MIE 124: Mini-Project 2

GENERAL INSTRUCTIONS

• You will submit one m-file for this homework. The file should be called
“firstname_lastname_mp2.m”.

• Please DO NOT submit .mlx files (LiveScripts) or text files (or any other files). These will
be returned with a zero grade.

• Do not forget to complete the Academic Honesty statement in a comment at the very top of
your m-files. Homework files missing this statement (a) will not be graded until the
completed statement is received, and (b) will receive significant point deductions.

• Following the Academic Honesty statement, insert comments with your name, student ID,
and a description of what the script does. Points will be deducted if this information is
missing.

• Use comments to distinguish between the various parts of the problem

Problem 1 (100 points)

In this homework, we will use root finding and numerical integration to solve engineering
problems in fracture mechanics.

Background: The field of fracture mechanics is concerned with the propagation of cracks in
materials. As you might expect, if a machine component develops a crack, the ability of this
component to sustain loads is compromised; if left undetected and/or unrepaired, such cracks can
propagate – grow in size – and cause catastrophic failure.
Under static loading conditions, whether a crack will propagate or not is determined by a
quantity called the stress intensity factor, denoted by K, which is a function of various
geometrical parameters and applied loads. As long as K is below some critical value, Kc (i.e.,
K<Kc) the component is safe from crack propagation. If K exceeds Kc, the crack will grow and
lead to an unsafe – possibly catastrophic – situation.
Under cyclic loading conditions – imagine, if you will, repeatedly applying and releasing
the load – things become worse and a crack can start propagating microscopically with each
cycle, even if K is far lower than Kc. This is called fatigue. You can see this for yourself quite
easily: take a paper clip and try opening it out and pulling it until it breaks – unless you have
super strength, this is unlikely to happen! Now, try repeatedly opening (until it bends) and
closing the paper clip a few times and you will cause it to develop a neck where it eventually
snaps. For machine components, it is possible to calculate how many such cycles the part can be
subjected to until a crack grows to the point that the component can no longer be used safely.
In this problem, we will explore static and cyclic failure using the numerical methods you
have learned thus far.

1
Let’s first consider the static loading case. Consider the situation illustrated in Figure 1: a metal
plate, subjected to tensile forces P, has a pre-existing center crack in it. The crack length is 2a;
we are interested in figuring out how large, a,1 could possibly become before the plate fails under
this (known) applied force P.

P
b

P
Figure 1: Symmetric center crack of length 2a in a plate of width 2b and thickness t (in the direction
normal to the plane of the sketch; not shown here) subjected to a tensile force P.

The stress intensity factor, K, for this particular case is given by the expression
P
K = F(α ) π a, (1)
2bt
where F(a) is, in turn, given by the expression
1− 0.5α + 0.326α 2 a
F(α ) = ,α = . (2)
1− α b

For our specimen, b=38 mm and t=6 mm. The maximum force this component encounters is P =
240 kN. For failure to occur, K must attain a critical value K= Kc; Table 1 lists typical values of
Kc for two steels.2

Table 1: Material constants for two steels

Material Kc C m
Man-Ten steel 200 MPa×m1/2 10-31 3.13
17-4 PH steel 120 MPa×m1/2 10-28 2.5

1
It is conventional to work with a – half of the crack size – rather than 2a, and we will follow this
convention here.
2
The units of K and Kc are decidedly odd but if you work consistently with SI units, everything should
work out fine. 1MPa=106 Pa.

2
Task 1: (Do this for each steel) Using Eqs. 1 & 2 and the information given to you above,
calculate the value of a required for failure to occur. Let us call this value acrit . What you have
found at this point is the maximum crack size, 2acrit, that the plate can tolerate under a load of
240 kN. Below this maximum crack size, the plate is safe from failure.

Now, let’s consider the dynamic loading case. Suppose the crack starts out at an initial size a0=1
mm. If you solved Task 1 correctly, you will see that a0<acrit , which means that the component
will not fail if we just load it statically to 240 kN. But now, the component is no longer loaded
statically: instead, we load it to 240 kN and unload it to 0 kN – this is one cycle – and we keep
doing this again and again.3 In this scenario, the crack grows gradually until a reaches acrit, at
which point the plate fails. The number of cycles Nfail required for the crack to reach acrit is given
by the expression
acrit
da
N fail = ∫ C[K(a)]
a0
m
. (3)

Representative values of C and m for the two steels of interest are given in Table 1. (The units of
C are ghastly and unimportant for now; just note that I have given you the value in SI units here;
m is a dimensionless number.) K was defined previously in Eq. 1; I have written it here as K(a)
so that you don’t forget the dependence on a!

Task 2: (Do this for each steel) Using Eq. 3 and the information given to you above, calculate N
(0 ≤ N ≤ N fail ) as a function of a for a0 ≤ a ≤ acrit . You might find it helpful here to think of Eq.
3 in the piecewise sense we discussed in Lecture 15 (Numerical Integration – Part 2), i.e., if the
crack has taken Nj cycles to get to length aj, then the cycles needed to get to length aj+1 is given
by
a j+1
da
N j+1 = N j + ∫ C [K(a)]
aj
m
. (4)

Use the trapezoidal rule to do these integrals with at least 100 uniform subintervals between a0
and acrit; you can use more subintervals if you like. You may not use Matlab’s trapz function!

Task 3: Once you have vectors of a and N, you can plot your results for all steels on a single
graph. Note that, while you can readily plot N vs. a, such a plot is less useful. For practical
applications, you would like to know how many cycles of life are left in the component when the
crack is at some size, as opposed to how many cycles it has taken for the crack to get to that size.
In other words, you want to plot Nfail - N(a) (Cycles remaining) vs. a. Label your axes, add a
title, add a legend indicating which line corresponds to which steel. Your graphs must be
legible: if one of the lines is basically flat along the x-axis, either use a semilog plot or use two
different y axes—do not make subplots or two separate figures, as we want to compare both
steels easily on a single plot.

3
If you’re wondering how such cycling might occur in real life, think about your car’s shock absorbers –
the springs repeatedly compress and expand on uneven roads.

3
In comments, answer the following questions for each steel (you can read data off your graph or
from the arrays):
(a) For a plate starting out with a (half-)crack size of a=1 mm, how many cycles of
loading/unloading can be applied until failure? (Round down to the nearest integer.)
(b) When you inspect the component after some time, you find that a=10 mm. How many
cycles remain in its usable life? (Round down to the nearest integer.)

For this assignment you will submit one single file that will contain a script portion and,
minimally, the two local functions described below.

The script portion will do the following (and nothing more!):

1. Set up the input variables.
2. Looping over the two steels,
a. Call a function to perform Task 1 and calculate acrit
b. Call a function to perform the numerical integration in Task 2
3. Produce plots for Task 3 (within or outside the loop is fine)
4. Answer questions in comments.
5. There should not be any other calculations whatsoever in the script portion – points will be
deducted for this. The functions described below will take care of all of the calculations.

Function for Task 1

1. Solve Eq. 1 to find acrit.
2. You may not use Matlab’s built-in solvers. Implement your own solver – you are allowed to
use any code from the m-files I have given you (cite this).
3. There should not be any other calculations whatsoever in this function – points will be
deducted for this.

Function for Task 2

1. This function will take all the necessary inputs to perform the numerical integration
(basically, a bunch of constants required for calculating the integrand, as well as a0 and acrit).
2. The outputs will consist of a vector a (which runs in uniform subintervals from a0 to acrit) and
a vector N (whose elements you calculate by numerical integration).
3. You may not use Matlab’s built-in integration routines (trapz, integral, ode23, ode45, etc.).
Implement your own integration routine – you are allowed to use any code from the m-files I
have given you (cite this).
4. There should not be any other calculations whatsoever in this function – points will be
deducted for this.

You can write as many sub-functions as you like beyond this (but don’t go overboard!)

4
Extra Credit [25 points]

Submit a separate m-file for this part – do not combine with previous script! There is a
separate submission portal for this part on Moodle.

• Name your file firstname_lastname_MP2_EC.m

• Comment your code liberally so that we can understand what you are doing. This is
especially important for extra credit problems, as you have a fair degree of latitude in
your approach.
• Address all questions for this part of the HW to the instructor and not the TAs
• Grading rubric:
o 25 points = Perfect – code works is well-structured, and has well-explained,
correct comments
o 15 points = OK – code works but disorganized/sloppy and or unclear explanation
of process
o 0 points = Wrong – Any error/bug whatsoever

Redo the assignment using the Matlab fzero function to solve for acrit and ode45 to integrate
Equation 3.