Você está na página 1de 8

How Stress Intensity Factors are Calculated

Background

Last year (in the JS course on Materials) we covered the concept of brittle fracture and
defined the parameters Kc (the fracture toughness) and K (the stress intensity). If you
don’t remember the basic theory you may need to revise it, from last year’s notes or from
Ashby and Jones Vol.I.

In the online textbook for the present course there is more information about fracture
toughness and stress intensity factors, showing that the simple equation we used in the JS
course is not always accurate, and giving examples of stress intensity factors for various
different situations. Many more examples can be found in books and online data bases,
covering a wide range of problems commonly encountered.

However, engineers need to be able to estimate K values for cracks in real components
and structures, for which the geometry and loading may not be similar to any of the
textbook examples. So how can they do that? In turns out that there are several different
methods – here we will look at some of the most commonly-used ones.

The Problem

The general problem is to find the value of K for a crack of arbitrary shape and size, in a
body of arbitrary shape and size, loaded with an arbitrary set of forces and restraints. The
value of K obtained will be a single value, with units of MPa(m)1/2. To be more accurate,
there are actually three K values, called KI, KII and KIII. These refer to the three loading
modes that can cause a crack to propagation, which are: tension (applied perpendicular to
the crack), in-plane shear and out-of-plane shear (shear forces applied parallel to the
crack, in either an in-plane sliding direction or an out-of-plane tearing direction). You can
find out more about these different loading modes in textbooks. The methods used to
calculate them are basically the same; here we will consider just the Mode I, tensile
value, which in any case is the most important in determining brittle fracture and fatigue.

Obviously the value of K will be affected by the magnitude and direction of the applied
forces, by any restraints applied to the body, and by the geometry of both the body and
the crack.

K can be found in one of two ways, either by experimental measurements on an actual


body containing a crack, or by making a computer model using finite element analysis or
similar methods. Nowadays we mostly use FEA because it is easier, quicker and more
versatile, but many of the methods described below can also be applied to physical
models.
Energy-Based Methods

The most commonly-used methods for calculating K are based on measurements of strain
energy changes in the material. Remember how the basic equation for brittle fracture –
the Griffith equation – is derived? We consider a piece of material containing a crack,
with a load applied to it. We then imagine what happens if the crack gets a little bit
longer, i.e. if the crack length increases from a to a+δa. The energy which the crack
needs in order to do this is given by GcδaB where B is the thickness of the body and Gc is
a material parameter in units of J/m2. We can show that Gc is related to Kc and to the
Young’s modulus E via the following equation:
E - Young’s modulus
K c = Gc E Gc - energy required to propagate the crack .
for weaker materials e.g. glass = surface
(1)
tension

Assuming that the loading points don’t move (and therefore that no energy goes into or
out of the body), this energy need for crack growth is obtained from the elastic strain
energy which is stored in the body. When the crack grows it releases some of this energy:
in the derivation of the Griffith equation we equate this energy to the energy needed for
crack growth to find the conditions for brittle fracture to occur.

Now, we can also use this strain energy to estimate K. We do this by defining a
parameter G which is the amount of strain energy released per unit of crack growth, the
so-called “strain energy release rate”. So if the total strain energy in the body is W then:

1 dW
G= amount of energy per unit area of crack (2)
B da
If G = Gc then failure occurs. But in any case we can link G to K through an equation
similar to eqn.1 above:

K = GE (3)

This is very useful because it gives us a way to calculate K for any situation. You simply
have to make the crack a little longer and measure the change in elastic energy in the
whole body. This is easy to measure externally: consider a graph of the applied load P
against extension U, as measured at the points where the load is applied (see figure 1
below). This will normally be a straight line provided the applied forces are not too large,
so that the stresses in the body are below the yield strength. The slope of this line will
depend on how long the crack is: if the crack gets longer then the slope will decrease
because the body is becoming less stiff, more compliant. Figure 1 shows two loading
lines, for crack lengths a and a+δa respectively. Now, the area between those lines is
equal to the change in elastic energy due to crack growth, so this is dW in eqn.2. Using
eqns 2 and 3 we can find G and K for this particular set-up. Because you can measure
these loading lines for any type of body, you can do these calculations easily, either on an
actual physical body or using an FE model.
Crack Length = a
Load, P

Crack Length = a+ a
a + delta a
slightly longer crack

Change in Strain
Energy W grey area = energy released by the crack growing

Displacement, U

Figure 1: Loading lines for a body containing a cracks of length (a) and (a+da). The
shaded area is the change in elastic energy due to crack growth.

Because the lines are straight, the area you measure will be proportional to the applied
force and displacement, which makes it easy to estimate K without having to make
measurements for lots of different load values. But the lines are not always straight. Some
materials show curved lines even in their elastic regions (especially polymers) and even
in metals the lines will curve if you exceed the yield strength. The good news is that the
same approach can be used to measure G, by taking the area between the lines. This is
shown in figure 2. In this case we normally refer to it as J rather than G, just to remind
ourselves that this is obtained from curved lines. Now of course the area will not be
simply proportional to the load so we would have to measure J using the actual loading of
interest. But we can still use eqn.3 (with J replacing G) to find K.

Values of G (or J) can also be obtained in a slightly different way, using a mathematical
trick which allows us to do it without extending the crack. I won’t go into the details: you
can find more about this in any textbook on fracture mechanics if you are interested but
you don’t need to know it. Basically one performs a line integral, along a line which
starts on one crack face and finishes on the other.
a
Load, P
a+ a
a+ delta a

Displacement, U

Figure 2: As fig.1, but for a material which shows non-linear behaviour. The elastic
energy can be obtained in the same way, but the strain-energy release rate is now called
J instead of G.

The function which you integrate on this line is a combination of the strain energy
density and local forces (tractions) at points along the line. The result is a value of J (or G
if the problem is a linear one) which you can convert into K using eqn.3 above. For this
reason J is often referred to as “the J integral”. Most commercial finite element software
packages have an inbuilt routine for calculating J automatically, so this is often the
easiest way to do it in practice. This method is usually done using finite element analysis
or by analytical calculation; it’s not suitable for experimental measurement.

Stress-Based Method
theoretical linear elastic solution

Another way to find K is to use the fact that a crack concentrates stress in a particular
way. We know (see last year’s notes) that the stress just in front of the crack is given by:
@ crack tip, r = 0, K is infinite
K K = stress intensity
σ (r ) = (4)
2πr

…where σ(r) is the stress at a distance r from the crack tip. This is illustrated below in
fig.3. The nice thing about this equation is it that it is true for any crack length, and
indeed for any shape of crack in any body. That’s to say, a given value of K implies a
given stress field surrounding the crack, however that value of K comes about. So if we
can measure the stress at a given distance ahead of the crack tip, then we know K.
However it only applies to relatively small distances from the crack tip: r must be very
much less than a. Experimentally it is possible to measure these stresses but not very
practical; however it’s very easy to get this stress information from an FE analysis. The
only problem is that it may not be very accurate to use a single value of stress at a single
distance, because the value given by the FE model may not be very accurate, especially
for points close to the crack tip. It’s better to measure the stress values at various
distances. In fact I’ve done some work to on this approach myself; I developed some
software that allows one to get accurate values of K by post-processing FE results: I
called it the Crack Modelling Method and it’s now used quite a lot in industry.

Crack Opening Displacement Method


strain

Another approach, instead of measuring the stress ahead of the crack tip, is to measure
the displacement behind it. As fig.3 shows, when you apply load to the material
containing the crack, this causes the crack faces to move, opening up the crack. This is
called Crack Opening Displacement (COD). The value of the COD at a given distance
along the crack faces is a function of the K. So if you can measure COD you can estimate
the K value. As with the stress-based method this is something which you can do
experimentally, but it’s pretty fiddly and certainly easier to do on an FE model.

Stress
(r)

COD
Distance, r

Figure 3: K can be found by measuring the stress ahead of the crack tip, or the
displacement behind it.

Notches Also

The above methods for measuring K were designed to be used when you have a crack
present in your component. However, it turns out that they can also be useful sometimes
when you have a notch or other type of stress concentration feature such as a corner, a
groove or some kind of defect like casting porosity. The reason this is useful is because
some notches have a similar effect to cracks, in that they can cause failure by fatigue or
brittle fracture. This will come up in Case Study 3, where we will see that it is sometimes
useful to find the K value for a notch. All of the above methods can be used to do that as
well.
hairline fractures will have a theoretical values of r = 0

Você também pode gostar