SR61 78
SR61 78
SR61 78
Washington
σx τxy τxz P
ME 354 τxy σy τ yz
τxz τyz σz
ε
Mechanics of
Materials σ
Laboratory σ
τ P
Course website: http://swhite.me.washington.edu/~jenkinsm/me354/
Michael G. Jenkins
Associate Professor, Mechanical Engineering
Rev 2.0
01 January 2001
Mechanical Engineering 354
Mechanics of Materials Laboratory
TABLE OF CONTENTS
Topic Page
1. Introduction ....................................................................................................1.1
Laboratory Procedure
Laboratory Report
2. Stress/Strain/Constitutive Relations...........................................................2.1
Stress
Strain
Constitutive Relations
1. INTRODUCTION
Mechanics of Materials is generally the name applied to a discipline in which the stress,
strain and deflections of loaded structural elements are considered. This set of notes presents
the laboratory aspects of this subject.
For nearly all design work it is necessary to know something of the elastic and, often,
plastic properties of the material to be used. While these properties are often available from
handbooks, sometimes particular properties of less common materials are needed, in which
case the engineer must perform his own tests. The performance of these typical tests in this
laboratory will give a better feeling for the significance of the various material properties and for
the accuracy with which these quantities can be determined.
The various sections of these notes are concerned with review of the subject of
mechanics of materials and related material properties including the laboratory application of
these principles to a simple structures. In various exercises, the stresses, strains and
deflections of a both simply-supported straight and curved beams are measured and
compared to analytical predictions. In another exercise, selected mechanical properties and
performance of representative engineering materials are measured using standardized test
methods and quantitatively compared to handbook values. The effects of stress
concentrations are the focus of another exercise in which photoelasticity is used to determine
stress raisers for comparisons to values obtained from compendiums. Fracture mechanics and
crack interactions are examined in a study of the load carrying reduction of cracks in components.
Time-dependent behaviour is evaluated through measurement and analysis of creep
deformation and cyclic fatigue failures. Structural instabilities such as column buckling are
compared to material strength in assessing engineering failures. Complex structures are
analyzed through experimental measurements and both simple and complex analytical
methods to assess the implications of oversimplifications in engineering analysis.
Laboratory Procedure
Mechanics of Materials Laboratory, ME 354, is intended to give an experimental
understanding and verification of the coursework covered in Mechanics of Materials, CIVE 220
(formerly ENGR 220) and Introduction to Materials Science, MSE 170 (formerly ENGR 170).
No one should be enrolled in this course who has not taken or is not currently taking MSE 170
and CIVE 220 or their equivalents.
1.1
Because of the nature of the laboratory experiments, it is necessary to conduct them as
a class activity with students either observing or directly participating in the exercises. In some
exercises, small by groups of students will conduct the experiment directly. In other exercises,
the instructor will take the lead in operating the equipment with some students participating as
assistants and others as observers and recorders. An instructor will always be available in the
laboratory to introduce the exercise, describe the operation of the equipment, discuss the
expected results and present the salient aspects of the analysis. Generally, students will b e
expected to work as teams when required but must complete written reports independently.
Some laboratory exercises require formal written reports. Other exercises require the
completion of pre-formatted lab reports and their transmittal to the instructor in the form of a
short memo report Still other exercises require only the completion of pre-formatted write-ups
without any additional writing. All lab reports, regardless of type, must be turned in to receive a
passing course grade. Missing lab reports at the time of assignment of final grades will mean
the assignment of a final grade of X for one quarter following the course, regardless of the
quality of the rest of the coursework, until all reports are in. Failure to complete missing lab
reports or to make other arrangements after one quarter has passed following completion of the
course will result in the conversion of the X to a 0.0.
Laboratory reports
Reports are the primary basis for the course grade. Examination grades and discussion
participation are also considered in the final grade. Grades are important to the student for a
relatively short time; report writing will be important to the student's total career.
The laboratory reports provide an opportunity for the student to sharpen writing skills
and to increase the awareness of writing standards. Future employers will require standards
and will judge your professional or technical ability in part on your reporting capabilities.
Sherman (Sherman et al, 1975) has stated "It would be an overstatement, perhaps, to say that
a career in a technical profession will be impossible if you cannot write effectively. It is no
overstatement, however, to say that weakness in writing is a handicap that will weaken your
qualifications for many desirable positions, and that skill in writing is an asset that can make your
professional advancement faster and easier."
Technical writing involves style, neatness, grammar, usage of words, spelling, and
format. Of these attributes, first five are generally applicable, whereas the sixth (i.e., format) is
specific to the particular application. In the course, the format is non arbitrary and is detailed in
the appendix.
Neatness: All written communications should be neat in their final form. Reports should
be machine generated. Original data may be in pencil and is always included in the appendix
of the report.
1.2
Grammar: Sentence structure, paragraph construction, and punctuation presumably
have been learned prior to taking ME 354. Errors in grammar will be noted by the grader so
that the student's writing skills will be improved.
Usage of Words: Misuse of words involves words and phrases that are problems for
many writers. A few examples of such "pairs" are: affect-effect, among-between, because-
for, fewer-less, like-as if, percent-portion, while-although, too-two, their-there.
Jargon is acceptable when properly used (i.e., not overused!). Specialized words are
acceptable to a particular profession but should not be used to impress an "outsider."
Debasing of the English language by the use of suffixes such as "ise" and "wise" is confusing
and unnecessary. Colloquialisms or contractions should not be used in formal technical writing.
Style: The style of writing is determined by the potential reader. A report may b e
formal or informal, childish or mature, personal or impersonal, stilted or admirable, wordy or
succinct. The formal laboratory report may be read by a teaching assistant or a professor, but it
should be written for an engineering manager. Properly written laboratory reports may b e
used for reference material; well-written reports will enhance this value.
Do not copy portions of these notes word for word in your report. Statements in "your
own words" will indicate understanding and descriptive conciseness ability. The use of future
tense or telling "what you are going to do" does not belong in a report of what you did.
Generally the tense of reports is such that anything in the report (e.g., tables, figures, section)
are referred to in the present tense. Anything done to produce the results of the reports is in
the past tense.
Traditionally, technical report writing has been conducted in the third person passive
voice (e.g., "The tests were conducted"). The use of "I" imparts a personal tone to the report
which is generally inappropriate. First person style emphasizes the writer's part in the
experiment or test rather than the material or equipment used. "I" and "we" are sometimes
used to reduce awkward or stilted language (such as using "one"). Such use should be kept to
a minimum, particularly in the Summary.
Spelling: Spelling words properly is a problem for many students. Incorrectly spelled
words, particularly simple words, indicate a juvenile approach to technical writing. To quote
Sherman (Sherman et al, 1975), "Even a weak speller, if he keeps a list of the words that he
misses, is usually surprised at its shortness.... He can often eliminate most of his errors b y
learning to spell no more than 40 or 50 words." Keep in mind that electronic spell checkers do
not have any bearing on word choice. The wrong word correctly spelled is still the wrong word.
Words frequently misspelled in this course include: yield, specimen, temperature,
Riehle, recommend, omission — to name just a few. The use of a word guide is strongly urged
for those students with a spelling problem.
1.3
2. STRESS, STRAIN, AND CONSTITUTIVE RELATIONS
∑F x =0 ∑Fy =0 ∑F z =0
(2.1)
∑M x =0 ∑M y =0 ∑M z =0
Internal forces are non external forces acting in a body to resist external loadings.
The distribution of these internal forces acting over a sectioned area of the body (i.e., force
divided by area, that is, stress) is a major focus of mechanics of materials. The response
of the body to stress in the form of deformation or normalized deformation, that is, strain is
also a focus of mechanics of materials. Equations that relate stress and strain are known
as constitutive relations and are essential, for example, for describing stress for a
measured strain.
Stress
If an internal sectioned area is subdivided into smaller and smaller areas, ∆A , two
important assumptions must be made regarding the material: it is continuous and it is
cohesive. Thus, as the subdivided area is reduced to infinitesimal size, the distribution of
internal forces acting over the entire sectioned area will consist of an infinite number of
forces each acting on an element, ∆A , as a very small force ∆F . The ratio of incremental
∆F
force to incremental area on which the force acts such that: lim is the stress which
∆ A→0 ∆A
can be further defined as the intensity of the internal force on a specific plane (area)
passing through a point.
2.1
Stress has two components, one acting perpendicular to the plane of the area and
the other acting parallel to the area. Mathematically, the former component is expressed
as a normal stress which is the intensity of the internal force acting normal to an
incremental area such that:
∆F n
σ = lim (2.2)
∆ A→0 ∆A
The general state of stress is one which includes all the internal stresses acting on
an incremental element as shown in Figure 2.1. In particular, the most general state of
stress must include normal stresses in each of the three Cartesian axes, and six
corresponding shear stresses.
Note for the general state of stress that +σ acts normal to a positive face in the
positive coordinate direction and a +τ acts tangent to a positive face in a positive
coordinate direction. For example, σ xx (or just σ x ) acts normal to the positive x face in
the positive x direction and τ xy acts tangent to the positive x face in the positive y
direction.
Although in the general stress state, there are three normal stress component and
six shear stress components, by summing forces and summing moments it can be shown
that τ xy = τ yx ; τ xz = τ zx ; τyz = τ zy .
y σy
τ yx
τyz τ xy σx
τ zy
τ zx τxz x
σz
z
Figure 2.1 General and complete stress state shown on a three-dimensional incremental
element.
2.2
Therefore the complete state of stress contains six independent stress components (three
normal stresses, σ x ;σ y ;σ z and three shear stresses, τ xy ; τ yz ; τ xz ) which uniquely
describe the stress state for each particular orientation. This complete state of stress can
be written either in vector form
σ x
σ y
σz
τ (2.4)
xy
τ
xz
τ yz
or in matrix form
σ x τ xy τ xz
τ σy τyz (2.5)
xy
τ
xz τ yz σ z
Force F
The units of stress are in general: = . In SI units, stress is
Area L2
N 6 N N
Pa = 2 or MPa = 10 2 = and in US Customary units, stress is
m m mm 2
lb lb kip
psi = f2 or ksi = 10 3 f2 = 2 .
in in in
Often it is necessary to find the stresses in a particular direction rather than just
calculating them from the geometry of simple parts. For the one-dimensional case shown
in Fig. 2.2, the applied force, P, can be written in terms of its normal, PN, and tangential,
PT, components which are functions of the angle, θ, such that:
PN = P cos θ
(2.6)
PT = P sinθ
The area, Aθ, on which P N and PT act can also be written in terms of the area, A, normal to
the applied load, P, and the angle, θ, such that
Aθ = A /cos θ (2.7)
The normal and shear stress relation acting on any area oriented at angle, θ, relative to
the original applied force, P are:
P P cos θ P
σθ = N = = cos2 θ = σ cos2 θ
Aθ A / cosθ A
(2.8)
P P sinθ P
τθ = T = = cosθ sinθ = σ cosθ sinθ
Aθ A /cos θ A
where σ is the applied unidirectional normal stress.
2.3
PN
Aθ
θ
P
θ
A = Aθ cos θ P
T
Figure 2.2 Unidirectional stress with force and area as functions of angle, θ
For the two dimensional case (i.e., plane stress case such as the stress state at a
surface where no force is supported on the surface), stresses exist only in the plane of the
surface (e.g., σ x ;σ y ; τ xy ). The plane stress state at a point is uniquely represented by
three components acting on a element that has a specific orientation (e.g., x, y) at the
point. The stress transformation relation for any other orientation (e.g., x', y') is found by
applying equilibrium equations (∑ F = 0 and ∑ M = 0 ) keeping in mind that
F n = σA and F t = τA. The rotated axes and functions for incremental area are shown in
Fig. 2.3. The forces in the x and y directions due to F n = σA and F t = τA and acting on the
areas normal to the x and y directions are shown in Fig. 2.4
By applying simple statics such that in the x'-direction, ∑ F x ' = 0 and
x'
∆Ax=∆A cos θ
θ
y'
∆A
∆Ay=∆A sin θ X
2.4
y x'
τxy∆Ax
y' θ σx' ∆A
σx ∆Ax θ
θ
τx'y' ∆A
θ
τxy ∆Ay X
θ
σy ∆Ay θ
2.5
( )
dσ sin2θ 2τxy
= 0 = σ x −σ y sin2θ + 2τ xy cos2θ ⇒ = tan2θ = (2.12)
dθ cos2θ σx − σy
There are two solutions for the principal stress angle (i.e., for maximum and minimum) so
that.
1 2τ xy
θ N1 = tan-1
2 σx − σy
(2.12)
1 2τ π
θ N2 = tan-1 xy
+ π = θN1 +
2 σ x − σy 2
cos2θ =
(σ x − σy /2 )
σ x − σ y 2
+ τ 2xy
2
Substituting the trigonometric relations of Eq. 2.14 back into Eq. 2.9 gives for the plane
stress case:
σ x +σ y σ x − σ y 2
σθ = σ 1,2 = ± +τ 2xy
2 2
(2.15)
τ xy
tan2θ p =
(
σ x − σ y /2 )
√( σx − σy )2 + τxy
2
2 τxy
2θ
(σx − σy )/2
2.6
Note that for the plane stress case in the x-y plane, σ z = 0 . Thus the 1 and 2 subscripts in
Eq. 2.15 are only for the x-y plane and are not necessarily σ1 and σ 2 for the three-
dimensional general state of stress. Therefore, ordering of σ1 and σ 2 of Eq. 2.15 is only
preliminary, until they are compared to σ z and ordered according to convention
σ 1 > σ 2 > σ 3 . For example, for a particular plane stress state σ1 and σ 2 found from Eq.
2.15 are 100 and 20 MPa, then the principal stresses are σ1 = 100, σ 2 = 20, σ3 = 0 MPa.
However, if σ1 and σ 2 found from Eq. 2.15 are 125 and -5 MPa, then the principal
stresses are σ1 = 125, σ 2 = 0, σ 3 = −5 MPa. Finally, if σ1 and σ 2 found from Eq. 2.14 are
-25 and -85 MPa, then the principal stresses are σ1 = 0, σ 2 = -25, σ 3 = −85 MPa.
Performing a similar substitution of the trigonometric relations of Eq. 2.14 back into
Eq. 2.10 gives for the plane stress case:
σ x −σ y 2
τ max = +τ 2xy
2
(2.16)
σave =
σ x +σ y
and tan2θ s =
( )
− σx − σy
2 2τ xy
Note that the τ max of Eq. 2.15 is only for the x-y plane. The maximum shear stress for the
three-dimensional stress state can be found after the principal stresses are ordered
σ 1 > σ 2 > σ 3 such that:
σ −σ
τ 1,3 = 1 3 (2.17)
2
Some general observations can be made about principal stresses.
a) In a principal direction, when τ =0, then σ 's are maximum or minimum
b) σ max and σ min ( τmax and τ min ) occur in directions 90° apart.
c) τ max occurs in a direction midway between the directions of σ max and σ min
+σy ,−τ
φ =2θ
σ
τ +σ x,+τ
σ σ
C= x + y R2 = (σx - C)2 + τ2 )
2
tan φ = -σ−τ
( x - C)
2.7
An interesting graphical relation occurs if the second equation in each of Eqs. 2.9
and 2.10 are squared and added together:
2
σ x +σ y σ x − σ y
σ = σθ = + cos2θ + τ xy sin2θ
x' 2 2
+
σ x −σ y 2
τ x'y ' =τ θ =− sin2θ +τ xy cos 2θ
2
=
σ x +σ y 2 2 σ x −σ y 2 2
σθ − +τ θ = +τ xy
2 2
(2.18)
≡
(x−h) 2 +y 2 = r2
The result shown in Eq. 2.18 is the equation for a circle (i.e., Mohr's circle) with radius,
2
σ x −σ y σ x +σ y
+ τ xy and displaced h= on the x=σ θ axis as illustrated in Fig.
2
r=
2 2
2.6. Examples of Mohr's circles are shown in Fig. 2.7. A procedure for developing Mohr's
circle for plane stress is shown in the following section.
τ max = σ1 − σ3
τ max for x-y plane 2
σ3 σ2 σ1
σ2 σ1
σ
σ
τ
τ
Mohr's circle for stresses in x-y plane Mohr's circle for stresses in x-y-z planes
Figure 2.7 Examples of Mohr's circle for plane and three dimensional stress states.
2.8
Graphical Description of State of
Stress
X
Fig. M1- Positive stresses acting
on a physical element.
+σy
2.9
σ2
Y
σ1
Direction
of +θ The direction of physical angle, θ, is from
θ the x-y axes to the principal axes.
X
Fig. M4 - Orientation of physical element
with only principal stresses
acting on it.
Principal
Axis Note that the sense (direction) of the
physical angle, θ, is the same as on the
Mohr's circle from the line of the x-y
stresses to the axes of the principal
Direction of θ
stresses.
Line of X-Y
Stresses
Fig. M5 - Direction of q from the line of x-y
stresses to the principal stress
axis.
Same relations apply for Mohr's circle for strain except interchange variables as
γ
σ ⇔ ε and τ ⇔
2
2.10
Recall that all stress states are three-dimensional. Therefore, a more general
method is required to solve for the principal stresses. One such method is to solve for the
"eigenvalues" of the stress matrix where, σ is the principal stress:
σ x −σ τ xy τ xz
τ σ y −σ τ yz (2.19)
xy
τ
xz τ yz σ z −σ
Finding the determinant for this matrix and grouping terms gives:
σ 3 − I1σ 2 + I2σ − I3 = 0 (2.20)
where the stress invariants, (I1 ,I 2 , I3 ) , do not vary with stress direction:
I1 = σ x + σ y + σ z
I 2 = σ x σ y +σ y σz + σ zσ x − τ xy − τ 2xz − τ 2yz
2
(2.21)
Note that if principal stresses are used in Eq. 2.21 for the σ terms then all terms
containing τ will be zero since by definition, principal stresses act in principal directions
on principal planes on which τ =0.
Eq. 2.20 can then be solved for the three roots which when ordered σ 1 > σ 2 > σ 3
are the principal stresses. Eq. 2.20 can be plotted as f (σ ) vs σ shown in Fig. 2.9 where
the principal stresses are the values of σ which occur when f (σ ) = 0.
f( σ )
σ3 σ2 σ1
σ
2.11
Strain
Whenever a force is applied to a body, it will tend to change the body's shape and
size. There changes are referred to as deformation. Size changes are known as
dilatation (volumetric changes) and are due to normal stresses. Shape changes are
known as distortion and are due to shear stresses.
In order to describe the deformation in length of line segments and the changes in
angles between them, the concept of strain is used. Therefore, strain is defined as
normalized deformations within a body exclusive of rigid body displacements
There are two type of strain, one producing size changes by elongation or
contraction and the other producing shape changes by angular distortion.
Normal strain is the elongation or contraction of a line segment per unit length
resulting in a volume change such that
A' B'− AB L f − Lo
ε= lim ≡ (2.22)
B →A along n AB Lo
2.12
εy
εyx
εxy εx
A
γ = εxy + ε
xy yx
Engineering shear strain,
Figure 2.10 General and complete strain state shown on a three-dimensional incremental
element.
Length L
The units of strain In general: = , In SI units, strain is
Length L
m m
for ε and or radian for γ and in US Customary units, strain is
m m
in in
for ε and or radian for γ
in in
Just as in the stress case it is often necessary to find the stresses in a particular
direction rather than just calculating them from the geometry of simple parts. For the two-
dimensional, plane strain condition (e.g., strain at a surface where no deformation occurs
normal to the surface), strains exist only in the plane of the surface (e.g., ε x ; εy ; γ xy ). The
plane strain state at a point is uniquely represented by three components acting on a
element that has a specific orientation (e.g., x, y) at the point. The strain transformation
relation for any other orientation (e.g., x', y') is found by summing displacements in the
appropriate directions keeping in mind that δ = ε L o and ∆ = γ h as shown in Fig. 2.11
Simply adding components of displacements in the x' direction,
∑ displacements in x 'direction for Q to Q * (see Fig. 2.12) gives
ε x' = ε x cos2 θ + ε y sin2 θ + γ xy cosθ sinθ
or (2.25)
ε + ε y ε x − εy γ
ε x' = x + cos2θ + xy sin2θ
2 2 2
2.13
y x'
∆ = γ dy
}
Q*
} δy = ε y dy
Q
y'
dy ds
θ
δ x= ε x dx
}
dx x
Figure 2.11 Rotated coordinate axes and displacements for x and y directions
y
x'
δ = ε x' ds dx
x' cos θ =
θ
{ ∆ = γ dy
Q*
sin θ =
ds
dy
θ ds
δy = εy dy
Q θ
δx = εx dx x
Figure 2.12 Displacements in the x' direction for strains/displacements in the x and y
directions
2.14
Just as for the stress in a body which is a function of the angle of rotation relative to
a given direction, it is natural to look for the angle of rotation in which the normal strain is
either maximum or minimum. A principal normal strain is a maximum or minimum normal
strain acting in principal directions on principal planes on which no shear strains act.
Because there are three orthogonal directions in a three-dimensional strain state there
are always three principal normal strains which are ordered such that ε1 > ε2 > ε 3 .
Also as in the stress case, the principal strains for the plane strain case
2
ε + εy ε − εy 2 γ xy
ε1,2 = x ± x +
2 2 2 (2.29)
γ xy
tan2θ p =
εx − εy
Note that for the plane strain case in the x-y plane, ε z = 0 . Thus, the 1 and 2 subscripts in
Eq. 2.29 are only for the x-y plane and are not necessarily ε1 and ε2 for the three-
dimensional general state of strain. Therefore, ordering of ε1 and ε2 of Eq. 2.29 is only
preliminary, until they are compared to ε z and ordered according to convention
ε1 > ε2 > ε3 .
For the shear strain case
γ max ε x − εy 2 γ xy 2
= + τ xy ,
2 2 2
(2.30)
ε + εy
ε ave = x and tan2θs =
(
− εx − εy )
2 γ xy
γ max = ε 1− ε 3
γ max /2 for x-y plane
ε3 ε2 ε1
ε2 σ
ε1 σ
γ/2
γ/2
Mohr's circle for strains in x-y plane Mohr's circle for strains in x-y-z planes
Figure 2.13 Examples of Mohr's circle for strain.
2.15
y
60°
c b c b
60°
45° a
a
x x
45° Rectangular 60° Delta
Figure. 2.14 Rectangular and Delta rosettes.
Note that the γ max of Eq. 2.30 is only the x-y plane. The maximum shear strain for the
three-dimensional strain state can be found after the principal strains are ordered
ε1 > ε2 > ε 3 such that:
γ 1,3 = ε 1 − ε3 (2.31)
As with stress, the complete strain state can be represented graphically as a Mohr's
circle. Examples of Mohr's circles for strain are shown in Fig. 2.13. Note that the same
procedure for developing Mohr's circle for the plain strain case can be used as with the
γ
plane stress case with by making the following substitutions: σ ⇔ ε and τ ⇔ .
2
An important application of the strain transformation relation of Eq. 2.26 is to
experimentally determine the complete strain state since stress is an abstract engineering
quantity and strain is measurable/observable engineering quantity. Equation 2.26
requires that three strain gages be applied at arbitrary orientations at the same point on
the body to determine the principal strains and orientations. However, to simplify the data
reduction, the three strain gages are applied at fixed angles usually in the form of 45°
rectangular rosette or 60° Delta rosette as shown in Fig. 2.14. The resulting equations to
determine the local (for the strain gage rosettes shown in Fig. 2.14) x-y strains in
preparation for determining the principal strains:
45° Rectangular 60° Delta
Rosette Rosette
εx = εa ε x = εa
(2.32)
1
εy = εc ε y = (2ε b + 2εc − εa )
3
γ xy = 2εb − (εa + ε c ) γ xy = 23 (ε b − ε c )
2.16
Constitutive Relations
If the deformation and strain are the response of the body to an applied force or
stress, then there must be some type of relations which allow the strain to be predicted
from stress or vice versa. The uniaxial stress-strain case is a useful example to begin to
understand this relation.
During uniaxial loading (see Fig. 2.15) by load, P, of a rod with uniform cross
sectional area, A, and length, L, the applied stress, σ ,is calculated simply as
P
σ = (2.33)
A
The resulting normal strain, εL , in the longitudinal direction can be calculated from
the deformation response, ∆L , along the L direction:
∆L
εL = (2.34).
L
Another normal strain, εT , in the transverse direction can be calculated from the
deformation response, ∆D , along the D direction:
∆D
εT = (2.35).
D
A plot of σ vs. εL (see Fig. 2.16) shows a constant of proportionality between stress
and strain in the elastic region such that
y = mx + b ⇒ σ = EεL (2.36)
∆σ
where E = is known as the elastic modulus or Young's modulus and Eq. 2.36 is
∆ε L
known as unidirectional Hooke's law.
Undeformed A
Deformed
P
P
D=Do Df
∆ D=Df -D
L=Lo
Lf
∆ L=Lf - L
Figure 2.15 Uniaxially-loaded rod undergoing longitudinal and transverse deformation
2.17
Ε
−ν
2.18
εz Stress εx εy εz
εx
εy σx σx Eσ x - νεx = -νEσ x - νεx = -νEσ x
σy εy
εz
σy - νε y = -νEσ y Eσ y -νε y = -νEσ y
εx
2.19
ε x 1 −ν −ν 0 0 0 σ x
E E E
ε y −ν 1 −ν 0 0 0 σ y
E E E
ε z −ν −ν 1 0 0 0 σ z
γ = E E E 1 (2.41)
xy 0 0 0 G 0 0 τ xy
γ
xz
0 0 0 0 1 0 τ
xz
G
γ yz
0 0 0 0 0 G τ yz
1
Although Eq. 2.41 is a very convenient form, it is not that useful since stress is
rarely known with strain being the unknown. Instead, strain is usually measured
experimentally and the associated stress needs to be determined. Therefore, the inverse
of the compliance matrix needs to be found such that
{σ } = [C ]{ε} (2.42)
−1
where [C] is the stiffness matrix such that [C] = [S ] . Expanding Eq. 2.42 gives
σ x E ν νE νE εx
1+ 0 0 0
σ y (1+ν) (1−2ν)
(1+ν)(1−2ν) (1+ν)(1−2ν) ε y
νE E 1+ ν νE
(1+ν)(1−2ν) 0 0 0
σz (1+ν ) (1−2ν ) (1+ν)(1−2ν) ε z
τ = νE νE E 1+ ν 0 0
0 γ xy (2.43)
(1+ν) (1−2ν)
(1+ν)(1−2ν)
xy (1+ν)(1−2ν)
0
τ xz γ xz
0 0 0 G 0
0 0 0 0 G 0
τ yz γ
0 0 0 0 0 G yz
Generalized Hooke's can be simplified somewhat for the special case of plane
stress in the x-y plane since σ z =0. Being orthogonal to the x-y plane, Since σ z is a also
a principal stress by definition, all the shear stresses associated with the z-direction are
also zero. Thus, the stress-strain relations for plane stress in the x-y plane become
E νE
0
σ x (1− ν 2 ) (1− ν 2 ) ε x
νE
σ y = 0 ε y
E
(2.44)
2 2
τ xy (1−0ν ) (1−0ν ) G γ
xy
For the special case of plane stress, although σ z =0, the strain in the z-direction is
not zero but instead can be determined such that
−ν
Plane stress : σ z = 0, ε z ≠ 0 = (ε + ε y ) (2.45)
1− ν x
2.20
Similarly for the special case of plane strain, although, ε z =0, the stress in the z-
direction is not zero but instead can be determined such that
Plane strain : ε z = 0, σ z ≠ 0 = ν(σ x +σ y ) (2.46)
For the special case of hydrostatic pressure, no distortion takes place, only size
changes. In this case the hydrostatic stress is
σH =
(σ x +σ y + σ x ) (2.47)
3
The ratio between the hydrostatic stress and the dilation is a special combination of
elastic constants called the bulk modulus.
σ
k= H =
(
σ x + σy + σ x
=
E )
( )
(2.49)
εV 3 εx + εy + εx 3(1− 2ν )
It is useful to know that elastic constants are related to the atomic structure of the
material and thus are not affected by processing or component fabrication. For example,
E is related to the repulsion/attraction between two atoms. The force-displacement curve
for this interaction is shown in Fig. 2.18. Since this is the uniaxial case, recall that
P ∆L x ∆σ dσ
σ = and ε = = . Therefore, since E is defined as E = = with
A L xe ∆ε dε
dP dx dσ x e dP dP
dσ = and ε = then E = = . Note from Fig. 2.18 that is the slope of
A xe dε A dx dx
the force-displacement curve, thus making the elastic modulus E fixed by atomic
interaction. From a materials standpoint, covalent and ionic bonds such as those in
ceramics are stiff leading to high elastic moduli in those materials. Metallic bonds are
intermediate in stiffness leading to intermediate elastic moduli in metals. Secondary
bonds such as those found in polymers are least stiff leading to low elastic moduli in those
materials.
For Poisson's ratio, it is useful to consider the sphere model of atomic structure as
shown in Fig. 2.19. Before deformation the triangle between centers has a transverse
side length of 3R , a longitudinal side length of R and a hypotenuse of 2R. After
longitudinal deformation, the transverse side length is 3R -dx, the longitudinal side
2.21
Figure 2.18 Repulsion-attraction force-displacement curve between two atoms.
length is R+dy and the hypotenuse is still 2R. Applying Pythagorean's theorem after
deformation gives
(2R )2 = ( )
2
+ (R + dy )
2
3R − dx (2.50)
Recalling that the longitudinal and transverse strains can be written in terms of the
deformations
εy = (
R + dy ) − R dy
= ⇒ dy = Rε y
R R
( )
(2.52)
3R − dx − 3R −dx
εx = = ⇒ dx = − 3Rε x
3R 3R
Combining the two terms for dy and dx in Eq. 2.52 and equating them to Eq. 2.51 gives
dy Rε y ε 1
= = 3 ⇒− y ≡ − =3 (2.53)
dx − 3Rε x εx ν
which shows that from a simple sphere model and expected deformations, Poisson's ratio
is fundamentally linked to atomic interactions giving a value of 1/3 which is the range of
many dense materials.
2.22
Figure 2.19 Undeformed and deformed sphere model for atomic structure and
determination of Poisson's ratio
2.23
3. BEAMS: STRAIN, STRESS, DEFLECTIONS
3.1
Figure 3.2 Various types of beams and their deflected shapes: a) simple beam, b) beam
with overhang, c) continuous beam, d) a cantilever beam, e) a beam fixed (or restrained)
at the left end and simply supported near the other end (which has an overhang), f) beam
fixed (or restrained) at both ends.
3.2
Figure 3.3 Section of simply supported beam.
The magnitudes and senses of Vr and Mr may be obtained form the equations of
equilibrium ∑ F y = 0 and ∑ MO = 0 where O is any axis perpendicular to plane xy (the
reaction R must be evaluated first from the free body of the entire beam). For the present
the shearing stresses will be ignored while the normal stresses are studied. The
magnitude of the normal stresses can be computed if Mr is known and also if the law of
variation of normal stresses on the plane a-a is known. Figure 3.4 shows an initially
straight beam deformed into a bent beam.
A segment of the bent beam in Fig. 3.3 is shown in Fig. 3.5 with the distortion highly
exaggerated. The following assumptions are now made
i) Plane sections before bending, remain plane after bending as shown in
Fig. 3.4 (Note that for this to be strictly true, it is necessary that the beam be
bent only with couples (i.e., no shear on transverse planes), that the beam
must be proportioned such that it will not buckle and that the applied loads
are such that no twisting occurs.
Figure 3.4 Initially straight beam and the deformed bent beam
3.3
Figure 3.5 Distorted section of bent beam
ii) All longitudinal elements have the same length such the beam is initially
straight and has a constant cross section.
iii) A neutral surface is a curved surface formed by elements some distance,
c, from the outer fibre of the beam on which no change in length occurs. The
intersection of the neutral surface with the any cross section is the neutral
axis of the section.
Strain
Although strain is not usually required for engineering evaluations (for example,
failure theories), it is used in the development of bending relations. Referring to Fig. 3.5,
the following relation is observed:
δy δ
= c (3.1)
y c
where δ y is the deformation at distance y from the neutral axis and δc is the deformation
at the outer fibre which is distance c from the neutral axis. From Eq. 3.1, the relation for
the deformation at distance y from the neutral axis is shown to be proportional to the
deformation at the outer fibre:
δ
δy = c y (3.2)
c
Since all elements have the same initial length, ∆x , the strain at any element can
be determined by dividing the deformation by the length of the element such that:
δy y δc y
= ⇒ ε = εc (3.3)
∆x c ∆x c
3.4
Figure 3.6 Undeformed and deformed elements
Note that ε is the in the strain in the x direction at distance y from the neutral axis and that
ε =ε x . Note that Eq. 3.3 is valid for elastic and inelastic action so long as the beam does
not twist or buckle and the transverse shear stresses are relatively small.
An alternative method of developing Eq. 3.3 involves the definition of normal strain.
An incremental element of a beam is shown both undeformed and deformed in Fig. 3.6.
Note once again that any line segment ∆x located on the neutral surface does not
changes its length whereas any line segment ∆s located at the arbitrary distance y from
the neutral surface will elongate or contract and become ∆s' after deformation. Then by
definition, the normal strain along ∆s is determined as:
∆s' −∆s
ε = lim (3.4)
∆s→0 ∆s
Strain can be represented in terms of distance y from the neutral axis and radius of
curvature ρ of the longitudinal axis of the element. Before deformation ∆s = ∆x but after
deformation ∆x has radius of curvature ρ with center of curvature at point O'. Since ∆θ
defines the angle between the cross sectional sides of the incremental element,
∆s = ∆x = ρ ∆θ . Similarly, the deformed length of ∆s becomes ∆s'= ( ρ − y ) ∆θ .
Substituting these relations into Eq. 3.4 gives:
( ρ − y ) ∆θ − ρ ∆θ
ε = lim (3.5)
∆θ→0 ρ ∆θ
3.5
Eq. 3.5 can be arithmetically simplified as ε = −y / ρ . Since the maximum strain
occurs at the outer fibre which is distance c from the neutral surface, εmax = −c / ρ = εc ,
the ratio of strain at y to maximum strain is
ε −y / ρ
= (3.6)
ε max −c / ρ
which when simplified and rearranged gives the same result as Eq. 3.3:
ε = εmax = εc
y y
(3.7)
c c
Note that an important result of the strain equations for ε = −y / ρ and εmax = −c / ρ = εc
indicate that the longitudinal normal strain of any element within the beam depends on its
location y on the cross section and the radius of curvature of the beam's longitudinal axis
at that point. In addition, a contraction (- ε ) will occur in fibres located "above" the neutral
axis (+y) whereas elongation (+ ε ) will occur in fibres located "below" the neutral axis (-y).
Stress
The determination of stress distributions of beams in necessary for determining the
level of performance for the component. In particular, stress-based failure theories
require determination of the maximum combined stresses in which the complete stress
state must be either measured or calculated.
Normal Stress: Having derived the proportionality relation for strain, ε x , in the x-
direction, the variation of stress, σ x , in the x-direction can be found by substituting σ for
ε in Eqs. 3.3 or 3.7. In the elastic range and for most materials uniaxial tensile and
compressive stress-strain curves are identical. If there are differences in tension and
compression stress-strain response, then stress must be computed from the strain
distribution rather than by substitution of σ for ε in Eqs. 3.3 or 3.7.
Note that for a beam in pure bending since no load is applied in the z-direction, σ z
is zero throughout the beam. However, because of loads applied in the y-direction to
obtain the bending moment, σ y is not zero, but it is small enough compared to σ x to
neglect. In addition, σ x while varying linearly in the y direction is uniformly distributed in
the z-direction. Therefore, a beam under only a bending load will be in a uniaxial, albeit a
non uniform, stress state.
3.6
Figure 3.7 Stress (force) distribution in a bent beam
Note that for static equilibrium, the resisting moment, Mr, must equal the applied
moment, M, such that ∑ MO = 0 where (see Fig. 3.7):
and since y is measured from the neutral surface, it is first necessary to locate this surface
by means of the equilibrium equation ∑ F x = 0 which gives ∫ σdA = 0 . For the case of
A
elastic action the relation between σ x and y can be obtained from generalized Hooke's
law σ x =
E
[
(1+ ν )(1−2ν )
( )]
(1−ν )ε x +ν ε y +ε z and the observation that ε y = εz = −νεx .
The resulting stress-strain relation is for the uniaxial stress state such that σ x = Eεx
which when substituted into Eq. 3.3 or 3.7 gives
ε σ
σ x =E c y = c y (3.9)
c c
Substituting Eq. 3.9 into Eq. 3.8 gives:
σc 2 σ 2
Mr = ∫ σdAy = ∫ y dA = x ∫ y dA (3.10)
A c A y A
Note that the integral is the second moment of the cross sectional area, also known as the
moment of inertia, I, such that
I = ∫ y 2dA (3.11)
A
3.7
Figure 3.8 Action of shear stresses in unbonded and bonded boards
Substituting Eq. 3.11 into Eq. 3.10 and rearranging results in the elastic flexure
stress equation:
My
σx = (3.12)
I
where σ x is the normal bending stress at a distance y from the neutral surface and acting
on a transverse plane and M is the resisting moment of the section. At any section of the
beam, the fibre stress will be maximum at the surface farthest from the neutral axis such
that.
Mc M
σ max = = (3.13)
I Z
where Z=I/c is called the section modulus of the beam. Although the section modulus
can be readily calculated for a given section, values of the modulus are often included in
tables to simplify calculations.
Shear Stress: Although normal bending stresses appear to be of greatest concern
for beams in bending, shear stresses do exist in beams when loads (i.e., transverse
loads) other than pure bending moments are applied. These shear stresses are of
particular concern when the longitudinal shear strength of materials is low compared to
the longitudinal tensile or compressive strength (an example of this is in wooden beams
with the grain running along the length of the beam). The effect of shear stresses can be
visualized if one considers a beam being made up of flat boards stacked on top of one
another without being fastened together and then loaded in a direction normal to the
surface of the boards. The resulting deformation will appear somewhat like a deck of
cards when it is bent (see Fig. 3.8a). The lack of such relative sliding and deformation in
an actual solid beam suggests the presence of resisting shear stresses on longitudinal
planes as if the boards in the example were bonded together as in Fib. 3.8b. The resulting
deformation will distort the beam such that some of the assumptions made to develop the
bending strain and stress relations (for example, plane sections remaining plane) are not
valid as shown in Fig. 3.9.
3.8
Figure 3.9 Distortion in a bend beam due to shear
The development of a general shear stress relation for beams is again based on
static equilibrium such that ∑ F = 0 . Referring to the free body diagram shown in Fig.
3.10, the differential force, dF 1 is the normal force acting on a differential area dA and is
equal to σ dA . The resultant of these differential forces is F 1 (not shown). Thus,
F1 = ∫ σ dA integrated over the shaded area of the cross section, where σ is the fibre
My
stress at a distance y from the neutral surface and is given by the expression σ = .
I
Figure 3.10 Free body diagram for development of shear stress relation
3.9
When the two expressions are combined, the force, F1 , becomes:
M M c
F1 = ∫ y dA= ∫ ty dy (3.14)
I I h
∆M 1 c dM 1 c
τ = lim ∫ ty dy = ∫ ty dy (3.17)
∆x →0 ∆x It h dx It h
Recall that V=dM/dx, which is the shear at the beam section where the stress is being
c
evaluated. Note that the integral, Q= ∫ ty dy is the first moment of that portion of the cross
h
sectional area between the transverse line where the stress is being evaluated and the
extreme fiber of the beam. When Q and V are substituted into Eq. 3.17, the formula for the
horizontal / longitudinal shear stress is:
VQ
τ = (3.18)
It
Note that the flexure formula used in this derivation is subject to the same
assumptions and limitations used to develop the flexure strain and stress relations. Also,
although the stress given in Eq. 3.18 is associated with a particular point in a beam, it is
averaged across the thickness, t, and hence it is accurate only if t is not too great. For
uniform cross sections, such as a rectangle, the shear stress of Eq. 3.18 takes on a
parabolic distribution, with τ =0 at the outer fibre (where y=c and σ =σ max ) and
τ =τ max at the neutral surface (where y=0 and σ =0) as shown in Fig. 3.11.
3.10
Y τ=0 σ = σ min
τ=0 σ = σ max
Figure 3.11 Shear and normal stress distributions in a uniform cross section beam
Finally, the maximum shear stress for certain uniform cross section geometries can
be calculated and tabulated as shown in Fig. 3.12. Note that a first order approximation
for maximum shear stress might be made by dividing the shear force by the cross
V
sectional area of the beam to give an average shear stress such that τ av ≈ . However,
A
if the maximum shear stress is interpreted as the critical shear stress, than an error of 50%
3V
would result for a beam with a rectangular cross section where τ max ≈ which is 1.5
2A
V
times τ av ≈ .
A
h τmax = 3V
2A
A=bh
τmax = 4V
3A
d 2
A=( π/4) d
do
di τmax = 2V
A
2 2
A=( π /4) (do - di )
Figure 3.12 Maximum shear stresses for some common uniform cross sections
3.11
Deflections
Often limits must be placed on the amount of deflection a beam or shaft may
undergo when it is subjected to a load. For example beams in many machines must
deflect just the right amount for gears or other parts to make proper contact. Deflections of
beams depend on the stiffness of the material and the dimensions of the beams as well as
the more obvious applied loads and supports. In order of decreasing usage four common
methods of calculating beam deflections are: 1) double integration method, 2)
superposition method, 3) energy (e.g., unit load) method, and 4) area-moment method.
The double integration method will be discussed in some detail here.
Deflections Due to Moments: When a straight beam is loaded and the action is
elastic, the longitudinal centroidal axis of the beam becomes a curve defined as "elastic
curve." In regions of constant bending moment, the elastic curve is an arc of a circle of
radius, ρ , as shown in Fig. 3.13 in which the portion AB of a beam is bent only with
bending moments. Therefore, the plane sections A and B remain plane and the
deformation (elongation and compression) of the fibres is proportional to the distance
from the neutral surface, which is unchanged in length. From Fig. 3.13:
L L +δ
θ= = (3.19)
ρ ρ +c
from which
c δ σ Mc
= =ε = = (3.20)
ρ L E EI
and finally
1 M
= (3.21)
ρ EI
which relates the radius of curvature of the neutral surface of the beam to the bending
moment, M, the stiffness of the material, E, and the moment of inertia of the cross section,
I.
Figure 3.13 Bent element from which relation for elastic curve is obtained
3.12
Equation 3.21 is useful only when the bending moment is constant for the interval
of the beam involved. For most beams the bending moment is a function of the position
along the beam and a more general expression is required.
The curvature equation from calculus is
1 d 2y / dx 2
= (3.22)
ρ
[ ]
3/2
1+ (dy / dx )2
which for actual beams can be simplified because the slope dy/dx is small and its square
is even smaller and can be neglected as a higher order term. Thus, with these
simplifications, Eq. 3.22 becomes
2
1 d y
= = y'' (3.23)
ρ dx 2
d2 y
EI 2 = M x = EIy'' (3.24)
dx
which is the differential equation for the elastic curve of a beam.
An alternative method for obtaining Eq. 3.24 is to use the geometry of the bent
beam as shown in Fig. 3.14 where it is evident that dy/dx = tan θ ≈ θ for small angles and
that d 2 y /dx 2 = dθ /dx . From Fig. 3.14 it can be shown that
dL dx
dθ = = (3.25)
ρ ρ
for small angles and therefore.
d2y dθ 1 M x d2 y
y'' = = = = ⇒ EI 2 = EIy'' = M x (3.26)
dx 2 dx ρ EI dx
For the coordinate system shown in Fig. 3.15, the signs of the moment and second
derivative are as shown. It is also important to note the following physical quantities and
beam action.
Figure 3.14 Bent beam from which relation for elastic curve is obtained.
3.13
Figure 3.15 Sign conventions used for deflection
deflection = y
dy
slope = = y'
dx
d2y
moment = Mx =EI = EIy''
dx 2
dM d3y (3.27)
shear = =EI = EIy' ' ' (for constant EI)
dx dx3
dV d4 y
load = =EI 4
= EIyiv (for constant EI)
dx dx
It is interesting to note that from Eqs. 3.24 and 3.26 can be written as
dθ
M x = EI (3.28)
dx
from which
θB xB M
x dx
∫ dθ = ∫ (3.29)
θA x A EI
Eqs. 3.28 and 3.29 show that except for the factor EI, the area under the moment
diagram between any two points along the beam gives the change in slope between the
same two points. Likewise, the area under the slope diagram between two points along a
beam gives the change in deflection between these points. These relations have been
used to construct the series of diagrams shown in Fig. 3.16 for a simply supported beam
with a concentrated load at the center of the span. The geometry of the beam was used to
locate the points of zero slope and deflection, required as the starting points for the
construction.
3.14
Figure 3.16 Illustration of various elastic relations for a beam in three-point loading
The double integration method can be used to solve Eq. 3.24 for the deflection y as
a function of distance along the beam, x. The constants of integration are evaluated by
applying the applicable boundary conditions.
Boundary conditions are defined by a known set of values of x and y or x and dy/dx
at a specific point in the beam. One boundary condition can be used to determine one
and only one constant of integration. A roller or pin at any point in a beam (see Figs.
3.17a and 3.17b) represents a simple support which cannot deflect (y=0) but can rotate
(dy/dx≠0). At a fixed end (see Figs. 3.17c and 3.17d) the beam can neither deflect or
rotate (y=0 and dy/dx=0).
Matching conditions are defined as the equality of slope or deflection, as
determined at the junction of two intervals from the elastic curve equations for both
intervals.
3.15
Figure 3.17 Types of boundary conditions
1) Select the interval or intervals of the beam to be used; next, place a set of
coordinate axes on the beam with the origin at one end of an interval and then
indicate the range of values of x in each interval. For example, two adjacent
intervals might be: 0≤x≤L and L≤x≤3L
2) List the available boundary conditions and matching conditions (where two or
more adjacent intervals are used) for each interval selected. Remember that two
conditions are required to evaluate the two constants of integration for each interval
used.
3) Express the bending moment as a function of x for each interval selected, and
equate it to EI dy2/dx2 =EIy''.
4) Solve the differential equation or equations form item 3 and evaluate all
constants of integration. Check the resulting equations for dimensional
homogeneity. Calculate the deflection a specific points where required.
3.16
Since the vertical shearing stress varies from top to bottom of a beam the deflection
due to shear is not uniform. This non uniform distribution is reflected as slight warping of
a beam. Equation 3.31 gives values too high because the maximum shear stress (at the
neutral surface) is used and also because the rotation of the differential shear element is
ignored. Thus, an empirical relation is often used in which a shape factor, k, is employed
to account for this change of shear stress across the cross section such that
−V
kAGy' = −V ⇒ y'= (3.32)
kAG
A single integration method can be used to solve Eq. 3.32 for the deflection due to shear.
The constants of integration are then determined by employing the appropriate boundary
and matching conditions. The resulting equation provides a relation for the deflection due
to shear as a function of the distance x along the length of the beam. Note however that
unless the beam is very short or heavily loaded the deflection due to shear is generally
only about 1% of the total beam deflection.
3.17
An example of the use of integration methods is as follows for a simply supported
beam in three-point loading. The loading condition, free body, shear and moment
diagrams are shown in Fig. 3.19.
a P b
Loading Diagram
a+b=L
P
R1 =Pb/L R2 =Pa/L
Pb/L
Shear Diagram
Pa/L
Pba/L
Pbx/L (Pbx/L)-P(x-a)
Moment Diagram
0≤x≤a x=a a0 ≤ x ≤ L
Figure 3.19 Loading condition, free body, shear and moment diagrams
There are two matching conditions: at x=a, y'1 =y' 2 and at x=a, y1 =y2
3.18
For 0 ≤ x≤ a (region 1) For a≤ x≤ L (region 2)
V= Pb/L V= Pa/L
M=Pbx/L M=(Pbx/L)-P(x-a)
Pbx Pbx
EIy1'' = −M = − EIy 2 '' = −M = − + P(x − a)
L L
-Pbx -Pbx2
∫ EIy1'' = ∫ L ⇒ EIy ' = + C1 −Pbx
∫ EIy2 '' = ∫
1
2L + P(x − a) ⇒
L
−Pbx 2 −Pbx 2 P(x − a)2
∫ 1 ∫ 2L + C1 ⇒
EIy ' = EIy2 ' =
2L
+
2
+ C2
−Pbx3
EIy1 = + C1x + C3
6L −Pbx 2 P(x -a) 2
∫ EIy2 ' = ∫ 2L
+
2
+C 2 ⇒
so that C1 = C2
Since C1 = C2 ,then C3 = C4
3.19
Applying the boundary conditons at x=0, y1 =0
1 −Pb 0 3
y1 = 0 = + C 0 + C
EI 6L
1 3
So C 3 = 0
and at x=L, y2 =0
1 −PbL3 P(L-a) 3
y2 = 0 = + + + =
C L C
EI 6L 2 4
6
y1 =
6EIL [
−Pbx 2
L − b 2 − x2 ] y2 =
6EIL [
−Pbx 2
L − b2 − x2 −]P(x -a )3
6EI
V=Pb/L V=Pa/L
−V Pb 1 −V Pa 1
kAG y' 1 = −V ⇒ y' 1 = =− kAG y' 1 = −V ⇒ y' 1 = =−
kAG L kAG kAG L kAG
3.20
Applying the boundary condition at x=0, y1 =0,
−Pb
y1 = 0 = 0 + C1
kLAG
−Pb
So C1 = 0 and y1 = x for 0 ≤ x ≤ a
kLAG
Pa Pa x
So C2 = and y 2 = − +1
kAG kAG L
Now the total deflection relation for both bending and shear is:
V=Pa/L
−Pbx 2 P(x - a )3
y1 = [
−Pbx 2
]
L − b 2 − x2 −
Pb
x
y2 =
6EIL
[L − b2 − x 2 − ] 6EI
6EIL kLAG
Pa −x
− +1
kLAG L
3.21
4. BEAMS: CURVED, COMPOSITE, UNSYMMETRICAL
Discussions of beams in bending are usually limited to beams with at least one
longitudinal plane of symmetry with the load applied in the plane of symmetry or to
symmetrical beams composed of longitudinal elements of similar material or to initially straight
beams with constant cross section and longitudinal elements of the same length. If any of these
assumptions are violated, the simple equations which describe the beam bending stress and
strain are no longer applicable. The following sections discuss curved beams, composite
beams and unsymmetrical beams.
Curved Beams
One of the assumptions of the development of the beam bending relations is that all
longitudinal elements of the bean have the same length, thus restricting the theory to initially
straight beams of constant cross section. Although considerable deviations from this restriction
can be tolerated in real problems, when the initial curvature of the beams becomes significant,
the linear variations of strain over the cross section is no longer valid, even though the
assumption of plane cross sections remaining plane is valid. A theory for a beam subjected to
pure bending having a constant cross section and a constant or slowly varying initial radius of
curvature in the plane of bending is developed as follows. Typical examples of curved
beams include hooks and chain links. In these cases the members are not slender but rather
have a sharp curve and their cross sectional dimensions are large compared with their radius of
curvature.
4.1
Fig 4.1 is the cross section of part of an initially curved beam. The x-y plane is the plane of
bending and a plane of symmetry. Assumptions for the analysis are: cross sectional area is
constant; an axis of symmetry is perpendicular to the applied moment; M, the material is
homogeneous, isotropic and linear elastic; plane sections remain plane, and any distortions of
the cross section within its own plane are neglected. Since a plane section before bending
remains a plane after bending, the longitudinal deformation of any element will be proportional
to the distance of the element from the neutral surface.
In developing the analysis, three radii, extending from the center of curvature, O’, of the
member are shown in Fig 4.1. The radii are: r that references the location of the centroid of the
cross sectional area; R that references the location of the neutral axis; and r references some
arbitrary point of area element dA on the cross section. Note that the neutral axis lies within the
cross section since the moment M creates compression in the beams top fibers and tension in
its bottom fibers. By definition, the neutral axis is al line of zero stress and strain.
If a differential segment is isolated in the beam (see Fig 4.2). The stress deforms the
material in such a way that the cross section rotates through an angle of δθ /2. The normal strain
in an arbitrary strip at location r can be determined from the resulting deformation. This strip has
an original length of Lo=r dθ . The strip’s total change in length, ∆L = 2( R − r)δθ / 2 . The normal
strain can be written as:
∆L ( R − r)δθ
. ε= = (4.1)
Lo r dθ
To simplify the relation, a constant k is defined as k = δθ / dθ such that the normal strain can b e
rewritten as:
R − r
ε=k (4.2)
r
4.2
−y
beam (i.e., ε = ). The nonlinear strain distribution for the curved beam occurs even though
ρ
the cross section of the beam remains plane after deformation. The moment, M, causes the
material to deform elastically and therefore Hooke’s law applies resulting the following relation
for stress:
R − r
σ = Eε = Ek (4.3)
r
Because of the linear relation between stress and strain, the stress relation is also
hyperbolic. However, with the relation for stress determined, it is possible to determine the
location of the neutral axis and thereby relate the applied moment, M to this resulting stress.
First a relation for the unknown radius of the neutral axis from the center of curvature, R, is
determined. Then the relation between the stress, σ , and the applied moment, M is
determined.
Force equilibrium equations can applied to obtain the location of R (radius of the neutral
axis). Specifically, the internal forces caused by the stress distribution acting over the cross
section must be balanced such that the resultant internal force is zero:
∑F x =0
dF
Now since σ = then dF = σ dA and
dA
F = ∑σ dA = ∫ σdA = 0
R
Ek ∫ dA -
r
∫ rr dA = 0 ⇒ R ∫ dA - ∫ dA = 0 (4.5)
r
Solving Equation 4.5 for R results in:
R=
∫ dA = A
∫ dA
r
∫ dA
r
(4.6)
where R is the location of the neutral axis referenced from the center of curvature, O’, of the
member, A is the cross sectional area of the member and r is the arbitrary position of
Table 4.1 Areas and Integrals for Various Cross Sections
4.3
the area element dA on the cross section and is referenced from the center of curvature, O’, of
the member. Equation 4.6 can be solved for various cross sections with examples of common
cross sections listed in Table 4.1
Moment equilibrium equations can be applied to relate the applied moment, M, to the
resulting stress, σ . Specifically, the internal moments caused by the stress distribution acting
over the cross section about the neutral must be balanced such that the resultant internal
moment balances the applied moment:
dF
Recall that since σ = then dF = σ dA
dA
if y is the distance from the neutral axis such that y = R - r
then dM = y dF or dM = y (σ dA )
Applying moment equilibrium such that∑M = 0 gives
M − ∑ y (σ dA ) = 0 or M =∑ y (σ dA )
Substituting the derived relations for y and σ gives
Again realizing that Ek and R are constants, Eq 4.7 can be expanded and grouped such that:
4.4
M = Ek R 2 ∫
dA
r
− 2 R ∫ dA + ∫ rdA such that
⇓ ⇓ ⇓
A
R2 2R A rA which gives
R
A
M = Ek R2 − 2R A + rA = EkA(r - R) (4.8)
R
Note that the third integral term in Eq 4.8, comes from the geometric determination of the
M
centroid such that r = ∫ r dA/A . Equation 4.8 can now be solved for Ek such that Ek = .
A(r - R)
Substituting this relation for Ek into Eq 4.3 gives:
R − r M R − r M(R − r)
σ = Eε = Ek = = (4.9)
r A(r - R) r Ar(r - R)
Substituting the relations involving y into Eqs 4.9 (y=R-r and r=R-y) along with an term for
“eccentricity” e= r -R gives:
M(R − r) My
σ= = (4.10)
Ar(r - R) Ae(R - y)
which is the so-called curved-beam flexure formula where σ is the normal stress, M is the
applied moment, y is the distance from the neutral axis (y=R-r), A is the area of the cross
A
section, e is the “eccentricity” (e= r -R) and R is the radius of the neutral axis (R = ).
dA
∫ r
Note that Eq. 4.10 gives the normal stress in a curved member that is in the direction of
the circumference (a.k.a. circumferential stress) and is nonlinearly distributed across the cross
section (see Fig 4.3). It is worth noting that due to the curvature of the beam a compressive
radial stress (acting in the direction of r) will also be developed. Typically the radial stress is
small compared to the circumferential stress and can be neglected, especially if the cross
section of the member is a solid section. Sometimes, such as the case of thin plates or thin
cross sections (e.g., I-beam), this radial stress can become large relative to the circumferential
stress. If the beam is loaded with forces (instead of pure moments) then additional stresses will
occur on radial planes. Because the action is elastic, the principle of superposition applies and
the additional normal stresses can be added to the flexural stresses obtained in Eq. 4.10.
4.5
Fig 4.3 Nonlinear stress distribution across cross section is a curved beam
The curved beam flexure formula is usually used when curvature of the member is
pronounced as in the cases of hooks and rings. A rule of thumb, for rectangular cross sections
for which the ratio of radius of curvature to depth ( r /h) is >5, shows that the curved beam
flexure formula agrees well with experimental, elasticity, and numerical results. If the flexure
formula used, a difference of 7% from the maximum stress determined from the curved beam
flexure formula can result at r /h=5. As this ratio increases (i.e., at the beam becomes less
curved and more straight), the difference of the maximum stress calculated from the flexure
formula for the straight beam and the curved beam flexure formula becomes much less.
4.6
A common machine element problem involving curved beams is the crane hook shown
in Fig. 4.4. In this problem, the load, F is 22,240 N, the cross sectional thickness, t=b=19.05
mm and the cross sectional width, h=W= is 101.6 mm. Since A=bh, dA=bdr and from Eq. 4.6:
A bh h (4.11)
R= = ro =
dA ro
∫ r ∫ dr ln
b
ri r ri
From Figs 4.4a and 4.4b, ri=50.8 mm, ro=152.4 mm, and A=5161.3 mm2.
Substituting the appropriate values into Eq. 4.11 gives R=92.5 mm. The eccentricity,
e=(101.6-92.5)=9.1 mm. Since the moment, M is positive such that M = F r ( r = radial
distance to the centroid where r =(ro+ri)/2).
In the case, the axial force, F, superposes an axial stress on the bending stress such
that
F My 22, 240 (22, 240 * 101.6)(92.5 − r)
σ= + = + (4.12)
A Ae( R − y ) 5161.3 5161.3(9.1) r
Fig 4.5 Stress distribution across the cross section of a crane hook
4.7
Unsymmetrical Bending
Another of the limitations of the usual development of beam bending equations is that
beams are assumed to have at least one longitudinal plane of symmetry and that the load is
applied in the plane of symmetry. The beam bending equations can be extended to cover
pure bending (i.e., bent with bending moments only and no transverse forces) of 1) beams
with a plane of symmetry but with the load (couple) applied not in or parallel to the plane of
symmetry or 2) beams with no plane of symmetry.
Fig 4.6 depicts a beam of unsymmetrical cross section loaded with a couple, M, in a
plane making an angle, α , with the xy plane, where the origin of coordinates is at the centroid of
the cross section. The neutral axis, which passes through the centroid for linearly elastic action
makes an unknown angle, β , with the z axis. The beam is straight and of uniform cress section
and a plane cross section is assumed to remain plane after bending. Note that the following
development is restricted to elastic action.
Since the orientation of the neutral axis is unknown, the usual flexural stress distribution
function (i.e., σ = E (εc / c ) y = (σ c / c ) y ) cannot be expressed in terms of one variable.
However, since the plane section remains plane, the stress variation can be written as:
4.8
σ = k1 y + k2 z (4.14)
The resisting moments with respect to the z and y axes can be written as
M rz = ∫ σdAy = ∫ k y dA + ∫ k zydA = k I
1
2
2 1 z + k2I yz
A A A (4.15)
M ry = ∫ σdAz = ∫ k yzdA + ∫ k z dA = k I
1 2
2
1 yz + k2I y
A A A
where I y and I z are the moment of inertia of the cross sectional area with respect to the z and y
axes, respectively, and I yz is the product of inertia with respect to these two axes. It will b e
convenient to let the y and z axes be principal axes, Y and Z; then I yz is zero. Equating the
applied moment to the resisting moment and solving for k1 and k 2 gives:
M cosα
M rZ = k1I Z = M cosα k1 =
IZ
(4.16)
M sin α
M rY = k2I Y = M sin α k2 =
IY
Substituting the expressions for k given in Eq. 4.16 into Eq. 4.14 gives the elastic flexure
formula for unsymmetrical bending.
M cosα M sin α
σ= Y+ Z (4.17)
IZ IY
Since σ is zero at the neutral axis, the orientation of the neutral axis is found by setting
Eq. 4.17 equal to zero, for which
cosα sin α
Y=- Z (4.18)
IZ IY
or
IZ
Y = -tanα Z (4.19)
IY
where Y is the equation of the neutral axis in the YZ plane. The slope of the line is the dY/dZ
and since dY/dZ= tan β , the orientation of the neutral axis is given by the expression
IZ
tanβ = - tanα (4.20)
IY
The negative sign indicates that the angles, α and β are in adjacent quadrants.
Note that the neutral axis is not perpendicular to the plane of loading unless 1) the angle,
α , is zero, in which case the plane of loading is (or is parallel to) a principal angle, or 2) the two
principal moments of inertia are equal. This reduces to the special kind of symmetry where all
centroidal moments of inertia are equal (e.g., square, rectangle, etc.)
4.9
Composite Beams
The method of "fibre" stress calculation for basic beam bending is sufficiently general to
cover symmetrical beams composed of longitudinal elements (layers) of different materials.
However, for many real beams of two materials (often referred to as reinforced beams) a
method can be developed to allow the use of the elastic flexure formula, thus reducing the
computational labor involved. Of course, the method is applicable to the elastic region only.
The assumption of plane sections remaining plane is still valid, provided that the
different materials are securely bonded together so as to give the necessary resistance to
longitudinal shearing stresses. Therefore, the usual linear transverse distribution of longitudinal
strains is valid.
The beam shown in Fig. 4.7 is composed of a central portion of material A and two
outer layers of material B. The beam will serve as the model for the development of the stress
distribution. The section is assumed to be symmetrical with respect to the xy and xz planes
and the moment is applied in the xy plane. As long as neither material is subjected to stresses
greater than the proportional limit stress, then Hooke's law applies and the strain relation is:
b
ε =ε
b a (4.21)
a
which, in terms of stress, becomes
σ σ
EB
b
= a
EA
()
b
a (4.22)
After rearranging, Eq. 4.22 becomes the relation for the stress distribution:
E b
σ b = σ a
B
(4.23)
EA a
From this relation, it is evident that the junction between the two materials where
distances a and b are equal, there is an abrupt change in the stress determined by the ratio,
E
n = B , of the two elastic moduli (see Fig. 4.7). Using Eq. 4.23, the normal force on a
EA
differential end area of element B is given by the expression:
4.10
where t is the thickness (width) of the beam at a distance b from the neutral surface. The first
term in parentheses represents the linear stress distribution in a homogeneous material A. The
second term in the parentheses may be interpreted as the extended width of the beam from
y=cA to y=cB if material B were replaced by material A, thus resulting in an equivalent or
transformed cross section for a beam of homogeneous material. The transformed section is
obtained by replacing either material by an equivalent amount of the other material as
determined by the ratio, n, of their elastic moduli. The method is not limited to two materials:
however the use of more than two materials in a beam might be unusual.
4.11
5. MECHANICAL PROPERTIES AND PERFORMANCE OF MATERIALS
Mechanical Testing
Mechanical tests (as opposed to physical, electrical, or other types of tests) often
involves the deformation or breakage of samples of material (called test specimens or test
pieces). Some common forms of test specimens and loading situations are shown in Fig
5.1. Note that test specimens are nothing more than specialized engineering components
in which a known stress or strain state is applied and the material properties are inferred
from the resulting mechanical response. For example, a strength is nothing more than a
stress "at which something happens" be it the onset of nonlinearity in the stress-strain
response for yield strength, the maximum applied stress for ultimate tensile strength, or
the stress at which specimen actually breaks for the fracture strength.
Design of a test specimen is not a trivial matter. However, the simplest test
specimens are smooth and unnotched. More complex geometries can be used to
produce conditions resembling those in actual engineering components. Notches (such
as holes, grooves or slots) that have a definite radius may be machined in specimens.
Sharp notches that produce behaviour similar to cracks can also be used, in addition to
actual cracks that are introduced in the specimen prior to testing.
5.1
Figure 5.1 Geometry and loading scenarios commonly employed in mechanical testing of
materials. a) tension, b) compression, c) indentation hardness, d) cantilever flexure, e)
three-point flexure, f) four-point flexure and g) torsion
Equipment used for mechanical testing range from simple, hand-actuated devices
to complex, servo-hydraulic systems controlled through computer interfaces. Common
configurations (for example, as shown in Fig. 5.2) involve the use of a general purpose
device called a universal testing machine. Modern test machines fall into two broad
categories: electro (or servo) mechanical (often employing power screws) and servo-
hydraulic (high-pressure hydraulic fluid in hydraulic cylinders). Digital, closed loop
5.2
control (e.g., force, displacement, strain, etc.) along with computer interfaces and user-
friendly software are common. Various types of sensors are used to monitor or control
force (e.g., strain gage-based "load" cells), displacement (e.g., linear variable differential
transformers ( LVDT's) for stroke of the test machine), strain (e.g., clip-on strain-gaged
based extensometers). In addition, controlled environments can also be applied through
self-contained furnaces, vacuum chambers, or cryogenic apparati. Depending on the
information required, the universal test machine can be configured to provide the control,
feedback, and test conditions unique to that application.
Tension Test
The tension test is the commonly used test for determining "static" (actually quasi-
static) properties of materials. Results of tension tests are tabulated in handbooks and,
through the use of failure theories, these data can be used to predict failure of parts
subjected to more generalized stress states. Theoretically, this is a good test because of
the apparent simplicity with which it can be performed and because the uniaxial loading
condition results in a uniform stress distribution across the cross section of the test
specimen. In actuality, a direct tensile load is difficult to achieve (because of
misalignment of the specimen grips) and some bending usually results. This is not
serious when testing ductile materials like copper in which local yielding can redistribute
the stress so uniformity exists; however, in brittle materials local yielding is not possible
and the resulting non uniform stress distribution will cause failure of the specimen at a
load considerably different from that expected if a uniformly distributed load were applied.
The typical stress-strain curve normally observed in textbooks with some of the common
nomenclature is shown in Fig. 5.3. This is for a typical low-carbon steel specimen. Note
that there are a number of definitions of the transition from elastic to plastic behavior. A
few of these definitions are shown in Fig. 5.3. Oftentimes the yield point is not so well
defined as for this typical steel specimen. Another technique for defining the beginning of
plastic behavior is to use an offset yield strength defined as the stress resulting from the
intersection of a line drawn parallel to the original straight portion of the stress strain
curve, but offset from the origin of this curve by some defined amount of strain, usually 0.1
percent ( ε = 0.001) or 0.2 percent ( ε = 0. 002) and the stress-strain curve itself. The total
strain at any point along the curve in Fig. 5.3 is partly plastic after yielding begins. The
amount of elastic strain can be determined by unloading the specimen at some
deformation, as at point A. When the load is removed, the specimen shortens by an
amount equal to the stress divided by elastic modulus (a.k.a., Young's modulus). This is,
∆σ
in fact, the definition of Young's modulus E = in the elastic region.
∆ε
5.3
Figure 5.3 Engineering stress-strain diagram for hot-rolled carbon steel showing
important properties (Note, Units of stress are psi for US Customary and
MPa for S.I. Units of strain are in/in for US Customary and m/m for S.I.
5.4
As shown in Fig. 5.3, often basic stress-strain relations are plotted using
engineering stress, σ , and engineering strain, ε . These are quantities based on the
original dimensions of the specimen, defined as
Load P
σ= = (5.1)
Original Area A o
The Modulus of Resilience is the amount of energy stored in stressing the material
to the elastic limit as given by the area under the elastic portion of the σ - ε diagram and
can be defined as
εo
σ ε
Ur = ∫ σ dε ≈ o o (5.3)
0
2
where σ o is the proportional limit stress and ε o is the strain at the proportional limit stress.
Ur is important in selecting materials for energy storage such as springs. Typical values
for this quantity are given in Table 5.1.
The Modulus of Toughness is the total energy absorption capabilities of the
material to failure and is given by the total area under the σ - ε curve such that
εf
(σ + Su )
Ut = ∫ σ dε ≈ o εf (5.4)
0
2
where Su is the ultimate tensile strength, σ o is the proportional limit stress and ε f is the
strain at fracture. Ut is important in selecting materials for applications where high
overloads are likely to occur and large amounts of energy must be absorbed. This
modulus has also been shown to be an important parameter in ranking materials for
resistance to abrasion or cavitation. Both these wear operations involve tearing pieces of
metal from a parent structure and hence are related to the "toughness" of the material.
5.5
The ductility of a material is its ability to deform under load and can be measured
by either a length change or an area change. The percent elongation, which is the
percent strain to fracture is given by:
L − Lo
%EL = 100εf = 100 f = 100 Lf −1 (5.5)
Lo Lo
where Lf is the length between gage marks at fracture. We should note that this quantity
depends on the gage length used in measuring L, as non uniform deformation occurs in a
certain region of the specimen during necking just prior to fracture, hence, the gage length
should always be specified. The percent reduction in area is a cross-sectional area
measurement of ductility defined as
A − Af
%RA = 100 o = 100 1− Af (5.6)
Ao Ao
where A f is the cross-sectional area at fracture. Note %RA is not sensitive to gage length
and is somewhat easier to obtain, only a micrometer is required. It should be realized that
the stress-strain curves just discussed, using engineering quantities, are fictitious in the
sense that the σ and ε are based on areas and lengths that no longer exist at the time of
measurement. To correct this situation true stress (σ T ) and true strain (ε T ) quantities are
used. The quantities are defined as:
P
σT = (5.7)
Ai
where A i is the instantaneous area at the time P is measured. Also
L
dL L
ε =∫ = ln
T
(5.8)
Lo
L Lo
or
A
dA A
ε = −∫ = ln o
T
(5.9)
Ao
A A
where L is the instantaneous length between gage mark at the time P is measured.
These two definitions of true strain are equivalent in the plastic range where the
material volume can be considered constant during deformation as shown below.
Since
AL = AoL o (5.10)
then
L Lo = A o A (5.11)
The constant volume condition simply says the stressed volume AL is equal to the
original volume AoLo. (Note this is only true in the plastic range of deformation, in the
5.6
elastic range the change in volume ∆V per unit volume is given by the bulk modulus B
E
(where B = and υ is Poisson's ratio).
3(1− υ)
Prior to necking, engineering values can be related to true values by noting that
L L + ∆L
ε T = ln i = ln o (5.12)
Lo Lo
then
ε T = ln(1+ ε) (5.13)
Ao L L o +∆ L
and since = = (5.14)
A Lo Lo
so
Ao
A= (5.15)
1+ ε
and since
P
σT = (5.16)
A
then
P
σT = ( 1+ ε) = σ(1+ ε ) (5.17)
Ao
where σ and ε are the engineering stress and strain values at a particular load.
True stress and true strain values are particularly necessary when one is working
with large plastic deformations such as large deformation of structures or in metal forming
processes. In the elastic region the relation between stress and strain is simply the linear
equation
σ = Eε (5.18)
and also
σ T = Eε T (5.19)
In the plastic region, a commonly used relation to define the relation between
stress and strain is given by
σ T = K (ε T )n = H(ε T )m (5.20)
where strength coefficient, K or H, is the stress when ε T =1 and m or n is an exponent
often called the strain hardening coefficient. Typically values for K or H and m or n are as
given in Table 5.2.
5.7
Table 4.2 Material constants m or n and K or H for different sheet materials
Specimen
Material Treatment n or m K or H Thickness
(MPa) (mm)
0.05% carbon rimmed steel Annealed 0.261 532 24
0.05/0.07% phosphorus 644 24
low-carbon steel Annealed 0.156
SAE 4130 steel Annealed 0.118 1168 24
Type 430 stainless steel 986 32
(17% Cr) Annealed 0.229
Alcoa 24-S aluminum Annealed 0.211 386 258
5.8
Figure 5.4 Logarithmic true stress-logarithmic (true) strain data plotted on
logarithmic coordinates
In the plastic region, the strains become sufficiently large that a finely graduated
scale used in conjunction with a pair of dividers to measure linear strain, or a micrometer
to measure lateral strain can be used.
In the U.S., generally a 2.0 in (50.8 mm) gage length is used to measure
deformations. The 2.0 in (50.8 mm) interval is often marked off with a special tool that
marks the interval with punch marks. These punch marks should be very light or fracture
will occur at this point. Alternatively an indelible marker can be used to avoid damaging
the surface of the test specimen.
5.9
Hardness
In the field of engineering, hardness is often defined as the resistance of a material
to penetration. Methods to characterize hardness can be divided into three primary
categories:
1) Scratch Tests
2) Rebound Tests
3) Indentation Tests
5.10
hardness dates from the late 1800's and is probably the most common hardness test in
the world.
The Brinell hardness number is obtained by dividing the applied force, P, in kg, by
the actual surface area of the indentation which is a segment of a sphere as illustrated in
Fig. 5.6 such that:
P 2P
BHN = HB = =
[ ]
(5.23)
πDt πD D − (D 2 −d 2 )
where D is the diameter of the ball in mm, t is the indentation depth from the surface in
mm, and d is the diameter of the indentation at the surface in mm.
Brinell hardness is good for averaging heterogeneities over a relatively large area,
thus lessening the influence of scratches or surface roughness. However the large ball
size precludes the use of Brinell hardness for small objects or critical components where
large indentations may promote failure. Another limitation of the Brinell hardness test is
that because of the spherical shape of the indenter ball, the BHN for the same material
will not be the same for different loads if the same size ball is used. Thus, geometric
similitude must be imposed by maintaining the ratio of the indentation load and indenter
diameter such that:
P1 P2 P3
= = = etc. (5.24)
D1 D2 D3
The Meyers hardness test is a variation on the Brinell hardness test and addresses
this lack of geometric similarity by using the projected area of the indentation such that:
P
MHN = HM = (5.25)
πd 2 /4
Although the Meyers hardness is less sensitive to applied load and is a more fundamental
measure of hardness, it is rarely used.
Vickers Hardness Test (a.k.a.. diamond pyramid hardness) In this test, the same
general principles of the Brinell test are applied. However, a four-sided diamond pyramid
is implied as an indenter rather than a ball to promote geometric similarity of indentation
regardless of indentation load (see Fig. 5.7). The included angle between the faces of the
pyramid is 136° which corresponds to a desired d/D ratio for the Brinell test of 0.25. The
resulting Vickers indentation has a depth, h equal to 1/7 of the indentation size, L,
measured on the diagonal. The Vickers hardness is obtained by dividing the applied
force by the surface area of the paramedical impression such that:
2P α
VHN = HV = 2 sin (5.26)
L 2
where P is the indentation load which typically ranges from 0.1 to 1 kg but may be as high
as 120 kg, L is the diagonal of the indentation in mm and α is the included angle of 136°.
5.11
Figure 5.5 Approximate relative hardnesses of metals and ceramics for Mohs scale and
indention scales.
The primary advantage of the Vickers hardness is that the result is independent of
load. However, disadvantages are that it is somewhat slow since careful surface
preparation is required. In addition, the result may be prone to personal error in
measuring the diagonal length along with interpretation of anomalies such as "pin
cushioning" for soft materials and "barreling/ridging" for hard materials.
5.12
P=3000 kg
Steel or or 500 kg
tungsten
carbide
ball D=10 mm
t
d
Side view
0 1 2
Rockwell Hardness Test The Rockwell test is the most widely accepted hardness
test in the United States. In this test, penetration depth is measured, with the hardness
reported as the inverse of the penetration depth. A two step procedure is used as
illustrated in Fig. 5.8. The first step "sets" the indenter in the material and the second step
is the actual indentation test. The conical diamond or spherical indenter tips produce
indentation depths, the inverse of which are used to display hardness on the test machine
directly. The reported hardness is in arbitrary units, but the Rockwell scale which
identifies the indentation load and indenter tip must be reported with the hardness
number (otherwise the number is useless). Rockwell scales include those in Table 5.3.
5.13
Table 5.3 Representative Rockwell indenter specifics
Rockwell Indenter Major Load
Scale (kg)
A Brale 60
B 1/16" Ball 100
C Brale 150
D Brale 100
E 1/8" Ball 100
F 1/16" Ball 60
M 1/4" Ball 100
* Brale is a conical diamond indenter
Some important points concerning Rockwell hardness testing include the following
1) Indenter and anvil should be clean and well seated.
2) Surface should be clean, dry, smooth, and free from oxide
3) Surface should be flat and normal
A primary advantage of the Rockwell hardness test is that it is automatic and self-
contained thereby given and instantaneous readout of hardness which lends itself to
automation and rapid through put.
5.14
Figure 5.8 Rockwell hardness indentation for a minor load and for a major load.
5.15
Note that for both these relations, there is considerable scatter in actual data, so
that these relationships should be considered to provide rough estimates only. For other
classes of material, the empirical constant will differ, and the relationships may even
become nonlinear. Similarly, the relationships will change for different types of hardness
tests. Rockwell hardness correlates well with ultimate tensile strength and with other
types of hardness tests, although the relationships can be nonlinear. This situation results
from the unique indentation-depth basis of this test. For carbon and alloy steels,
conversion charts for estimating various types of hardness from one another as well as
ultimate tensile strengths are contained in an ASTM standard, ASM handbooks and
information supplied by manufacturers of hardness testing equipment
Torsion
The torsion test is another fundamental technique for obtaining the stress-strain
relationship for a metal. Because the shear stress and shear strain are obtained directly
in the torsion test, rather than tensile stress and tensile strain as in the tension test, many
investigators actually prefer this test to the tension test. Since all deformation of ductile
materials is by shear, the torsion test would seem to be the more fundamental of the two.
The torsion test is accomplished by simply clamping each end of a suitable
specimen in a twisting machine that is able to measure the torque, T, applied to the
specimen. Care must be used in gripping the specimen to avoid any bending. A device
called a troptometer is used to measure angular deformation. This device consists of two
collars which are clamped to the specimen at the desired gage length. One collar is
equipped with a pointer the other with a graduated scale, so the relative twist between the
gage marks can be determined. The troptometer is useful for measuring strains up to and
slightly past the elastic limit. For larger plastic strains, complete revolutions of the collars
are counted.
The test, then, consists of measuring the angle of twist, θ (radians) at selected
increments of torque T (N-m). Expressing the twist as θ '= θ /L, the angular deflection per
unit gage length, one is able to plot a T - θ ' diagram that is analogous to the load-
deflection diagram obtained in the tension test. To be useful for engineering purposes, it
is necessary to convert this T - θ 'diagram to a shear stress (τ ) - shear strain (γ ) diagram
similar to the previous normal stress ( σ ) - normal strain (ε ) diagram. Of course, one can
also convert the τ - γ diagram to a σ - ε diagram as will be shown later.
5.16
Figure 5.10 Torsion of cylindrical test bar
5.17
Figure 5.11 Elastic Shear Stress Distribution
5.18
2πτr 3 2πτ r r 4 πτ rr 3
T= r a da = = (5.35)
r o r 4 2
and the shear stress at the outermost fibers is
2T
τr = 3 (5.36)
πr
Note that Eqs. 5.33 to 5.34 applies only in the elastic region. When the metal starts to
deform plastically, the shear stress distribution is no longer linear, but is as shown in Fig.
5.12.
The relation between T and τ is no longer the same. To evaluate this relation we
begin as before, noting that the torque at a very thin ring of radius a is again given by
dT = 2πτa2 da (5.37)
So the total external torque resisted across the section is then
a
T = 2π ∫ τa2 da (5.38)
o
aθ
The shear strain relation γ = at any radius a is still valid, however, and substituting this
L
in Eq. 5.38, we obtain
τγ 2L2 L
T = 2π ∫ r 2 dγ (5.39)
o θ θ
The shear stress T at any radius a is also a function of γ only, i.e.
τ = f (γ ) (5.40)
so the expression for torque T can be written in terms of γ only as
γr
Tθ3 = 2πL3 ∫ f (γ ) γ 2dγ (5.41)
o
5.19
Differentiating both sides of this equation with respect to 9, one obtains
d
dθ
( ) 2 dγ
= Tθ3 = 2πL3 f(γ r ) γ r r `
dθ
(5.42)
rθ
since γ r = then
L
dγ r r
= (5.44)
dθ L
( )
and substituting these quantities in the equation for d dθ Tθ3 and working out the
5.20
At the typical point P at which it is desired to obtain the shear stress, observe that
dT PC
θ = BC and that = such that T = AP. Substituting these quantities, the result is
dθ BC
1 PC PC + 3AP
τ= BC + 3AP or τ = .
2πr BC
3 2πr3
With this last relation it is then a simple matter to obtain values of τ at various θ
rθ
positions of the plastic part of the T - θ curve. Remember that γ = , the complete τ - γ
L
curve can be obtained.
For the materials approach, it is possible to again make the valid assumption that
rθ
γ = . However, τ is determined as one of two functions of γ depending on whether the
L
internal stress state is in the elastic or plastic range. However, calculating this internal
stress state requires a priori knowledge of material properties usually determined from a
E
tensile test. In particular, E is required to calculate G = , σ is required to calculate
2(1+ υ) o
σ K
τ o = o , K and n are required to calculate Kτ = ( n +1)/2 and n for shear equals n for
3 3
tension. Once these relations are established, then it is possible to calculate the shear
stress from the shear strain for the elastic or plastic condition as follows.
K
where G is the shear modulus, Kτ = ( n +1)/2 in which K and n are the strength coefficient
3
and strain hardening exponent from the tension test, respectively. The shear strain at
yield can be determined from an effective stress-strain relation from plasticity such that
τ
γo = o (5.48)
G
σ
where τ o = o in which σ o is the "yield" strength from a tension test.
3
For any given T-θ combination, it is possible to calculate the shear strain at the
rθ
surface of the specimen (that is, r=R) as γ = . Comparing this shear strain to that
L
calculated in Eq. 5.48, allows the choice of either Eq. 5.46 or 5.47.
Note that when the shear stress at r=R is plastic, the total torque, T, required to
produce the deformation, θ , will have two components: an elastic torque, T e and a plastic
torque, Tp since the shear stress across the cross section of the specimen will have both a
5.21
plastic part and an elastic part as shown in Fig. 5.14. The relation for T can then be
written as:
Ttotal = Telastic + T plastic (5.50)
where
τo γo
Telastic = ∫ τ dA r = ∫ Gγ dA r (5.51)
0 0
τR γR
Tplastic = ∫ τ dA r = ∫ Kτ γ n dA r (5.52)
τo γR
For convenience it is possible to rewrite the integration variables in Eqs. 5.51 and 5.52 in
terms for the specimen radius, r, only such that
ry
τy
Telastic = ∫r y
2πr 3dr (5.53)
0
K rθ'
R n
Tplastic = ∫ 3 3
2πr 2 dr (5.54)
ry
θ θ θ τo τ
where θ' = and ry = γ o = = θ' o .
L L L G G
Equations 5.53 and 5.54 can be solved either closed form or numerically for any
combination of T and θ .
γ =r θ /L
τ =f( γ )
Elastic Plastic
τ =Gγ τ =Kτ γ
n
r=0 ry r=R
Radial distance, r
Figure 5.14 Shear stress and shear strain as functions of radial distance
5.22
Once the shear stress-strain curve is obtained, engineering properties are easily
calculated. A few of the more important quantities will be discussed. As in the tension test,
yield strengths for shearing stress can be defined, such as a proportional limit or an offset
yield strength. The Modulus of Rupture is the total area under the τ - γ curve determined
at r=R and represents the total energy absorption abilities of the material in shear.
Figure 5.15 Mohr's circles for the tensile test and torsion test
5.23
As in the tension test, the Modulus of Resilience is the area under the elastic
portion of the τ - γ curve such that
γo
Ur = ∫ τ dγ (5.55)
o
Similarly, the Modulus of Toughness is the area under the total τ - γ curve such
that
γf
Ut = ∫ τ dγ (5.56)
o
The Modulus of Rigidity (or Shear Modulus), G, is the slope of the τ - γ curve in the
elastic region and is comparable to Young's Modulus, E, found in tension. Recall that the
∆τ E
relation between E and G is G = = .
∆γ 2(1+υ )
The true shear stress-strain curve can be compared to the tensile true stress-strain
curve by converting the normal values to shear values. The conversion is as follows:
σ
Elastic range: τ equivalent = ; γ equivalent = 1.25ε (5.57)
2
σ
Plastic range: τ equivalent = ; γ equivalent = 1.5ε (5.58)
2
That these values are correct can be seen from Mohr's circle of stress and of strain
for the elastic and plastic ranges (Fig. 5.15). Knowledge of Poisson's ratio, υ , is needed
for Mohr's circles of strain for the tensile test. For mild steel in the elastic range, υ = .0.30;
in the plastic range, υ =0.5 as a result of the constant volume assumption.
Impact
The static properties of materials and their attendant mechanical behavior are very
much functions of factors such as the heat treatment the material may have received as
well as design factors such as stress concentrations.
The behavior of a material is also dependent on the rate at which the load is
applied. Polymeric materials and metals which show delayed yielding are most sensitive
to load application rate. Low-carbon steel, for example, shows a considerable increase in
yield strength with increasing rate of strain. In addition, increased work hardening occurs
at high-strain rates. This results in reduced local necking, hence, a greater overall
material ductility occurs. A practical application of these effects is apparent in the
fabrication of parts by high-strain rate methods such as explosive forming. This method
5.24
results in larger amounts of plastic deformation than conventional forming methods and,
at the same time, imparts increased strength and dimensional stability to the part.
In design applications, impact situations are frequently encountered, such as
cylinder head bolts, in which it is necessary for the part to absorb a certain amount of
energy without failure. In the static test, this energy absorption ability is called
"toughness" and is indicated by the modulus of rupture. A similar "toughness"
measurement is required for dynamic loadings; this measurement is made with a
standard ASTM impact test known as the Izod or Charpy test. When using one of these
impact tests, a small notched specimen is broken in flexure by a single blow from a
swinging pendulum. With the Charpy test, the specimen is supported as a simple beam,
while in the Izod it is held as a cantilever. Figure 5.16 shows standard configurations for
Izod (cantilever) and Charpy (three-point) impact tests.
A standard Charpy impact machine is used. This machine consists essentially of a
rigid specimen holder and a swinging pendulum hammer for striking the impact blow (see
Fig. 5.17). Impact energy is simply the difference in potential energies of the pendulum
before and after striking the specimen. The machine is calibrated to read the fracture
energy in N-m or J directly from a pointer which indicates the angular rotation of the
pendulum after the specimen has been fractured.
Figure 5.16 Charpy and Izod impact specimens and test configurations
5.25
mass, m
h1
h2
IMPACT ENERGY=mg(h1-h2)
Figure 5.17 Charpy and Izod impact specimens and test configurations
The Charpy test does not simulate any particular design situation and data
obtained from this test are not directly applicable to design work as are data such as yield
strength. The test is useful, however, in comparing variations in the metallurgical structure
of the metal and in determining environmental effects such as temperature. It is often
used in acceptance specifications for materials used in impact situations, such as gears,
shafts, or bolts. It can have useful applications to design when a correlation can be found
between Charpy values and impact failures of actual parts.
Curves as shown in Fig. 5.18 showing the energy to fracture as measured by a
Charpy test indicate a transition temperature, at which the ability of the material to absorb
energy changes drastically. The transition temperature is that temperature at which,
under impact conditions, the material's behavior changes from ductile to brittle. This
change in the behavior is effected by many variables. Metals that have a face-centered
cubic crystalline structure such as aluminum and copper have many slip systems and are
the most resistant to low-energy fracture at low-temperature. Most metals with body-
centered cubic structures (like steel) and some hexagonal crystal structures show a sharp
transition temperature and are brittle at low temperatures.
Considering steel; coarse grain size, strain hardening, and certain minor impurities
can raise the transition temperature whereas fine grain size and certain alloying elements
will increase the low temperature toughness. Figure 5.18 shows the effect of heat
treatment on alloy steel 3140 and 2340. Note that a transition temperature as high as
about 25°C is shown. This material, then should not be in service below temperature of
25°C when impact conditions are likely to exist.
5.26
Figure 5.18 Variation in transition-temperature range for steel in the Charpy test
Any of these criteria are usable. Perhaps the most direct criteria for a particular
metal is to define the transition temperature as that temperature at which some minimum
amount of energy is required to fracture. During World War II, allied Victory ships literally
broke in two in conditions as mild as standing at the dock because of the use of steel with
a high-transition temperature, coupled with high-stress concentrations. It was found that
specimens cut from plates of these ships averaged only 9 J. Charpy energy absorption at
the service temperature. Ship plates were resistant to failure if the energy absorption
value was raised to 20 J at the service temperature by proper control of impurities.
5.27
Plasticity Relations
Power
σο σο Linear
σο
E E
εT εT εT
Rigid-Perfectly Plastic Elastic-Perfectly Plastic Elastic-Linear Hardening
Elastic-Power Hardening
Figure 5.19 Mathematical approximations of plot curves
The hardening- flow curve is the most generally applicable type of flow curve. This
type of plastic deformation behaviour has been modeled two different ways: Simple
Power Law and Ramberg-Osgood.
In the Simple Power Law model, the stress strain curve is divide into two discrete
region, separate at σ = σ o such that:
Elastic : σ = Eε (σ ≤ σ o ) (5.59)
Plastic : σ = Hε n (σ ≥ σ o ) (5.60)
5.28
For the Ramberg-Osgood relation, σ o is not distinct "break" in the stress-strain curve, but
is instead calculated from the elasticity and plasticity relations such that
1
H 1− n
σo = E (5.63)
E
General stress-strain relations can be developed for deformation plasticity theory
such that the effective stress is
1
σ eff = (σ 1 − σ 2 )2 + (σ2 −σ 3 )2 +(σ 3 − σ1 )2 (5.64)
2
and the effective total strain is
σ
ε = +εp (5.65)
E
where the effective plastic strain is
2
εp = (ε p 1 − ε p 2 )2 + (ε p2 − ε p 3 )2 + (εp 3 − ε p1)2 (5.66)
3
The resulting effective stress-effective strain curve is independent of the state of stress
and is used to estimate the stress-strain curves for other states of stress. Of particular
importance for are equations which allow correlation of plasticity relations for tension and
torsion such that:
τ = Kτ γ n (5.67)
K
where Kτ = ( n +1)/2 in which K and n are the strength coefficient and strain hardening
3
exponent from the tension test, respectively. In addition,
σ
τo = o (5.68)
3
where σ o is the yield strength from a tension test.
Strain
Hardening
σο
εp εe
Strain
Figure 5.20 Elastic-plastic stress-strain curve
5.29
6. STRESS CONCENTRATION AND STRESS RAISERS
It is very important for the engineer to be aware of the effects of stress raisers such
as notches, holes or sharp corners in his/her design work. Stress concentration effects in
machine parts and structures can arise from internal holes or voids created in the casting
or forging process, from excessively sharp corners or fillets at the shoulders of stepped
shafts, or even from punch or stamp marks left during layout work or during inspection of
parts.
6.1
almost uniform stress across the cross section, as if the groove didn't exist at all. This
plastic flow action is the reason why notches and holes in ductile materials may not lower
the ultimate strength when the specimen is tested statically, and is why stress
concentrations are sometimes ignored when designing with ductile materials. If the
groove is sufficiently deep, the large amount of material adjacent to the groove may
prevent any plastic flow action from occurring, and the specimen will fail at a stress higher
than an ungrooved specimen, stress being based on the reduced section area as shown
in Fig. 6.2. This is an instance when a stress concentration can be dangerous in a ductile
material.
Very little of the energy-absorbing plastic flow will occur with such a severe notch,
and such a member may fail in a brittle manner with a small shock load. In addition it
should be remembered that any grooves at all are dangerous in ductile materials if the
load fluctuates in magnitude, since fatigue crack initiation is a surface phenomenon and
the resulting fatigue strength is strongly influenced by surface finish.
The effects of a discontinuity in a brittle material are very much different than in a
ductile material. With these materials, no stress relieving plastic flow action is possible
and the full value of the stress concentration is valid right up to the fracture strength. For
these materials, then, we expect the fracture strength to be reduced from the unnotched
fracture strength by the value of K. In fact, one method for determining K is to use brittle
plaster test specimens with notches of various severity.
Design with brittle materials must be done with a great deal of care to avoid
undesirable failures. Generous fillets are used, holes eliminated, and attachments
carefully worked out. Considerable care must be taken to avoid even surface scratches
during fabrication.
Experimental Techniques
Elastomer Models: Geometric models can be used when the concern is with the
elastic behavior of materials. Metal specimens are not particularly good for demonstrating
elastic behavior and stress concentration effects because they are so stiff. It is much
easier to visualize elastic behavior if an elastomer specimen is used. There is, however,
one important difference between the behavior of most elastomers, such as rubber, and
that of metal. The stress-strain relation is linear, elastic (to yielding) for metal and is
nonlinear, elastic for rubber. This difference is offset by the large, easily measured
strains, which occur in rubber.
Usually a square grid of lines is printed on the surface of the specimen. A loading
frame can be used on which the specimen is stretched to approximately twice its original
length. The shape of the grid network is then carefully observed while the specimen is in
6.2
the frame. At the junction of large and small portions of the specimen, it can be observed
that the strains are significantly greater than those removed from the junction. In fact, the
exact region of maximum strain can be seen on the deformed grid and the stress (strain)
concentration factor can be calculated.
Brittle Coating: One of the most straight-forward methods of experimental stress
analysis involves the use of a brittle coating. During testing, brittle materials fracture with
a clean, square break that is always oriented so the fracture surface is normal to the
direction of the largest principal stress. The brittle coating technique utilizes this property
to gage the magnitude and direction of stresses in a loaded member.
The use of brittle coatings in stress analysis has a long history, but its real
beginning was in the observation that hot-rolled steel with a mill-scale coating would
behave in a most unusual manner when stressed. In tension tests, for example, the mill
scale would crack in a geometric pattern indicating principal stress direction. In a tensile
test the cracks appear normal to the direction of load, while in a torsion test the cracks
appear in a 45° helix pattern.
Figure 6.1 Stress concentration factors for a stepped, flat tensile specimen.
6.3
Figure 6.2 Stress concentration factors for a circumferential grove in a tensile specimen.
The usefulness of the brittle oxide mill scale is limited by the fact that the yield point
of the material must be exceeded before cracking occurs. Today a much more sensitive
brittle coating known as Stresscoat™ is available. This material is a patented mixture
which can be sprayed on the structure to be analyzed and after drying will crack at strain
levels as low as 400 µ m/m.
Stresscoat™ possesses many of the important characteristics of a brittle material,
however, it also has several limitations which must be allowed for during usage. One of
these is that the Stresscoat™ must dry for several hours after application before it can be
used. In addition, although Stresscoat™ is now available in aerosol cans, the grade to be
used depends on the temperature of the test room only. The material is simply sprayed on
to an average film thickness of about 0.01 mm or, with practice, until the correct uniform
yellow shade is obtained. At the same time the model is sprayed a number of calibration
specimens are also sprayed and all are allowed to dry in the test environment.
At the time of testing the Stresscoat™ is first "calibrated" by loading the calibration
specimens as a cantilever beams in a special loading fixture. A series of fine cracks
normal to the long axis of the beam will be evident and the last crack nearest the loading
6.4
cam is marked with a soft pencil; the strain level at this crack, as indicated when the bar is
held in a fixture, is the material sensitivity. This is true because the strain level in a
cantilever beam is a maximum at the rigid end and decreases uniformly to zero at the
loaded end. Somewhere, then, along the length of the bar the strain will decrease to a
level that is insufficient to crack the coating. The last crack appearing nearest the loading
end is the critical level of strain.
A typical Stresscoat™ test is as follows: An estimate of the maximum load to be
applied is made and load increments to reach this load decided upon. Because the
Stresscoat™ is sensitive to the duration of load application, a loading interval is used.
The specimen is loaded to the level of the first interval, inspected for cracks and then
unloaded within the time interval allotted. The specimen is then allowed to remain
unloaded for about five min before loading to a load increased by the desired increment.
Each loading inspection and unloading cycle must be done within the same time interval,
probably 100 s per interval is reasonable. As the crack pattern progresses with
increasing loads, the locus of points of crack tips is marked with a grease pencil. These
marked lines are points of known strain value as found from the calibration bar. The most
critical cracks in this experiment are the initial cracks that form at the reduced section.
These will be the first cracks that form and considerable care should be exercised in
obtaining the load at which they initiate.
The calibration bars are loaded in one-second intervals, and the sensitivity thus
obtained is corrected to the actual sensitivity caused by the longer loading cycle in the
model by using a creep correction chart supplied by the instructor.
Photoelastic Technique: The photoelastic technique is one of the most powerful of
experimental stress analysis techniques. The photoelastic technique is valuable because
it gives an overall picture of the stress field, quickly showing regions of stress
intensification. In addition, the direction of principal stresses is also easily determined.
Like all experimental techniques, photoelasticity requires some practice to yield accurate
results, in particular, the determination of the principal stresses σ 1 and σ 2 on the interior
of the model requires considerable effort. Often, one is interested only in determining the
stress on the boundary of the model where one of the principal stresses is zero.
In the photoelastic method a model of the shape to be investigated is made from a
suitable transparent material. The model is then loaded in a manner similar to the actual
part and an accurate description of the stress magnitude and direction is obtained by
measuring the change in optical properties of the transparent model. These changes in
properties are measured by viewing the model in a special equipment called a
polariscope, so named because polarized light or light vibrating in a single plane only, is
used.
6.5
The property of the model material that makes it suitable for stress field studies is
termed birefringence. The effects of this property are as follows:
2. These split polarized beams are out of phase by an amount that is dependent of
the difference of the principal stresses, i. e. to (σ 1 −σ 2 ) at a point on the loaded
model.
The theoretical background of photoelasticity is beyond the scope of these
laboratory notes, although numerous references are also available on this experimental
technique. Simply note how the engineer can quickly use photoelasticity to determine
stress concentrations Typically, the polariscope is used in what is termed a circularily
polarized light configuration. In this configuration the model is located between the
polarizing elements as sketched in Fig. 6.3.
Note in the Fig. 6.3 that special filters called polarizers are used, one at each end of
the polariscope. Inside these filters is another set of polarizing filters called quarter-wave,
(λ 4) , plates. These elements can be arranged so the background light is either light,
called light field, or completely extinguished, called dark field.
6.6
When the loaded model is viewed in this type polariscope, a fringe pattern termed
an isochromatic pattern is apparent. These patterns are the loci of constant principal
stress difference. That is, if we know the calibration constant f of the photoelastic material
then
t
f = (σ1 − σ 2 ) (6.1)
N
where f is the stress-optical coefficient, N is the fringe order, t is the model thickness, and
σ 1 and σ 2 are the plane-stress principal stresses.
Each dark band (See Fig. 6.4) for a dark field arrangement corresponds to an
integral (0, 1, 2, 3, etc.) fringe order. In this experiment, we are simply interested in the
maximum fringe order at the radii. The fringe order can be determined in at least two
ways. One method is to count the fringe order to the point of interest by beginning at a
point of zero fringe order such as a free unloaded corner. At such a corner σ 1 = σ2 = 0 .
hence σ 1 = σ2 = 0 and N must be zero. The second method is to observe to increase in
fringe order at the point of interest as the model is slowly loaded from zero load.
Once the maximum fringe order has been determined at the edge of the notch,
including estimates of fractional fringes orders, the stress can be calculated such that:
N
(σ 1 − σ 2 ) = f (6.2)
t
where f is the stress-optical coefficient determined previously, N is the fringe order, t is the
model thickness, and σ 1 and σ 2 are the plane-stress principal stresses in which one of the
plane-stress principal stresses is equal to zero at the free surface of the notch edge.
6.7
Figure 6.4 Photoelastic model as viewed in polariscope. Fringe value is 0 at external
sharp corners and 3 in narrow leg. Note that the model is a uniaxially loaded
tensile specimen.
6.8
7. FRACTURE
ALLOWABLE ALLOWABLE σο
STRESS, σ STRESS, σ
High K Ic a t = transition crack length
Low K Ic between yield and fracture
Figure 7.1 Cracks lower the material's tolerance (allowable stress) to fracture.
7.1
σmax
Repulsion Attraction
Separation distance, x
where E is the elastic modulus and x e is the equilibrium distance (a.k.a., lattice spacing)
between the atoms. Assuming λ /2 =x e and solving Eq. 7.2 for σ max gives.
E
σ max = (7.3)
π
where σ max is the maximum cohesive strength (i.e., greatest strength potential) of a
material.
Another approach is to evaluate the work done in pulling the atoms apart:
λ /2
πx
x x
7.2
in real materials must be preventing them from reaching the strength potential predicted
by the maximum cohesive strength.
In engineering design, stress raisers such as holes can raise the local stress to
levels much greater than the remote or applied stress. For example, for a plate with a
hole in it subjected to uniaxial, uniform tensile stress (see Fig. 7.3) the elasticity solutions
in polar coordinates with variables r and θ are:
σ R 2 3R2
σ rr = 1− 1+ −1 cos2θ
2 r2 r 2
(7.7)
σ R2 3R 2
σθθ = 1 + + 1 + cos2θ
2 r2 r2
At the edge of the hole for θ=0°, Eqs. 7.7 can be rewritten in Cartesian coordinates as:
σ R2 3R 2
σ xx = σ rr = 1 − 2 2
2 r r
(7.8)
σ R2 3R 2
σ yy = σθθ = 2 + 2 + 2
2 r r
Finally at the edge of the hole, where r=R, the stresses are:
σ xx = σ rr = 0
(7.9)
σ yy = σθθ = 3σ
σ
The stress concentration factor can be defined as kt = local . For the stress in the y-
σ remote
direction, k t is 3. Thus, the local stress at the edge of the hole is three times the remotely
applied stress regardless of the size of the hole as long as the hole diameter is small
relative to the width of the place.
In the case of an elliptical hole in uniaxial, uniform tension (see Fig. 7.3), the stress
in the y-direction can be related to the major and minor axes such that:
c
σ yy = σ 1 + 2 = σ 1+ 2
c
(7.10)
a ρ
a2
where the radius of the ellipse tip is ρ = . The stress concentration factor for the
c
elliptical hole is no longer constant as it is for the circular hole but is a function of the
ellipse geometry:
c c
kt = 1 + 2 ≈ 2 (7.11).
ρ ρ
Note in this case that as the minor axis 2a shrinks to zero (e.g., as in a crack), ρ goes to
zero and the stress concentration factor in Eq. 7.11 goes to infinity.
7.3
σ σ
2
y y a
_
r ρ= c
θ 2a
R x x
2c
σ σ
Figure 7.3 Plates with circular and elliptical holes, subjected to uniform tensile stresses
Thus, elliptically-shaped features can have local stresses which are much greater
than remote stresses. If these local stresses approach the maximum cohesive strength on
a microscopic scale, then failure can initiate at such features ultimately leading to failure
of the material.
However, stress analyses such as these do not lend themselves to application in
engineering because infinite stress concentration factors are not usable in design.
Therefore, E. Orowan proposed the following approach:
Combining Eqs. 7.10 and 7.6 (simplifying per Eq. 7.11) gives:
Eγ s c
σ max = σ yy ⇒ = σ 2 (7.12)
ao ρ
If fracture occurs at an atomistically-sharp ellipse tip, then let ρ ⇒ ao and the
remote stress required to cause brittle fracture is:
Eγ s
σf = (7.13)
4c
7.4
A. A. Griffith proposed a different solution by assuming neither a crack shape or a
stress solution. Instead, Griffith stated that the criterion for brittle fracture is:
"A crack will propagate when the decrease in elastic strain energy is as least equal
to the energy required to create new crack surfaces."
The starting point for Griffith's solution is to evaluate the strain energy in the volume
surrounding the area in a plate (see Fig. 7.4) of thickness, t, and width, >>>W≈ ∞ , under
uniform uniaxial tension, σ, into which a crack could be introduced. For a crack of length,
2c, the stored elastic strain energy is:
1
UE = σεV =
2
σ2
2E
2
( )
πc t ≈
σ2
E
πc 2t ( ) (7.14)
The resistance to fracture is related to the energy required to create new fracture
surfaces, Us, which is a function of the fracture surface energy, γ s, and the surface area of
the crack, A=2 (2ct), such that:
Us = γ s A = γ s (2(2ct)) = γ s (4ct) (7.15)
The change in stored energy from the uncracked panel to the cracked panel is:
σ2
∆U = U s − U E = γ s A = γ s (4 ct) −
E
( )
πc 2 t (17.16)
>>>W
t
2c
σ
Figure 7.4 Uniformly stress plate of thickness, t, and width, W, into which a crack of length,
2c, will be introduced.
7.5
The three energy terms, Us, UE and ∆ U, are plotted as functions of half crack
length, c, in Fig. 7.5. It is apparent that the critical condition for unstable, brittle fracture is
d∆U
reached when c=cc and =0 at the peak of the ∆ U curve. Differentiating Eq. 17.6,
dc
setting it equal to zero, and solving for the stress at which brittle fracture occurs results in
the following:
2γ s E
Plane stress (σ z = 0) σ f =
πc
(17.17)
2γ s E
Plane strain (εz = 0) σ f =
(1− ν 2 )πc
It is interesting to compare Eqs. 17.13 and 17.17 and note that they only differ by
factors of 1/4 and 2/π under the radical, respectively. However, from a physical standpoint
Eq 17.17 is more "satisfying" because it is based purely on an energy balance, making no
assumptions about crack shapes or stress solutions.
Although Eq. 17.17 gives a physical relation for the critical fracture stress, it has
little engineering application because it is specific to the case of uniform stress and infinite
relative dimensions. G. Irwin attempted to place an engineering sense on the
understanding of crack/structure/material interactions. The resulting discipline is known
as Fracture Mechanics, and when dealing with non plastic situations, Linear Elastic
Fracture Mechanics (LEFM).
Us
∆U
cc Crack Length, c
Ue
Figure 7.5 Graphical representation of the Griffith energy balance
7.6
Irwin used the Griffith approach as the basis for his subsequent development, first
denoting a new engineering term, G , as the strain energy release rate and crack driving
force:
dU dU
G= ≈ (7.18)
dA tda
where U is the potential energy of the cracked panel, a is now defined as the half crack
length (note, a=c), and t is the plate thickness. In this context, G ≈2γ s such that the Griffith
relation for fracture stress is written in LEFM as:
GE
σf = (7.19)
πa
The utility of G for engineering purposes can be illustrated by an example of a
center cracked panel subjected to a uniform, uniaxial tensile stress (see Fig. 7.6). In this
case, it can be shown that:
G = P 2
1 dC
Fixed load for t = 1
2 da
(7.20)
1 2 dC
Fixed displacement G = - P for t = 1
2 da
dC δ
where is the compliance change (compliance is C = ) with crack extension, da, P is
da P
the applied load, and δ is the displacement. For a known geometry, the dimensionless
compliance, EBC (E is elastic modulus and B=t is the specimen thickness) is plotted
versus dimensionless crack length, a/W (W is the specimen width). At the critical
condition when the plate fractures, P=Pc and the dimensionless compliance is
δ
EBCc = EB c (see Fig. 7.6) From the master compliance curve for the particular
Pc
dCc
geometry, is determined and the critical G at fracture is:
dac
1 dC
G c = Pc2 c (7.21)
2 dac
Although Eq. 7.21 gives a readily accessible engineering quantity for critical
fracture, it still lacks utility for general applicability to engineering design. An alternative
approach is to look at the stress at the crack tip. The near field stresses for a through
crack in an infinite plate, subjected to a uniform, uniaxial tensile stress are (see Fig. 7.7):
7.7
P Pc
P, δ
W δc
B=t
δ
Test Results
EBC
2a
EBCc _dC
_c
EBW
da c
a /W
c
a/W
P, δ
Master Compliance Curve
Figure 7.6 Illustration of method for determining critical strain energy release rate
θ θ
cos 1-sin sin +.....
a 3θ
σ xx = σ
2r 2 2 2
θ θ
cos 1+sin sin +.....
a 3θ
σ yy = σ
2r 2 2 2
a θ θ 3θ
σ xy = τ xy = σ cos sin cos +.....
2r 2 2 2
(7.22)
σ zz = 0 and τ yz = τ zx = 0 for plane stress
or
σ zz = ν(σ xx + σ yy ) for plane strain
Note in Eqs. 7.22 that as r ⇒ 0 , the stresses predicted by the equations, all become
infinite, thus rendering the equations unusable for calculating the critical stress at the
crack tip for engineering design purposes. Irwin, however, took a different approach and
defined a stress intensity factor, K, which could uniquely define the stress state at the
crack tip, without the need to determine the actual stress such that:
K = σ πa (7.23)
7.8
y σyy σxy
σ
r σxx
θ
2a
Figure 7.7 Near field stress state for a crack in an infinite plate subjected to uniform,
uniaxial tension.
Irwind did not chose the epression for K arbitrarily. Note that K and G are related:
GE GE σ 2πa
If σ f = then σ 2 = and G =
πa πa E
(7.24)
2
K
But, since K = σ πa then G =
E
Again, note that although Eqs. 7.25 still predict infinite stresses at the crack tip, the
stress intensity factor can now be used to define the stress state at the crack tip. From an
engineering standpoint, the stress intensity factor can be used like stress to predict the
critical condition at fracture:
K
Fracture OCCURS if FS ≤ 1 where FS= c (7.26)
K
7.9
where Kc is a material property known as fracture toughness and K is the stress intensity
factor at a particular combination of crack length and stress in the component. Several
things are important in Eq. 7.26.
The first is related to the "mode" of the fracture or loading. There are three modes
designated via Roman numerals as: I (opening), II (sliding) and III (tearing) as illustrated
in Fig. 7.8. Note that the most critical mode is Mode I because the crack tip carries all the
stress whereas in Modes II and III some of the stress is carried by interaction of the
opposing crack faces.
A second point is that the stress intensity factor defined in Eq. 7.23 is for the special
case of idealized crack in an infinite plate. Real cracks are affected by the geometry of the
component, the applied stress field, and other factors. Thus, Eq. 7.23 can be generalized
as
K = Yσ πa = ασ πa = Fσ πa (7.27)
where Y, F and a are geometrical correction factors which maintain the uniqueness of the
stress intensity factor by accounting for the particular geometry. For example, a center
through crack in a plate of finite width W, has a Y of ~1.12. In other words, the stress
intensity factor is about 12% greater in a finite width plate than for an infinitely wide plate.
Geometrical correction factors are found from closed form solutions, finite element
analyses, and experimental methods such as photoelastcity. Many handbooks of stress
intensity factors for common and not so common geometries have been complied. Some
common geometries are shown in Fig. 7.9
MODE I
OPENING MODE II
MODE SLIDING
MODE
MODE III
TEARING
MODE
7.10
Figure 7.9 Common and practical stress intensity factors
7.11
Plane Stress Plane Strain
Fracture Toughness, Fracture Toughness,
Kc K Ic
Bs
Material Thickness, B
Figure 7.10 Fracture toughness as a function of material thickness.
The third point is that the most critical fracture toughness of the material must be
measured so as to be geometry independent. Specifically, the plane stress fracture
toughness is a function of material thickness and is an important consideration when
determining fracture toughness in sheet material. Plane strain fracture toughness is
independent of material thickness and is less than the plane stress fracture toughness as
illustrated in Fig. 7.10.
When conducting fracture toughness tests, especially of metals, three critical
aspects must be addressed if a valid test is to result. The first is a valid stress intensity
factor for the particular geometry being tested. Geometries often used include center
cracked panel, three- or four-point flexure and compact tension (see Fig. 7.9). The
second is that the dimensions are in proper proportion to assure a valid stress intensity
factor and a thick enough specimen for plane strain conditions. ASTM recommends a
thickness such that:
K2 2
B ≥ 2.5 Ic (7.28)
Sys
where Sys is the yield strength of the material and KIc is the plane strain fracture
toughness. Finally, a plasticity requirement must be met where for a valid test:
Pmax
≤ 1.1 (7.29)
PQ
where Pmax is the maximum load recorded during the fracture test and PQ is the load at
which the load-displacement curve deviates from linearity.
Various trends in fracture toughness are related to material type. For example, as
yield point increases, fracture toughness generally decreases since yield points close to
the ultimate tensile strength of the material indicate low ductility and tendency to brittle
fracture (see Fig. 7.11)
7.12
Yield Strength, Syp
Figure 7.11 General relation between fracture toughness and yield strength
The type of atomic bonding can indicate tendency to brittle fracture (see Fig. 7.12).
In general, covalent and ionic bonds as are found in ceramics, glasses and polymers
usually leas to lower fracture toughness. Metallic bonds usually lead to higher fracture
toughness. Some materials, such as composites, may have relatively high apparent
fracture toughness values although the applicability of LEFM to these materials is
debatable.
Design philosophies can take several forms, all based on the basic relation that
fracture will occur when the stress intensity factor in the component is equal to the fracture
toughness of the material such that:
20-200
MPa√m
10-100
MPa√m
0-10
MPa√m 1-5
MPa√m
7.13
For example, in the design of a nuclear pressure vessel:
a) Material is chosen for certain properties (corrosion resistance, etc.).
This fixes KIc.
b) In the component, allow for the presence of large flaw since these are
detectable and correctable. This fixes Y and ac.
c) Design for the allowable stress, σ, to accommodate K Ic, Y and ac.
In a different example, an aerospace application:
a) Material is chosen for certain properties (high yield strength, low density,
etc). This fixes KIc.
b) In the component, fix the design stress, σ, for high performance or high
payload to weight ratio, etc. This fixes σ.
c) Use nondestructive testing to find a before it reaches critical Y and ac.
Another example is the clever use of cracks to provide a warning before
catastrophic fracture can occur. In the leak before break philosophy, the thickness of a
pressure vessel is chosen so that a crack will penetrate the wall before it can reach a
critical size to cause brittle fracture. In this way, the presence of the crack can be easily
spotted without expensive (and often unreliable) non destructive test methods.
7.14
8. TIME DEPENDENT BEHAVIOUR: CREEP
where D is the diffusion rate, Do is a constant, Q is the activation energy for atomic motion,
R is the universal gas constant (8.314J/mole K) and T is the absolute temperature. Thus,
diffusion-controlled mechanisms will have significant effect on high temperature
mechanical properties and performances. For example, dislocation climb, concentration
of vacancies, new slip systems, and grain boundary sliding all are diffusion-controlled and
will affect the behaviour of materials at high temperatures. In addition, corrosion or
oxidation mechanisms, which are diffusion-rate dependent, will have an effect on the life
time of materials at high temperatures.
Creep is a performance-based behaviour since it is not an intrinsic materials
response. Furthermore, creepis highly dependent on environment including temperature
and ambient conditions. Creep can be defined as time-dependent deformation at
absolute temperatures greater than one half the absolute melting. This relative
T (abs )
temperature ( ) is know as the homologous temperate. Creep is a relative
Tmp (abs )
phenomenon which may occur at temperatures not normally considered "high." Several
examples illustrate this point.
8.1
Creep Stress Rupture
Strain
T/Tmp >0.5
Conceptually a creep test is rather simple: Apply a force to a test specimen and
measure its dimensional change over time with exposure to a relatively high temperature.
If a creep test is carried to its conclusion (that is, fracture of the test specimen), often
without precise measurement of its dimensional change, then this is called a stress
rupture test (see Fig 8.1). Although conceptually quite simple, creep tests in practice are
more complicated. Temperature control is critical (fluctuation must be kept to <0.1 to
0.5°C). Resolution and stability of the extensometer is an important concern (for low
8.2
creeping materials, displacement resolution must be on the order of 0.5 µm).
Environmental effects can complicate creep tests by causing premature failures unrelated
to elongation and thus must either mimic the actual use conditions or be controlled to
isolate the failures to creep mechanisms. Uniformity of the applied stress is critical if the
creep tests are to interpreted. Figure 8.2 shows a typical creep testing setup.
The basic results of a creep test are the strain versus time curve shown
σ
schematically in Fig. 8.3. The initial strain, ε i = i , is simply the elastic response to the
E
applied load (stress). The strain itself is usually calculated as the engineering strain,
∆L
ε = . The primary region (I) is characterized by transient creep with decreasing creep
Lo
dε
strain rate ( = ε˙ ) due to the creep resistance of the material increasing by virtue of
dt
material deformation. The secondary region (II) is characterized by steady state creep
(creep strain rate, ε˙min = ε˙ ss , is constant) in which competing mechanisms of strain
hardening and recovery may be present. The tertiary region (III) is characterized by
increasing creep strain rate in which necking under constant load or consolidation of
failure mechanism occur prior to failure of the test piece. Sometimes quaternary regions
are included in the anlaysis of the strain-time curve as well, although these regions are
very specific and of very short duration.
8.3
Constant Load
Constant Stress
dε
dt
Primary Secondary Tertiary
I II III
Time, t tf
where β is a constant for transient creep and k is related to the constant strain rate. A
"better" fit is obtained by:
ε = ε i + εt (1− exp(rt)) + tε˙ss (8.3)
where r is a constant, εt is the strain at the transition from primary to secondary creep and
ε˙ss is the steady-state strain rate. Although no generally-accepted forms of nonlinear
strain-time relations have been developed, one such relations is:
8.4
σ3 > σ >σ Τ3 > Τ >Τ
2 1 σ >σ 2 1 Τ >Τ
2 1 2 1
σ Τ
1 1
Time, t Time, t
ε = ε i +Bσ mt + Dσ α (8.5)
The steady state or minimum strain rate is often used as a design tool. For
example, what is the stress needed to produce a minimum strain rate of 10-6 m/m / h ( or
10 -2 m/m in 10,000 h) or what is the stress needed to produce a minimum strain rate of 10 -
7 m/m / h ( or 10-2 m/m in 100,000 h). An Arrhenius-type rate model is used to include the
where n is the stress exponent, Q is the activation energy for creep, R is the universal gas
constant and T is the absolute temperature.
To determine the various constants in Eq. 8.7 a series of isothermal and iso stress
tests are required. For isothermal tests, the exponential function of Eq. 8.7 becomes a
constant resulting in
ε˙ ss = ε˙min = Bσ n (8.8)
Equation 8.8 can be linearized by taking logarithms of both sides such that
logε˙ss = log ε˙min = log B + n log σ (8.9)
8.5
= + +
εi εi
Time, t Time, t Time, t Time, t
Log-log plots of ε˙min = ε˙ ss versus σ (see Fig. 8.6) often results in a bilinear
relation in which the slope, n, at low stresses is equal to one indicating pure diffusion
creep and n at higher stresses is greater than one indicating power law creep with
mechanisms other than pure diffusion (e.g., grain boundary sliding).
For iso stress tests, the power dependence of stress becomes a constant resulting
in
ε˙ ss = ε˙min = C exp
−Q
(8.10)
RT
Equation 8.10 can be linearized by taking natural logarithms of both sides such that
Q1
lnε˙ss = ln ε˙min = ln C − (8.11)
RT
1
Log-linear plots of ε˙min = ε˙ ss versus (see Fig. 8.7) results in a linear relation in which
T
−Q
the slope, , is related to the activation energy, Q, for creep.
R
log σ
Figure 8.6 Log-log plot of minimum creep strain rate versus applied stress showing
diffusion creep and power law creep.
8.6
. -Q/R
1/T
Figure 8.7 Log-linear plot of minimum creep strain rate versus reciprocal of temperature
showing determination of activation energy.
The goal in engineering design for creep is to predict the behaviour over the long
term. To this end there are three key methods: stress-rupture, minimum strain rate vs. time
to failure, and temperature compensated time. No matter which method is used, two
important rules of thumb must be borne in mind: 1) test time must be at least 10% of
design time and 2) creep and/or failure mechanism must not change with time,
temperature or stress.
Stress-rupture This is the "brute force method" is which a large number of tests are
run at various stresses and temperatures to develop plots of applied stress vs. time to
failure as shown in Fig. 8.8. While it is relatively easy to use these plots to provide
estimates of stress rupture life within the range of stresses and lives covered by the test
data, extrapolation of the data can be problematic when the failure mechanism changes
as a function of time or stress as shown by the "knee" in Fig. 8.8.
Minimum strain rate vs. time to failure This type of relation is based on the
observation that strain is the macroscopic manifestation of the cumulative creep damage.
As such, it is implied that failure will occur when the damage in the material in form of
creep cavities and cracks resulting from coalesced creep cavities reaches a critical level.
This critical level of damage is manifested as the failure which can be predicted from the
minimum strain rate and the time to failure such that.
ε˙min tf = C ≈ εf (8.12)
8.7
Τ1
Stress, σ Τ2 >Τ1
or log σ
Τ3 > Τ2 > Τ1
Change in
failure
mechanism
Time to failure, tf
or log tf
Figure 8.8 Stress rupture plots for various temperatures
8.8
σ1 < σ2 < σ3
Allowable
log t f Q log e σ
R
1/T
The design problem is to determine the allowable stress to give 2000 h life at 200°C. For
this alloy, the activation energy, Q, is 150.5 kJ/mole. Using Q=150,500 J/mole, R=8.314
J/mole K, tf=2000 h, and T=473 K, P SD is calculated as -13.21. Substituting this value of
PSD into Eq. 8.15 gives an allowable stress of 25 MPa.
σ 1< σ2 < σ3
Allowable
log t f σ
1/T
Q log e
-C
R P = T (log t f - C)
LM
8.9
9. TIME DEPENDENT BEHAVIOUR: CYCLIC FATIGUE
9.1
Fatigue cracks, then, begin at a point of high-stress concentration. The severity of
those stress concentrations will vary, even with carefully prepared laboratory specimens.
In addition, the rapidity with which the crack propagates will vary. The cracks are irregular
and will follow various paths around regions of stronger metal. A consequence of this
manner of crack propagation is wide variation in time to failure of a number of seemingly
identical test specimens loaded with the same load. For this reason a number of
statistical procedures have been developed for interpreting fatigue data.
The basic mechanism of a high-cycle fatigue failure is that of a slowly spreading
crack that extends with each cycle of applied stress. In order for a crack to propagate, the
stress across it must be tension; a compression stress will simply close the crack and
cause no damage. One way, then, of preventing fatigue or at least extending the fatigue
life of parts is to reduce or eliminate the tensile stresses that occur during loading by
creating a constant compressive surface stress, called a residual stress, in the outer
layers of the specimen. We can picture how it is possible to induce a constant
compressive stress of this type by a simple analogy. Consider a bar with a small slot cut
in the surface. If we force a wedge tightly into this slot, the wedge and the material in the
bar itself adjacent to the slot will be in a compressive stress state. If a tensile stress is now
applied to the bar, no tensile stress can exist in the region of the slot until the compressive
stress caused by the wedge is overcome. In other words, the tensile stress in the region
of the slot will always be less than in a region removed from the slot by an amount equal
to the magnitude of the residual compressive stress.
Favorable compressive stresses can be induced in parts by more practical
methods than cutting slots. One of the most common methods is that of shot peening.
Shot peening is a process in which the surface of the part is impacted by many small steel
balls moving at high velocity. This process plastically deforms the surface of the part and
actually tends to make it somewhat bigger than it was. The effect is the same as the
wedge. A now larger surface is forced to exist on a smaller sublayer of material with the
net result that a compressive stress is induced in the outer layer making it more resistant
to fatigue failure.
9.2
Figure 9.2 One type of rotating beam reversed stress testing machine (a.k.a. R.R. Moore
rotating bending fatigue test machine)
σm
σ max
σ ∆σ
σa t
σmin
Figure 9.3 Examples of stress cycles for completely-reversed and tension-tension loading
9.3
σ max = Maximum stress
σ min = Minimum stress
σ + σ min
σ m = Mean stress = max
2
∆σ = Stress range = σ max − σmin
∆σ
σ a = Stress amplitude = = (σ max − σm ) = (σm − σmin )
2
Note: tension = + σ and compression = − σ. Completely reversed R= −1, σm = 0.
σ
R = Stress ratio = min
σ max
σ a 1− R
A = Amplitude ratio = =
σ m 1+ R
Some test machines turn at 10,000 RPM. and are equipped with a counter that
counts once for each 1000 revolutions. The machines are also equipped with an
automatic cutout switch which immediately stops the motor when the specimen fails.
Fatigue test data are obtained in a number of ways. The most common is to test a
number of different specimens by determining the time to failure for a specimen when
stressed at a certain stress level. Figure 9.5 shows schematic representation of these test
results on a type of graph called an S-N curve which shows the relationship between the
number of cycles N for fracture and the maximum (or mean or amplitude or range) value
of the applied cyclic stress. Generally, the abscissa is the logarithm of N, the number of
cycles, while the vertical axis may be either the stress S or the logarithm of S. Typical test
data are shown plotted in Fig. 9.6. in which the data can sometimes be represented by
two straight lines.
Figure 9.5 Schematic representation of S-N curves for ferrous and non ferrous materials.
9.4
The fatigue strength is then the stress required to fracture for a set number of
cycles. The usual trend is for the fatigue strength to decrease rapidly in the cycle range of
10 3 to 106 . In the range of 105 to 106 the curve flattens out and for some materials such
as steel actually becomes flat, indicating this material will last "forever" in fatigue loading,
if the applied stress is kept below a certain value. The stress corresponding to failure at
an infinite number of cycles is called the endurance limit (or endurance limit) of the
material and is often designated as Se or σe. It should be recalled, in light of the previous
discussion, that a considerable amount of scatter will exist and values such as endurance
limit will be only approximate when a small number of test specimens are used.
Figure 9.6 Typical S-N curves for determining endurance limits of selected materials
under completely reversed bending.
9.5
Many times, in preliminary design work, it is necessary to approximate the S-N
curve without actually running a fatigue test. For steel it has been found that a good
approximation of the S-N curve can be drawn if the following rules are used.
1. Obtain the ultimate tensile strength Smax of the specimen ( σ max in a simple
tension test).
3. Join these points together to form a S-N diagram similar to that shown in Figs
9.5 or 9.6
Many factors affect fatigue failures. Not only do fatigue failures initiate at surfaces,
but stress raisers create stresses which are greatest at surfaces. Fatigue factors
Recall the stress concentration factor:
σ
kt = LOCAL (9.1)
σ REMOTE
where σ LOCAL is the maximum local stress at the stress raiser and σ REMOTE is the remote or
net stress. The effect of the stress raiser on fatigue strength can be evaluated by first
defining a fatigue strength reduction factor:
σ eUN − NOTCHED
kf = (9.2)
σeNOTCHED
where σ eUN − NOTCHED and σ eNOTCHED are the endurance limits for un-notched and notched
fatigue specimens as illustrated in Fig. 9.7.
S σeUN-NOTCHED
σe NOTCHED
log N f
Figure 9.7 Illustration of endurance limits for notched and un-notched fatigue tests.
9.6
A notch sensitivity factor can be defined which relates the material behavior, kf, and
the component parameter, kt such that
k −1
q= f (9.3)
kt − 1
where q=0 for no notch sensitivity, q=1 for full sensitivity. Note that q increases as notch
radius, r, increases and q increases as SUTS increases
Generally, kf<< kt for ductile materials and sharp notches but kf≈kt for brittle
materials and blunt notches. This is due to i) steeper dσ/dx for a sharp notch so that the
average stress in the fatigue process zone is greater for the blunt notch, ii) volume effect
of fatigue which is tied to average stress over a larger volume for blunt notch, iii) crack
cannot propagate far from a sharp notch because steep stress gradient lowers KI quickly.
In design, avoid some types of notches, rough surfaces, and certain types of loading. As
mentioned, compressive residual stresses at surfaces (from shot peening, surface rolling,
etc.) can increase fatigue lives.
Endurance limit, Se =σe , is also lowered by factors such as surface finish (ma), type
of loading (mt), size of specimen (m d ), miscellaneous effects (mo ) such that:
σe
σa
Goodman
Soderberg
σm S
ys
Suts
9.7
The mathematical expression for this effect is the equation of the line such that:
σ
σ a = σ e1 − m (9.5)
SUTS
which is known as the Goodman line. If SUTS is replaced with Sys, then Eq. 9.5 is known
as the Soderberg relation.
If a factor of safety, FS, and / or fatigue factors, kf, are used then the following
expressions results for either brittle or ductile materials.
σe σm
Brittle σa = 1 − (9.6a)
FS •k f (SUTS /(k f ≈ k t )FS)
σe σm
Ductile σa = 1 − (9.6b)
FS •k f (SUTS / FS)
N1 N2 N3 N
+ + = ∑ j =1 (9.7)
Nf1 Nf2 Nf3 Nfj
where N1 and N f1 , etc. are the actual number of cycles and number of S-N fatigue cycles
at stress, σa1 etc. (see Fig. 9.10) respectively.
σa3
σ a2
σa1
σ t
N1 N2 N3
9.8
σa3
σa σa2
σ a1
N N
f 3 f 2 N f1 Nf
Figure 9.10 Illustration of cycles to failure on S-N curve for different stress amplitudes
S-N curves are generally used to design for infinite life in fatigue. That is, the
fatigue data are used to chose stresses such that fatigue cracks will never develop.
However, the fatigue mechanism is such that is possible to analyze the growth of fatigue
cracks using linear elastic fracture mechanics. This is useful in extending the service life
of components when a fatigue crack is noticed, rather than discarding a part which might
still have useful life.
The fatigue process (see Fig. 9.11) consists of 1) crack initiation, 2) slip band crack
growth (stage I crack propagation) 3) crack growth on planes of high tensile stress (stage
II crack propagation) and 4) ultimate failure. Fatigue cracks initiate at free surfaces
(external or internal) and initially consist of slip band extrusions and intrusions as
illustrated in Figure 9.12. When the number of slip bands reaches a critical level
(saturation) cracking occurs (see Figure 9.13). These slip band cracks will grow along
directions of maximum shear for only one to two grain diameters (dg ), when the crack
begins growing along a direction normal to maximum tensile stress. Although, fatigue
striations (beach marks) on fracture surfaces represent successive crack extensions
normal to tensile stresses where one mark is approximately equal to one cycle (i.e. 1
mark≈1N), beach marks only represent fatigue cycles during crack propagation and do
not represent the number of cycles required to initiate the crack. Thus, summing the
beach marks does not represent the total number of fatigue cycles.
9.9
Slip Band Tensile Stress
Crack Propagation Crack Propagation
Slip Band
Crack Initiation
1-2 dg
During fatigue crack propagation (stage II may dominate as shown in Fig 9.14)
such that crack growth analysis can be applied to design: a) cracks are inevitable, b)
minimum detectable crack length can be used to predict total allowable cycles, c) periodic
inspections can be scheduled to monitor and repair growing cracks, d) damage tolerant
design can be applied to allow structural survival in presence of cracks.
The most important advance in understanding fatigue crack propagation was
realizing the dependence of crack propagation on the stress intensity factor. The
mathematical description of this is known as the Paris power law relation:
da
= C(∆K )m (9.8)
dN
where da/dN is the crack propagation rate (see Fig. 9.15), ∆K = F(∆σ ) πa in which F is
the geometry correction factor, a is the crack length, and C and n are material constants
found in stage II of the da/dN vs ∆K curve (see Fig. 9.14)
9.10
Saturation
Cracking
Number of Cycles N
Figure 9.13 Slip band saturation and the onset of fatigue cracking
To predict the crack propagation life of a component, Eq. 9.8 can be rearranged
such that:
Nf af f a
da da
∫N dN = ∫a C(∆K)m a∫ C(F∆σ πa)m
= (9.9)
i i i
Assuming that F can be approximated as nearly constant over the range of crack growth
and assuming that m and C are constant, then:
a(1− − a(1i − (m / 2 ) )
(m /2))
Nf = f
(9.10)
[ ] [1− (m /2) ]
m
C F(∆σ) π
where ai is the initial crack length which is either assumed (~2 dg ) or determined by non
1 KIc
2
I II III
log
da/dN m
∆ Kth log ∆ K
Figure 9.14 Crack growth rate versus stress intensity factor range.
9.11
a
da/dN
Figure 9.15 Crack length vs. number cycles relations from which crack propagation rate is
found.
If F is a function of crack length, i.e. F(a,W, etc.), then numerical integration must be
used such that:
.
Nf af af
da da
∫N dN = a∫ C(∆K)m = a∫ C F(a,W,etc)∆σ πa m (9.11)
i i i[ ]
Crack propagation rates are also highly sensitive to R ratios primarily because
crack propagation only occurs during tensile loading. Thus, the longer the crack stays
open (i.e. R ≥0) the more time out of each cycle that crack propagation occurs. Thus, the
more negative R, the more tolerant the crack is of ∆K and vice versa (see Fig. 9.16)
log ∆ K
Figure 9.16 Effect of R-ratio on crack growth rates.
9.12
10. COMPRESSION AND BUCKLING
P P
L F=K∆ x
Figure 10.1 Model of a slender column under axial load with a restoring force/moment
10.1
P
Unstable
Pcr=KL
Stable
−θ 0 +θ
Figure 10.2 Illustration of stable and unstable conditions along with critical load
The ideal column (pinned at both ends as shown in Fig. 10.4) can be modeled
mathematically using the equation of the elastic curve. Note that assumptions for the
ideal column include: i) column is initially perfectly straight, ii) load is applied through the
center of the cross section, iii) material is homogeneous and linear elastic, and iv) column
buckles and bends in a single plane. The elastic curve equation is:
10.2
P P
P
x
M
L
P
v
The smallest value of P (in other words, the value of P reached first when loading from
P=0 to some critical load) is obtained when n=1, so that the critical load for the column
(known as the Euler load) is:
π 2EI
Pcr = (10.8)
L2
10.3
Note that this critical load is independent of the strength of the material, but rather
depends on the column dimensions (actual length, L, and smallest moment of inertia, I)
and material stiffness, E. Note also that load-carrying ability increases as the moment of
inertia increases, as elastic modulus increases but at length decreases. Thus, short, fat
columns made of stiff material will have very low tendency to buckle. A key to
understanding buckling is to create a design whereby the required (i.e., applied) design
load is always much less than the theoretical buckling load. The reasons for this are that
the theoretical buckling load is determined based on an ideal column with many inherent
assumptions. Thus, a large safety margin must be placed between the design load and
the calculated critical buckling load.
Equation 10.8 is the solution for an ideal column with pinned/pinned end
conditions. If other boundary conditions are applied then the following critical load
equations result for the indicated end conditions.
π 2EI
Pcr = Pinned −Pinned
L2
π 2EI
Pcr = Free −Fixed
4L2
4π 2EI (10.9)
Pcr = Fixed −Fixed
L2
2.046π 2EI
Pcr = Pinned − Fixed
L2
The equations in Eq. 10.9 are a bit cumbersome because four different relations
need to be remembered. So instead an effective length is defined such that Le = KL .
Now a single equation can be used to replace the four equations of Eq. 10.9 and effective
length constants, K, for each end condition can be used to define the effective length. The
effective length constants are as shown in Fig. 10.5 and the single equation is:
π 2 EI π 2EI
Pcr = = (10.10)
L2e (KL)2
For purposes of design, it is more useful to express the critical loading condition in
terms of a stress such that:
P π 2EI π2E
σ cr = cr = ⇒ σ cr = (10.11)
A AL2e (Le / k)2
where k= I / A is the smallest radius of gyration determined from the least moment of
inertia, I, for the cross section. The term, Le /k is known as the slenderness ratio and
contains information about the length and the cross section.
10.4
Figure 10.5 Effective length constants for various end conditions
It is interesting to investigate the case where the applied stress, σ , is equal to the
critical buckling stress, σ cr , and the generalized yield strength, σo . At this point there is a
transition between yield and buckling:
π 2E
σ = σ o = σ cr ⇒ σo = (10.12)
(Le / k )2
If Eq. 10.12 is solved for Le /k , the resulting relation marks the combination of
length and cross section at which the compressive behaviour transitions from yielding to
buckling. This relation is known as the minimum slenderness ratio:
Le π 2E
= (10.13)
k min σo
This transition can be illustrated by plotting the relation between stress, σ , and
slenderness ratio, Le /k , as shown in Fig. 10.6. Note that in reality there is no sharply
divided transition between yielding and buckling. Instead the σ vs. Le /k curve can be
divided into three regions as shown in Fig. 10.7. Region I is the short-column region in
which general yielding occurs when σ =σo . Region II is an intermediate-column region
in which i) the column may yield or may buckle and ii) empirical relations are used to
approximate the resulting curve. Region III is for long columns and buckling will occur.
10.5
σo
σ π2 E
σ=
( Le/k )2
Le/k | min
Le/k
Regions I and III are straight forward to determine (either apply σ =σo . in Region
π 2E
I or σ cr = in Region III). Region II is material/ geometry/loading dependent and
(Le /k)2
empirical relations are often used. One approach is to fit a parabola to the σ vs. Le /k
curve from σ =σo . to σ =σo / 2 such that.
(Le / k )
2
L L
σ = σ o 1− 0≤ e ≤ e
2
for
( e min )
2L / k k k min
and (10.14)
π E2
Le Le
σ = for >
(Le / k)2 k k min
σo Parabolic Approximation
Figure 10.7 Stress vs. slenderness ratio relation with parabolic fit in Region II
10.6
In summary, note that buckling loads and stresses are sensitive to the following
1) stiffness of material
2) length of column
3) cross section dimensions
4) cross sectional shape
5) end conditions
6) initial eccentricity
7) eccentric loading
8) end conditions
Thus, the moral of the story is to always keep design loads well below the calculated
critical buckling loads.
10.7
11. Structures: Complex Stresses and Deflections
Engineering structures may take many forms, from the simple shapes of square
cross section beams to the complex and intricate shapes of trusses. Regardless of the
shape another important aspect of structures is that often the stresses, strains, and
deflections of the structure do not lend themselves to simple and straight-forward
analyses of simple components such as those used for materials testing (e.g., the
uniaxially-loaded and uniformly-stressed tensile specimen). Further complicating the
analyses of engineering structures is the need to apply failure criteria to evaluate the
probable success (or non success) of the design.
Failure Criteria
Engineering failure can be broadly defined as the "inability to perform the intended
function." An obvious failure is a broken part (unless of course the intended function is to
fail as in the case of shear pins or explosive bolts!) which is known as fracture. However,
excessive elastic or plastic deformation without fracture can also constitute a failure. In
addition, a component with too much or not enough "give" such as with too compliant or
too stiff of a spring-like component can be a failure. A cracked component such as in a
pressure vessel would constitute a failure if a leak occurred. Thus, failure criteria can be
based on stress, strain, deflection, crack length, time or cycles, or any other engineering
parameter we choose to apply.
The most common failure criteria are stress-based. The basic premise is that
failure will occur in the component or structure when the combined stress state is equal
that which caused failure in the same material subjected to a uniaxial tensile test. Two
primary types of stress-based failure criteria are used: yield (for ductile materials with
%el>5) and fracture (for brittle materials with %el<5). In both cases, one can consider
failure to occur at the onset of non linearity in the tensile stress strain curve (the yield point
in ductile materials and ultimate tensile strength in brittle materials).
Fracture Criterion: The simplest failure criterion is that failure is expected when the
greatest principal stress reaches the uniaxial tensile strength of the material. Thus, the
Maximum Normal Stress fracture criterion (a.k.a., Rankine) can be specified as:
[ ]
Fractures if MAX σ1 , σ 2 , σ 3 ≥ SUTS (11.1)
where the function MAX indicate the greater of the absolute values of the principal normal
stresses. Note that it is assumed that the ultimate strength of the material is the same in
compression or tension.
11.1
σ
σ1=-Suts σ1=Suts
σ2=Suts
"safe"
σ
σ2=-Suts
This fracture criterion can be represented in a plane stress state (σ z =0) where σ 2 is the
ordinate and σ 1 is the abscissa. As shown in Fig. 11.1, any combination of σ 1 and σ 2
that falls within the square box (i.e., FS=1 for Eq. 11.2 where ±SUTS=σ 1 or ±SUTS=σ 2 ) is
"safe" and the perimeter is fracture.
Yield criteria: There are two relatively well-accepted yield criteria: Maximum Shear
Stress criterion (a.k.a., Tresca) and Octahedral Shear Stress criterion (a.k.a., Distortional
Energy or Von Mises). Each is discussed as follows.
The simplest yield criterion is that yield failure is expected when the greatest shear
stress reaches the shear strength of the material. Thus, the maximum shear stress yield
criterion can be specified as:
(σ − σ 2 ) (σ − σ 3 ) (σ − σ 3 ) σo
Yields if MAX τ12 = 1 , τ13 = 1 , τ 23 = 2 ≥ τ = (11.3)
o
2 2 2 2
where the function MAX indicates the greater of the absolute values of the principal shear
stresses.
A factor of safety can be defined based on Eq. 11.3 such that:
Yields if FS ≤ 1
τ o =σ o /2
where FS= (11.4)
(σ − σ 2 ) , τ = (σ1 − σ 3) , τ = (σ 2 − σ 3)
MAX τ 12 = 1
2 13
2 23 2
11.2
σ2
σ1=σο
σ1 − σ2 =−σo σ2 =σο
"safe" σ1
σ1 − σ2 =σo
σ2 =σο
σ1 =σο
Figure 11.2 Failure envelope for maximum stress criterion
This yield criterion can be represented in a plane stress state (σ z =0) where σ 2 is
the ordinate and σ 1 is the abscissa. As shown in Fig. 11.2, any combination of σ 1 and σ 2
that plots within the parallelogram (i.e., FS=1 for Eq. 11.4 where ±σ o =(σ 1-σ 2 ), ±σ o =σ 1
or ±σ o =σ 2 ) is "safe" and the perimeter is yielding.
A more complicated yield criterion is that yield failure is expected when the
octahedral shear stress,τ h , reaches the octahedral shear stress at yield of the material,
τ ho . Thus, the octahedral shear stress yield criterion can be specified as:
Yields if τh ≥ τ ho
(11.5)
where τh =
1
3
(σ 1−σ 2 ) + (σ 2 −σ 3 ) + (σ 3 −σ 1)
2 2 2
and
2
τ ho = σ (11.6)
3 o
when the stress state of a uniaxial tensile test at yielding (σ 1=σ o , σ 2 =σ 3 =0) are
substituted into the relation for τ h given in Eq. 11.5. If Eq 11. 5 and 11.6 are set equal to
each other the yield criterion can be expressed in terms of normal stresses:
1
(σ 1 −σ 2 )2 + (σ 2 −σ 3 ) + (σ 3 −σ 1) 2
2 2
Yields if τh ≥ τ ho ⇒ ≥ σ (11.7)
3 3 o
11.3
σ2 2
σo σο2=σ12− σ1σ2 +σ2
σο
"safe" σο σ1
σο
Figure 11.3 Failure envelope for maximum stress criterion
Yields if FS ≤ 1
σo σo
where FS= = (11.8)
σH
1
2
(σ 1−σ 2 ) + (σ 2 −σ 3 ) + (σ 3 −σ 1 )
2 2 2
where σ H is the effective stress based on the octahedral shear stress criterion.
This yield criterion can be represented in a plane stress state (σ z =0) where σ 2 is
the ordinate and σ 1 is the abscissa. As shown in Fig. 11.3, any combination of σ 1 and σ 2
that plots within the ellipse (i.e., FS=1 for σo2 = σ12 − σ 1σ 2 + σ 22 is "safe" and the
perimeter is yielding.
The usefulness of the three failure criteria presented here is shown in Fig 11.4 for
the failure envelopes for the plane stress case where σ 1 and σ 2 are normalized to SUTS
or σ o . Note for the brittle material (cast iron) that the actual failure points follow the
maximum normal stress criterion envelope (i.e., FS=1) and for the ductile materials (steels
and aluminums) that the actual failure points fall between the maximum shear stress and
octahedral shear stress criteria envelopes (i.e., FS=1). Since the maximum difference
between the two yield criteria is about 15%, it is often advisable to err on the side of
conservatism and use the simpler maximum shear stress criterion for ductile materials.
11.4
Figure 11.4 Failure criteria and failure points plotted on normalized plane stress
coordinates
Combined Stresses
The previous discussion of failure criteria was based on two premises: 1) the
uniaxial tensile behavior of the material was known (i.e., SUTS or σ o .) and 2) the principal
stresses based on all the coordinate stresses was known.
When determining the mechanical properties and performance of a material, such
as its yield strength, it is desirable to choose a fundamental test that will give the required
property in the most direct manner. Thus, for yield strength, a simple one-dimensional
tensile test specimen is usually used. Figure 11.5 illustrates the fundamental tensile test
and shows a free body diagram of an infinitesimal element with the one-dimensional
stress σ z acting on it
On the other hand, in a realistic situation, the engineer is usually faced with a two-
or three-dimension load condition in which, at any point, P, the loaded member may be
subject to a combination of tension, compression and shear stresses as idealized in Fig.
11.6.
11.5
Figure 11.6. General three-dimensional stress state
Figure 11.6 shows an arbitrary body loaded with forces P and moments M. At a
point O, shown enlarged at the right, an infinitesimal three-dimensional element can be
acted upon by normal stresses σ x , σ y and σ z acting in the x, y and z directions, as
shown. Often, the stresses in the z directional are zero, or much smaller than σ x or σ y .
This is a condition called plane stress, and the analysis is simplified.
Thus, in order to determine the margin of safety of a loaded structure, it is
necessary to relate the two- or three-dimensional stress state that usually occurs, with the
fundamental strength, like tensile yield strength, that is obtained in the laboratory. This
relationship is defined through the previously-discussed failure criteria of which there are
a number in addition to those already discussed.
Prerequisite to applying a failure criterion, is to deduce from the general two- or
three-dimensional element in which both shear and normal stresses are present, the
principal stresses σ 1 and σ 2 and the maximum shear stresses τ max . These stresses can
be determined either analytically or experimentally.
The analytical calculation of principal stress and maximum shear stress involves
the superposition of normal and shear stress to determine the total stress acting at a
critical point. Thus, normal stresses are computed from P/A for simple tension and Mc/I for
bending. Shear stresses are computed from Tr/J for torsion and VQ/It for direct shear. For
thin-walled, pressurized cylindrical pressure vessels the hoop stress is given by pr/t and
the axial stress by pr/2t.
11.6
Figure 11.7 A general two-dimensional stress state and stresses resulting from
element rotation.
Since the final stress state is independent of the order in which the loads are
applied, the total stress existing at a point with a combined loading consisting, perhaps, of
tension, torsion, pressure and shear can be found by simply adding normal stresses
together and shear stresses together taking proper account of their sign. Finally, after
considering the stresses caused by each load, a two-dimensional element at the point
considered may appear as shown in Figure 11.7.
Then an application of Mohr's circle of stress will give the principal stresses σ 1 and
σ 2 and the maximum shear stress τ max . Stress is the quantity that causes failure, yet we
realize from earlier work, that one cannot measure stress directly. This is because stress
is related to the force in a part, which, except in very simple cases, is not easily measured.
On the other hand, strain is easily measured and these values can easily be converted to
stress values.
Calculation of principal strains can be performed as follows. If we consider only a
two-dimensional object, the unknown strains are the normal or elongation strains ε x and
ε y and the shear strain γ xy . Since electrical resistance strain gages can measure only
normal strains (ε ), not shear strains ( γ ), we need to measure three normal strains at a
point to determine the three strains ε x , ε y , and γ xy at a point. This follows directly from
the equations that give strain in a direction "a" ( ε a ), oriented at an angle θ from the x axis,
as shown in Figure 11.8, when strains along the x and y axes, ε x and ε y , and the shear
strain γ xy are known. Thus:
εa =
1
2
( ) 1
2
( )
ε x + ε y + ε x − ε y cos2θ + γ xy sin2θ
1
2
(11.9)
11.7
Figure 11.8 Orientation of "a" direction and the x-y axes
1. Assume we are using a rosette with strain gages oriented along lines a, b, and
c oriented at angles α and β apart, in this case, α = β = 45°. It is desired to
find the maximum strains ε 1 and ε 2 .
2. Along an arbitrary horizontal axis x - x, lay off three vertical lines a, b, and c
corresponding to the three measured strains ε a . ε b , ε c measured from the y
axis. The y - y axis will be the γ 2 shear strain axis.
11.8
Figure 11.9 Mohr's strain circle as determined graphically from strain valves
ε a . ε b , and ε c .
3. Let the b gage direction lie between the a and c directions as shown. Then
from a convenient point F on line b, lay off lines whose directions correspond
to the direction of the gauging lines of the rosette, maintaining the same
directional sense as in the rosette. These lines intersect the lines a and c at
points A and C.
6. From points C, B and A, draw radii to O. Draw the strain axis ε horizontal
through O. These radii will have the same angular orientation sense as the
corresponding gauging lines of the rosette; the angle between the radii will be
twice the actual angle between the gages.
7. The point A, B, and C on the circle give the values of ε and γ 2 for the three
gages.
8. Values of principal strains are determined by the intersection of the circle and
the ε axis. The angular orientation of ε 1 from gage A is shown as 2θ .
11.9
Thus this simple graphical technique results in a Mohr's circle of strain. Strain
values at any angular orientation can be found. Once the principal strains are found the
principal stresses follow directly from the Hooke's relations, considering Poisson's effect:
σ 1 νσ 2
ε1 = − (11.10)
E E
σ 2 νσ 1
ε2 = − (11.11)
E E
σ1 =
(ε1 +νε 2 )E
(1 −ν 2 )
(11.12)
σ2 =
(ε 2 +νε 1)E
(1−ν 2 )
(11.13)
11.10
Figure 11.10 Relatively simple engineering component subjected to a complex loading
condition
In general, truss members are slender and can support little lateral load.
Therefore, major loads must be applied to the various joints and not the members
themselves. Often the weights of truss members are assumed to be applied only at the
joints (half the weight at each joint). In addition, even though the joints are actually rivets
or welds, it is customary to assume that the truss members are pinned together (i.e., the
force acting at the end of each truss member is a single force with no couple). Each truss
member may then be treated as a two force member and the entire truss is treated as a
group of pins and two-force members.
Truss Members
Joints
Figure 11.11 Example of a Simple Truss
11.11
Y
Z X
11.12
2P
P L
B
A
√2 L y
L
L
C D x
L
CD D
P
∑ F = 0 ⇒ ∑ F x = 0 = −FCD ⇒ FCD = 0 →
(11.17)
∑ F y = 0 = P +FBD ⇒ FBD = P ↑
and since F BD pulls on the joint, then the joint must pull back on the member so member
BD is in tension.
2P
P L
A B
√2 L L
L
C
D
P L P
3P
11.13
Using the method of joints, ∑ F =0 at joint C such that
AC
√2
1
1
CD
P C 1
∑ F = 0 ⇒ ∑ Fx = 0 = P −
F ⇒ FBC = 2 P ←
3P 2 BC
(11.18)
1
∑ Fy = 0 = 3P − F − FAC ⇒ FAC = P ↓
2 BC
Since FAC pushes on the joint, then the joint must push back on the member so member
AC is in compression. Furthermore, since FCB pushes on the joint, then the joint must
push back on the member so member CB is in compression.
AB
P A
11.14
For axially loaded members, the displacement is:
N N
∆ N = ∫ U L dx (11.20)
EA
where NU is the axial force in the member due to a unit load applied at the point and
direction of interest, NL is the actual force in the member due to the actual applied load on
the structure, E and A are the elastic modulus and cross sectional area of the individual
member. The integral sign signifies that the calculated quantities for each member are
summed via integration to give the final total deflection at the point and direction of
interest.
For members subjected to bending moments, the displacement is:
MU M L
∆M = ∫ EI
dx (11.21)
where MU is the bending moment in the member due to a unit load applied at the point
and direction of interest, ML is the actual bending moment in the member due to the actual
applied load on the structure, E and I are the elastic modulus and cross sectional moment
of inertia of the individual member. The integral sign signifies that the calculated
quantities for each member are summed via integration to give the final total deflection at
the point and direction of interest.
For members subjected to torsion, the displacement is:
T T
∆ T = ∫ U L dx (11.22)
GJ
where T U is the torque in the member due to a unit load applied at the point and direction
of interest, TL is the actual torque in the member due to the actual applied load on the
structure, G and J are the shear modulus and polar moment of inertia of the individual
member. The integral sign signifies that the calculated quantities for each member are
summed via integration to give the final total deflection at the point and direction of
interest.
For members subjected to transverse shear, the displacement is:
VU VL
∆v = ∫ GA
dx (11.23)
where VU is the transverse shear in the member due to a unit load applied at the point
and direction of interest, VL is the actual transverse shear in the member due to the actual
applied load on the structure, G and A are the shear modulus and cross sectional area of
the individual member. The integral sign signifies that the calculated quantities for each
11.15
member are summed via integration to give the final total deflection at the point and
direction of interest.
The total deflection due to each of these contributions can then be found by adding
the individual contribution such that
N N M M V V T T
∆ t = ∫ U L dx + ∫ U L dx + ∫ U L dx + ∫ U L dx (11.24)
EA EI GA GJ
An example of the unit load method applied to the simple truss example is shown
in Fig. 11.14 in which only the axial loading contributions are required since truss
members are pinned and no bending moments, transverse shear, or torque can be
carried in the members.
11.16
Figure 11.14 Example of application of unit load method to find a deflection in
a simple truss
11.17
12. Pressure Vessels: Combined Stresses
Cylindrical Vessels: A cylindrical pressure with wall thickness, t, and inner radius,
r, is considered, (see Figure 12.1). A gauge pressure , p, exists within the vessel by the
working fluid (gas or liquid). For an element sufficiently removed from the ends of the
cylinder and oriented as shown in Figure 12.1, two types of normal stresses are
generated: hoop, σ h , and axial, σ a , that both exhibit tension of the material.
12.1
Figure 12.1 Cylindrical Thin-Walled Pressure Vessel
For the hoop stress, consider the pressure vessel section by planes sectioned by planes
a, b, and c for Figure 12.2. A free body diagram of a half segment along with the
pressurized working fluid is shown in Fig. 12.3 Note that only the loading in the x-
direction is shown and that the internal reactions in the material are due to hoop stress
acting on incremental areas, A, produced by the pressure acting on projected area, Ap.
For equilibrium in the x-direction we sum forces on the incremental segment of width dy to
be equal to zero such that:
∑F x =0
2[σ h A] − pA p = 0 = 2[σ h t dy] − p 2r dy
or solving for σ h (12.1)
pr
σh =
t
where dy = incremental length, t = wall thickness, r = inner radius, p = gauge pressure,
and σ h is the hoop stress.
Figure 12.2 Cylindrical Thin-Walled Pressure Vessel Showing Coordinate Axes and
Cutting Planes (a, b, and c)
12.2
Figure 12.3 Free-Body Diagram of Segment of Cylindrical Thin-Walled Pressure Vessel
Showing Pressure and Internal Hoop Stresses
For the axial stress, consider the left portion of section b of the cylindrical pressure
vessel shown in Figure 12.2. A free body diagram of a half segment along with the
pressurized working fluid is shown in Fig. 12.4 Note that the axial stress acts uniformly
throughout the wall and the pressure acts on the endcap of the cylinder. For equilibrium
in the y-direction we sum forces such that:
∑ Fy = 0
σ a A − pAe = 0 = σ a π ( ro2 − r 2 ) − p πr 2
or solving for σ a
p πr 2
σa =
π ( ro2 − r 2 )
substituting ro = r + t gives (12.2)
p πr 2 p πr 2 p r2
σa = = =
(
π [r + t] − r 2
2
) π (r 2
+ 2 rt + t 2 − r 2 ) (2rt + t )
2
since this is a thin wall with a small t,t 2 is smaller and can
be neglected such that after simplification
pr
σa =
2t
12.3
Figure 12.4 Free-Body Diagram of End Section of Cylindrical Thin-Walled Pressure
Vessel Showing Pressure and Internal Axial Stresses
Note that in Equations 12.1 and 12.2, the hoop stress is twice as large as the axial
stress. Consequently, when fabricating cylindrical pressure vessels from rolled-formed
plates, the longitudinal joints must be designed to carry twice as much stress as the
circumferential joints.
since this is a thin wall with a small t,t 2 is smaller and can
be neglected such that after simplification
pr
σa = = σh
2t
Note that for the spherical pressure vessel, the hoop and axial stresses are equal
and are one half of the hoop stress in the cylindrical pressure vessel. This makes the
spherical pressure vessel a more “efficient” pressure vessel geometry.
12.4
Figure 12.5 Free-Body Diagram of End Section of Spherical Thin-Walled Pressure Vessel
Showing Pressure and Internal Hoop and Axial Stresses
12.5
ri 2p ro2
σh = 1 +
(ro2 − ri2 ) r 2
ri 2p
σa = (12.4)
(ro2 − ri2 )
ri 2p ro2
σr = 1 −
(ro2 − ri2 ) r 2
where ro=outer radius, ri=inner radius, and r is the radial variable. Equations 12.4 apply
for any wall thickness and are not restricted to a particular r/t ratio as are the Equations
12.1 and 12.2. Note that the hoop and radial stresses( σ h and σ r ) are functions of r (i.e.
vary through the wall thickness) and that the axial stress, σ a , is independent of r (i.e., is
constant through the wall thickness. Figure 12.6 shows the stress distributions through
the wall thickness for the hoop and radial stresses. Note that for the radial stress
distributions, the maximum and minimum values occur, respectively, at the outer wall
( σ r =0) and at the ( σ r =-p) as noted already for the thin walled pressure vessel.
Equations 12.4 can be generalized for the case of internal and external pressures
such that
r 2p - r 2p − r 2 r 2 (p − pi )/r 2
σh = i i o o 2 i o2 o
(ro − ri )
ri 2pi - ro2po
σh = (12.5)
(ro2 − ri2 )
ri 2pi - ro2po + ri 2 ro2 (po − pi )/r 2
σh =
(ro2 − ri2 )
where po=is the outer gauge pressure and, pi=inner gage pressure.
12.6
Combined Loading
Typical formulae for stresses in mechanics of materials are developed for specific
conditions. For example
P
Axial Loading, σ = N
AN
My VQ
Beam Bending, σ = and τ =
I It
PT
Direct Shear, τ = (12.6)
AT
Tr
Torsional Shear, τ =
J
pr pr
Pressure Vessels, Shear, σ h = , σa = , σr = 0
t 2t
Often, the cross section of a member is subjected to several types of loadings
simultaneously and as a result the method of superposition can be applied to determine
the resultant stress distribution caused by the loads. In superposition, the stress
distribution due to each loading is first determined, and then these distributions are
superimposed to determine the resultant stress distributions. Note that only stresses of
the same type and in the same direction can be superimposed. The principle of
superposition can be used for the purpose provided that a linear relationship exists
between the stress and the loads. In addition, the geometry of the member should not
undergo significant change when the loads are applied. This is necessary in order to
endure that the stress produced by one load is not related to the stress produced by any
other loads. The following procedure is taken from (Hibbeler, 1997)
12.7
12.8