0% found this document useful (0 votes)
37 views3 pages

Project 3

This document describes using the finite element method to model the deflection of beams under an applied force. It provides the mathematical formulation of the problem and outlines steps to develop a numerical solver, including deriving the weak form, discretizing the problem, implementing a solver for a single element, extending it to multiple elements, and assessing accuracy.

Uploaded by

igerhard23
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
37 views3 pages

Project 3

This document describes using the finite element method to model the deflection of beams under an applied force. It provides the mathematical formulation of the problem and outlines steps to develop a numerical solver, including deriving the weak form, discretizing the problem, implementing a solver for a single element, extending it to multiple elements, and assessing accuracy.

Uploaded by

igerhard23
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

2.097J / 6.339J / 16.

920J
Numerical Methods for Partial Differential Equations
Massachusetts Institute of Technology – Fall 2017

Handed Out: Oct. 16, 2017 Due: Oct. 30, 2017

Finite Element Method - Deflection of Beams

Problem Statement

Small elastic deformation of isotropic solid materials in two dimensions under a force F~ (~x) is
governed by the linear elasticity equation

∂ σ~x (~u) ∂ σ~y (~u) ~


+ +F =0, x, y ∈ Ω (1)
∂x ∂y

where the deformation, a vector field ~u = (ux , uy ), is the unknown to be solved. Both σ~x = (σxx , σxy )
and σ~y = (σyx , σyy ) are vector fields in the two-dimensional space. Together they are known as the
Cauchy stress tensor. The Cauchy stress tensor depends on the deformation ~u = (ux , uy ) in the
following way:

     ∂ux 
σxx 1 0 0 ν ∂x
1−ν 1−ν  ∂ux 
 = E
σxy  0 0
2 2   ∂yy 

(2)
0  ∂u
 1−ν 1−ν
σyx  1 − ν 2 0 
2 2 ∂x 
σyy ν 0 0 1 ∂uy
∂y

where E is the Young’s modulus and ν is the Poisson ratio of the material. The material of choice is a
titanium alloy with the pliable name Ti-6Al-2Nb-1Ta-0.8Mo. You can assume E = 1.18×1011 N/m2
and ν = 0.31 for the reminder of the project. We will consider two types of boundary conditions
on the domain boundary:

1) Dirichlet boundary conditions on ∂ΩD :

ux = uy = 0. (3)

2) Free boundary conditions on ∂ΩN :

nx σ~x + ny σ~y = ~0, (4)

where ~n = (nx , ny ) is the normal vector of the boundary of the domain.

1
Questions

1) (50 pts) Mathematical foundation

a. (20 pts) Derive the weak form of the problem considering only the free boundary condi-
tion (4). Show that the weak form is symmetric and positive semi-definite.
b. (10 pts) Discretize the weak form derived in Part a. in a square domain x ∈ [−1, 1], y ∈
[−1, 1]. Reduce the system of equations to a matrix form using the following bilinear
basis functions. (Represent each entry of the matrix and the right hand side analytically
in terms of the basis functions and the force (Fx , Fy )).

(1 + x)(1 + y) (1 + x)(1 − y)
g−− := , g−+ := ,
4 4 (5)
(1 − x)(1 + y) (1 − x)(1 − y)
g+− := , g++ :=
4 4

Figure 1: The basis functions in Equation (5).

c. (5 pts) If a two-dimensional Gauss quadrature is used to integrate each entry of the


matrix in Part b., how many points will be necessary to compute an exact answer? Nu-
merically compute the matrix for a single element, assuming the free boundary conditions
for all boundaries.
d. (5 pts) Now consider a square domain x ∈ [x0 − a, x0 + a], y ∈ [y0 − a, y0 + a]. How would
the result of Part c. change?
e. (5 pts) How does the weak form derived in Part a. change if part of the domain boundary
∂ΩD satisfies the Dirichlet boundary condition (3), while the rest of the domain boundary
∂ΩN satisfies the free boundary condition (4)?
f. (5 pts) How does the discretization derived in Part b. and matrix computed in Part c.
change if the left boundary, x = −1, y ∈ [−1, 1], has the Dirichlet boundary condition
while the rest of the domain boundary satisfies the free boundary condition?

2
2) (50 pts) Stepwise, build up the capability of your numerical solver.

a. (15 pts) Program a finite element solver for a single element, assuming Dirichlet boundary
condition on the left boundary, x = −1, y ∈ [−1, 1], and free boundary condition on all
other three edges. The force exerted in the domain is a force in the negative y direction
located at x = 1, y = 0. That is, Fx ≡ 0, and Fy is a Dirac delta function with magnitude
−8 × 104 N at (1, 0). Plot the computed deflec ion ux and uy of the domain.

b. (20 pts) Build on top of your solver to accommodate beam with a single row of N
elements. The domain is now Ω = (x, y) : x ∈ [0, 2N ], y ∈ [−1, 1]. Your solution and
test functions are in the space of continuous functions that are bilinear in each element
{x ∈ [2i − 2, 2i], y ∈ [−1, 1], i = 1, . . . , N }. Apply the same boundary conditions as
in Part a., i.e., Dirichlet boundary condition (Equation 3) at x = 0, y ∈ [−1, 1], and
free boundary condition on other parts of the boundary. Apply the same concentrated
external load F~ as in Part a. on the right end of the beam, at x = 2N, y = 0. Plot the
deformation field ux and uy for N = 2, 4, and 10.

c. (15 pts) Make your solver capable of simulating a beam with multiple elements in the y
direction. Consider the same geometry, boundary condition, and force as in Part b., but
discretized using M × N by M smaller square elements for M > 1. Plot the deformation
ux and uy for (N = 10, M = 2), (N = 10, M = 4), and (N = 10, M = 8). Also plot the
deflection at x = N, y = 0 as a function of M .

3) (20 Bonus pts) Quantify the performance of your solver.

a. (5 pts) Estimate the order of accuracy of your solver, i.e. the convergence rate of the
error. You may assume a high fidelity solution (using a large M ) being the analytical
solution to reference of your error.
b. (15 pts) Discuss and explore how the accuracy of the solver can be improved.

You might also like