from numpy import * def legendre_interval(count): from math import sqrt if count <= 1: p = array([0.0]) w = array([2.0]) return p, w elif count == 2: p = array([-1.0 / sqrt(3.0), 1.0 / sqrt(3.0)]) w = array([1.0, 1.0]) return p, w elif count == 3: p = array([-sqrt(3.0 / 5.0), 0.0, sqrt(3.0 / 5.0)]) w = array([5.0 / 9.0, 8.0 / 9.0, 5.0 / 9.0]) return p, w elif count == 4: p = array([-sqrt((3.0 + 2.0 * sqrt(6.0 / 5.0)) / 7.0), -sqrt((3.0 - 2.0 * sqrt(6.0 / 5.0)) / 7.0), sqrt((3.0 - 2.0 * sqrt(6.0 / 5.0)) / 7.0), sqrt((3.0 + 2.0 * sqrt(6.0 / 5.0)) / 7.0)]) w = array([(18.0 - sqrt(30.0)) / 36.0, (18.0 + sqrt(30.0)) / 36.0, (18.0 + sqrt(30.0)) / 36.0, (18.0 - sqrt(30.0)) / 36.0]) return p, w else: p = array([-(1.0 / 3.0) * sqrt(5.0 + 2.0 * sqrt(10.0 / 7.0)), -(1.0 / 3.0) * sqrt(5.0 - 2.0 * sqrt(10.0 / 7.0)), 0.0, (1.0 / 3.0) * sqrt(5.0 - 2.0 * sqrt(10.0 / 7.0)), (1.0 / 3.0) * sqrt(5.0 + 2.0 * sqrt(10.0 / 7.0))]) w = array([(322.0 - 13.0 * sqrt(70.0)) / 900.0, (322.0 + 13.0 * sqrt(70.0)) / 900.0, 128.0 / 225.0, (322.0 + 13.0 * sqrt(70.0)) / 900.0, (322.0 - 13.0 * sqrt(70.0)) / 900.0]) return p, w def legendre_quad(order): from numpy import zeros (p, w) = legendre_interval(order) size = len(p) xi = zeros(size * size) eta = zeros(size * size) weight = zeros(size * size) for i in range(size): for j in range(size): xi[i * size + j] = p[i] eta[i * size + j] = p[j] weight[i * size + j] = w[i] * w[j] return xi, eta, weight def iso_quad(element_nodes, xi, eta): from numpy import array from numpy import sum from numpy.linalg import det from numpy.linalg import inv shape = array([ (1.0 - xi) * (1.0 - eta) / 4.0, (1.0 + xi) * (1.0 - eta) / 4.0, (1.0 + xi) * (1.0 + eta) / 4.0, (1.0 - xi) * (1.0 + eta) / 4.0 ]) # bilinear shape functions shape_dxi = array([ -(1.0 - eta) / 4.0, (1.0 - eta) / 4.0, (1.0 + eta) / 4.0, -(1.0 + eta) / 4.0 ]) # derivatives of the shape functions in the first parametric direction shape_deta = array([ -(1.0 - xi) / 4.0, -(1.0 + xi) / 4.0, (1.0 + xi) / 4.0, (1.0 - xi) / 4.0 ]) # derivatives of the shape functions in the second parametric direction x = element_nodes[:, 0] y = element_nodes[:, 1] jacobi = array([ [sum(shape_dxi * x), sum(shape_dxi * y)], [sum(shape_deta * x), sum(shape_deta * y)] ]) # Jacobi matrix jacobian = det(jacobi) inverted_jacobi = inv(jacobi) shape_dx = inverted_jacobi[0, 0] * shape_dxi + inverted_jacobi[0, 1] * shape_deta shape_dy = inverted_jacobi[1, 0] * shape_dxi + inverted_jacobi[1, 1] * shape_deta return jacobian, shape, shape_dx, shape_dy def thermal_force_quads(nodes, elements, thickness, elasticity_matrix, alpha_t, tfunc=None, gauss_order=3): from numpy import sum, array freedom = 2 dimension = len(nodes) * freedom force = zeros(dimension) element_nodes = 4 (xi, eta, w) = legendre_quad(gauss_order) alpha = array([alpha_t, alpha_t, 0.0]) for element in elements: fe = zeros(element_nodes * freedom) vertices = nodes[element[:], :] for i in range(len(w)): (jacobian, shape, shape_dx, shape_dy) = iso_quad(vertices, xi[i], eta[i]) b = array([ [shape_dx[0], 0.0, shape_dx[1], 0.0, shape_dx[2], 0.0, shape_dx[3], 0.0], [0.0, shape_dy[0], 0.0, shape_dy[1], 0.0, shape_dy[2], 0.0, shape_dy[3]], [shape_dy[0], shape_dx[0], shape_dy[1], shape_dx[1], shape_dy[2], shape_dx[2], shape_dy[3], shape_dx[3]] ]) if tfunc is None: fe = fe + thickness * b.transpose().dot(elasticity_matrix).dot(alpha) * w[i] * jacobian else: therm = tfunc(sum(vertices[:, 0] * shape), sum(vertices[:, 1] * shape)) fe = fe + thickness * therm * b.transpose().dot(elasticity_matrix).dot(alpha) * w[i] * jacobian for i in range(element_nodes * freedom): ii = element[i / freedom] * freedom + i % freedom force[ii] += fe[i] return force