMatrixFrobeniusNorm = 2.1978108e+09;
int testCiarlet = assembleAndCompare(basis, parameters, x, expectedEnergy, expectedGradientTwoNorm, expectedGradientInfinityNorm, expectedMatrixFrobeniusNorm); int testCiarlet = assembleAndCompare(basis, parameters, x, expectedEnergy, expectedGradientTwoNorm, expectedGradientInfinityNorm, expectedMatrixFrobeniusNorm);
return testCiarlet + testLog + testSquare; return testCiarlet + testLog + testSquare;
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment