The Finite Element Method, TMMI37, Autumn 2013

modified 2013-09-20

Laboratory Problem IIa

Below follows a description of Lab IIa, which
 Consists of three parts (Part 1, Part 2 & Part 3)
 Is to be evaluated/reported by peer student evaluation and reporting to the
teacher
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Laboratory problem IIa; Part 1

Simple 2-dimensional elastostatic FE-analysis,
or
What does the stress state look like at an elliptical hole in a
uniaxially loaded thin plate?

Introduction

We will here, as a simple application of FEM for 2-dim. elastostatic problems,

study the stress state at an elliptical hole in a uniaxially loaded thin plate.

There are a number of reasons why we have chosen this problem

 It is a geometrically simple problem, for which the FE-results can be

compared with dittos based on handbook expressions

 You will have the opportunity to take advantage of the symmetry in the
problem (good to practice), a procedure which often reduces the
computational cost substantially

 You will get an insight in how stress concentrations appear at holes in

loaded structures. More generally one may say that

Stress concentrations appear at all rapid geometry changes in loaded

structures, which is an issue that has to be taken into consideration in
the design work!

 It is an appropriate problem for studying so called “mapped meshing”

technique, where the structure is split into 4-sided regions such that it can
be meshed with nicely looking rectangular/4 sided elements whose
sizes can be easily controlled (see the figure below).
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Aim

The aim of the problem is first of all that you are to analyze a simple detail by
FEM, and that you at the same time will get some ideas about the stress state at
an elliptical hole, how symmetries may be utilized and how to work with a
mapped mesh.

Task

As mentioned above, the task is to study a uniaxially loaded thin plate with an
elliptical hole located symmetrically at its centre. More specifically, you are to
analyze the stress concentration that appears at the hole, which is to be done
both analytically (by a handbook formula, see below) and numerically (by
FEM).

It is up to you to choose the dimensions of the plate (width D and thickness t)

and the load P, see below. However, you are to make the plate long enough,
such that the effect of the hole is negligible at its ends, and that you do not get
plastic yielding at the hole. Concerning the geometry of the hole, we would like
you to study three different cases, namely

I) D/2a=2, a/b=2 II) D/2a=4, a/b=1/2 III) D/2a=4, a/b=1

P D 2a P

2b
t

You are to use all symmetries in the problem, and study the smallest possible
part of the plate with relevant boundary conditions. Concerning element type
and meshing, you can for some case(s) let ANSYS handle that, while you for at
least one case are to use a mapped mesh. For the mapped mesh case, we would
also like you to study how an increase in mesh density (more elements) will
affect your FE-result in comparison to the handbook solution. Finally, we would
like you to consider what the stress concentration analytically will be for a
circular hole which is small compared to the plate size, and what happens with
the stress state when the hole becomes more and more like a transverse crack.
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Analytical expression

The stress concentration at the hole is given by the following analytical

expression (from Roark’s formulas for stresses and strains, 6:th Edition,
page 735), valid for 1/2 ≤a/b≤ 10)

 max  k nom ,  nom  P /(t ( D  2a ))

k  K1  K 2 (2a / D)  K 3 (2a / D ) 2  K 4 ( 2a / D )3
K1  1  2a / b
K 2  0.351  0.021 a / b  2.483(a / b)
K 3  3.621  5.183 a / b  4.494(a / b)
K 4  2.27  5.204 a / b  4.011(a / b)

How to carry out the work

A suggestion for how to carry out the work can be found below (initially you
work in the toolbox DesignModeler, and then change to the toolbox
Mechanical)
 Sketch a rectangle with chosen dimensions (consider the symmetry when
choosing dimensions)
 Create a 2D-body (in the Concept-menu in DesignModeler)
 Create a new sketch, where you do an ellipse with chosen radii
 Use the tool Extrude to ”cut out” the hole in the plate
 Mesh, apply forces and restraints, solve the problem and study the result
(all in Mechanical)
What a ”free mesh” and a mapped mesh may look like is shown below, where
we in the latter case obviously can avoid a dense mesh in regions where it is not
necessary, i.e. in our application far from the hole! Of course you do not have to
do exactly as shown below, and are free to test your own ideas (the illustrations
also give some hints concerning how to handle the symmetry )
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

In order to achieve a mapped mesh, you do as described below

 Start from the plate geometry (with the hole)
 ”freeze” it by using the command Freeze (in the Tools-menu in
DesignModeler)
 Define lines in a new sketch, that in a wanted way split your plate
 Split the plate into two sub-bodies by using the option Slice Material in
the Extrude-tool (this requires that the body, to be split, is frozen)
 Redo the above procedure (create a new line and split the body with it) as
many times as needed
 The so obtained sub-bodies are then put together in a common part, by
first marking them and then using the command Form New Part (in the
Tools-menu)
 Now you switch toolbox to Mechanical, where you mesh your new part
(consisting of a number of sub-bodies). By this you will get a mesh which
is continuous across the sub-body boundaries.
 By choosing the command Mapped Face Meshing and apply it for the
complete plate (under Mesh in the ”activity tree”), you may then choose
to have 4-noded elements (quadrilaterals) everywhere.
 By the command Sizing (under Mesh) you may then change the element
size for the complete plate or for parts of it
 Furthermore, by the same command (Sizing), you can also give a “bias” to
the element distribution along some edge, i.e. to make the mesh density
increase in areas where a lot happen

Reporting

For a description of the reporting, see the end of this document.

 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Laboratory Problem IIa; Part 2

Limitations of different element types, or

Can I always trust my results?
Introduction
It is of course a fact that one must always check the relevancy of the obtained
FE-results. Errors can for instance be introduced via mistakes in the geometry
creation (dimensions etc) or via mistakes in the specification of material data.
Furthermore, an insufficiently locked structure (it can rotate and/or translate as a
rigid body) may automatically be given a locking somewhere by the program
(this can be checked by checking that the reaction forces are in equilibrium).
Finally, it is in fact the case, as you will see, that the FE-formulation itself can
work improperly for certain situations.

Here you are to study how different element types behave in a simple
application, more precisely a cantilever beam subjected to a load applied at its
free end. As you will see, some element formulations work not as well as others.
By this we can draw the conclusion that

It is important to know that each element type has its limitations,

i.e. there are situations for which it does not work well!

Goal
Even if we in the course do not have time to go any deeper into element type
limitations, you are to see that they do exist! Specific details can be found in the
FE literature.
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Task
As was mentioned previously, the task is here to study a cantilever beam
subjected to a transversal load at its free end.

You are to analyze the deflection of its free end, analytically as well as by
different (membrane) element types. More specifically, you are to analyze a
beam of length α , height 5 cm and width 2 cm (it will clearly be slender enough
for the beam theory to work well). Furthermore, you are to use 10 rectangular
elements in your analysis. For that/those cases where you obtain a poor result (a
too small deflection), you are to refine your mesh and study how the results
converge towards the correct solution.

The length α is
1.01m for group 1
1.02m 2
…
1.11m 11
etc

Element types to test

(Choose 3D-analysis in the Project Menu, but model your beam as a 2D-
body with the “surface from sketch”-tool!)

a) The 4-noded element that is standard/default in ANSYS in a 2-

dimensional analysis (a higher order element).
b) The 4-noded membrane element PLANE42. This element type must be
requested by the user by adding a small command file in the
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20
MECHANICAL-part of ANSYS, and more specifically under
“Geometry”, see below.

By the command file you overwrite the default choice of element type
in Workbench. The rows you are to add look like

ET,1,PLANE42,,1,3
R,1,0.02
MP,EX,1,2E11
MP,NUXY,1,0.3

What you do here, is to say that you would like to use the element type
PLANE42, of thickness 0.02, with the Young’s modulus 2E11, and
with the Poisson’s ratio 0.3. Further details can be found via the
ANSYS-help menu. It is to be noted that you here actually use the 4-
noded bilinear iso-parametric QUAD4-element that we have studied
on Lecture 7 and Teaching Class 6.

Reporting

For a description of the reporting, see the end of this document.

 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Laboratory Problem IIa; Part 3

A closer look at isoparametry and

numerical integration
Introduction
As the third part of this laboratory work you are to take a closer look at
numerical integration for element stiffness calculations. From the lectures and
teaching classes we know that we use Gauss-quadrature for this purpose, and
that a minimum integration order is needed in order to obtain the correct result.
The task is here to write a MATLAB-program that calculates the element
stiffness matrix for the 4-noded bi-linear iso-parametric QUAD4-element that
we studied on Lecture 7 and Teaching Class 6. The program is to be written in a
general fashion, in that it from provided input data (nodal coordinates, Young’s
modulus and Poisson’s ratio) calculates the element stiffness matrix, and is to be
checked against the specific element on Teaching Class 6.

Task

As was mentioned previously, the task is here to write a MATLAB-program for

calculation of the element stiffness matrix for a QUAD4-element, where 4-point
integration is to be used (c.f. Teaching Class 6).

Reporting

For a description of the reporting, see the end of this document.

 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Reporting
The reporting/assessment of this lab work is to be carried out by both peer
student evaluation and reporting to the teacher. More precisely

 The laboratory work is to be carried out in groups of 2 persons (possibly

one group with 3 persons)

 Each lab group is to carry out the tasks for Part 1, Part 2 and Part 3
presented above.

 Each lab group evaluates another lab group for the lab tasks listed below
(Group A examines Group B, Group B examines Group C, Group C
examines Group D, etc, i.e. two groups are not to evaluate each other). In
the evaluation you are to carefully check that

o for Part 1:
the same maximum normal stress is found analytically and by FEM
for the three different hole-geometries. Observe! Each group make
a separate model for each of the cases (in the same ANSYS project)
and save them as a basis for your evaluations. The mapped mesh
shall be of the type shown in the Figure above (with “nicely”
looking elements). The analytical expression is probably most
easily handled by MATLAB. An ANSYS-plot of the normal stress
(with elements shown) for a mapped mesh, where the maximum
stress can be seen, is to be sent in to the teacher- the associated
analytical stress concentration value is to be given on the form
below.

o for Part 2:
the same deflection of the free end of the beam is found both
analytically (by using the elemental case for a cantilever beam) and
by ANSYS (with standard higher order elements), and that an
approximate result is found by ANSYS using a sufficiently fine
mesh of QUAD4-elements. An ANSYS-plot of the latter case,
where the lateral deflection value can be seen, is to be sent in to the
teacher- the associated analytical deflection value is to be given on
the form below.
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

 The peer student examination is to be reported on the form/sheet which is

found on the next page, which is to be handed in BOTH electronically
(by e-mail to Kjell S) and in printed form (in the bookshelf in the
corridor of Solid Mechanics) no later than 18:00, Friday, October 18,
2013.

 This is to be handed in

o The form below

o Part 1: A normal stress plot from ANSYS (with elements shown)

for a mapped mesh, where the maximum stress can be seen (only
electronically)

o Part 2: A deflection plot from ANSYS where the value for the
lateral deflection of the beam is seen for the case of “sufficiently”
many QUAD4-elements (only electronically).

o Part 3: The MATLAB-program file, directly runnable as a .m-file,

which is to be prepared with the following input data: L=1m, t=1
cm, E=2E11Pa and υ=0.3 (only electronically). No peer student
evaluation is to be done for Part 3!

Note that the peer student evaluation aims at giving you an experience in critical
evaluation of another group’s simulation work, and is to be done carefully!
Furthermore, it is the responsibility of both groups that the reported results are
correct! A poor report with obviously erroneous results may have a negative
effect on the final grade of the course (for both groups)!

GOOD LUCK!
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Peer student evaluation

Lab IIa, TMMI37, Autumn 2013
(fill in the form electronically!)
We

Name personal number e-mail

1
2
3

have got our work evaluated and approved by the following persons

Name personal number e-mail

1
2
3

Regarding Part 1, for the case chosen, we have found the following analytical
stress concentration value (c.f. ANSYS plot)
……

Regarding Part 2, we have found the following analytical max deflection value
(c.f. ANSYS plot)
……

Which positive learning outcomes of the peer student evaluation have you
experienced?
….

Which negative learning outcomes of the peer student evaluation have you
experienced?
….
 The Finite Element Method, TMMI37, Autumn 2013
modified 2013-09-20

Please Observe! Include the prepared MATLAB-file (as a “.m” file, extended
for handling stresses) and plot according to previous page, as separate files
attached to your email (and not in zipped archives)! If the groups do not agree,
please contact the teacher!