Scribd
Upload
6 views

Assignment - 4

Uploaded by

Vishwa S me23e002
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
6 views

Assignment - 4

Uploaded by

Vishwa S me23e002
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Department of Applied Mechanics and Biomedical Engineering, Indian Institute of Technology, Madras.

Fundamentals of Finite Element Analysis (AM 5450)


Assignment 4 – 1D FE Solutions
Submission Date: 9 Sept 2024

Instructions:
I. Use scientific non-programmable calculator wherever necessary.
II. MATLAB/PYTHON scripts used to prepare plots/figures/results should be attached with your answer.
III. Scan all the pages of your hand-written answers, MATLAB/PYTHON scripts, output results, and figures into
a PDF file.

Definitions –
 p-type FE approach – It is an approach where improvement in the solution of the field variable 𝑢̃(𝑥) is sought
by implementing higher degree polynomial approximation for the solution.
 h-type FE approach – In order to improve the solution of the field variable 𝑢̃(𝑥) number of finite elements
(with the same order) are increased until the desired relative error threshold is achieved.
 Combination of p-type and h-type – For rapid convergence to the solution, number of higher-order finite
elements (based on higher degree polynomial approximation) are used; although, this approach requires
more computational resources.
 Relative error RE for the solution 𝑢̃(𝑥) obtained with N and N-1 finite elements is computed as follows.
𝑢̃𝑁 − 𝑢̃𝑁−1
𝑅𝐸 =
𝑢̃𝑁
 A solution quantity 𝑢̃(𝑥) (or another quantity say strain 𝑢̃′(𝑥)) is said to be converged if the Relative error
in that quantity attains a desired preset threshold criteria i.e. RE <= 1e-6. (Use this criteria for this
assignment)
Question
Find the displacement field 𝑢̃(𝑥) and strain 𝑢̃′(𝑥) for the governing differential equation of the tapered bar as shown
in Figure 1 below using the three approaches mentioned in A and B programmed in MATLAB/PYTHON.

Figure 1 Axially loaded bar with linearly varying cross-section area A(x)

Assume 𝐸 = 200 GPa, 𝐴0 = 3.1416𝐸 − 04 m2, 𝐴𝐿 = 1.9635𝐸 − 05 m2, 𝐿 = 2 m, 𝑃 = 2 kN, and 𝑞(𝑥) =
𝑐 = 1 kN/m.
𝑑 𝑑𝑢
(𝐸𝐴(𝑥) ) + 𝑞(𝑥) = 0
𝑑𝑥 𝑑𝑥
with the boundary conditions given as, 𝑢(0) = 0 and 𝐴𝐸𝑢’(𝑥)|𝑥=𝐿 = 𝑃.

Page 1 of 2
A. Use the Galerkin’s method by assuming polynomials of order 1 to NP, where NP is the converged polynomial
order at which the residual function 𝑒(𝑥) = 1E-6 (approximates to 0) for all values of 𝑥 in the domain. (a) Plot
the displacement field 𝑢̃(𝑥) and strain 𝑢̃′(𝑥) as a function of 𝑥 in MATLAB/PYTHON for each polynomial order
1 to NP. (b) Plot the residual function 𝑒(𝑥) by substituting back the obtained 𝑢̃(𝑥) for each polynomial order
1 to NP.
B. Use the Galerkin’s method with a 2-noded finite element form of assumed solution with the linear
approximation 𝑢̃(𝑥) = 𝑁1 𝑢1 + 𝑁2 𝑢2 . Use the Mapping to convert the interpolation functions Ni to a local
element CS ranging over -1 to +1. (a) Find out the number of such linear elements NLE required to obtain a
converged nodal displacement at the tip point and at the mid-point of the tapered bar. (b) Find out the relative
error for the strain 𝑢̃′(𝑥) in the elements carrying nodes of the tip and mid-point of the tapered bar when NLE
number of elements are used. (c) Plot the solutions of displacement field 𝑢̃(𝑥) and strain 𝑢̃′(𝑥) over the
length of bar (separate plots in MATLAB/PYTHON) as the number of elements increase upto NLE (Show min. 5
plots).
Which approach among A and B seems to be better? Answer based on following points –
1) Considering the computational complexity and time required.
2) Accuracy of the obtained solution and its derivative with minimum computational effort.

Page 2 of 2

You might also like