FirstPrinciples
BodyUnderExternolLooding
When performing engineering analpis, you are viftually alr^ra),sconcemed with how a body will behaveunder extemal loading. Newton,s
28
Chopter 2t Fundomentats
Iaws,or the laws that will most generally govem this behaior, are listed
below.
.
Thi.rd.Lau: Acton and reaction forces between interacting bodies will be equal in magnitude, collinear, and opposite in direction.
The most important engineering equation arising from these laws follows:
8q.2.1 r= N
where Fis the resultant force vctor, rz is the massof the body under
consideratron, and a is its acceleration vector.
Becauseacceleration is the time deri\,?tive of rclocity (do/dt),
d, G =
^
nE,constrtutes the kneu momtnturn\/ectorof a body, the above equation
can also be wntten a5follovls,
ttq.z.z
F = 4 i= c
Fig.2.1.Genenlfteebody
di agir8m (a). Resultant forces
anclmonents (b), Second
*q;;Q
(a)
(b)
G)
29
Fittt Prin.iples
The most useful tool for understanding and implementing the loads
and constmints, or boLndaryconditionsthat govern a body s beha\ior, is
the Fee boq d.iagram.'fbe general free bod,""diagram above (a) reprcsents lhe body in space removed from its operating s,r'steltl.Ali externally applied loads and reaction forces are represented with vectoft on
the body. If the body isir, equililirium,all theseforce vectors mus! add up
co zero, both in magnitude and direction.
In the most general sense,externally applied loading on a lhreedimensional gid body cannot only alter iLstEnslation, but its rotation as well.
Refeming to resultant forces and moments (b) anct the second la$'
equivalent (c) in Fig. 2.1, the conesponding spatial equalions of
motion for a rigid body follow:
Eqs.2.3 lr = o
La= n
where IF and !M are the force and mornent vector sums, respectively,
of all externally applied loading, including reactions, and H is the d't?gr.,Idr mommtum vector of the body. Both tM and H must be calcdated
about the same point on the body.
Iig. 2.1 sho\,'Efree body motion where this point corresponds to G the
center of gravrty ofthe body. For constrained motion, it corresponds to
q the fixed point about which the body rotales. In Eqs. 2.3, the time
deritive of H is a complex quantif to deal rith mathematically, but
sumce it to sal that H is a function of both the angular velocity and
angular acceleration of the body. Its inertia component is not the mass
of the body but its mass Mnart of inaria rc"ror (I), which is a 3 x 3
massprotlucts af inzrmatrix comprised of massmomenlsofineftia (Iii),
^nd
,ia (1i), derived with respect to the body coordinale axes.These quantities describe how the mass of a rigid body is distributed with respect to
the chosen a..es.The general equations for these quantities follow:
where ?,J, and & are any combination of the three coordinaLe a.\es cho
Choprer2t Fundo,mono,lt
30
,.,
'
V{Em^^
(b)
r,= ni+'t+k'
with i denoting the spring stiftress, t t}Ie damPer coemcint' and Ir' a
t aod t the b;y's resultant apPlied force, Position, velocity, and accel_
emtion along the x axis.
Constrainingthe body to pla ar motion [seeFig 2'2(b)], Eqs 23simpliry to fie fouo$,ingexPression.
>F = tuc
luo=toa
In Ore above equahon, ac is the vectonal acceleration of the center of
the
raviw (c.s.) oi tfr. Uoay. EMc is rhe sum of all momenis about'
axis
an
i"ro.'ooini. l. i" *t. massmoment oI inertia ot the body about
," tft! phne of motion through the ( g. and d is dre bodv's
",t.if accelerauon.
angular
Barring dyramic anal)ses,FEA will alwaysdeal with bodies in equilib'
rium. 6y deEnition, a body in such a state must have zero acceleration'
so that Ore result of all extemally appljed forces must be zero This tyPe
of anatsis is called statia Although *fs condition sounds very limiting,
Eq.2'6
3l
AreoMomenls
of Inertiq
There are many g?es ofapplied loadsthar causea continuouslydistributed force over an area.In thesecares,it is often necessary
to calculate
the resultantmoment causedby this force about an a_riseither on or
normal to the plane of fhe area. This is known as t}:,eava monentof iner,
ti r and is the soluiion to ttle inte.T^l l@istane)2d(area).
For example,Iig. 2.3(a) showsa submergedvertical wall subiecrto a
disrribuLed
pres\urerp, rhari. propordonalro r]ledep$ rl'b;lon rJre
horizontal surface line. The total moment experiencedbv the wall
abourrhesurfa.eline is ,l)':d,t.wheref is Lheconstanr
o[proporrionaf
ity. In dre samemanner, as will be discussedlater in this chapter,an
cl/.ri( beamunder pure bendrngIFig.2.J{brl wittderetopin is cro.s
secaiona linear distribution of normal force intensif/ (or stress,o) that
is p.opoftional to the verticaldistance(1), from a neuoal axis.Hence,
rhe rolalmomenron rhi' cros.serLionrvillonte againbe,{ly1d,r.
I finaj
example concernsan elasticbar under torsion lsee lig. 2.3(c)1.This
torsionalmoment lvill causea distribution of tangentialshearstressr,
that is proportional to the radial distance(r), ftom the shaftcenter.In
this case,chetotal moment is AjlZA.
32
"g'O
Deftnitions
ano
Fig.2.4. Rectangalar
ol inertia@).
polarmoments
Parallelexistheorcm(b)
o)
is known as ihe nctatgulat
The 6rst ti{o examPlesin Fig 2 3 utitize whal
Fig
24(a) These are area
in
a sectilon,illustrated
.f i"".,ij
stttem
"t
"r"."o
.f i"".,lo .lout a rectangular or Cart'Ji47'coordidate
il.."tt
following
the
of interest, and are given by
;;;.;;.;i;"
"ection
equadon.
Lq.z./
t , = lt . A
Ir= l''dA
33
I,= Ic,+Ad,'
If
Eq. 2.11
Iq+ A d
J, = J",+ At
where the subscript , dcDotes dle terfirs about the centroidal axes.
when a geometrically complex section can be divided into a number of
simple ones, it is pocsible to obtain a resulting fotrlrorir, moment of inertia. This is done by adding the individual cenEoidal moments of inertia
together with a parallel axjs dreorem correction for each of the simple
sec0onsasfollows:
F4. 212
I, = lt",*2e,a,2
Ir= >4,+>^d:
8 2 . 1 3 r .= > tc.+ t^ t
If dre body of interest has a section that is completely asymrnetric, or
tlte cooidinate axes chosn are placed in zuch a way that the body is
a.syrnmetricwith respect to dis coordinate qtstem, a product of inerha
resuls. This is expressedby thc next equation.
F4. 2.14
1,, = IrydA
Note tlEt as soon as one of the axesbecom$ a st'nmetry axis, thfu inertia term becomes zero.
Using the product ofinertia, it i3 po$ible to mathemadcally rotate a set
of axcs about a poittt and compute new moment of inertia terms as
functio$ ofboth the initial idertia terms and the anqle of roadon. The
Chcpter 2, Fu ndd.nentot 5
34
it is possible solvefor
angle is the only variablein the equations'so
marlmum
ofto
ii." J,l."l anqlc rcU (hat give" the are"an orienauon
expressron
and minimumlnerda.a" describedin lhe follohlng
2t"..
Eq.2.15 tn2.:L= r:t_r
argle ilefines the axis
Eq. 2.15 ields two angles that differ by tr/2 One
is the minimum
oithe ma"imr,m moment of inerga, and the other
inntia Norc
axesol
it i." *. .".'.^g"l"r a(esare called $e pnn'ipal
side ol Lq z l) rs zero'
rhat if a cbo.en axisis a sFnme(n axis the righ(
be ze(o or tt z' tnt\
and the , orrespondingangle ol rolaoon musl
atts'
of uh;ch is a
n"an" *"t ;J a ia of axi is chosm,at l&'st on'
'mmeh)
thesemlltt bePinciqaLa,es'
the principal moments ot
For an arbitrary set of axes' the magnitudes of
inertra are as follows.
aq.z.tar..,.^,^=$tl
( an simPlvu(ilize
lt is useful Io note thar manv CAD and FEA PackiEes
ol comPlex
propenies
a sketchedse(tion Lo calculale cros*serdonal
rhesr
iddiLron
a r'rserdefinedcoordinate s)slem ln
g".*;;
"u.*
hdve an exlen'ive librarv of ProPerlieslorundard
iackage, uruallu
shapes.
ondStrqin
Stress
Ir is sumrisinghow many engrneersengagein
srressanel\'iswidrout
swuctu'ar
rodenne
aLre
beine
""""
has many fac"I1""J1ff:,#i;::.il:,il;
solution
{
FE
in
an
sousirt
at hand requires adrance
q*antity for the
Problem
;;".;;;;;:"*t
oPlions \4an) rexts inrludingmo'r i ollege
*."i.an.
"f 'f'. "i"ilable
deuri rl vou lrsn to
mechaniis bools. will i over dris malerial in more
if vou arelo
pu'..* i,. Vot .un n."er Inor' too much aboulrhesetopics
irtilizefinireelementanalFisin )our career'
35
WhatIs Stress?
When a body is subjected to an applied load, a stress state is caused
inside the body. The sEess can be described as dre intemal force
exerted by either of any two adjacent sections of the body upon the
other, across an imaginarl plane of separation. When the forces are
parallel to this plane, the stressis called .rfiearstrert(1). When the forces
are [ormal to it, the stressis called zornNa,strerJ(o ).
Subdividing the body into many imaginary stress elements is usefirl at
ftis point (see Fig. 2.5). For the body to be in static equilibrium, both
shear and normal stressesmust act on each one of these elementl in
such a way as to place it in static equfibrium. If *re normal stress is
directed toward the element on which it acts,it is called .onprerdu stress
and, by convenion, is [egative in iElue. If it is directed awayfrom that
elemnt, it is called ,mtirrstress and j5 positive. All of dresestressesresult
fiom the cohesivenature of the body's materia.l;if the body came apart
with no resfutanceunder applied loading, it would experience no stress.
Fig.2.5. Generalstess
elements In equilibrium,
1'
Chapter 2t Fvndomentak
36
Principal Stresses
Tt'J.;ifr
;':T;::
{iTfi;:
*r::l ;:*;ru:::':,iJ;::'J:'
*+Ji:;"','."Jil#$';,:1pi+*;t,"4
f:!#l: .i
(o/.) ml'
referred to as the ma\imum^
f'T,n::"::::':l''.1:'i
'":lxi:i*::,,*.1t,'3;;;i;pfi
I
onrv'
funcrionof loading not geomery
whenoneoftheprin,]rp"r
Y""* I liii;ihi.,Ii:i*1:1#;$i:::
;:,::iT;f."liIff
itttr:{::::.1i::t*Htn.ii*i**
ire in *rai praneald idenucalon any.p:
Lothat plane,althougbsrain usualtl ls
::l#ffi
,$;i":iJ
::#,rl:s#iltrf
l.*:i'.'#'ff
be ignored
Fig.2.6.GeneatPtane
Oi e ntation af PtinciPal
stess fO. Onantatiot
(a)
(b)
(c)
iffi lixf'fl$
H*tr,"$"'&'iT.1i:J:::1x,,'ilS.llilll;f
the follolving quadratrc equatron'
it
i1
,t
lt
{.
8q.2.17 ct,r=s]:?+
37
ond Snoh
',''tt
The soludon of this equation lsee Fig. 2.6(b)] alsumes that you have
accds to the local normal (o! oJ and shear (TJ stressdata. Note dra[
the angle be[weenrle principal stressorientaion and the measured
stressstate is denoted b'yS.
For the same biaxial condition, there is an additional element orientation of interest for which dre shear stressesare a maraimum, although
dre corespondrng lrormal stressesate not zero (see c in previous illug
tration). This odentation i! 45' alvayfrom the orientation of principal
stress,and iE stressstate is given by the next two equations.
t;:
+\,,
6:+ 6,
fu. 2.19 a = z
\
=41;43
38
rl
_\
x
'J
(r)
ptanestess(a)andfiiaxialstess(b)'
Fig.2.7. Mohlsclrcledla{en for
introduced above may b readily
All elemental stress stale exPressions
very
with the use of Mo hr'scircb diagram'fhis
i".iJ
J..pr"..nted
;;;.;;p;i.^r,".*.91'*="lT:18'J"t;l;5litri#.':l;
case
dition (a), and the general tfla$al
#il?il
;;f;il
iDformadononceyou are.comfortabl:^T9'*:
Slrorn
,1_,
ffirr
llrll*
1,i
i
ltilt
6oD
li
il I
Ftg.2.8.Defomatlonof a
unilom barunderunla,Ja
Ioadlng.
ofan elemen!
Asdepictedin Fig.2 8, the changeitr size
is
*,rt *.p*, to its original size knownas r.nrl
i"'iiiJt
.l"o1" (e) This quantitv is related.o the
:;;;;;iy
ot tota strain(6) in a bar of lengxh I
;;;'"I;;;il"
under uniaxial loadinS'
39
Stressand Stuoin
E q .2 .2 1
A uniformly distributed stresstakesPlacein the normal, crosssectional
area A awayfrom the ends of a bar under a unia\ial loading 4 and it is
calculated based on the following equation.
Eq.2.25 6 = !
In this case,making use of Eq. 2.23,the total elongationof the bar will
reduce to the next equation.
Eq.2.26 6, = +
..
::=
.j
:..
Eq.2.27   ;
Note the negative sign, which indrcates "narrowing" of the bar under
Len.ionand bulging'of rhebar underi ompres'ion.
It tums out that dle three elastic constants mentioned aboveare related
by the following equation.
Eq.2.28 E = 2G(r+r)
Principal Snain
The strainsthat occur in the direction of princiPalsressesar known as
principal strains.Note t)tat all shear strains will be zero for this element
orientatron.Hence, if you were to expeimentally obtain the p.incipal
Chdp.fP 2t Fundomen'd]ls
40
E (a1+ vez)
E ,(1' ) + vE (% + % l
 ;;;T
E 211v)+ rE tr + r)
a,= ;2" ^,
El(1 v) + vE(l+ 2)
o, = _=__if,l]
FundomentolSfressSfofes
Y;,'.',"i
ffiT.:fJji,#Ti:"T#l$
:ffirydi"i:#
'ii:.L,'J:'il:[.ffi
*'i'"ltr'.*;';;;;,^1'LJ'1L"'r".'
il*i*#:*":;*::'**q;g;ni.d#
ol tre tox
Dletenessand reference, each
fiJ.;..r*n
FEAsoludon
.upa*'of itscoresPondingripreienotive
Slrcss in Flexure
Fig. 2.9. FEAof a
4l
Stressdnd Srr(,i'
Fig.2.10.Stmight
Neufal Aas
Sr6s Field
Assume a matedal thal is both isotropic and homogeneous, such that its
properties are di.ection independent and do not change throughout its
exoanse (seeFigs. 2.9 and 2.10) Assume as well that this material obeys
Hooke's law. Nirr', consider an initially straight beam of constant cross_
section, which is made up of this matedal and is subjected to a pute
bending moment (M.If ils crosssectionsremain planar and either one
of the siction's principal ares coincides with the plane ofbending, the
normal stressesdeveloped follow
nq.z .z e c =  \ !
In Eq. 2.29, l denotes the area moment of inertia of the beam's cross
section about its transverse a{is, and ) is the normal distance away from
.J1e neutral axk il tr'e plane of bending. The neuffal ais is defrned bI
the intenection of the plane ofbending with the longitudinal surface of
zero {iber sress, or neutral surJacz.lhe localion of the neutral surface
may be calculated by {inding the horizontzl crosssectional line about
$'hich the area moments are equal.
Note that the sign notation of Eq. 2 29 follows the use of a righthand
coordinate slttem. It fotlows that this flexural stress will be a ma'{imum
at )'M = ., $here dle sign of c will golern the direction of the stressper
this equation.
For certain materials, it is useful to measure a fbxutuI, or bendingmodu'
lui (t d using the following equatron:
Eq.2.J0
EBR=
Chdplff 2t Fundam.ntdlt
42
Cdrrcidal A$
l
is originallv curvedasseenin Fig 2 l '
lf the beamunder consideradon
sec*re
of
axis
t" longer located by the cenuoidal
*tt
iil'".tili
(P)'
former can be
the
and
;" " *paraied by a dist'n'e
;;:il;
by thefol(
t .*s radiusofcu'uarurc tJ ascalcuta*d
i."J"iiyl
""",r
lowingequanon:
fu2.31 ,^=4
J;
z.sz"=ffi
no.
Eq.2.33 ct= 7;
^.
='"o
Aer.
43
Stressin Shear
Fig.212. FEAaf a
cantileverbeamin shear,
t/
1=
vo
tb
Chdprer2, Fsndom.ntals
/t4
fabte2.2.
'i#i i,i.
"no, "+"u
lomutasfonseteaad"ta4"gj* gg
1^* = 1V
tA
solid
Reclngulat,
2V
l\olbw
CirculaI,
Slress in lorsion
Fig. 2.14.FEAof a round
G
ttl
its longitudinal
When a torque (?) is applied to a1 elasticbeamabout
axis For a
torsional
developsawavfrom the
J; ;;;;;:4.i.;
'trisi
inFigt,f,;[5tT;.:*1ffi]t:t,fX:
(houow
orsorid
;;;;;;;."
45
Eq.2.35 r = L
Here, r is the radius from the torsional a{isand /is the section,s Dolar
areJ momenr otinenid. OfcoursF, rhe maimum ror.iorat
irres,
"herr
will occur at r^o"= r, Lhe outer radius of the bar. The anzular
deflecLion (0' ar rhc end ofa solid round bar of lengrh / follow\:
Eq.2.36 o = #
where Gis *re matedal'smodulusof rigidity.
A note must be made here regading Eq. 2.36.Alrhough rhis equation
aspresentedmay not be usedfor beamswith crosssectionsthat ar.enot
pedectly circular (hollow or solid), by subsriruringthe mriable r< for /
the equationmay be genemlizedfor all other sections.Krepresentsthe
secior\'s tot'sionalstilfnerr/artor This factor is equal to./for round sections, yet ir is lessthan Jffor all other sections.When assigningline element propertiesin FEA,you must input K Be&?reofthe fact that many
FEA preprocesso$confuse the nomenclature by referring to K asf
However,by erroneouslyinputting the section'spolar area for its torsional stiffnessfactor, you will effectivelyundentiffen the line elenents
in torsion.aavalues
for a riety of bea.msectionsare tabulatedin man_v
en8ineering rexrbools, norablv Raarh\ Fotmuln"lor Sres ond Strainby
varren C. Young (Mccraw Hill, 1989). Refer to Chapter 7 for more
information on line elementproperties.
It is very difiicult to obtain formulas for both stressdistribution and
maximum stressin beamswith noncircular sections.Hence, most are
done experimentally. The following is an approximate formula for a
rectangularsectionbeamofwidth (zr) and thickness(r).
Eq.2.i7
r^.,
= 4f :  r ' t r l
ChoPtcr 2 t Fvndomentals
46
Stressin Ptessure
Fig.2.16.FEAol a thicl<
cyIinder uncler Pressu
under
Fia.2.17.CYlincler
fi
(a).
Press
p;sure
(b).
cylkdets
P' 
{al
\"/
*t;::t.' l'''T1"ilid,r:;"ff
cvlinder
A capped
TS [ff Hil5
',l"xul*:?;ll#r''i:.J[fli;l;'rai* ap'i"*"t"'
"lli"*
r:;'J[I;
TiJiflI *"#ffii'J"Tt
rhe end caps are calcilated as tollowsi
*ffi.:Hl#i
47
fus.2.38 c, =
p {i
p.to
t,
2 2.
.2
to tPoP,)/l
7
2
2 2.
..2
p i t i  p . r o +t , f . \ p o  P i ) / f
,;,i
For a thiowalled cylinder with rhickness (t) lessthan about onetwenrieth of its radius, the maximum tangential and constant longitudinal
stressesunder intemal pressure (p) and negligible outside prelsure are
calculated via the next equations.
Of coune, the radial stress in this case would simDlv be t}Ie i[temal
pressureon re inside ofthe cylinder and zero on the outside.
It is usefii to note that if the pressure created due to an equallength
cylindrical pres*fit r9s known, Eqs. 2.38 could be used for obtainiAg
dre stress state on both the outside and inside cylinders. Hence. this
pressureis given here asa function ofthe radial inrerference(6):
where now [see Fig. 2.17(b)], z; is the iffide cy'inder's intemal iad.ius,
r, is the outlide cylinder's extemal ndius, R is the transition radius, and
E, Eo,v, and va are dle inside and outside material Young's moduli and
Poisson'sratios, respectively.
ChapFJr 2, Funddmcntals
48
S'/,.essin Contact
Fig.2.18.FEAof two sqhercs
1, .
r,
o[ nvo bodiesbeing
The resuldnq suessstale develoPedwitiin each
.omplex nonJinearphenomenon*rat can onl) be
;.i..";;;",n;;;"
y * clo'sedform for a h'ndtul ofsPecialcaseslnthis
i.rJr""a'"i"it"i.
from the forced contact of two spheres
t;*ttresulting
H';;
;;;;;;
.lrinders foll be presented (see Fig 2 19) '
.qJr."g,f'
"ri*.
d Ne pressedtogether
When t\,vosolid sPheresof diamete6 d I a]r,d 2
a
contact circular area ot
into
ruitf, u ior.. 4 tftiit .ontact Point tums
radius a given bY
;;;ii;;,
49
where E1, E2 v 1 and v2 are the respective elastic constants of the two
spheres'materia.ls.
At rhe center of *ris area,a maximum pressurep'G
will occur of the following rnagnitude.
Eq.2.4J p^",= :+
zfta'
Placing a coordinate sFtem at this center poilt, with the contact circle
on the ry plane and the z axis norma.lto eidrer one of the spheres,the
principal stressescoincide wirh rhe3eaxesand are a firnction of the z
coordinate according to the following equation:
a, = :P",*=
where v is the Poisson'sratjo oI rhe sphere under consideradon.The
maximum principal shear sfess is Lhengiven by rhe next equadon.
E+ 2.47 ,^* =
?!bt
Placing a similar coordinate s'6tem on this contact region, but this time
orienting rhe x axis parallel to the axes of the cylinders, the resuhing
normal pdncipal stesses are given by the following equations.
Chdptar 2, Fundom.nLlt
50
a,=.",^lE il
Es.2.4B
.,=o*1,!;^,)ffi';]
in Eqs 2'49'
The resulting principal shear stressesare Provided
Eqs 2.49
6,O,.
t', = 'zto o.
o"lo,
\t'= "f
change(AZ)' it
If an unconstminedbody is subjectedto a temPerature
ptop".t"naily in all directionsaccordingto the [ext equa"iii"*pa
uon.
5'
Eq. 2.50
z, = e, = z= d(a?)
6 = d@r)E
Thus far, *ris section has concemed itself with very idealized seometries. Yet, real parts will usually have design fearuies.such as'holes,
notches, fillets, steps,grooves,and so forth, that v.ill causethe idealized
scss distribution to develop highly localized regions of concentmted
stressin their vicinities. T\ese stressconcotrarrbri will also aDDearclose
to unplanned irregulariLiesin rhe part, su(h as cracks and pia. To
account for this, the folloyring themetirelstress
csncentratianfactorqor Kt
is introduced that depends on dre tt?e of stressinvolved.
Eas,2.52
K. = !!:
K.,
!!J
Chcpter 2t Fondo''lentols
52
j*: [:,:'r*'{l
.ir:r*+:.iiriiil
lil,"r#.]i,il
i
u,a,ion
;hcp.rrinrh"r,aJ,
l:
i::iil :i, Jll:.::"i.:.Iil;"l"ilil
:Iilq::i:i::'ilii:i!:il:#ffr,*ffr:h:.s:T
i"n*#;i.1il;;I:Jil::Ti.T:;l""ii*,""r."..i".1'l*"rvi"
witir Eqs. 2 52
an"alysis
ProPerlies
Moleriol
in vaious references l!'rth r1lry1ng
lvlaterial Properties can be found
t*H*il:ri"[idl:irlii:r';"ril"f
rl]#
:?;ir:r
, r'"nu*'.o""
lil".'fi1i,#l;,tl::iiJ':.*:';;;'.', "r*'
lT
""..1*i.otr.""gi"='"g:l:1"1':1fr
.i5iffi
::?ffix;:fi'J;';
ii
p.op.'""'
io.
*'ffi:::T,;#l:xli;l:li:'::it?*;;;;
to
"u,",i.r
proPerties
lhese
q".tu."able, consideriesultsbasedon
:',ii*rn**H:'H#!;il'{:+3;l15,
""*"ii..,
formid un<IerantrciPatedloading cc
similar to the Part being studred'
Typesof Moteriols
Mdbridl
Ptoperties
53
\ommonMotenol rrooertrcs
In the previous section, three material dependent properties were presented: the modulus of elasticity (E), modulus of rigidity ( c), and pois_
son's ratio (v). These properties remain virtuall), constanr for all
materials of the same type. The properties introduced in this section
are more manufacturing process dependenr and can vary greatly
berhcen tuo bodiesof rhe same marerialrypewhich rneremanutaiured
differendy.
Arguably, the most useful of all mate al prope{ies to have accessto is
ilr \r/Z,?gi.\ore Lhar the unirs
sllengr]r rre lhe same ir5 rhose ot
"t
stiress.Yet, $.hereassress in a body
is al ,aysa function of dre applied
loading and crosssecrion.\r rcng$ is an in herenr properrvol rhe Lod1 .
matedal/manufacrudng process and governs the overall peformance
of its design.
Strength and other g?ical material prope{ies are ofrcn obtained from
a standard tensile test which subjects a sample bar to unialial stress,
extracting deformation versus applied load data. This information is
then plotted ol1 a stresssrain diagram to illusrrate *re rclationship
between the rious panmeters ofinterest. Ifa high degree ofaccurac],
is required, matedal properties used in a stuctural analysisshould be
obained trom tesringunder ( ondirionssimilar ro rhe aciual operarine
condiLionsof rhe pafl or .vsrem.Thesc .per aring r onrlirion. should bi
kno$'n, even if such a level of accuracy is not necessary,in order to
adjust the data to compensate. Conditions to consider include operattng temperature, strain rate, material grain or flow direction with
respect to loading, and tonion versus tension ve$us bending.
Chc,oter 2: Fun'dqmentcls
54
Stain t
A q?ical stressstrain plotrellects mgineering J,rsss(see fig. 2 22). This is
a somewhat inaccurate qualtity because it is calculated by dividing the
applied load (l) by the onginal cross seclional area (A). T)?ically' a
Ioialized, irreversible decrease in crosssectional arca (or naching)
occurs in r rensile rest.I his oc.urrence i\ rhosn in dn engineerinq
, ur.{re\hen the }lres. appear. lo decter:c $iih in( reasinB
sLresssrrain
strain. In actualiry, the reduced crosssectionalarea causesthe resulting
\lre\s sldlc to leep ri.ing ar a funclion ol 5u?in. lhc tnu \ltP\\'tnin
.r,ze. which accounLs for the sPecimen's necking, t?ically shows
qreater stess values thaD the engrneenng stressHowever, because it is
liflicult to track the change in the crosssectionalarea of the sPecimen'
the engineedng stressstraincurve is used almost exclusively
As seen in lig. 2.22, fie sbessstrainrclation ill initially be linear up to
point P, known as the proportional lirnit 'this Portion of the curve is govirned bv Hooke's law and Eq. ?.24. Then' although not proPonional
an),mor;, the material *ill conrinue to behave elastically up to Point t,
t}.e ek)stic limit. Beyond this point, permanent deforrnation will be
obsened upon removal of the load.
P oint Y is the )ield point of the material, corresponding to its )i,ld rhsr'8'i
(S^). Some malerials, such as ferous mater:ials, have a distinct yield
pJint or "kree" on the stresssfain cune. For these materials, Points E
and lwill coincide. Yet, for many other materials, the field point is less
clear and is often detellI]lnedby an offselnetl,od Using this method, tsis
Mclericl Properties
55
Eq.2.5J s,, =
7,,
where Tz is the maximum point LIon the torquetwist diagram, ris the
radiusof r he bar. and / is .econd momenr ol ared.
DuctileversusBrittleMoteriolBehovior
According to the previous discussion,a bodyis said to have yielded or to
hare underqone plasric delormaLjon il ir doe. nor regain irs origin.il
shape when a load is removed. The resulting deformation is called per
nanm! rpl.ll permanenrser i5 obriindble, the marerialis.aid to exhibir
dr,rlilio. A measure of ducriliqv comes from rhe percent elongation, or
srrain at failure, of a tensile test specimen. Brittle materials will have a
much lower elongation and area reducdon than ductile ones. Hence.
the amount of necking, and the corresponding dip in the engineering
stressstraincur've,is indicative ofthe ductility of *re matedal.
Ior ductile maLerials, the ultimate tensile and compressive srengths
have .rpprorimareh thc same ab\oluLeralue. Brirde marerialson'rhe
other hand are stronger in compression than in tension.
I
Choptat 2, Fvn do''entatt
56
Modulus of rupflrle
strength.
The str'uctural anal)st should have a feel for whether the material being
studied will behave as ductile or britde at the temperatures aIId strain
ntes expected. Most materials become more britde as strain rate
increases and as temperafilre decreases.The method of results evaluation and failure quantities used are dependent on this property.
Rules of thumb used to determine if bdtde or ductile behavior should
be expected are summarized below.
.
SofelyFoctor
Having inEoduced both stress and strength, now rs a good ime to
present a quantity that relates the two in the design processjtle JdJAzlJAz,or (n). This number is defined as the quotient of the sffength divided by
the stressin a part, and it provides a! indication of the level of conidence in not only the accuracy of the inputs used and their representation in the analvsir.but ako in dre accuracvof the ana&sistool itself.
Ea.2.54
i = ttreryth
57
Foilure Modes
Modes
Fsilure
ltsInterpretotion
Resu
The fiISt stepin resultsinterpretationis to revrewthe goalsset Ibrth at
the beginning of the study.These should tell you where to look and
lvhat to look for. In mostcases,you will be looking for someevidenceof
failure or assur:rncethat failure is unlikely. \4tth this in mird; a few
words on the nature of failurc in engineering design are warranted
before sDeciticfailure predictorsare discussed
ChdpF.r 2, Fundo,mentclt
58
.
lose
Fatipue. Parts that are subject to variable loading will
cycles
stre'ngth with time and may fail after a certain number of
Bodiesunder load Sradudllvdelofm over Lime The dpptt
Creep.
pnl'nodutu\ propern is denved lrom emPirical creep data lor
uarious matetials and may be used to compensate for the effects
ofcreeP.
modesmar
ln most engineeringproblems lwo oI more of theselailure
rhi operaLingcondirionsoI the srsrem lr i' imPort:)nt
be possible"given
t., ieview thi data for all occurrences of any potential failure'
FailureTheories
Classic
inreresr' ehoo'ing
ln caseswhcre lailure due lo yielding or tracrureis of
failure.PreJ. .or,"., srressquandtre' and apPllng rhe aPpropriarc.
and wtoelv
general
more
of
the
dic(or or lheory i' imPortanr. Sereral
theones
failure
used failure theories are discussedbelow These classic
or excessrveelas'
exclude existing macroscopic cracks, buckling, creep'
failure
tic failure and aie primarily concemed with material
unifying
These theories are not deri?bte laws but tend to Provide
are.frequendy
data
material
Published
data
.l.o.rno of."p.ti.ntal
enSlneerdetermined by testing in uniaxial stressstates,whereas actual
\4ureo"ver'
\trest
srale'
ine probLemsare lreduendybiaxjaloI comple\
"mar.rial
ma1 peform in a ducdle manner al one lemPelature o'
a
i"^a;"n.o"aiL;"n, v.r lail in a brirde mode al rnother' lr is e:.rremelv
f'r lhe
i,troo.Lnr ro understand (he load parh and malerial beha\ior
oanicular Lestconditjon regardles'oflhe mean" ofanal!sis'
59
Fdilure Modes
Ec 255 6".=l
The beaury of the above equation is that it represents the entire stess
state, no matter how complex il is.
Fig. 2.23 depicts the three ductile failure theofies for a plane suesssituation. The sress states defined by the locus oi points enclosed by the
Chopter 2t Fundcmenldls
60
Fiq.2.23.Conqansan
of threeductilefailure
II
Failutc ltl6des
6'
CwltnLMolv thzory.Fra:ctore occu$ when the maxiown and
minimum principal stres6escombine for a condition which satisfies the following:
E q.2.56 ?  ? z t
where S!, and S@represent the ultimate tensile and compressive
strengths, and both 03 and S@are alwap negative, or in comPressron.
Although this theory is applicable to both ductile arld bdttle
materials, it is applied more frequendy to britde materials
becarxe they are stronger in compression. When compression js
dominant (od >> o), the Mohr cdterion is the most reliable predlctor.
Mod.ifui Mohr thztry.Fmcture occu6 as defined in the CoulombMohr theory except in the fourth quadrant condition where o/
is in tension and 02 is in compression. In this situation, the
mateial is somewhatstronger than a CoulombMohr plot would
suggesL The impact of the modified Mohr theory is shown in
Fis.2.24.
Chdp'tr 2t Fvnda'.enldls
62
i:;f;;1i::l.ll;il.T;'il::,',"1T:,Iff.?::tdj:*i
.ared pol)gon. Thesc suess strtesillusLn
*te mJ\imum not'rnal'tre's'
in a pl'ane.ue'., ondilion ds prcdi'le'l bl
Mohr theories' respectively
i"',fiuu.t'.,
^"a odifred
.T::;
x:'j.'.T.T[T:::Xff""iilii:i;?l';::;i?:"t*'lii':f
.ort..t*tive prediction and is acceptablefor design'
f"ta,rl
ot"
OtherFoilureTheories
discussion are presented
Failure theones excluded in the prerious
below.
Buckling
t rii i, tas(rrrm L,
whi'h is assigned
irl"''."irri' and an effe'rire lengthlactor A
"i
of the columnends Common
ii...ang irr" .""straint conditions
in lig 2 25
.nJ.onditionsonatheir coresPondingK\tlues areshown
Fcifur. Modct
63
FRFd
f iFdFF.
Fitd<uided
Fi&dFikn
t = lL
\l
a", = frzE
(L,/r)2
EC.2.60
L, F"
ch.pt',r 2 Fun&n
ntg,tt
tL' / t\'
ffi;""t;;;.,
;dL
in Fig 2 26 and
The resultinggeneralcurve of od versusLrlr is shown
Note r}ratthe
accuracy
i*u"." *ti??a *rttough testingwith exceilent
of
,5g,Zt n"tni t'tPoloto,into the nonEulerregion
.fr"J""
"Gq
nonconser tive critical stresss'Hence' it must be
the curve provibes
a tnear FEA code for executingbuckling P^rot>
"r
"i"16?,it",'ii"*"
for columns that do not satisryEq' 2 60'
;;"d;;;;enatil
on thesenonEuler columff
ttte oPeratilg stresses
* i""i*
ii"i**,
point, uu&ting is not a Possibllailure mode'
..*d"ilf"ttlr.e
"aa
Fig.2.26. C ical stess
in @lwrmsas a hmctlon
of slendemssleio.
L'/t
in buctling ,
Fig. 2.2? sho*s FEA result! for a comPlex model
Fdilure Modes
65
Fig.2.27.BucklingFEAaf a
conplex shelI structure.
Fig. 228.Eccentrically
Maxinun unit laad tor
differentvaluesof load
eccenficity and column
slendanessratio (b).
'
"rn
ll
li
\"
L./ /
(b)
:.:
Chcpler 2, F undo,l'i,entcts
66
matenal
Hence, it is consewativel,v assumed that failure constitutes
cqudlion
follo$ing
vield'
rh'
vieldine in tornpression l hrs as'umPLion
e( entrF
ror Lhe'ma',imumunir load rClr Lhalr an be suPPorledb\ an '
cally loaded column:
z.azf,=s,"lr."4*(?.iE)f
nq.
disiance from the neuwhere eis the eccentricity of the load and c is the
thar in this case r is
Note
irJ *i. to the surface Iiber in compression
(orre\Ponds,lo
Lhe
mdiu'Lt gwarionir
na n..".*afy,ft..inimum
'
urs )otc also.rnat
as.ociaredwi(h the a"Yisdround whi(h bending oc(
cannor be solved exPticitl\ lor P '41hen'e '1 graphic"l
,f,l' ."r.,"""
n?rLd solurron rn
.otuLion i" mosr often uLilized Frg 2 28{b\ shohs "
eccentricity and
,i."tl"o. of pu,t"t ofunit load; functions of both
Note that as either the slenderness ratio increasesor
"lendem.ss.atio.d.creases, these cunes aslnptodcatly approach Euler's
the e.certtriclty
hlperbola.
Fdtigue
l os c(k
FA. 2.29.fYqicalS'N
diagramfor steet.
=l <
,;,..,,.
l l i shc)!l e
#dJll:',:
q,,
s,
lno
10r
10:
1or
101
10J
li}
l0'
r0'
number otloading
Fadguedara in lhe form of materialstrengrhver"us
rotaF
, vtlis. or 5N rratr, are generall)deri\ ed trom a re\cre b(ndinS
lhe
ho$
Note
for
sreel
char
L
:N
rng beam test.Fig.2.29thows a rypical
Foilurc Modes
67
strength ofsteel drops at a certain rate up unlil about 1,000 cvcles and
then continues to drop at a higher rale. This change in slope separates
'ivhat are considered lolF and high qtb
fatigu.,failures. ,\s the cyclic load
ing of the specimen continues past this point, th_estrength of the marerial will stabilize somewhere be${een 10" and 10' cycles,the number of
cycles lhar cmpidcally represent "infinite" life. The corresponding
enAu:rance
ot fati.gue Linit (.Se)is defined as rhe mar.imum cyclic stress
rvhich a part can sustain for an "infinite" number ofqclcs. Note that lor
nonferrous metals and alloys, the strengLh of the mateial never stabilizes but keeps decreasing \'\,irhtime. Hcnce these mateials do not have
an enduGnce limi!,
The endurance limit of the actual rotating beam specimcn is usually
designated a,s5;'. For lerrous alloys with an ultimate strength below ?00
ksi, Sp'is approximately half of this strength. For ferous allo,\''swirh a
strength above 200 ksi, E'is approximately equai to 100 ksi. Because
nonfe ous metals and alloys lack an endurance limi!, a fatigu( strength
(S/') is usually reporrcd for 50(10') cycles of reversed sress. This
strength is ofteD as low as 1/4 S,, for some aluminum allop.
Of course, one must find a \^,avto corelate the endurance strength ofa
part to that of the test specimen. This is accomplished lia several modifying factors, hich are all lessthan or equal to uniq/, as seen in the ne].r
s" = kakbk
"k,tk.s"
Here, Aais a sur{ace factor, A, is a size factor, A. is a load factor, ld is a
tempcralure factor, and id is an all e compassing, other mjscellaneolrs
effects fector. Numbers for some of these factors can be readih
oboined in the literature while otheN are rarelv available.All are menlioned here simplv to point out the difflcr tv of executing predictive
fatigue anal)'sis.
E.2.53
Chcpt r 2, Fondcrm.nlc,t
68
ing a good handle on the lalues for these parameters, successfulanalysis is impossible.
Severalrnetho& are alailable to relate ryclic loading data to fatigu Me.
All zuch theones are estimatesbased on empirical data alrd should only
be used for initial designestimates.As a result ofsensitivity to so many fac_
ton, nothing can replace actual fatigue testing in reahtic envhonments
to guarantee a reliable designfor a part exPeliencil8 cydic loading.
To obtain the fatigue strength at Ncycles for a Part exPeriencing dlrrt_
nd.ting or completcly reversed stress, you can curve_fit the SN curve
using the following equation:
Eq. 2.64
sr = aNb
a = ::::!0.9S .,
I
b = ,;toe___
N=
\;l
Fb. 2.30.slnusoidal
tucluatngst/ass,
amplifudeversua6me,
Tbne
Fcilure Modes
69
% t6:! = \
E,r.2.65 t != 
where ,lt is the number of cvclesofstress O, appiied to the part, and
,\,,is
*le,fatigue life co_rrespondingro d,. Manson."
f ieajity fouira
in the liteEture h uses a graphical approach and
etfr"a
is qpicalty pieferred
becauseit correlates to empirical dara more consistenrly.
\{rhen a ductile material is subjected to a fatigue_qrpeloading, there
are
basic structural changes that occrrf. In chronological order, t:hechanges
are summarized below.
Chdr,ler 2t Fu ndstrl,enlols
a:
"il:,:fi{#,j,1;:,;.":;:::l'$i,J::1.::H[::T'J
comPletelYrcversible'
anptane;,:j:f.1fJ:iiii:,1::"'JJ,',:.'J.".',:"'
3.(ra..k
sro,th
across the secuon
* \\'henthe''':1"::i:':i5'Tti.,"'Jl
(Iuctik
JaitT
'
lnc alrP s"
4 Littinate
s i / r l h d l c J n r l u l 'r l q r d l l r
c r o*
r er lr un
Io !
by ductile failure
th s'dmPleruptures
^rrl
'
r\,F\liii;';l;ill':"}:t:Il:
spec,r(
nn,
d,d
.e.don
A,,houeh,hi.
*' "^'Tli:;l';:].'"lJl^""i*""J""'*..''*''
i)i.1lii,'
".'
"'
beros
In"'''ig'lin'
I
"''''l::::l:i";
";'"'"mmxri/ed
'" :"".;":
:",::,;;;'1"
i;i:i:..:;::
"'r'r::'"':il:l,S

s rrdi n
$l ' i ( h ' ^' rqe rrrtl " ' l u " ' ei rton
'n'l
rii" J"t"ut" r'tig"" tir''
resis'1ant
' ,,*""*01'1*1;1I.;:i":::.il'li::t'(rack
beiauseroughne:
.
creasc
prottsse"
H:rrdening
.
surtaceronrtitianxns
'l1l""l",ll"u
Pro's(!'lv
and
plaring
r^f;itc
lalisur srrengrh
to
'orrusron
t *l;l*.;Ul
an.m
' En'Jh
A
5lrenglh
c
:;'U:l'::iXi":iil::'::$:
'omr
nn o\i n IoI tqt"'
^ll(d'
nij::l'id$'[l]$
J'.i
r;;*i
*
*,{***i**
:tr.::I'ii;..;Ti:i*n:l
;;l*;*11.fr
ff:::rJ:':T::;":';:,"::
8 q 2 6 9 c= e t+ n fi
il!"1:il;';
a'
scd
erpre
;nbe
FoifuF ttlo&c
ofvisct:ttlfi
where1 is *re matedal'scoefficieIlt
,11it"'TT:T:;
oPerating
i' n."''tion or o' uPPliedloadand
ffit..l;
;:;::
"
be obtainedexperimntally'
al'ailablewith some FEA codesbul
Vrscoelasticanalpis solutions are
informadon
l"t"rial studiesto ProvidProPerty
JJ.i.ii.to.a
i"r.l".",,
r, **t
i.'r"**
i"i'#iJ""
Dr hour.
ftom a time
*t."lor"f .reepprocess
crccPrar. ot ru7o
,rt. ,".* whiih producesa minimum
Chapler 2, Fundo,li',ento'ls
72
AD b a rp n t \o f P l tP 4 tr' ) ,{ ' dl /u'
Anolysis
Dynornic
DlTramicanalysesas aPplied to FLA.involve loads and corresponding
,oponr. .4,". that \Ery \vidl time Stricdy speaking,such anaryses
because
shJuld be referred to asvibratron and time resPonseanalyses,
of
realm
in
the
is
not
large displacement,comPletelyrigid body motion
FEAVibntion and time responseanalysescan be subdividedinto the follor'
ing three relatedcategories.
.
FrequencyresPonseanalYsis
Transientresponseanalysis
sys_
The first of the categories involves the y'ea'i'rd'ion of the d)namic
tem. This analysischaracterizesthe systemin the absenceof external
last
loadinq and servesto define its d)'namicproPerties Conversely'the
under
involve
slstems
wo uri krto*tt as forcedresponsean^lyses These
or
externallyappliedioading tunctions,which can be either liequency
al
Limedependenr.The anatlsisnPe requiredlo 'ohe the Problem
6"p"na on rhe t\p" of informarion$e( i' necdedIo reacha
1,u16!v;i1
desigDdecision
ModalAnolysis
which
The building block of all dynamic analyses is rlle mulal analJsis'
modt
slap"'
rcDot.'' the notutol lftqunriet znd corresPonding lnwipul
d
o[ rhe.\stem under efa]uation. ln other tords when Pertorming
morlal anatvsis.vou solve for the distinct deformation shapes that the
frequenvibrating systemwitl assume at each of its preferred oscillating
a
simple
aid
of
the
cies. Th"ese concepts are better presented with
example.
73
Dynami. AnclYsis
ofa
Fig.2.32.Frcevibratian
frrct
mode
cantileverbeam,
l
Refering to Fig. 2.32, it is intuitive that a thin beam {ixed at one end
will librate or fluctuate most easily about its fixed point with no additional "nodej' or bends (inflection Points) in its deformed shape The
natuml frequency ((Dr) corresponding to this mode shaPe is essentially
the oscillatorv speed with which the beam moves from one extreme to
the other and back. This speed is defined by two tundamental Ph)sical
pammetels of the bea:^ mass(m) and rigidiry G) or "sPringback"
Eq.2.70 o"* J;
The mass contribution to this equation is understood by considering
inertia. The more mass (inertia) that &e beam has, the harder ir is for
the beam to change directions when fluctuating' and consequently, the
slower the motion. Springback is the force that resiststhe displacement
of the beam from its equilibrium position. When the beam is bent past
this position and then released, its material elasticity tries to snap it
bacLinto place. Inertial effects prcvent the beam frorn immediately
returning to its equilibrium Point. Consequendy, the bean overshoots
its mark, rerurning to a sprin8back condition on the other side' and
the cycle begins again. The more rigidity, the faster this happens'
The intemction of these !!!0 Parameten balance out to provide a constant oscillation speed, which is the first natural ftequency of the slstem. This first natural frequency is the lowest sPeed at lehich the beam
will vibrate after all external excitations are removed, a state known as
free vibation
}o\emed by the following equation:
^nd.
Chopr'r 2, Fundcmenlols
74
Eq. 2.71
e+ @,zo= o
s = csii(o,'+\t')
Eq.2.73 0 =
and it describes the fifst oscillatorl mode of the beam
Note that the o, units are radians per unit time. It is often more conve_
nient to descnbe this natural frequency in terms of cyclesper unit time
(rycles per second is common) using the following nerv wriable.
8q.2.71
Either of these t$'o descriptrons ofthe natural frequency can be used to
calculate the time required for the system to complete one ftrll cycle of
oscillation at this f.equenry. This is known as r}]'e system's natural period.
Eq. 2.75
1^ = t2n
rePresents
In reality, no system is free to i'ibrate indefrnitely. ,ardrg
loss
at
a
molecular
level or
inefficiencies of the material due to energY
of the slstem due to component interaction. In general, damPing
deca)s the vibration of the slstem and recums it to its equilibrium Position in a time peiod that depelds on its danLqtngcoelfcient(c) lhis
coemcient is proportional to dre velocilv 6 of lhe system ard modifles
Eq. 2.71 a5follows:
Eq.2.76
=o
6+2(,@,A
+ 0'.20
Dynamic Andtysis
75
where tl]'e darnpingratia (() has been conveniently introduced as a measure of the severity of the damping. This latio is calculated by
Eo.2.77 L =
'
zman
L
fu .2 .7 8 x= 6Lt
a genemlsolution to Eq. 2.76is found asfollows:
F 4 .2 .7 9
o = A t?'
,{_.,t._,,.",
ce 1tt
undedamryd in cases
. whetego=Aoand6a
: (bl
o)
The three distinct ranges of damping ratio lalues that characterize ststem behavior can be categorized as follows.
.
76
Eq.2.52 u, = a^,14
NoCein Eq. 2.81 thar two nerv constants, Cand \r, have been mathematicall) inroduced to obtain an equation similar in form .o Eq. 2.72. The
value ofthe constants is still determined from initial conditions.
The rate of decav ofan rmderdamped sysFm is represented by the ratio
of consecutive xmplitudes as follows:
6" 2g
'
lz
 e\'1"1/"lt4)
A.*r
Dtlnomie Andtysis
tional and three roladonalto deflne its position ar all times as pafi
of
the overall system. Of course, many of these tariables can be fixed rlue
to gcomerric and/or boundary condition constraints. Such set of grzd,r_
ali.ed noftlinates descrjbes general morion b1 recognizing const;inL
This set rvill have the minimum number of coordinares require.l
to
deflne thc system dvnamically; this Dumber must thus be equal to the
degrees oflieedom of the system.
Fig. 2.34.Discretizedbean
modelwith3D spingand
78
2.s4Mi+r,fl;,t{;J.f:;j"
Ec.
;3liiJ={l}
the vectotialacceleratidn
(M)
modifies
tlre rtar,t majsm,tttt
'nafivt
where
'i
positidn
(K)
modifiesthe oector;al
tSr.t** nnX,*trlratn:{
il
'natris
".i
trt.
krxt
*',*'',
,x2
*!rtr,l * A
79
Dynd'''i. AnqtYsis
1,'i
lxzl
where Xl and X2 are the amplitudesof oscillationof mJand m2 respectively. X is kaown as the system's eigm ectoror agrt rroda Substituting
this solution into Eq. 2.84,the resulting mat x form equation aPPean
as follol"'s.
,
E q.2.86 l K d' \{ l x= l
[ rt , I , , o a ,
 ,
tk2t,rtba.
J x r] = 1f0
l . f
l ^)
LU I
8q.2.87
,a
:!
::.
For rhi. rso deBreeol lreedom sr.rem.Fq. 2.87 vieldsa quadratirin o'
The roots of lhis quadratic are the natural jlequencies of the system,
also kno\a'n as its ngenualues.Ilell e, there are two natural frequencies
for this system,each associatedwith one ofits two eigenmodes.
It is txeful to note that for this s'stem, and any other linear elastic sys1em for that matter, the deflected shape at any given time will be a linear combination of all i!! eigenmodes as follows:
E q.2.8 8x  L x , ' = x E
****"'mru
ff******W
fj;ii:fr**y,;l#:ix,.::;,n:.:"il$s:.i
g1*gg:'*'tru*+
FrequencyResponse
Anolysi
s
:;.i,;;;
il_;T::'5
ii"d,;Tl'i.tJ.:ilil;:;ii:.";1.:tt;i
.*l*
r##:ffir*.,:il:*
*,,"";.,.:,,,*.
l;*,r'',*;##i
illd$Hlffi
;i
'*
;x*T:::(*:il1.""1,f
, udo0"0,,.'*'edegree
offreedom
sysrem
8l
Oyndmi. Anatysis
Fig.2.37.Danqed single4egree
of heedom sYstemsubiactto
Fo snot
ni + ci + k, = Fostniot
Note that this equation is Eq. 2.5 in the presence of a harmonic forcing
function 4=F,sir{Dt, where 4 is the ampliiude of dris force, and (D is it!
driving fiequency. By making the same substitutions made in the free
vibration deriations, this equation becomes the following.
Eq.2.9O i +21$^t+ a^', = :s;n(!t
The solutiol of t}te above equation is a sum of two terms: l]te transiznt
,rolzri.oawhich is the decaying, free vibmtion solution (Eq 2.81), and
the steai4statzsolxttidn,which is any solution to the complete equation.
By assumingsolution tt?es similar to those used in the deriration of the
free vibration equations, a complete solution to the underdamped s)stem is found as follows:
Eq.2.el
, ={t,(.9l'.L,,f,If"'
Eq.2.s2
I rrll I
chdptet 2, Fu ndcmenldl 5
82
ffii
ude Ull,
Fig.2.38.Magnitication
factoranclPhaseangle
vercusfrequencYBno
for differentdamqing
of
ratiasofsingledegrce
aD
83
Dyncmic Anolysit
Tabte2.3.ReprcsenhtivedamPingratios
(lnelastic
Enge)
Melals
withloinis
slruclures
lVelal
ines
steel
lransrnission
/
Allminum
du ngeadnquake
bulldings
Large
0.010.05
can be
For an zdegree o freedom system, all of the above concepts
lor $c
e),oandedin"matrL{fonn. usins a derjrarion 'imilat lo lhrt used
follor":
nFc re\pon5e.l he marrL\ equalion cqui\dlenr ro Eq 2 86
chdpter 2t Furrdd,ntent"ls
84
ol fieedom syslcm
\r alrernate approach for sohing an ddegree
inlolves mnslering Eq. 2.94 into modal space E((D)with fie help ofLq.
2.88 as seenbelo$.
Erl. 2.95
anabsis,rses rhe
Hence, this approach , kr;.o\\n as modallequenq respanse
its
generalized
coordinates, to
of
the
structure,
i
stead
of
mode shapcs
dcscribe moLion. Note that for Eq. 2.95 to hold, the solution to el'ery
single one ofthc sLructure's mode shapesmust be oblained prior to the
anallsis. Of coursc, this is prohibitive fbr sFtenls with a large number of
degrees oflieedorn. Yet, Eq. 2.95 turns out to be a good approximarion
as long as every modc up to a f.cquenqr at least two to threc l;mes the
highest opcraling frequencv of the s,vstemis used. Becausethis number
of modes rvitl, in most cases, be less than the number of generalized
coordinates thar describe the s)stem, using (his method is generall,v
m ch mole numcricallv efficient than direct anal)'sis,although not as
Depending on the FLd package you use, you might be able to select thc
method for carling out your analysis.If the model is small, or subject
to only a fel! excitation frequencies, or subject to high excitation fte
quencics, the direct melhod should be rserl. Thjs nelhod should also
be used if a higher level of accuract is deemed necessarl.Yet, for large
models or models subject to many cxcitation frequencies, thc modal
method will prove to be the betier choice.
85
Dyfi3ntic Andlysis
86
Chapi.r 2t FvndomGn'!. t
E7
5o$tficrY
Summrry
This chapter should be thought of as a Pocket calculator/dictionrry
torr lo I drr\ oul J Lom'
ro
s.r.l*erruu shoul.l nuI e\Pe( t ir Jrone rllo\L
p1".'nn,fl.i. o..n"ut. lou ro definc "ten engincerinc 'uir'liriun r :r
Con
llouta pioue !o be an extremcl]' helptul and concise companion
in
thc
arc
you
$'henel'er
available
Prosirter miking rhis chapter reediiy
presented
fund'unentals
engineering
basic
The
IEe
ofcona"cti"g
l."s
in this chapter should be referenced ar all stagesofthe analysisProcess
time m
At Lhe beginning ofthe FEA process,,voushould alr{a,vstake the
so llill
doing
u.rae..tand the"nature of thi problem at hand because
usuullr s.rrelorr rime l.lrer b' helpine 'oLt aroid unnecean 'al'ul'
rhe andv(r'l
riorr. U hrr are (hr lunddmenlal princiPlesthat will gotern
miSht find
you
How can they be qualifred and quantified? Sometimes
ptoblem
that a simple manual calculadon is all vou need to solve lhe
Try nol ro lose perspective ar this initiai stage'
phe\{hen setling uP ,volrr FEA modet, try to enYision the fundamental
sate will be
nomena to b"ee.raluaredin the analpis. What kind ofstress
Will
modFl;
oi
dre
,au.e.l br rh. geometn and bound.r^ '"ndirions
or
bc
!implilied
il5ell
rl,ere b" bendins, lor.ionl Cdn rlre geomeLn
a
broken into a sum ofsimpler, more manageable shapes?AttemPl Prerts numert_
diction of the resulling slressstate conto$r or even some ot
resul$
eYaluatrng
$'hen
ca.lmagnitudes. This rill prove inv.duable
draw yourNer.t, veriS that the boundary conditions make sense Alwa,vs
confiHoN
state
equilibrium
the
self a free lodv cliagram and e\aluate
biinq
rror
Lion(
dre
re'r'
.l.n' ,,. u.u ,h., ui.xpecr.d and or unde'ired
loads
drc
in,.oau..a by ,.ur r"pre.errurion ol rhe run'traint5l {re
b}' their rep
rhemse\'es adding or subtracting stiffness to the strucrure
to leave out
decide
you
svstem
resentation? Rem;mber, anY part ofthe
aoal,vsis
of the model must bc accurately accounted for in the
.z
i
t
:
t .:
IFA'
As mentroned above, when you hal'e obtained results in vour
Set,il
alrulaLion,,r.p' r" . t'..f, lhF\c b\ meansol a Iough manurl '
lo br anle
r.u ere rn Lh. bcllpark.Of r our \e \oucdnnor e\pecl\oul}ell
are
io solv. every probl.m accuratell after all, this is the reason you
a
closer
off,
usine IL\. Yet, ifvour calculation is an order ofmagnitude
insplction is warranted. Remember Lhat engineering arralysishas been
taking place since long before lhe introdu.tion of the {inite element
88
Choprer2, Fundomentdls
method. l'EA is a relatively new tool. When used with care and forethought, it can certainlyprovide excelleDtresultsin an extremelyeflficient manner. Yet, as with any powedul root, if it is usd blindly, the
magnitudeof the error can bejust asmagnificent.
The material coveredin this chapter wasintentionally conciseand, in
addition to servingasa useful rcference,it wasmeant to remind you or
inlroduce\ou lo somebasiccngineeringcon(ep(s.lr is hightyier ummended that you explore thesetopicsin more depth by usingyour college textbook, the FnA software documentation, or references
mentioned in the Bibliography. Armed wirh the appropriate background information on the technology and underlying phlsics, you will
bc better equippedto succeedin the excitingworld ofsimulation based
designusing finite elementanaltsis.