Homec4science
Diffusion GooseFEM (main) Browse

rGOOSEFEM/include/GooseFEM/ElementQuad4Planar.h
4ff8d7c67eb4main

Last Change
Raw File

ElementQuad4Planar.h
View Options

/**
* Quadrature for 4-noded quadrilateral element in 2d (GooseFEM::Mesh::ElementType::Quad4),
* in a Cartesian coordinate system.
* The different with ElementQuad4.h is that here the tensors live in 3d and are assumed plane
* strain.
*
* @file ElementQuad4Planar.h
* @copyright Copyright 2017. Tom de Geus. All rights reserved.
* @license This project is released under the GNU Public License (GPLv3).
*/
#ifndef GOOSEFEM_ELEMENTQUAD4PLANAR_H
#define GOOSEFEM_ELEMENTQUAD4PLANAR_H
#include "config.h"
#include "detail.h"
namespace GooseFEM {
namespace Element {
namespace Quad4 {
/**
* Interpolation and quadrature.
* Similar to Element::Quad4::Quadrature with the only different that quadrature point tensors
* are 3d ("plane strain") while the mesh is 2d.
*
* Fixed dimensions:
* - `ndim = 2`: number of dimensions.
* - `tdim = 3`: number of dimensions or tensor.
* - `nne = 4`: number of nodes per element.
*
* Naming convention:
* - `elemmat`: matrices stored per element, [#nelem, #nne * #ndim, #nne * #ndim]
* - `elemvec`: nodal vectors stored per element, [#nelem, #nne, #ndim]
* - `qtensor`: integration point tensor, [#nelem, #nip, #tdim, #tdim]
* - `qscalar`: integration point scalar, [#nelem, #nip]
*/
class QuadraturePlanar : public QuadratureBaseCartesian<QuadraturePlanar> {
public:
QuadraturePlanar() = default;
/**
* Constructor: use default Gauss integration.
* The following is pre-computed during construction:
* - the shape functions,
* - the shape function gradients (in local and global) coordinates,
* - the integration points volumes.
* They can be reused without any cost.
* They only have to be recomputed when the nodal position changes
* (note that they are assumed to be constant under a small-strain assumption).
* In that case use update_x() to update the nodal positions and
* to recompute the above listed quantities.
*
* @param x nodal coordinates (`elemvec`).
* @param thick out-of-plane thickness (incorporated in the element volume).
*/
template <class T>
QuadraturePlanar(const T& x, double thick = 1.0)
: QuadraturePlanar(x, Gauss::xi(), Gauss::w(), thick)
{
}
/**
* Constructor with custom integration.
* The following is pre-computed during construction:
* - the shape functions,
* - the shape function gradients (in local and global) coordinates,
* - the integration points volumes.
* They can be reused without any cost.
* They only have to be recomputed when the nodal position changes
* (note that they are assumed to be constant under a small-strain assumption).
* In that case use update_x() to update the nodal positions and
* to recompute the above listed quantities.
*
* @param x nodal coordinates (`elemvec`).
* @param xi Integration point coordinates (local coordinates) [#nip].
* @param w Integration point weights [#nip].
* @param thick out-of-plane thickness (incorporated in the element volume).
*/
template <class T, class X, class W>
QuadraturePlanar(const T& x, const X& xi, const W& w, double thick = 1.0)
{
m_x = x;
m_w = w;
m_xi = xi;
m_nip = w.size();
m_nelem = m_x.shape(0);
m_N = xt::empty<double>({m_nip, s_nne});
m_dNxi = xt::empty<double>({m_nip, s_nne, s_ndim});
m_thick = thick;
for (size_t q = 0; q < m_nip; ++q) {
m_N(q, 0) = 0.25 * (1.0 - xi(q, 0)) * (1.0 - xi(q, 1));
m_N(q, 1) = 0.25 * (1.0 + xi(q, 0)) * (1.0 - xi(q, 1));
m_N(q, 2) = 0.25 * (1.0 + xi(q, 0)) * (1.0 + xi(q, 1));
m_N(q, 3) = 0.25 * (1.0 - xi(q, 0)) * (1.0 + xi(q, 1));
}
for (size_t q = 0; q < m_nip; ++q) {
// - dN / dxi_0
m_dNxi(q, 0, 0) = -0.25 * (1.0 - xi(q, 1));
m_dNxi(q, 1, 0) = +0.25 * (1.0 - xi(q, 1));
m_dNxi(q, 2, 0) = +0.25 * (1.0 + xi(q, 1));
m_dNxi(q, 3, 0) = -0.25 * (1.0 + xi(q, 1));
// - dN / dxi_1
m_dNxi(q, 0, 1) = -0.25 * (1.0 - xi(q, 0));
m_dNxi(q, 1, 1) = -0.25 * (1.0 + xi(q, 0));
m_dNxi(q, 2, 1) = +0.25 * (1.0 + xi(q, 0));
m_dNxi(q, 3, 1) = +0.25 * (1.0 - xi(q, 0));
}
GOOSEFEM_ASSERT(m_x.shape(1) == s_nne);
GOOSEFEM_ASSERT(m_x.shape(2) == s_ndim);
GOOSEFEM_ASSERT(xt::has_shape(m_xi, {m_nip, s_ndim}));
GOOSEFEM_ASSERT(xt::has_shape(m_w, {m_nip}));
GOOSEFEM_ASSERT(xt::has_shape(m_N, {m_nip, s_nne}));
GOOSEFEM_ASSERT(xt::has_shape(m_dNxi, {m_nip, s_nne, s_ndim}));
m_dNx = xt::empty<double>({m_nelem, m_nip, s_nne, s_ndim});
m_vol = xt::empty<double>(this->shape_qscalar());
this->compute_dN_impl();
}
private:
friend QuadratureBase<QuadraturePlanar>;
friend QuadratureBaseCartesian<QuadraturePlanar>;
template <class T, class R>
void gradN_vector_impl(const T& elemvec, R& qtensor) const
{
GOOSEFEM_ASSERT(xt::has_shape(elemvec, this->shape_elemvec()));
GOOSEFEM_ASSERT(xt::has_shape(qtensor, this->shape_qtensor<2>()));
qtensor.fill(0.0);
#pragma omp parallel for
for (size_t e = 0; e < m_nelem; ++e) {
auto u = xt::adapt(&elemvec(e, 0, 0), xt::xshape<s_nne, s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto dNx = xt::adapt(&m_dNx(e, q, 0, 0), xt::xshape<s_nne, s_ndim>());
auto gradu = xt::adapt(&qtensor(e, q, 0, 0), xt::xshape<s_tdim, s_tdim>());
// gradu(i,j) += dNx(m,i) * u(m,j)
gradu(0, 0) = dNx(0, 0) * u(0, 0) + dNx(1, 0) * u(1, 0) + dNx(2, 0) * u(2, 0) +
dNx(3, 0) * u(3, 0);
gradu(0, 1) = dNx(0, 0) * u(0, 1) + dNx(1, 0) * u(1, 1) + dNx(2, 0) * u(2, 1) +
dNx(3, 0) * u(3, 1);
gradu(1, 0) = dNx(0, 1) * u(0, 0) + dNx(1, 1) * u(1, 0) + dNx(2, 1) * u(2, 0) +
dNx(3, 1) * u(3, 0);
gradu(1, 1) = dNx(0, 1) * u(0, 1) + dNx(1, 1) * u(1, 1) + dNx(2, 1) * u(2, 1) +
dNx(3, 1) * u(3, 1);
}
}
}
template <class T, class R>
void gradN_vector_T_impl(const T& elemvec, R& qtensor) const
{
GOOSEFEM_ASSERT(xt::has_shape(elemvec, this->shape_elemvec()));
GOOSEFEM_ASSERT(xt::has_shape(qtensor, this->shape_qtensor<2>()));
qtensor.fill(0.0);
#pragma omp parallel for
for (size_t e = 0; e < m_nelem; ++e) {
auto u = xt::adapt(&elemvec(e, 0, 0), xt::xshape<s_nne, s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto dNx = xt::adapt(&m_dNx(e, q, 0, 0), xt::xshape<s_nne, s_ndim>());
auto gradu = xt::adapt(&qtensor(e, q, 0, 0), xt::xshape<s_tdim, s_tdim>());
// gradu(j,i) += dNx(m,i) * u(m,j)
gradu(0, 0) = dNx(0, 0) * u(0, 0) + dNx(1, 0) * u(1, 0) + dNx(2, 0) * u(2, 0) +
dNx(3, 0) * u(3, 0);
gradu(1, 0) = dNx(0, 0) * u(0, 1) + dNx(1, 0) * u(1, 1) + dNx(2, 0) * u(2, 1) +
dNx(3, 0) * u(3, 1);
gradu(0, 1) = dNx(0, 1) * u(0, 0) + dNx(1, 1) * u(1, 0) + dNx(2, 1) * u(2, 0) +
dNx(3, 1) * u(3, 0);
gradu(1, 1) = dNx(0, 1) * u(0, 1) + dNx(1, 1) * u(1, 1) + dNx(2, 1) * u(2, 1) +
dNx(3, 1) * u(3, 1);
}
}
}
template <class T, class R>
void symGradN_vector_impl(const T& elemvec, R& qtensor) const
{
GOOSEFEM_ASSERT(xt::has_shape(elemvec, this->shape_elemvec()));
GOOSEFEM_ASSERT(xt::has_shape(qtensor, this->shape_qtensor<2>()));
qtensor.fill(0.0);
#pragma omp parallel for
for (size_t e = 0; e < m_nelem; ++e) {
auto u = xt::adapt(&elemvec(e, 0, 0), xt::xshape<s_nne, s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto dNx = xt::adapt(&m_dNx(e, q, 0, 0), xt::xshape<s_nne, s_ndim>());
auto eps = xt::adapt(&qtensor(e, q, 0, 0), xt::xshape<s_tdim, s_tdim>());
// gradu(i,j) += dNx(m,i) * u(m,j)
// eps(j,i) = 0.5 * (gradu(i,j) + gradu(j,i))
eps(0, 0) = dNx(0, 0) * u(0, 0) + dNx(1, 0) * u(1, 0) + dNx(2, 0) * u(2, 0) +
dNx(3, 0) * u(3, 0);
eps(1, 1) = dNx(0, 1) * u(0, 1) + dNx(1, 1) * u(1, 1) + dNx(2, 1) * u(2, 1) +
dNx(3, 1) * u(3, 1);
eps(0, 1) = 0.5 * (dNx(0, 0) * u(0, 1) + dNx(1, 0) * u(1, 1) + dNx(2, 0) * u(2, 1) +
dNx(3, 0) * u(3, 1) + dNx(0, 1) * u(0, 0) + dNx(1, 1) * u(1, 0) +
dNx(2, 1) * u(2, 0) + dNx(3, 1) * u(3, 0));
eps(1, 0) = eps(0, 1);
}
}
}
template <class T, class R>
void int_N_scalar_NT_dV_impl(const T& qscalar, R& elemmat) const
{
GOOSEFEM_ASSERT(xt::has_shape(qscalar, this->shape_qscalar()));
GOOSEFEM_ASSERT(xt::has_shape(elemmat, this->shape_elemmat()));
elemmat.fill(0.0);
#pragma omp parallel for
for (size_t e = 0; e < m_nelem; ++e) {
auto M = xt::adapt(&elemmat(e, 0, 0), xt::xshape<s_nne * s_ndim, s_nne * s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto N = xt::adapt(&m_N(q, 0), xt::xshape<s_nne>());
auto& vol = m_vol(e, q);
auto& rho = qscalar(e, q);
// M(m*ndim+i,n*ndim+i) += N(m) * scalar * N(n) * dV
for (size_t m = 0; m < s_nne; ++m) {
for (size_t n = 0; n < s_nne; ++n) {
M(m * s_ndim + 0, n * s_ndim + 0) += N(m) * rho * N(n) * vol;
M(m * s_ndim + 1, n * s_ndim + 1) += N(m) * rho * N(n) * vol;
}
}
}
}
}
template <class T, class R>
void int_gradN_dot_tensor2_dV_impl(const T& qtensor, R& elemvec) const
{
GOOSEFEM_ASSERT(xt::has_shape(qtensor, this->shape_qtensor<2>()));
GOOSEFEM_ASSERT(xt::has_shape(elemvec, this->shape_elemvec()));
elemvec.fill(0.0);
#pragma omp parallel for
for (size_t e = 0; e < m_nelem; ++e) {
auto f = xt::adapt(&elemvec(e, 0, 0), xt::xshape<s_nne, s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto dNx = xt::adapt(&m_dNx(e, q, 0, 0), xt::xshape<s_nne, s_ndim>());
auto sig = xt::adapt(&qtensor(e, q, 0, 0), xt::xshape<s_tdim, s_tdim>());
auto& vol = m_vol(e, q);
for (size_t m = 0; m < s_nne; ++m) {
f(m, 0) += (dNx(m, 0) * sig(0, 0) + dNx(m, 1) * sig(1, 0)) * vol;
f(m, 1) += (dNx(m, 0) * sig(0, 1) + dNx(m, 1) * sig(1, 1)) * vol;
}
}
}
}
void compute_dN_impl()
{
#pragma omp parallel
{
array_type::tensor<double, 2> J = xt::empty<double>({2, 2});
array_type::tensor<double, 2> Jinv = xt::empty<double>({2, 2});
#pragma omp for
for (size_t e = 0; e < m_nelem; ++e) {
auto x = xt::adapt(&m_x(e, 0, 0), xt::xshape<s_nne, s_ndim>());
for (size_t q = 0; q < m_nip; ++q) {
auto dNxi = xt::adapt(&m_dNxi(q, 0, 0), xt::xshape<s_nne, s_ndim>());
auto dNx = xt::adapt(&m_dNx(e, q, 0, 0), xt::xshape<s_nne, s_ndim>());
// J(i,j) += dNxi(m,i) * x(m,j);
J(0, 0) = dNxi(0, 0) * x(0, 0) + dNxi(1, 0) * x(1, 0) + dNxi(2, 0) * x(2, 0) +
dNxi(3, 0) * x(3, 0);
J(0, 1) = dNxi(0, 0) * x(0, 1) + dNxi(1, 0) * x(1, 1) + dNxi(2, 0) * x(2, 1) +
dNxi(3, 0) * x(3, 1);
J(1, 0) = dNxi(0, 1) * x(0, 0) + dNxi(1, 1) * x(1, 0) + dNxi(2, 1) * x(2, 0) +
dNxi(3, 1) * x(3, 0);
J(1, 1) = dNxi(0, 1) * x(0, 1) + dNxi(1, 1) * x(1, 1) + dNxi(2, 1) * x(2, 1) +
dNxi(3, 1) * x(3, 1);
double Jdet = detail::tensor<2>::inv(J, Jinv);
// dNx(m,i) += Jinv(i,j) * dNxi(m,j);
for (size_t m = 0; m < s_nne; ++m) {
dNx(m, 0) = Jinv(0, 0) * dNxi(m, 0) + Jinv(0, 1) * dNxi(m, 1);
dNx(m, 1) = Jinv(1, 0) * dNxi(m, 0) + Jinv(1, 1) * dNxi(m, 1);
}
m_vol(e, q) = m_w(q) * Jdet * m_thick;
}
}
}
}
constexpr static size_t s_nne = 4; ///< Number of nodes per element.
constexpr static size_t s_ndim = 2; ///< Number of dimensions for nodal vectors.
constexpr static size_t s_tdim = 3; ///< Dynamic alias of s_tdim (remove in C++17)
size_t m_tdim = 3; ///< Number of dimensions for tensors.
size_t m_nelem; ///< Number of elements.
size_t m_nip; ///< Number of integration points per element.
array_type::tensor<double, 3> m_x; ///< nodal positions stored per element [#nelem, #nne, #ndim]
array_type::tensor<double, 1> m_w; ///< weight of each integration point [nip]
array_type::tensor<double, 2> m_xi; ///< local coordinate per integration point [#nip, #ndim]
array_type::tensor<double, 2> m_N; ///< shape functions [#nip, #nne]
array_type::tensor<double, 3> m_dNxi; ///< local shape func grad [#nip, #nne, #ndim]
array_type::tensor<double, 4> m_dNx; ///< global shape func grad [#nelem, #nip, #nne, #ndim]
array_type::tensor<double, 2> m_vol; ///< integration point volume [#nelem, #nip]
double m_thick; ///< out-of-plane thickness
};
} // namespace Quad4
} // namespace Element
} // namespace GooseFEM
#endif
c4science · Help