PDHonline Course G204 (6 PDH)

Design for Static Strength

Instructor: Robert B. Wilcox, P.E.

2020

PDH Online | PDH Center

5272 Meadow Estates Drive
Fairfax, VA 22030-6658
Phone: 703-988-0088
www.PDHonline.com

An Approved Continuing Education Provider

www.PDHcenter.com PDH Course G204 www.PDHonline.org

Design for Static Strength

.
Robert B. Wilcox, P.E

1.0 Prerequisites

It is assumed the student has a working knowledge of stress analysis and is able to
determine the state of stress at the critical location of the part. It is also assumed that
the materials being evaluated are linear elastic materials, and that the loading is
static.

2.0 Material Properties and Testing

When a body is subjected to a purely axial load, it will elongate. This elongation is
called normal strain, ε, and it is defined as the change in elongation, ∆L, divided by
the original length, L. Thus normal engineering strain is defined as:

ε = ∆L/L

The axial load, call it "P" will create a state of normal stress, σ, in the material,
defined as the load divided by the cross sectional area, "A" of the part. In the case of
a simple bar in tension then, the normal stress is:

σ = P/A

Axial stress and strain are linearly related to one another and as long as the material
is not stressed beyond a certain point, called the yield strength, it will behave in a
linearly elastic manner under loading. That means it stretches under the load, but if
the load is removed, it returns to its original length with no permanent, or plastic,
deformation.

The stress and strain are related to one another by Hooke's Law, given by:

σ = Eε

Where E is the Young's Modulus, a material constant specific to each material. In a

sense, the Young's Modulus defines how inherently stiff the material is. Most steels,
for example, have a Young's Modulus of around 3x107 psi, while most aluminum
alloys have a Young's Modulus of around 1x107 psi. This means that under a given
load, if you have two identical parts, one made from steel, and the other from
aluminum, the aluminum part will deform or stretch three times as much as the steel
part.

© Robert B. Wilcox Page 2 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

One of the most fundamental tests of the static strength of materials is the tensile
test. Many times, the results of the static test are all the designer will have to go on to
make decisions about the strength of the designed part. The object of the test is to
obtain the stress-strain diagram for the part. In this test, a specimen is subjected to
increasing axial load, and the deformation is measured precisely and plotted against
the applied load/area or applied stress.

When a ductile material undergoes a tensile test, the stress-strain diagram looks
something like this (although the shape varies for different materials):

The yield strength of the material is shown as sy and the ultimate strength is denoted
su. Units for the yield and ultimate strength of the material are generally given in psi,
ksi, or MPa. (1 MPa = .14503 ksi = 145.03 psi). Hooke's Law applies in region 1,
while the material is experiencing elastic deformation.

A ductile material will experience a

significant amount of plastic
deformation prior to failure. One of the
quantities measured in a tensile test is
the percent elongation which is the
difference between the original length
and the length at rupture, divided by the
original length, x 100. Materials with a
percent elongation of 5% or more are
generally considered ductile, and those
with 5% or less are considered brittle.

The second curve is for a tensile test of

a brittle material. Note the absence of

© Robert B. Wilcox Page 3 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

significant yielding (non-linear part of curve) prior to failure.

Another measure of ductility related to percent elongation is percent reduction in

area. This is a measure of how much necking occurs in a tensile specimen prior to
final failure. The higher the number, the more ductile the material is considered to be.

Strength of the material is thus usually specified as tensile yield strength and
ultimate tensile strength. Other strength quantities include compressive strength,
shear strength and torsional yield strength. For many materials compressive strength
is nearly equal to tensile strength, but for some, such as brittle materials like cast
iron, there may be a large difference. Brittle materials tend to be stronger in
compression than in tension, and consequently the compressive strength must
usually be considered separately for brittle materials if compressive loading is a
possibility.

Hardness of the material is another of the basic properties. Most hardness testing
involves the application of a concentrated compressive load on an indenting probe.
The size and/or depth of the indentation is measured, and converted to the
appropriate hardness scale. (Hardness scale conversions are easily found online). In
general for most metals, harder = stronger in terms of tensile strength. An increase in
hardness, however, is usually obtained at the expense of ductility. Thus, harder
materials may be stronger from a tensile strength perspective, but for many
applications, ductility is a highly desirable property. One of the reasons ductility is
desired is it represents the ability of a material to absorb energy in the form of plastic
strain.

Impact strength relates to how much energy a material can absorb as it fractures.
Most impact strength tests are done by swinging a weighted pendulum to strike a
notched bar of material, and measuring the height of swing of the pendulum post-
strike. The change in height of swing is kinetic energy absorbed by the specimen
during the process of fracture. Two popular tests are called the Charpy impact test
and the Izod impact test. Results are usually reported in ft-lb of energy absorbed.
The higher the ft-lb, the higher the impact strength. Impact strength is highly
dependent on temperature, and this must be part of the overall design considerations
when impact loading is possible.

3.0 Statistical Variations

Virtually all materials are subject to variations in chemistry, processing variations,

impurities, etc., all of which lead to variations in material properties like tensile
strength and yield strength. The designer must acknowledge and deal with these
variations. Other factors subject to statistical variations could include loading,
tolerances on machined or extruded parts, thicknesses of coatings, etc. It is up to the

© Robert B. Wilcox Page 4 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

designer to consider statistical variation as required, as part of his or her design

methodology.

The important concept to understand is that for a given value, say, tensile yield
strength, the values will vary, and they will probably (although not always) vary
according to a normal distribution. There are relatively straightforward ways to
determine if a set of data from a sample follows a normal distribution, and how well it
fits. (For more on this, consult a statistics text). Assuming your data fits the normal
distribution, what you are really interested in is how much variability there is, and
what conclusions can be made relevant to applicability of stated values (such as
tensile yield strength) to your design situation.

Statistical analysis usually involves taking a sample of values from a population, and
extending the results of the sample to the overall population.

First a few definitions:

N
1
population mean: μ=
N
∑x1
j

1 N
sample mean: x = ∑ xj
N 1

=∑
2
x 2
sample variance: s 2
x −x
N

( x) ⎤
1/ 2
⎡ ∑
2

⎢∑ x −
2
⎥
sx = ⎢ N ⎥
sample standard deviation: ⎢ N −1 ⎥
⎢ ⎥
⎣ ⎦
The mean, standard deviation and variance are all easily calculated with most
scientific calculators.

The Normal (or Gaussian) Distribution

The variations in many populations (sets of data) will be found to fit a particular type
of probability distribution called a normal distribution. The normal distribution is the
familiar bell-shaped curve which represents the frequency and variation of data
points from the mean.

© Robert B. Wilcox Page 5 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

The total area under the curve which describes the standard unit normal distribution,
from x = -∞ to x = +∞, equals one square unit. The area between any two points
represents the relative proportion of the distribution that lies between those two
points.

The value "z" is defined as the standardized variable, and is given by:

x−μ
z=
sx
Through the standardized variable, the distribution of any normally distributed
population data may be related to the unit normal distribution. The equation for the
unit normal distribution using the standardized variable z is:

− z2
1
f ( z) = e 2

2π
For any value of z, the area under the curve can be determined by numerical
methods, the Excel function "normdist()", scientific calculator, or by looking up the
value in standard statistical tables. The area found represents the proportion of the
data which will lie above or below the value specified.

Example: A machined lot of 500 brackets is sampled, and the mean of the load
capacity is found to be 5000 lb, with a standard deviation of 250 lb. The acceptable
minimum load capacity is 4500 lb. How many brackets will fail this criteria? (Assume
a normal distribution).

© Robert B. Wilcox Page 6 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

Solution: First, calculate the standardized variable, z:

x−μ
z= = ( 4500-5000 )/250 = -2
sx
Since the normal distribution is symmetrical, we can use the positive value for z.
From a standard table for z = 2.0, we find P(z) = 0.9773.

Since P(z)+Q(z) = 1, we know that Q(z), the area of the left side "tail" is:

© Robert B. Wilcox Page 7 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

Q(z)= 1-0.9773 = 0.0227

So this is the proportion of area under the curve to the left of -z.

Then, the total number of brackets with a load capacity of less than 4500 lb will be:

0.0227 x 500 = 11.375

Thus, about eleven or twelve brackets will be expected to fail the criteria.

4.0 Heat Treatment

Metals are often subjected to various heat treatment processes to obtain properties
which are desired for a particular design, such as strength, toughness, hardness,
ductility, fatigue resistance, etc. Heat treatment usually involves a number of specific
steps involving temperature and time, and which cause changes to the crystalline
structure of the metal being processed. Some of the more common processes are:

Annealing: Slow cooling of hot material. It may be done for a number of reasons
including softening, removal of residual stresses, purification, increasing ductility, or
to produce certain changes in crystalline structure.

Hardening: Heating and then quenching, usually of steels. Increased hardness is

generally accompanied by increased strength.

Normalizing: Heating to slightly above the critical temperature (steels) and slowly
cooling in air. Generally used to refine grain structure, remove residual stresses, and
improve toughness.

Quenching: Rapid cooling of hot material by immersion in cool liquid or gas.

associated with formation of martensite.

Spheroidizing: Application of certain sequences of heat and cool to produce spheroid

carbides in steels. Generally used to improve machinability of steels.

Tempering: Reheating to a temperature below the critical temperature, followed by a

specific cooling rate. Generally used to achieve a certain properties like hardness,
strength and ductility. In aluminum alloys, designated by the suffix "T".

5.0 Stress Concentrations

© Robert B. Wilcox Page 8 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

A stress concentration is any discontinuity in a part which elevates the stresses in the
region of the discontinuity. Examples include cross-holes, grooves, changes in
section, slots, threads, bosses, flanges, etc.

Here is an example from a finite element analysis (FEA) assembly contact model of a
small high pressure tool cylinder/housing with a cross-hole (oil port) and a threaded
rear end cap. The cross-hole is a stress concentration, as well as the thread in the
end of the cylinder.

For many common changes of section or features such as cross-holes, slots, etc.,
there are standard charts available for determining stress concentration. Prior to the
widespread availability of FEA, stress concentrations were determined theoretically
or experimentally by techniques such as strain gauging, brittle coating and
photoelastic techniques.

Stress concentrations can be though of as bottlenecks for field lines of stress.

Considering a plate with a hole in tension, imagine the stress field as straight lines
running through the part. The hole causes these lines to deflect, elevating the stress
field in the region of the hole. In fact, this analogy to field lines deflecting is not just a
good visualization tool. It can be used on real parts to reduce the stress

© Robert B. Wilcox Page 9 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

concentration by "guiding" the stress field lines around the discontinuity at a gentler
rate of change. A part can actually be made stronger sometimes by removing
material. Here is an example from an FEA model. The stress concentration, Kt, is
less where the added smaller holes have been placed to redirect the stresses around
the larger hole.

The stress concentration factor for normal stress, Kt, is defined as the ratio of the
actual maximum stress in the region of the discontinuity to the nominal stress through
net section or gross section of the part. Pay particular attention to whether net
section or gross section is referenced when using stress concentration charts. Net
section is the reduced section in the area of the discontinuity (in the case above, it
would be the width of the bar minus the hole diameter times the thickness). Gross
section would be the width of the bar without the hole times thickness.

σ max
Kt =
σ0
For shear stress concentration factor, use Kts instead of Kt.

τ max
Kts =
τ0
Application of stress concentration factors in design depends to a high degree on the
ductility of the material being evaluated. Since stress concentration is a localized
effect, in a ductile material subjected to static loading, the material will yield
plastically in the area of the stress concentration when it is first loaded. For this
reason, as long as the material is ductile and the loading is static, it is not necessary

© Robert B. Wilcox Page 10 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

to apply the stress concentration factor when computing stresses. For brittle
materials, however, the stress concentration factor should be used even in static
design scenarios because a brittle material will not yield, and thus there will be no
plastic relief of stresses around the stress concentration. So for brittle materials and a
single stress concentration, simply multiply the nominal stress by the appropriate
stress concentration factor - but first adjust the applied stress concentration factor to
account for the effects of notch sensitivity.

We can define the fatigue stress concentration factor (which is also used as an
effective stress concentration factor for ductile materials), Kf as

K f = 1 + q ( K t − 1)

Where q is the notch sensitivity factor, a material constant. For example, for most
cast iron materials the notch sensitivity is low, from 0 to 0.2. We can typically use 0.2
for cast irons in general and be designing conservatively. For brittle materials with
stress concentrations then, we can look up the notch sensitivity factor, q, and
calculate Kf, then use Kf to either reduce the allowable stress, or increase the applied
stress. From a design standpoint it makes no difference which approach is used, and
in fact both approaches are commonly used.

Other cases where Kt (or Kf) should be applied in static loading are for highly cold-
worked materials, very high strength materials, and case hardened materials.

It should be noted also that if a part will be subject to repeated (fatigue) loading, than
for both ductile and brittle materials a fatigue stress concentration factor Kf is used.
(This course will not address fatigue loading).

6.0 Temperature

Elevated or reduced temperatures will significantly affect the material properties of all
engineering materials. If elevated or reduced temperatures are part of the design
parameters, they must be considered and accounted for. The best method of design
verification is actual testing of actual parts in the environmental conditions they will
be subjected to.

Reduced temperatures, in addition to changing material properties in general, may

also cause a transition from ductile to brittle behavior in otherwise ductile materials.
The temperature at which this transition takes place is called the nil ductility transition
temperature (NDT or NDTT). Essentially, the impact strength and elongation of a
material undergoes rapid decreases in the region of the nil ductility temperature.
Design of critical parts and structures must take this into account and provide a
margin of temperature difference between the NDT and the minimum temperature

© Robert B. Wilcox Page 11 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

the structure or part will be subjected to in service. (If designing critical parts or
structures, be sure to comply with the appropriate codes and standards which govern
the design of such parts.)

Elevated temperatures in general reduce the strength of metals, and long term
exposure to elevated temperatures can cause permanent deformations, even when
stresses remain below the yield strength of the material. This phenomenon is known
as creep, and is a significant factor when designing with non-linear materials such as
plastics, or parts subjected to long-term significant elevated temperatures. In
essence, elevated temperatures can cause materials which behave in a linear elastic
way at room temperature, to behave like a non-linear plastic. It is highly
recommended that creep properties be well understood for the particular material
and design conditions and verified by long-term testing. The following figure shows
the effects of elevated temperature on tensile and yield strength of AISI low alloy
steels.

(from MIL-HDBK-5H, 1998)

© Robert B. Wilcox Page 12 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

7.0 Corrosion

Corrosion is the deterioration of the properties of a material as a result of interacting

with the environment to which it is exposed. Three basic type of corrosion which may
affect parts from a static strength point of view are simple environmental corrosion,
galvanic corrosion, and stress corrosion.

Environmental corrosion is caused by chemical interaction with liquids, solids, or

gases in the environment. Corrosion is a significant factor in the design of any
structures, machines or vehicles which will subjected to outdoor elements, elevated
temperatures, corrosive chemicals, marine environments, radiation, ozone, etc.
Environmental corrosion can be mitigated by selecting appropriate materials resistant
to corrosion, by surface coating, or surface maintenance. One of the most commonly
used tests to determine resistance to corrosion is a salt-spray test, of which there are
many variants.

Galvanic corrosion is caused by a flow of current between dissimilar metals in

electrical contact. It causes surface damage in the form of uniform corrosion and/or
chemical attack. Anywhere two dissimilar metals are in contact, galvanic corrosion
must be considered a possibility.

Galvanic corrosion can be mitigated by application of non-conducting coatings

between parts, or by sealing openings in cracks between parts (thus preventing
electrolytic compounds from entering). Other methods include increasing the
size/surface area of the anodic member (more actively corroding member) relative to
the cathodic (more inert) member, cleaning, or choosing metals which are close in
the galvanic series.

© Robert B. Wilcox Page 13 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

(from: www.corrosionsource.com)

Stress corrosion or stress corrosion cracking is the phenomenon of failure of

normally ductile materials under the combined influences of tensile stress and a
corrosive environment. Under the influence of tensile stresses, be they applied
stresses due to load or residual stresses due to processing, materials may corrode
and crack, that ordinarily would not be considered susceptible to corrosion. The
mechanism of stress corrosion cracking is similar the that causing galvanic corrosion.
Stress levels of 50-75% of yield strength are usually required for stress corrosion
cracking to take place, but in some cases lower stresses can still cause it.

Stress corrosion cracking can be mitigated by heat treatment to reduce or eliminate

the residual stresses from processing. Shot peening is another, preferred method for
removing surface residual tensile stresses and replacing them with a protective layer
of residual compressive stress. Design changes which prevent the development of
residual or assembly stresses (such as reducing the interference of a press-fit, or
reducing installation torque on a tapered plug) may be a simple fix as well, as long as

© Robert B. Wilcox Page 14 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

function is not adversely affected. Protective coatings to prevent corrosive agents

coming in contact with exposed stressed surfaces may also be used.

8.0 Safety Factors and Allowable Stress

The factor of safety in a static design, in terms of design stress or allowable stress,
sa, and material strength, S is:

n = S/sa

For example, if we have a part where tensile yield strength is the governing factor,
with a tensile yield strength Sy = 90,000 psi, and our allowable stress is 30,000 psi,
the n = 90,000/30,000 = 3.0.

Accounting for uncertainty in the design material strength and load variability may be
achieved by using a combined safety factor, nc.

nc = ns nl
Where ns is a factor which relates to uncertainty in the strength of the material or part
and nl is a factor which relates to uncertainty in the loading of the part. If, for example
you know the yield strength of the material is 100,000 psi with a standard deviation of
5000 psi, and you want 99% reliability relevant to material strength, then you could
use statistics to determine a factor ns which would account for material variation
(assuming a normal distribution applies). We can either calculate or use tables to find
the z value which corresponds to 99% of the area under the curve (thus leaving 1%
for the "tail"). From the chart on page 6, we can interpolate to find z = 2.326 for a P(z)
of 99%. We can then compute the reliability factor, Kr from the following formula:

sx
Kr = 1 − z
μ

5,000
Kr = 1 − 2.326 = 0.8837
100,000
And then:

ns = 1 / Kr = 1 / 0.8837 = 1.132
as the factor to account for the yield strength variation.

© Robert B. Wilcox Page 15 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

If you want to also account for a variation in loading of +30 percent, for example, you
could apply a load variability factor of 1.3, in combination with the material factor of
1.132 to get a total factor:

nc = ns nl = (1.132)(1.3) = 1.472
Thus, the allowable tensile stress using the combined safety factor would follow from:

nc = S / σ a

then σ a = S / nc = 100,000 / 1.472 = 67,930 psi

It then may be further desired to apply an overall safety factor once load variability
and material variability have been accounted for, depending on the nature of the
application.

It is up to the designer to consider every possible angle when applying safety factors.
Some of the issues to consider include, but are not limited to:

1) Consequences of failure with respect to safety and functionality

2) Governing codes

3) Sources of load variability (impact, human error, unexpected overload, machine

vibration, etc.)

4) Environmental influences (temperature, corrosion, weather related issues, etc.)

5) Part geometric variability, tolerances

6) Accidental damage or misuse/abuse

7) Degree of verification testing

8) Material variability

9) Functional consequences of over design (for example, most airplane structural

parts could not possibly carry a 4:1 static factor of safety - the plane would be too
heavy to fly - but in the hydraulic power tool industry, a 4:1 static factor of safety is
fairly common practice)

10) Ductile or brittle materials

11) Can failure be anticipated, observed, avoided, inspected, monitored

© Robert B. Wilcox Page 16 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

12) Cost of replacement

13) Collateral damage to dependent parts or assemblies

9.0 Combined Stress Failure Theories - General

If the state of stress at the critical location(s) involves a single stress in one direction,
(a uniaxial stress), then it is fairly simple to relate the tensile test properties to the
stress and determine if a part will fail or not. When, as is usually the case, stress is
multiaxial, (in the absence of or prior to actual
product testing), it becomes necessary to relate
the results of properties obtained by uniaxial
tests to the multiaxial stress condition and
make a decision about the integrity of the part.
There are a multitude of theories to this effect,
called combined stress theories. Not all will be
presented here - just the ones commonly used.
Over the course of time, tests have shown that
certain failure theories fit the experimental data
better than others depending on the properties
of the materials being analyzed. Selection of
which failure theory to use depends largely on
whether the material is ductile or brittle. In the case of ductile materials, it also
depends on whether or not there is a significant difference between the tensile yield
strength, syt, and the compressive yield strength, syc. In the case of brittle materials,
it similarly depends on whether or not there is a significant difference between the
ultimate tensile strength, sut and the ultimate compressive strength, suc. Of course, if
you are governed by a specific design code, you must follow the appropriate code,
even if it conflicts with the table presented here. In general, remember, ductile
materials have elongation > 5% in 2", and brittle materials have elongation <5% in 2".
Also note that in any of the following theories which use principal stresses, they are
arranged such that:

σ1 f σ 2 f σ 3

© Robert B. Wilcox Page 17 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

Selection of Appropriate Static Failure Theory - Ductile Isotropic Materials

Tensile Yield Strength (syt ) vs. Appropriate Failure Theory

Compressive Yield Strength (syc)

Relationship
syt approximately equal to syc Distortion Energy Theory (a.k.a. Von Mises - Hencky
Theory, - equivalent to Octahedral Shearing Stress
Theory)
Maximum Shear Stress Theory (a.k.a. Tresca-Guest
Theory)
syc significantly different than syt Mohr's Theory

Selection of Appropriate Static Failure Theory - Brittle Isotropic Materials

Ultimate Tensile Strength (sut ) vs. Appropriate Failure Theory

Ultimate Compressive Strength (suc)

Relationship
sut approximately equal to suc Maximum Normal Stress Theory

suc significantly different than sut Modified Mohr's Theory

10.0 Distortion Energy (Von Mises - Hencky, or Octahedral Shearing Stress)

Theory

This theory uses the principal stresses, s1, s2, s3; and the uniaxial failure strength
sf. The formula is based on the theory that failure will occur if the distortion energy
per unit volume which is the result of a multiaxial state of stress equals or exceeds
the distortion energy per unit volume at the instant of failure in a uniaxial stress test
(such as the tensile test). Thus failure is predicted to occur if the following expression
is true:

1
2
[ ]
(σ 1 − σ 2 )2 + (σ 2 − σ 3 )2 + (σ 3 − σ 1 )2 ≥ σ f 2

for the common situation of bi-axial stress the von Mises based failure criteria
reduces to (fails if true):

σ 12 − σ 1σ 3 + σ 32 ≥ σ f 2

And for the special case of bending and torsion, failure occurs of the following is true:

© Robert B. Wilcox Page 18 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

σ x 2 + 3τ xy 2 ≥ σ f 2

11.0 Maximum Shear Stress (Tresca-Guest) Theory

While the distortion energy theory is preferred, the maximum shear stress theory is
easy to apply, is conservative, and is used in many design codes. Maximum shear
stress theory is only used to predict onset of yielding. The maximum shear stress
theory states that failure will occur if any of the following expressions are true:

In principal shear terms: τ1 ≥ τ f or τ1 ≥ τ f or τ1 ≥ τ f

σf
Where: τ f = = the failure value for the principal shear stress corresponding to
2
the failure value for the principle normal stress from the uniaxial test.

And the equivalent expression in principal stress terms states that failure will occur if
any of the following are true:

σ 1 − σ 2 ≥ σ f or σ 2 − σ 3 ≥ σ f or σ 3 − σ 1 ≥ σ f

The maximum shear stress theory states that the yield strength in shear is one-half of
the tensile yield strength.

12.0 Mohr's Theory

For the unusual case of ductile materials with significantly different values of tensile
and compressive yield strengths (uneven ductile materials), the graphical method of
Mohr's Theory may be used. In this method, the results of uniaxial tensile test(s),
compressive test(s), and if available, torsional shear test(s) can be graphed,
constructing circles for each test on the t-s plane in a manner similar to that used
to construct the familiar Mohr's circle.

© Robert B. Wilcox Page 19 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

The best curve tangent to the Mohr's circles from the uniaxial tests is constructed.
Then, failure is predicted to occur in a multiaxial state when the largest conventional
Mohr's circle associated with the state of stress in the part becomes tangent to or
exceeds the boundaries defined by the tangent curve to the Mohr's circles created
from the uniaxial test specimen data.

13.0 Modified Mohr's Theory

The modified Mohr's theory fits

experimental results well and is
used for brittle materials where
there is a significant difference
between the ultimate
compressive strength and the
ultimate tensile strength (uneven
brittle materials). It is best
explained graphically.

The failure criteria vary by where

the stress point (determined from
the maximum and minimum
principal stresses) lies according
to the graph shown. If the state
of stress lies inside the green
zones it is in the "safe" zone and
the factor of safety n >1. If it is in

© Robert B. Wilcox Page 20 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

the red zones it is in the failure region and n < 1. This type of problem can be solved
graphically by constructing the graph and then determining if the particular state of
stress at the critical point(s) falls in the green range. The border between green and
red represents n=1. It may also be solved numerically according to the following
formula:

1/n = maximum of : ( σ1/St, - σ3/Sc , σ1/St - ( σ1+ σ3)/Sc)

14.0 Maximum Normal Stress Theory

This theory uses the principal stresses, s1, s2, s3; the tensile failure strength st, and
the compressive failure strength sc. Failure is predicted to occur if either of the
following expressions are true:

s1 or s2 or s3 P st s1 or s2 or s3 O sc

The maximum normal stress theory should only be used for even brittle materials. It
is overly conservative for ductile materials, and can be unsafe for uneven materials in
general.

© Robert B. Wilcox Page 21 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

15.0 Combined Failure Theory Examples

15.1 Example: Application of Combined Stress Failure Theories - Ductile

Materials

For a torsion bar, the loading is as shown. At what load "P" will the part yield at point
"A" if the bar is made from a steel with 18% elongation and tensile and compressive
yield strength = 81 ksi?

Since this is a ductile part, and no design codes are telling us to do otherwise, let's
use the von Mises (distortion energy) theory. Also, since the material is ductile and
this is a static loading case, we will not apply any stress concentration factors. This is
also a special case of bending and torsion, so we can bypass Mohr's circle and use
the formula:

σ x 2 + 3τ xy 2 ≥ σ f 2

So we need to find the shear stress:

Tr
τ xy =
J
Where T = 15P, r = 0.5 and J = the polar moment of the 1" bar, which is given by :

© Robert B. Wilcox Page 22 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

1 30P
J = πr 4 substitution gives τ xy =
2 πr 3
For the bending stress, sx we have:

Mc
σx =
I
where c = distance from the neutral axis, M is the bending moment (in this case M =
14P) and I = the moment of inertia of the bar, which is given by:

1
I = πr 4 .
4
Then, substitution gives:

56 P
σx =
πr 3
So we can now write the failure criteria in terms of the load P, as:

2 2
⎛ 56P ⎞ ⎛ 30P ⎞
⎜ 3 ⎟ + 3⎜ 3 ⎟ ≥ σ f
2

⎝ πr ⎠ ⎝ πr ⎠
Rearranging, we get:

π 2 r 6σ f 2
P≥
5836
Substituting in 0.5 for r, and 81 ksi for the failure strength, we get:

P = 416.4 lbf for the load which will yield the bar.

There are many online resources available for calculating Mohr's circle principle
stress and von Mises stress, like the one here:

http://www.aoe.vt.edu/~jing/MohrCircle.html

© Robert B. Wilcox Page 23 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

You provide the normal stresses and shear stress, and it provides the principle
stresses and von Mises stress. The Mohr's circle for the above problem was run and
the results look like this:

© Robert B. Wilcox Page 24 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

15.2 Example: Application of Combined Stress Failure Theories - Brittle

Materials

For a pin with a full radius groove, the loading is as shown. If P = 20,000 lbf and T =
2000 in-lb, what is the safety factor? Assume no material or load variability. The pin
is made from an ASTM 40 cast iron with tensile strength = 42 ksi and compressive
strength of 140 ksi. Notch sensitivity, q = 0.2.

Since this is an uneven brittle part, and no design codes are telling us to do
otherwise, let's use the modified Mohr's theory. Also, since the material is brittle we
will apply the stress concentration factors, adjusted for the notch sensitivity. For the
loading shown at point "A" we will have a shear stress due to torsion from T, and A
compressive normal stress due to the axial load P. Looking at the torsion, we know
for a round bar that

Tr 1
τ xy (max) = ; and we know J = πr 4
J 2
2T 2(2000)(in − lb)
so τ xy = = = 10,190( psi )
πr 3
π (0.5) (in )
3 3

and K ts = 1.27 so τ xy (corrected ) = (1.27)(10,180) = 12,940 psi

© Robert B. Wilcox Page 25 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

And our FEA model verifies this…

We do not, however want to use the full value of the Kt calculated either from the
FEA or from stress concentration charts due to the low notch sensitivity of the
material. We adjust the Kt using the formula for notch sensitivity:

K f = 1 + q ( K t − 1) = 1 + (0.2)(1.27 − 1) = 1.054

so our actual shear stress for design purposes is:

τ xy (corrected ) = (1.054)(10,190) = 10,740 psi .

Now to find the corrected axial compressive stress. First we can calculate a nominal
compressive stress in the net section simply by dividing force by area:

( −20,000)(lbf )
σx = = −25,460 psi
π 0.52 (in 2 )
And we compute Kf from the given Kt (for axial loading) and the notch sensitivity:

K f = 1 + q ( K t − 1) = 1 + (0.2)(1.79 − 1) = 1.158

Then, σ x (corrected ) = (1.158)(−25,460) = −29,480 psi .

© Robert B. Wilcox Page 26 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

We construct a Mohr's circle, either manually or with a program and get the
maximum and minimum principle stresses for use in the modified Mohr's theory.

This gives is s1 = 3500 psi and s3 = -32,980 psi. From the modified Mohr's theory we
have:

1/n = maximum of : ( σ1/St, - σ3/Sc , σ1/St - ( σ1+ σ3)/Sc)

σ1/St = 3500/42000 = .0833 ;

- σ3/Sc = -(-29,385)/140000 = 0.236

σ1/St - ( σ1+ σ3)/Sc = 0.0833 - (-.211) = 0.294

so we choose 1/n = 0.294, or n = 1/.294 = 3.40 = factor of safety against static

failure.

16.0 Application of FEA Stresses

A major concern is how to apply the results of FEA derived stresses in the field of
static failure analysis. You can't just blindly take the FEA results and use them,
especially for ductile materials. FEA (if done properly) will determine the actual
localized stresses in all of various discontinuities in a part. In the case of high stress
gradients which are highly localized, like at a cross-hole in a bar or plate, it would be
overly conservative to fully apply those stress values when designing for static
loading with ductile materials. For brittle materials, on the other hand - you want to
apply the full FEA results (corrected for notch sensitivity), and make sure that the
mesh used is good enough to capture the elevated stresses in the critical areas.

FEA can effectively be used to determine the stress concentration factor, then the
static failure analysis can be conducted by hand methods.

Take the torsion bar we looked at for the ductile static loading example, for instance.
A good FEA on this part run at the maximum (failure) load we determined, 416.4 lbf,
gave the following result in the area of the fillet:

© Robert B. Wilcox Page 27 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

Obviously, stress concentration effects are showing in the FEA, which is reporting
peak stress of 117 ksi, vs. the 81 ksi yield strength. What does a designer do about
this? It depends on whether the material is ductile or brittle.

For ductile materials, we want to determine the nominal stress, since we won't be
applying the stress levels that result from the stress concentration in the failure
theory analysis.

One way to use the FEA results is to conduct a hand calculation if it is possible for
the geometry, in conjunction with the FEA, and determine, at least approximately
what the nominal stresses are in the net section of the part just before the high stress
gradient effects of the stress concentration begin to appear. Or, if hand calculations
are impractical, use the appearance of a high stress gradient to indicate the
beginning of the stress concentration. A good indicator of high stress gradient in a
contour plot is a very narrow band of color (like the yellow band in the previous
figure). It may be helpful, if you know the approximate nominal stress to re-scale the
contour colors around the nominal stress value to increase the resolution of the color
bands. Once you know where the high stress gradient begins you can pick a node
just outside that range and use that value for the stress for application to the failure
theory, such as von Mises-Hencky.

Another option for ductile materials, (or to get a baseline for calculating Kt for brittle
materials) is to eliminate the stress riser from the geometric model prior to the FEA
meshing. This actually works well, and reduces analysis time for the FEA run.
Be sure when leaving out features that the overall stiffness of the part is not
significantly affected. The results now show 80.5 ksi - very close agreement to failure
stress determined by the failure theory calculations.

© Robert B. Wilcox Page 28 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

And one more option, which is effectively the same as the last option, is to
intentionally use a course mesh in the area of the stress riser. If the mesh resolution
is low enough to render the stress riser ineffective, then the elevated stresses
associated with the discontinuity will never appear in the solution.

For brittle materials, use FEA results from a fully converged, and detailed mesh
with a fine enough resolution to determine maximum stresses to determine Kt. Then
use hand calculations, or one of the previously mentioned techniques to find the
nominal stress just outside the range where the high stress gradient begins, then
determine Kt from the ratio of the maximum stress from the high resolution mesh,
and the nominal stress from the low resolution mesh or hand calculation. Then
correct Kt for notch sensitivity and proceed using an applicable failure theory (usually
modified Mohr's theory covers all brittle material cases, unless codes require
something else).

In applying any of the previous methods, experience level and an understanding of

the mechanics of materials is essential, as some judgment is involved. When
eliminating details or using a course mesh, be sure that the mesh is still fine enough
to fully capture nominal stress levels, and that the overall structure is not softened or
stiffened significantly due to the changes in the mesh from the actual part geometry.
This is particularly important if deflection is a major design consideration. In this case,
the fully resolved mesh model can be used for deflection output, and another model
with the stress riser removed by one of these methods can be used for the stress
results.

Again, the good engineer never went wrong by verifying the FEA with hand
calculations, and with plenty of tests on actual parts under actual or closely
simulated conditions of use.

© Robert B. Wilcox Page 29 of 30

www.PDHcenter.com PDH Course G204 www.PDHonline.org

17.0 Bibliography

Collins, Jack A. "Failure of Materials in Mechanical Design." Second edition. New

York: John Wiley & Sons, 1993.

Cook, R. "Finite Element Modeling for Stress Analysis." New York: John Wiley &
Sons, Inc., 1995.

Lipson, C. "Basic Course in Failure Analysis" Cleveland: Penton, 1980.

Roark, R. & Young, W. "Formulas for Stress and Strain." New York: McGraw-Hill,
1982.

Shigley, J. & Mitchell, L. "Mechanical Engineering Design." New York: McGraw-Hill,

1983

Teng, Y. "How Linear FEA Helps in Fatigue Analysis." Machine Design; Penton
Media - 2004.

© Robert B. Wilcox Page 30 of 30