Course Topics:

Composite beams

Composite columns

Levels of analysis

Frameworks

Torsion

Connections

Stability fundamentals
1
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
Consider two rectangular beams resting on each other, but not attached in anyway:
(non-composite)
b
MEXT = 2 MINT
2C h
MEXT MEXT
= 2 (1 / ρ) EI

= (E / ρ) 2I
Same for both
members
Inon-comp
2C ρ
bh 3 2b(2c )
3
This strain discontinuity 4bc 3
MINT is the heart of the matter! 2I = 2 = =
12 12 3
ρ
If σmax = σy, then there is a limit
on MINT:
MINT M INT c σ yI
σy = → (M INT )max =
I c
2σ y I σ y I non − composite
∴ (M EXT )max = 2(MINT )max = =
c c
2
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
…if we eliminate the strain discontinuity at the interface between beams…
(composite)
b

2C
MEXT
h

θ Describe strain distribution within composite

ρ+c
ρ cross-section: (sin θ ≈ θ)
MINT
ρ-c B* AB = ρ sin(θ )
A*
A B
A * B * = (ρ − c ) sin(θ )
∆L ρ sin(θ ) − (ρ − c ) sin(θ ) c sin(θ )
ε= = =
L ρ sin(θ ) ρ sin(θ )
c
ε = ⇒ Strain varies linearly with depth
ρ
Thus, elementary beam theory applies:
3
1 b(4c ) 64bc 3 16bc 3 σ y Icomposite
M INT = EI composite ; I composite = = = (M EXT )max = (MINT )max =
ρ 12 12 3 c

3
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior

Consider the differences in stiffness and strength that exist between the
composite and non-composite conditions:

16 bc 3
I composite 3 The composite beam is four times stiffer
= 4 bc 3
=4 than the non-composite case.
I non − composite
3

This stiffness result can be used as a basis for comparing strength:

(M EXT − composite )max = σ y 4I non − composite =

2σ y I non − composite
2c c

2σ y I non − composite
(MEXT − composite )max c The composite beam is twice as strong
= =2
(M EXT − non − composite )max σ y I non − composite as the non-composite beam.

4
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
…if we eliminate the strain discontinuity at the interface between beams…
(composite)
b MINT MINT
b

2C h shear connection
MEXT
h
h ∆ε
MINT

∆ε must be made small enough to be considered negligible for practical purposes.

One strategy for achieving this condition is through the specification of mechanical
shear connection – the most common means for this is the use of the headed shear
stud:

The “headed” geometry is

useful in controlling the tendency
for inter-laminar separation, as
well as for controlling withdrawal
of the stud from concrete at ultimate
when the stud is bent over…

5
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
It is common in steel building and bridges for the concrete floor / deck to be made
to act compositely with underlying steel beams / stringers for the resistance of
transverse loading:

Question:
How much of the floor / deck
plate can be counted on for
? participation in the composite
wide?
How section?

6
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
Consider the distribution of compressive stresses in the slab portion of the composite
beam system: (these are nonlinear due to the phenomenon known as “shear-lag”)

σ(x)

Bf
X
ts

B’

B’ (σ max ) B ' ts ∫
= ts σ ( x ) dx

BE = Bf + 2B’
7
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior

Effective width and the transformed section: (assuming full composite action)
BE

ts

Objective: (used in elastic response calculations)

Transform the concrete slab area into an
equivalent area of steel, while preserving
original deck thickness, ts.
Bf
Btransformed = BE / η ; where η = Esteel / Econcrete
All of the section property techniques from
Statics and Deformable Body Mechanics Modulus of concrete is computed as
are applicable here.
1 .5
E concrete = w c fc′ , Wc1.5 is in # / ft3 & fc’ is in ksi

8
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
Explore the notion of “compositeness” – tied to issue of strain compatibility
slip
slip
MSlab
slab
MSlab slab

MBeam
MBeam
beam beam

Non-composite Partially-composite

slab

Slab and beam behave independently Slab and beam are now interacting and
of one another. The neutral axis for the the neutral axes are migrating towards
slab and beam occur at their respective one another as slip is reduced.
centroidal axes.
MComposite
beam
Slab and beam act as one. There is a
single neutral axis and no interfacial
Fully-composite slip.
9
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Behavior
Of course, strain compatibility does not imply
stress compatibility…the difference in elastic
Strain Stress
modulus becomes evident here.

slab

MComposite
beam

Fully-composite

Some interfacial treatment is clearly required to practically enforce this interfacial strain compatibility.
The bond at this interface must be durable and able to maintain effect through cyclic loading and
environmental influences. Frequently, chemical means for achieving adhesion are brittle and not
robust enough to continue functioning in the event of an overload. In metal structures mechanical
connectors are typically employed as the means to transmit the required shear forces across the
interface – NOTICE THE REFERENCE TO FORCE. In design, some degree of interfacial slip is
tolerated as a means to simplify requirements on constructability; while at the same time achieving
desirable composite member response at ultimate loading.
10
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
W(x)
Consider a simply supported composite beam…

-Vertical shear stresses

equilibrate transverse
Shear, v(x) applied loading
-Horizontal shear stresses
equilibrate vertical shear
stresses. Their distribution
through the cross-section
varies according to: vQ
It

This behavior would imply that the mechanical shear connectors should be spaced non-uniformly across the
beam longitudinal axis – this is difficult to ensure at the time of construction, and since the greatest concern is
for ultimate response, shear stud ductility is exploited and a uniform spacing is used. The shear stud ductility
permits minor violations in strain compatibility (at the ends in this case) to permit load sharing among the studs.
At ultimate, all material elements would have the same stress distribution.
11
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
AISC design assumptions and limitations associated with Specification Chapter I:
•In spite of the foregoing, strain is assumed to vary linearly through the
cross-sectional depth
•Material limitations:
•Fy < 75ksi for structural and reinforcing steel
•3ksi < fc’ < 10ksi for normal weight concrete
•3ksi < fc’ < 6ksi for light weight concrete
•Nominal composite strength based on plastic stress distribution†
•Steel has reached its Fy
•Concrete stress has attained 0.85fc’ with a strain not exceeding 0.3%
•For composite sections, the following capacity reduction factors apply:
•For beams
φb = 0.9
•For columns
φc = 0.75
•For composite beams, the effective compression flange width is the sum
of the effective widths for each side of the beam centerline
taken as the
smaller of:
•1/8 of the beam span (taken as the clear distance between support C.L.s)
•1/2 of the distance to the centerline of the adjacent beam
•The distance to the edge of the slab (i.e. the end of overhanging slab portion)
†This may not be true for a built-up section. However, all W-shapes possess
compact webs when rolled from A992 steel. 12
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
Three possibilities exist in relation to the internal stress distribution for fully plastic
composite beam sections (concrete assumed to be ineffective in tension):
0.85fc’ 0.85fc’ Cc
slab slab
Cc
Fy Cs

Ts Ts
beam beam

Fy Fy
PNA in concrete PNA in steel flange
0.85fc’ Cc
slab To locate the PNA, begin by assuming
Cs the concrete force to be the smallest of:

Fy •As Fy (tension yielded steel section)

•0.85fc’Ac (all concrete in comp.)

Ts
beam
• ΣQn (combined stud capacity)
Fy
PNA in steel web 13
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
Example assuming a fully composite section (i.e. shear studs do not control):

120”

4.5” To locate the PNA, begin by assuming the concrete

force to be the smallest of:

•As Fy = (13.0 in2)(50ksi) = 650k  Controls

20.66” •0.85fc’Ac = 0.85(3ksi)(120”)(4.5”) = 1377k

W21x44 • ΣQn
assume to be infinite for now…

Thus, plastic neutral axis is in the concrete.

6.5”

From statics, we know that the longitudinally directed forces must balance, and thus Cc = Ts = 650k.

The foregoing may be used to compute the depth of Whitney’s stress block in the concrete:
650kips
a= = 2.12"
0.85(3ksi )120"

The section plastic capacity can then be computed as Mn = Cc(dC) + Ts(dT) = 650kips[10.33” +4.5”-(2.12/2)]
Mn = 8950k”
φMn = 8055k”
14
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
Now try a heavier section:
120”

4.5”

To locate the PNA, begin by assuming the concrete

force to be the smallest of:
21.51”
•As Fy = (32.7 in2)(50ksi) = 1635k
W21x111 •0.85fc’Ac = 0.85(3ksi)(120”)(4.5”) = 1377k  Controls

• ΣQn
assume to be infinite for now…

12.34”
Thus, plastic neutral axis is in the steel. But is it in the flange
or the web?

Recall the requirement for statical equilibrium along the longitudinal axis: C = T

Note that even if we assume the entire web and the bottom flange to resist the concrete force, we only get
Tf = 12.34”(0.875”)50ksi = 540k & Tw = 1635kips – 2(Tf) = 555k. However, Tf + Tw = 1095k < Cc = 1377k.
Thus it is clear that some portion of the top flange must also participate in resisting the concrete deck force.
The portion of the top flange in compression may be arrived at by solving the longitudinal equilibrium equation:
1377kips + 12.34”(x)50ksi = 1635kips – 12.34”(x)50ksi
x = 0.21”
15
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
We may now arrive at the cross-sectional capacity by summing moments about the
top of the top flange of the steel:
Cc = 1377 kips

slab O
Fy Cs = 12.34”(0.21”)50ksi = 130 kips

Ts = 1635 kips – 130 kips = 1505 kips

beam

Fy

Conc. moment Steel comp. fl. moment Steel tension fl. moment

Mn = MO = 1377k(0.5)4.5” - 130k(0.5)0.21” + 12.34”(0.875”)50ksi(21.5”-0.875”/2) +

0.55”(21.5” – (2)0.875”)50ksi[(21.5” – (2)0.875”)0.5 + 0.875”]

= 20294 k” Steel tension web moment

φMn = 18265 k”
16
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
Within the context of modern building practice, the economics are such that composite
beam strength is typically controlled by shear stud capacity, V’ = ΣQn

Qn = 0.5 Asc fc′E c ≤ Rg R p Asc Fu

Where Asc is the cross-sectional area of a shear stud, and the “R” factors are
reductions factors described in “user note” on page 16.1-87 of the Spec.

As a point of departure, we could use the foregoing capacity equation to arrive at a

suitable combination of Asc and number of studs to satisfy the requirements of the
first two examples. However, a more efficient technique for selecting the appropriate
shear studs would be to use the design aid provided on T3-21 on p. 3-207 of the
Manual.

In the first example, we needed to transfer 650 kips across the interface of each shear
span (assuming a simply supported beam geometry). If we require ¾” studs then from
T3-21 we note that for a plain normal weight concrete deck, Q = 21 kips / stud. Thus,
for the entire beam we require: 650kips
# of studs = = 31studs
shear − span
21kips
stud
17
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
As pointed out previously, it is the shear stud capacity that normally controls; a
condition known as partial composite action. This nomenclature may be deceptive
since the “partial” modifier refers to the ability of the section to develop the requisite
shear capacity for full plastification of all elements within the cross-section. The strain
compatibility condition is not being addressed.

Consider now our first example if, instead of the steel section governing the cross-section
capacity, the shear studs control with Cq = ΣQn = 400 kips:

We begin by finding a statically appropriate cross-sectional depth for Whitney’s stress

block. 400kips
a= = 1.31"
0.85(3ksi )120"
This decrease in strength creates a condition wherein the steel must assist in resisting
compression resulting from a shift in the plastic neutral axis. In other words, as a result
of the stress distribution in the section, only 400 kips in tension can be accommodated
across the concrete-steel interface:
650k − 400k
Acompression − steel = = 2.5in 2
2(50ksi )
2.5in 2 2.5in 2
Accounts for the fact that the same steel x= = = 0.385" < t f = 0.45"
gained in compression is lost in tension. Bf 6.5"
 1.31"   0.385" 
M pc = 650kips (10.33" ) + 400kips  4.5"−  − 250kips   = 8204k "
 2   2 
18
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
Design the following:
•30’ x 30’ panel
•Beams spaced 10’ o.c.
•Be = 90” (1/4 span)
•5” slab thickness built using 2” decking profile
•Normal weight concrete with fc’ = 3ksi
•A992 steel
•WLL = 80 psf, WDL = 80 psf
•A W16 rolled section is called for based on existing requirements for
connection geometry
Based on this geometry,
90” let’s assume that “a” is
1.0”
3.0”
5.0” Y2

W 16x?? Y2 ≡ the distance from the centroid of the

assumed rectangular compressive stress
block in the concrete, to the top of the top
flange [5” – 0.5(1”)] = 4.5” 19
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
We can now evaluate the loading on our structure by factoring the service loads
to arrive at reasonable loads leading to the activation of the governing limit state
(e.g. concrete crushing, steel yielding, horizontal shear failure in the stud region, etc.)

Wu = 1.2 (WDL) + 1.6 (WLL) = 224 psf

Since this is a simply supported beam subjected to a uniformly distributed loading,

we arrive at the required moment capacity as:

Wu L2 0.224ksf (10')(30' )
2
Mu = = = 252k '
8 8
It is now very convenient to use the design aids provided in Table 3-19 of the manual:

•On page 3-183, we enter the table with our Y2 and Mu and find that a W 16x26
handles our requirements with the least weight.
•Further, we note form the table that the PNA falls at “position 6,” as described in
Figure 3-3c on page 3-30 of the Manual.
•We also note that Y1 indicates the location of the PNA relative to the top of the
top flange (i.e. Y1 = 2.04”).
20
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design

Based on Table 3-19, we also see that our required horizontal shear transfer force, ΣQn,
in each shear span, is 145 kips.

Knowing the required horizontal shear capacity at the interface, we may now arrive at
the depth of concrete needed to equilibrate the force:

145kips
a= = 0.63"
0.85(3ksi )(90" )

We can now arrive at our actual Y2: Y2 = 5” – 0.5(0.63”) = 4.69” > 4.5” (assumed)

We can now go back to our table and linearly interpolate between a Y2 of 4.5” & 5”
to see if our factored capacity remains adequate:

274k '−269k ' 274k '− x

= → Mu = 270.9k ' > 252k’ O.K. (of course, we could
5"−4.5" 5"−4.69" have done this by inspection!)

21
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
It is important to investigate the design problem of the construction condition:

•WLL is assumed to be 0.20 klf

•WDL from steel beam self-weight is roughly 0.03 klf
•Concrete deck dead load can be computed as follows

girder spacing and NOT BE

(ALL of the concrete must be carried)
deck thickness
span

30' (10' )(0.333') = 100ft 3

specific weight of
concrete
 lbs 
100ft 3 145 3  = 14.5kips
 ft 

14.5kips
WDL = = 0.483klf ≈ 0.5klf
30'

22
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE BEAMS: Design
We must now factor the construction loads and check to see if the bare steel can
handle the loading:

Wu = 1.2(0.5klf ) + 1.2(0.03klf ) + 1.6(0.2klf ) = 0.956klf ≈ 1.0klf

1klf (30')
2
Mu − construction = = 112.5k '
8

the strength provided by a 30’ long, but fully braced, W 16x26 bare steel beam
is φMn = 166 k’ > 112.k’ OK

Deflections should be checked as well:

4
5(0.041667 k )(30' x12 ) L
∆= in = 1.04 ≈ OK
384(29000 ksi )301in 4 360

Use a W 16x26, A992 member for the steel beam component

23
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Behavior
The two commonly encountered composite column types are either “encased” or “filled.”
In each of these cases, a rolled steel section is combined with concrete that is reinforced
and confined in some way.

Unconfined concrete

Partially confined concrete

Highly confined concrete*

Rolled Section

Longitudinal reinforcing

Confining steel*

Encased Composite Column Filled Column

*The effect of confinement in circular CFTs may be questionable as concrete strengths become
very large1 (e.g. fc’ > 14ksi); although at lower concrete strengths, the confinement of circular CFT
sections is believed to be superior to that of rectangular hollow sections filled with concrete2.
1O’Shea, M.D., Bridge, R.Q. (2000) “Design of Circular Thin-Walled Concrete Filled Steel Tubes,” Journal
Oo Structural Engineering, ASCE, 126(11), pp. 1295-1303.
2 AISC Specification Commentary 24
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Behavior
Concrete material response under confinement:
σ
Compression Response

Confined

Unconfined

Tension Response

Not only does confinement enhance strength characteristics of concrete itself, but it also
leads to significant improvements in composite column ductility; especially for high
strength concrete in encased columns* (i.e. concrete with fc’ > 10ksi)
*El-Tawil, S., Deierlei, G.G. (1999) “Strength and Ductility of Concrete Encased Composite Coumns,”
Journal of Structural Engineering, ASCE, 125(9), pp.1009-1019
25
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Behavior

Behavioral model for internal stress distribution in composite columns:

•Full strain compatibility occurs within composite cross-sections (i.e. all cross-
sectional elements are perfectly bonded together.)
•Maximum concrete compressive strain is 0.3%
•Concrete is ineffective in tension
•Strain hardening response of steel is ignored
•A single numerical factor is used to convert cylinder strength, fc’, into in-situ
concrete strength
0.85 is used for encased columns and rectangular CFTs;
while 0.95 is used for circular CFT - in recognition of the better confinement
afforded by the circular cross-section.
•As composite columns become slender, inelastic and elastic flexural buckling
become manifest in the member. As it is that the existing column curves for rolled
sections do not directly apply to composite columns (as a result of the increasing
complexity in the interactions between cross-sectional components at incipient
Buckling) a modified flexural resistance is employed when computing the flexural
buckling capacity.

26
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design

Limitations on proportions of composite columns:

Encased columns

•The cross-sectional area of the steel core must be greater than 1% of the gross
area. (ASIDE
it is noted that when concrete strengths approach 16 ksi, steel
core areas on the order of 8% to 16% significantly enhance structural ductility and
post-peek strength response in encased columns – although such high concrete
strengths are not currently permitted by AISC – see paper by El-Tawil and Deierlein.)

•Concrete encasement of the steel core shall be reinforced with continuous

longitudinal bars and lateral ties or spirals. The minimum transverse reinforcement
cross-sectional area shall be at least 0.009 in2 / in of tie spacing.

•The minimum ratio of continuous longitudinal rebar area to gross cross-sectional area
shall be 0.4%
this is done to preclude the possibility of an overly, under-reinforced
section which may precipitate a sudden flexural failure.

27
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design

Limitations on proportions of composite columns:

Filled columns

•The cross-sectional area of the steel HSS must be greater than 1% of the gross
area.

•While it is that higher ratios are possible (when justified with testing or analysis), the
maximum b/t ratio for rectangular HSS used in composite columns must conform to:
b E
≤ 2.26
t Fy

•The maximum D/t ratio for a round HSS filled with concrete shall be 0.15 E/Fy . Higher
slenderness ratios are permitted, when justifiable with testing or analysis.

28
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design
Composite column compressive strength:

When Pe > 0.44Po then inelastic flexural buckling controls the column capacity
and the resulting nominal resistance is:
 P
 o


φPn = 0.75Po 0.658  Pe 
 
 

When Pe < 0.44Po then elastic flexural buckling controls the column capacity
and the resulting nominal resistance is:
Encased columns
Ψ = 0.85
φPn = (0.75 )0.877Pe Filled rectangular columns
Ψ = 0.85
Filled circular columns
Ψ = 0.95
Where,
Po = As Fy + Asr Fyr + ΨAc fc′

π 2 (EI eff ) Reinforcing steel minimum specified yield strength

Pe =
(KL )2 Total reinforcing steel cross-sectional area
29
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design
Composite column compressive strength:

An effective flexural rigidity is employed in the calculation of the flexural buckling

load of composite columns as a means for adjusting the existing AISC column
curve so as to more closely agree with empirical and theoretical results involving
composite column buckling. It is noted that this approach is not the most accurate.
The Eurocode, for instance, has adopted a series of three new column curves for
use with composite columns.
Accounts for reductions in concrete section stiffness due to
cracking that occurs during the activation of the stability limit
The current AISC equation for EIeff is: state – the form of the reduction is partially empirical.

EI eff = E s I s + ΓE s I sr + CE c I c The contribution of the reinforcing bars is halved in recognition

of the fact that at least some of the bars will be yielded, and
thus will not be contributing to the flexural rigidity.

where Γ is taken as 0.5 for encased columns and 1.0 for filled columns

and where C is equal to C1 for encased columns, and C3 for filled columns

 As   As 
C1 = 0.1 + 2  ≤ 0.3 C3 = 0.6 + 2  ≤ 0.9
 Ac + As   Ac + As 
30
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design
Example: encased column

22”
0.85fc’ Fy Fyr
#8

22”

9” W 14x90
1.5”
1.5”

W 14x90: 4 - #8 rebar: Normal weight concrete encasement:

As = 26.5 in2 Asr = 3.16 in2 Ac = [(22”)2 – 26.5 in2 – 3.16 in2] = 454.3 in2
Is = 999 in4 Is = 4[0.049087(1”)4 + 0.79in2(9”)2] = 256 in4 Ic = {[22”(22”)3] / 12} - 999 in4 – 256 in4 = 18,266 in4
Fy = 50 ksi Fy = 60 ksi fc’ = 4 ksi

Assume K = 1.0 and column length = 14’

31
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design
Example: encased column

Determine the axial strength:

Po = AsFy + AsrFyr + 0.85fc’Ac

Po = 26.5 in2(50ksi) + 3.16 in2(60ksi) + 0.85(454.3 in2)4ksi = 3,059 kips

Determine the buckling strength:

π 2EI eff
Pe =
(KL )2

EI eff = E s I s + 0.5E s I sr + C1E c I c

 As   26.5 
C1 = 0.1 + 2  = 0.1 + 2  = 0.21 ≤ 0.3
 Ac + As   454 . 3 + 26 . 5 

1 .5
 lbs 
E c = 145 3  4ksi = 3,492ksi
 ft  32
Advanced Behavior of Metal Structures – Dr. Earls
COMPOSITE COLUMNS: Design
Example: encased column

EIeff = 29,000ksi(999in4) + 0.5(29,000ksi)256in4 + 0.21(3,492ksi)18,266in4 = 46,077,823 k-in2

Pe =
(
π 2 46,077,823k − in 2 ) = 16,113kips
2
  12"  
14'  
  1' 

Identify nature of column response: elastic or inelastic buckling:

Pe = 16,113 kips > 0.44Po = 0.44(3,059 kips) = 1,346 kips
inelastic buckling cotrols, (I2-2)

 P
 o

   3,059 kips  
 

Pn = Po 0.658 
P
e  
= 3,059kips 0.658  16,113 kips  
= 2,825kips
   
   

φPn = 0.75 (2,825 kips) = 2,119 kips Filled columns are analagous…

33
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Material Nonlinearity Only – MNO (First-Order Inelastic):
EXAMPLE:

∆ Pcollapse ∆=∞
P P
1 2
Hinging

L L
EI

EI
3EI
K=
L3

1 2

PL3 Hinging ∆
∆= Hinging
3EI

34
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Material Nonlinearity Only – MNO (First-Order Inelastic):

W W1 W2

∆ ∆1st Hinge Collapse

Mp Mp Mp
1st hinge at wall 2nd hinge can only
form after 1st hinge
rotates plastically
Mmax
Mp Mp Mp
Uniformly Load, W

Deflection at incipient collapse

1st hinge

Deflection, ∆
35
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric nonlinearity (Second-Order Elastic):

P P P
∆ ∆’
H H H

M=HL M=HL+∆P M = H L + ∆’ P

Second-Order Converged
P P Moment P Second-Order
Moment
For the given structure, if the combination of H & P do not result in
a convergent ∆, then the structural configuration is unstable. 36
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric nonlinearity (Second-Order Elastic):
EXAMPLE: plane frame
P

Y, u2

2”
1” X, u1
Pcr = 68 kips
1” 20”

frame element
cross-sections 300.00
Equilibrium Trace

250.00

200.00
Standard Linear
Analysis
150.00
Geometrically non-linear
100.00 (softening response)

50.00

0.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
-50.00

37
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (Second-Order Inelastic):

P P P
∆ ∆’
H H H

plastic hinge

M = H L < Mp M = H L + ∆ P < Mp M = H L + ∆’ P = Mp
Collapse!
Geometric
P P Second-Order P
Moment

38
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (Second-Order Inelastic):
EXAMPLE (Ziemian’s [BS ’84, MS ’85,PhD ’90] Frame – Gravity Loading Only):
Statically indeterminate
to degree six – when six
plastic hinge form, the
DL 1.4 k/ft. & LL 2.0 k/ft structure will become a
mechanism.

W 14 x 109
W 14 x 120

15’ – 0”
W 8 x 13

DL 2.1 k/ft. & LL 5.0 k/ft

20’ – 0”
W 14 x 132

W 14 x 120
W 8 x 15

E = 29,000 ksi
Fy = 36 ksi

20’ – 0” 48’ – 0”

39
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (Second-Order Inelastic):
EXAMPLE (Ziemian’s Frame – Factored Gravity Loading Only):

Output from MASTAN2

(Ziemian and McGuire)

40
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (Second-Order Inelastic):
EXAMPLE (Ziemian’s Frame – Factored Gravity Loading Only):

hinge 4,5

hinge 3
hinge 6 hinge 2

hinge 1

41
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity: Mastan2 approach

F = K E + Kσ + K M
T
Tangent Stiffness ( ∇φ ) K E
KM =− T
Matrix K E ∇φ ( ∇φ ) K E ∇φ

P 2 2 4 2 2 6 2 2 4
P M M  P  M   My  P  M  P   My   My  M 
f(P, Mx, My) φ , x , y =  + x  +  + 3.5    x  + 3    + 4.5    x 
 Py M px M py   Py   M px   M py   Py   M px   Py   M py   M py   M px 
                   

∇φ M21 Zero-length plastic hinges

M22
F21

F11 M11 F22

EI
Warp 1 M12
F31
F12
F32 Warp 2
Mx
M31
My M32

42
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (summary of results for Ziemian’s Frame):

43
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity (summary of ADINA results for Ziemian’s Frame):

- denotes plastic hinge

44
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity: ADINA approach with beam elements

M P 
< 0.2 → α ( P ) =  1 − 0.5
P 
 Mn
Mn Pn  Pn 

P 9 P 
1.0 ≥ 0.2 → α ( P ) =  1 −  Mn
Pn 8  Pn 
H1-1b
-1a α (P )
M φ
H1 Mn φy

P 8  Mx My 
+  +  ≤ 1.0
Pn 9  Mnx Mny 

P M My 
+ x +  ≤ 1.0
2Pn  Mnx Mny 

P
Pn
45
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Overview on “Levels of Analysis”
Geometric and material nonlinearity: ADINA approach with beam elements

When the user inputs moment-curvature data for specific axial loads, for use in a
nonlinear analysis within ADINA, no cross-section data or material properties need to
be added in the usual sense. Rather than entering material properties and section
dimensions directly, ADINA infers these from the moment-thrust-curvature response
data input by the user.
P σy A α ( P ) Mn
Axial Force

Moment
c ε M
≈φ =
EA c EI EI

Axial Strain Bending Strain

46
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Discussion of “Second-Order Effects”
Designers oftentimes differentiate between two general types
of second-order effects within frameworks:
P
∆
Member – level (P-δ)
[emanates from member out-of-
straightness]

Frame – level (P-∆)

[emanates from inter-story drift]

47
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Discussion of “Second-Order Effects”

Initial imperfections leading to second-order effects in frameworks:

Member – level (P-δ): Max. permissible member out-of-straightness – L/1000

[Section 7.13.1.2 AISC Code of Standard Practice for Steel Buildings and Bridges, March 18, 2005]

Frame – level (P-∆): Max. permissible out-of-plumbness – (1/500) x Story Height

[Section 7.13.1.1 AISC Code of Standard Practice for Steel Buildings and Bridges, March 18, 2005]

In order to account for these nonlinearities, as well as nonlinear effects induced by

deformations under load, inelastic material behavior, thermally induced residual
stresses, etc., one may employ a robust commercially available FE software system
(such as ADINA) to perform an analysis where all of these features are incorporated
into the model. However, at present there is not much guidance (none?) on how to take
stochastic quantities, such as initial imperfections, and include them in the modeling
supporting a particular design instance (for example).

48
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Discussion of “Second-Order Effects”
Another approach would be to perform a second-order elastic analysis that
employs the actual structural loads, plus additional “notional loads” that AISC
recommends as a means for accounting for the previously mentioned sources
of nonlinearity, as well as: patterned gravity loading, temperature gradients
across the structure, foundation settlement, uneven column shortening…”
AISC (2005). The members would then be checked using the strength provisions
within Specifications for Structural Steel Building (AISC 2005).

The notional loads are computed as being 0.002 x times the summation of gravity
effects felt by the given story. The notional loads are positioned to be laterally
acting concentrated forces imposed at the top of the story under investigation.

½ Pgravity ½ Pgravity

0.002 (Pgravity)

USE K = 1.0 FOR ALL MEMBERS!!

49
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Design Example
Consider Ziemian’s Frame:
DL 1.4 k/ft. & LL 2.0 k/ft

W 24 x 102

W 14 x 109
W 14 x 120

15’ – 0”
W 8 x 13

DL 2.1 k/ft. & LL 5.0 k/ft

W 36 x 170

20’ – 0”
W 14 x 132

W 14 x 120
W 8 x 15

E = 29,000 ksi
Fy = 36 ksi

20’ – 0” 48’ – 0”

The frame was proportioned such that it should achieve it’s structural collapse
load under the factored gravity condition (1.2DL + 1.6LL).

50
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Design Example
An additional notional load of 0.001Y
is imposed here to eliminate the need
0.003 [ 816” (0.4076 k / in.) ] = 0.9978 Kips for checking τb in equation A-7-2. This
allows for the uniform reduction in the
0.003 [ 816” (0.8772 k / in.) ] = 2.1474 Kips elastic modulus described below.

Factored Gravity – Roof: 0.4076 k /in.

Roof Notional Loading
W 24 x 102

W 14 x 109
W 14 x 120

15’ – 0”
W 8 x 13

Factored Gravity – 2nd Story: 0.8772 k /in.

2nd Floor Notional Loading

W 36 x 170
E* = 23,200 ksi

20’ – 0”
W 14 x 132

W 14 x 120
W 8 x 15

This reduced modulus accounts

for the stiffness reductions of
(A-7-2 &3): the former reflects
the erosion of the Euler buckling
load due to member imperfections
while the latter accounts for the
so called “spread of plasticity”

20’ – 0” 48’ – 0”

* Use 29,000 ksi when checking serviceability

51
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Design Example
Analysis results: using MASTAN2
Apply notional loads in such a way as to create the
most critical condition…

S8

S6
S2

S4
S7

S3
S1

S5
NOTE: All beam sections are compact and all members are continuously braced in the out-of-plane
direction. All member cross-sections are oriented to have the strong axis resist in-plane loading.

52
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Design Example While Cb is not needed in this example since
members are fully braced against LTB, it is still
provided as a reminder that it may be needed
Analysis results: using MASTAN2 in other applications.

S1 P = 59k M = 29.1k” Cb = 1.313 S8L P = 1.8k M = 5887k” Cb = 3

P = 62k M = 62.8k” Cb = 1.294 P = 0.9k M = 5805k” Cb = 3

S2 P = 25k M = 80.0k” Cb = 2.222 S8R P = 74k M = 12320 Cb = 1.9

P = 25k M = 89.6k” Cb = 2.551 P = 75k M = 12410 Cb = 1.909

S3 P = 658k M = 2856k” Cb = 1.55 Tensile axial loading is conservatively assumed

P = 658k M = 4027k” Cb = 1.55 to be zero.

Moment gradient coefficient, Cb, is limited to being

S4 P = 200k M = 6674k” Cb = 2.257 less than or equal to 3.
P = 200k M = 6809k” Cb = 2.259

S5 P = 330k M = 2931k” Cb = 1.603 Interaction Equation Results:

P = 327k M = 1942k” Cb = 1.19 S1: 0.67 (H1-1a)
S2: 0.31 (H1-1b)
S6 P = 108k M = 6739k” Cb = 2.262 S3: 1.04 (H1-1a)
P = 107k M = 6772k” Cb = 2.259 S4: 1.08 (H1-1b)
S5: 0.69 (H1-1a)
S7L P = 1.3k M = 16990 Cb = 2.767 S6: 1.14 (H1-1b)
P = 0k M = 16420 Cb = 2.798 S7L: 0.79 (H1-1b)
S7R: 1.26 (H1-1b)
S8L: 0.60 (H1-1b)
S7R P = 0k M = 26520 Cb = 1.713
S8R: 1.30 (H1-1b)
P = 0k M = 27260 Cb = 1.735

53
Advanced Behavior of Metal Structures – Dr. Earls
FRAMES: Additional Thoughts…
•Frame analysis employing notional loads is valid only for tiered buildings with
orthogonal framing between floors and columns.

•When designing foundations, do NOT include notional loading when computing

base shear – foundation moments should include notional load effects, but the base
shear should not.

•If the sway deformations of the structure are complex, then it is important to do a first
order analysis of the structure, under factored loading, in order to understand the sway
deformation response. Once this is known, apply the notional loads in directions that
are additive to the deformations observed in the first order analysis (i.e. your notional
loads will not necessarily all be in one direction within a single framing plane)

•Notional loads primarily are meant to account for P-∆ second-order effects (despite
what Appendix 7 says!). If your drifts are limited to be less than story height / 500,
then you may want to use a smaller notional load.

•P-δ second-order effects are accounted for in the geometric stiffness matrix in the
elastic second-order analysis; as well as through the use of design provisions
employing a “column curve” (complimented by the 0.8EA reduction).

54
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory

Review of elementary mechanics

Torque
Moment acting about the longitudinal

axis of a structural element

Consider a circular shaft

differential area, da
Reactive torque Equilibrium
(at a cut) shear stress
on the cut
ρ

c
c
Torque = (force )x (dis tan ce ) = ∫ ρ (τ )da
0
Applied torque 55
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory

Review of elementary mechanics

Does a cross-sectional slice within the circular shaft warp under the action of
a uniform torque (i.e. does a slice that was originally flat distort and become non-flat)?

Let’s do a thought experiment –

Reactive torque If cross-sectional slice warps then we may

imagine a point on the circumference of the
yellow section, ω, that moves away from an
observer standing at location β as a result of
warping distortion. However, to an observer
α at α the loading is identical to that observed
ω at location β, and yet the point on the yellow
β section, ω, appears to move in the opposite
direction from what is observed at location β.
Applied torque hmmm…this is not consistent.

Q.E.D - Proof by contradiction

56
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory

But wait, according to this proof warping can never occur with a constant torque…

Observation regarding previous proof –

The conclusions from the earlier proof by

contradiction are predicated on an axi-
Reactive torque symmetric cross-section. Meaning that it
is perfectly acceptable to have a single point
that is at once moving towards α, and away
from β, since the views are preserved from
α section to section (i.e. no matter which end
the observer is at, they will see the same
ω
β RELATIVE point moving towards or away)

Applied torque Preservation of consistency does not

preclude the occurrence warping in general;
it only precludes it for the axi-symmetric case
57
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory
Reactive torque

γ
A For small twist, γ, the circumferential
A’ distance between A and A’ is γ L.

Applied torque
A
…and so, γ L = φ ρ
Considering, now, the section itself, A’ φ
φρ γL
the circumferential distance between ∴γ = ;φ =
A and A’ is φ ρ. L ρ
ρ
0 to c
φc γ max L
⇒ γ max = ;φ =
L c
substituting for φ in the γ expression
reveals the linear variation of shear
through the cross-section
58
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory

γ
Reactive torque

γ
A
A’

γ is a shear strain that can be used

L
to compute shear stress by applying
Hooke’s Law.
Applied torque
Shear Stress, τ

Shear Modulus, G

Shear Strain, γ
59
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – engineering theory

Now we may relate shear stress to Torque:

ρ ρ
Gγ = Gγ max ⇒ τ = τ max
c c τ max J Tc
T = ;τ max =
ρ
∫ ∫c
c J
T = (τρ )da = τ max ( ρ )da
τ max
=
c ∫ ρ 2da Compute angle of twist:

 cφ 
polar moment of inertia, J, or G  J
St. Venant’s torsion constant, KT Gγ max J  L  JGφ
T = = =
c c L
LT
∴φ =
JG

60
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Theory of Elasticity
See for instance:
Timoshenko, S.P, Goodier, J.N. (1934) Theory of Elasticity, 3rd ed., McGraw Hill, p 295.

St. Venant torsion uniform torsion ≡ unrestrained cross-sectional warping and

torque is constant along member length

Considering equilibrium of internal forces, compatibility in deformations, a linear-elastic

material model, and recognizing that shear does not develop on a free surface, we
arrive at the governing differential equation for torsion:

∂ 2ψ ∂ 2ψ ∂γ
2
+ 2 = −2G
∂x ∂y ∂z

Where G is the shear modulus, φ is the angle of twist, and ψ is a so-called

“stress function.” Solution of the torsion problem consists of simply finding
a stress function that simultaneously satisfies the foregoing governing equation
as well as an prescribed boundary conditions in the problem at hand ☺

…not a simple matter in most cases.

61
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
Warping may occur, but it is unrestrained…cross-section is thin-walled: b >> t

In such a case, there is a fortuitous correspondence in form between the previously

mentioned governing equations for torsion in an linear-elastic body, and the equations
describing the deformation of a thin membrane stretched over an open frame of
dimensions b and t: end effects are neglected
z in this approach…
y
b

y q α
x MSV t F ž F
žo
Membrane stretched

∂z
over opening b x t
∂y
MSV
ž ≡ height of membrane at cross- τmax
sectional mid-line 0
q ≡ pressure acting on membrane
F ≡ circumferential force holding
the membrane to the frame
62
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
Analogous expressions between cross-sectional distortion in thin-walled members
and a membrane stretched over an opening with the same proportions as the cross-
section:

MEMBRANE TORSION



∂ 2z ∂ 2z q ∂ 2ψ ∂ 2ψ ∂γ
+ = − + = −2G
∂x 2 ∂y 2 F ∂x 2 ∂y 2 ∂z
deflection, ž stress function, ψ
q ∂γ
2G
F ∂z



∂z ∂z
slope ; stresses τxz; τyz
∂x ∂y

∫∫


volume V =
∫∫ zdxdy torque MSV = 2 ψdxdy

63
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy

height of membrane at cross-sectional mid-line:

y


4zo y 2
z=
t2

ž slope of this parabolic expression:

žo



dz 8zo y  Slope, and thus stress,
= 2 varies linearly across
dy t the plate thickness

slope is max at y = ± (t / 2):

4z
(τ SV )max = o
t
64
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
ends are neglected

∑Fž = 0: q t b – 2 b F sin(α) = 0
y

projected membrane area

α
F ž F
žo For small membrane deformations (i.e. α < 15˚):

sin(α ) ≈ α =


( 2) = 4z

dz t
o
dy t

which, upon substitution, leads to:

 4zo  q 8zo
qtb − 2bF  =0→ =
 t  F t

65
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
The volume under the membrane can be computed as:
t
t
y 2
2
2 3 2

4z y 
4z y 
α
∫
2b zo − o2 dy = 2b zo y − o 2 
0
t  3t 
0
F F


ž 
zo t  2tzo b
=  zo t − b =
žo  2  3
Based on the membrane analogy, this volume is equal to ½ MSV and thus,


4tzo b
MSV =
3
Going further with the membrane analogy…


q ∂γ
3MSV q 8zo
= 2G , zo = , and = 2
F ∂z 4tb F t
 3MSV 
8 
 4bt  6MSV ∂γ
lead to : = = 2G
t2 t3 ∂z
 ∂γ 
t 3bG 
⇒ MSV =  ∂z 
3 66
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
We may now isolate a St. Venant torsion constant (same as polar moment of inertia):
t 3b
KT =
3

The maximum shear stress in the thin-walled element may then be described by:
(τ SV )max = tMSV
KT
In the case of a more complicated cross-section composed of multiple straight, thin-
Walled elements:
i =n
t ij MSV 1
(τ SV ) =
KT
; where we now have KT =
3 ∑b t ;
i =1
3
ij ij n = number of cross - sectional plate elements

It is pointed out that the KT developed in this way is only valid for thin-walled, open
sections; and then is only valid when applied to straight elements making up the
cross-section. While sharp corners in the cross-section can lead to stress raisers
that create a condition of extremely high local shearing stress, Professor Bruce
Johnston (1936) showed that the fillets in hot-rolled shapes are sufficiently smooth
to result in only modest stress increases – the resulting shear stresses are of the
same magnitude as the shear stress in the straight elements joining at the fillet.
67
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – Membrane Analogy
Computing KT:
i
1 i =n
1
tij
KT =
3 ∑
i =1
bij t ij3

bij n

Example: I – shaped cross-section

MSV t
τ=
t KT

d
w

1
KT =
3
[
2bt 3 + (d − t )w 3 ]
b
68
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping
We will now focus on the kinematics associated with warping of a non-circular
cross-section subjected to uniform torsion

NOTE: just because a section warps, does not mean it is experiencing “warping
torsion.” The latter has a specific meaning that will be addressed later in course…

Consider a thin-walled open cross-section:

x, y) ρ
t Q(
ta
en a
ang
T x
C
φ
MSV

y
E
ρo
- denotes centroid
S (xo, yo)
- denotes torsion center
69
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping
ρ and ρo are positive when the observer
positioned at Q(x,y), and walking in the
direction of integration denoted by “s”,
sees the centroid and torsion center,
respectively, on the left.
fs ight
Line o

,y )
t at (x ρ
n Q
nge a
Ta x
C
φ
MSV

y
E
ρo

S (xo, yo)

70
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping
Note, all discussions and formulations that
follow are referred to the middle plane of the
cross-sectional plate components, and thus
the effects of St. Venant shear are zero at this
z - coordinate location within the cross-sectional plate
thickness; thus the given differential cross-
sectional element under investigation maintains
its right angles.

dz
A

C
B’
B C’
dw
D’ D ds

C
r
Differential warping deformation is seen here as the
component of the deformation from B to B’ acting z
along the z – axis. Warping is positive when acting
in the positive z – direction (i.e. positive in elongation).
x
y After application of the torque
φ
dφ cross-sections z + dz will be
S deformed through an angle dφ
MSV with respect to the neighboring
cross-section dz away, at z.
71
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping
Consider the kinematics of the patch A B C D within x-y and s-z planes:
ds φ
rdφ
B, D
D’ A, C
x – y plane β tangent
C’
β
90˚ proje
ction line

s – z plane

r
 dφ  B’
dw = − r sin β ds
 dz  ρo
since B A
r sin β = ρo then, φ
dφ

 dφ 
dw = − ρo  ds D’
 dz  dw dz
d C’
C S
rdφ sin β
rdφ sin β
dz 72
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping

After integration of the differential warping deformation term, we obtain:

dφ s
w = wo −
dz ∫
0
ρ o ds

where “w” is the warping deformation of any point on the middle line at a
distance “s” from the edge “O” and “wo” is the deformation of point “O”.

Let us now define the so-called “double sectorial area” or “unit warping with
respect to the torsion center, S
s

∫
ωo = ρods
0

Leading to the following form for the warping deformation within a cross-section:

w = w o − φ ′ωo

73
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping

ds dy

dx Cross-sectional segment
f C(0,0)
γ
x γ
γ
g h e
Q(x,y) γ d
Geometric measures:

i Cd = y fd = x sinγ ed = y cosγ
tangent

Qi = ρo fe = ρ Qg = x – xo

Qd = x gS = yo – y gh = gS cotγ
S ( xo , yo )
dy dx dx y
= sin γ = cos γ = cot γ
ds ds dy
Calculating ωo:
dy dx
From the foregoing, we note that ρ = fe = fd − ed = x sin γ − y cos γ = x −y
ds ds
dy dy dx dx dx dy
and ρo = Qi = (Qg + gh ) sin γ = [(x − xo ) + (y o − y )cot γ ]sin γ = x − xo −y + yo = ρ + yo − xo
ds ds ds ds ds ds
Qg gh
74
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: St. Venant torsion – cross-sectional warping

dx dy
Recognizing that ρo = ρ + y o − xo may be integrated from O ( x1 , y1 ) to s ( the
ds ds
location of Q ( x , y ) measured along the curvilinear coordinate line), yields:

s s x y

∫ ∫
ωo = ρods = ρds + y o dx − xo dy
0 0 ∫x1 ∫ y1

Integrate Integrate
wrt “s” wrt “x” & “y”

Integration is carried out to the location of point Q

whose coordinates are s , x , and y.

Thus obtaining: ωo = ω + y o x − y o x1 − xo y + xo y1

It can then be seen that the relative warping of one cross-sectional point to any other
point in the same cross-section is proportional to ωo

75
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion

If warping deformations are constrained within a member (e.g. by welding thick

plates to the end of an I-beam), or if the applied torque is not uniform along the
member length (e.g. a concentrated torque applied at mid-span), then the torsional
response is characterized as being one of: “warping torsion”, or “non-uniform torsion”
(each of thee terms is equivalent).

As a result of non-uniform torsion, additional longitudinal and shear stresses arise.

These stresses are additive to the stresses induced by flexure, axial loading, and
St. Venant’s torsion.

Thus the longitudinal warping stress may be expressed according to Hooke’s Law:

dw
σ w = Eε w ; ε w =
dz

The warping strain term may be expanded on in terms of earlier results…

76
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion

Considering the earlier expressions for warping deformations:

d (w o − φ ′ωo )
εw = = w o′ − ωoφ ′′
dz

Represents the change in d (ωo )

= 0 Since member is prismatic
warping deformation of dz
“O” as a function of the
longitudinal coordinate “z.”

Thus the longitudinal warping stress may be expressed as:

σ w = Ew o′ − Eωoφ ′′

77
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion
Consider equilibrium in a cross-sectional strip “s” by “dz”:

s s

∫ ∫
τ w tdz + t (σ w + dσ w )ds − tσ w ds = 0
0 0
σw t
dz

τw = 0
O
dσ
s

∫
τ w t = − t w ds
0 dz σw ) t
( σw + dσ

τw t dz
It is noted that under the action of a torque, C
axial force and bending moments within the
cross-section must be zero and thus we may z
state the requirements for self-equilibrium in x
warping stress as: E E y
P =0=
∫ σ tds = E ∫ (w ′ − ω φ ′′)tds
0
w
0
o o
The x, y, and z axes are
E E principal centroidal
Mx
∫ 0 ∫
= 0 = yσ tds = E (w ′ − ω φ ′′)ytds
w
o
o o

E E

∫ ∫
Integration carried out over
My = 0 = xσ tds = E (w ′ − ω φ ′′)xtds
w o o Entire cross-section
0 o
78
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion
Considering the foregoing integral expressions for axial and moment equilibrium,
the meaning of several expressions may be noted:

due to x, y, and z being the cross-sectional principal centroidal axes,

E E E

∫ xtds =∫
0 0 ∫
ytds = xytds = 0
0

From the definition From the definition of

of a centroid the product of inertia

…and recalling mechanics of materials,

E E E

∫ ∫ ∫
2
tds = A, x tds = I y , y 2tds = I x
0 0 0

79
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion
Warping products of inertia
(used to locate the shear center)
…considering also the following… Iωy Iωx
xo = , yo = −
Ix Iy
Identifies the location of a point within a
E E cross-sectional plane (not necessarily

∫ ωxtds = I ∫ ωytds = I
0
ωx
0
ωy
the centroid) through which a transverse
shear force may be applied to induce
bending without inducing torsion in the
member

It is subsequently seen that cross-sectional equilibrium considerations yield:

φ ′′(− xoI x + Iωy ) = 0

φ ′′(y oI y + Iωx ) = 0

…and since φ’’ ≠ 0, thus the coordinate of the torsion center may be arrived at:
Iωy Iωx
xo = , yo = −
Ix Iy
80
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion

It is thus observed that the “torsion center” and “shear center” are the same point.

NOTE: this statement is only strictly true if the axial loads and bending moments
are zero. If this is not the case then the “shear center” location stays put, but the
“torsion center” migrates.

Considering again now the equation expressing the equilibrium of axial loads:
E E
P =0=
∫
0 ∫
σ w tds = E (w o′ − ωoφ ′′)tds
0

It is observed that, Cross-sectional area, A

E E φ ′′ E

∫0 ∫
w o′ tds = φ ′′ ωotds
0
w o′ =
A ∫ ω tds
0
o

the foregoing result may then be used in the previously obtained equation for
warping normal stress: σ w = Ew o′ − Eωoφ ′′

81
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion

Thus the new expression for warping normal stress is:

1 E 
σ w = Eφ ′′
A 0 ∫
ωotds − ωo 

A new cross-sectional property may be defined, the normalized unit warping:
1 E
ωn =
A 0 ∫
ωotds − ωo
Thus leading to an even more compact form for warping normal stress:

σ w = Eωnφ ′′
Recalling the previously derived expression for warping shear flow:
s dσ w
τwt = − t∫0 dz
ds

Integration to a point of interest on the cross-section yields (with substitution for σw):
s

∫
τ w t = − tEωnφ ′′′ds
0

82
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Non-uniform or warping torsion
…yet another cross-sectional property may now be defined, the warping statical
moment:
s

∫
Sw = ωntds
0

Thus permitting an even more compact warping shear flow expression:

τ w t = −ESwφ ′′′

Based on the foregoing, the following observations are made:

Goes as
σ w → φ ′′
τ w → φ ′′′
Mw → φ ′′′

83
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion
Consider the contribution to the total torsional moment, Mz, attributable to warping
shear flow arising out of a condition of non-uniform torsion:

∫
Mw = τ w t ρods
0

Force per length Moment arm about S.C.

Substitution of the previously obtained warping shear flow expression yields:

E
 s 
Mw = −Eφ ′′′ ∫0 ∫
ρo  ωntds ds
 0 
Simple integration by parts may now be applied as follows:
s

∫
u = ωntds, dv = ρods
0

s
⇒ du = ωntds, v = ρods = ωo
∫0

84
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion
Applying the usual integration by parts equation,

∫
uv − vdu
1 E  E
0
E

Results in,
 ∫
A 0 ∫
ωotds  ωntds −
0 ∫
0
(ωn )2 tds
*
 E E E

 0 ∫ 0 ∫
Mw = −Eφ ′′′ ωo 0 ωntds − ωoωntds 


1 E
From an earlier result ωn =
A 0 ∫
ωotds − ωo

As a cross-sectional property, this term is a constant

* Thus leading to: A

1 E  E E E

A 0 ∫ ∫
ωotds  tds − ωotds = 0
0 0 ∫ ∫
⇒ ωn tds = 0
0
85
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion
As a result of the foregoing, all but one term is annihilated thus simplifying the
Expression for warping torsion:

E
Mw = −Eφ ′′′
∫
0
(ωn )2 tds

A new cross-sectional property
the warping moment of inertia, Iω

or,

Mw = −EIωφ ′′′

Thus, by superposition, the expression governing differential equation for the

response of structural members possessing elastic thin-walled open prismatic
cross-sections is:
M z = Msv + Mw
M z = GKT φ ′ − EIωφ ′′′ Valid for the case of concentrated end-torques
producing a constant torque condition along the
members longitudinal axis, z.

86
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion
A similar expression may be obtained for the case of a distributed torque
(i.e. a torque that is uniformly distributed along the longitudinal member axis).

Consider equilibrium on the free body drawn below:

− M z + mz dz + M z + dM z = 0
dz
dM z Mz
⇒ = − mz
dz

Thus the governing differential

mz
equation for the case for with a
distributed torque can be arrived at
by simply differentiating the earlier
governing equation, yielding:
z
Mz + dMz
IV
GKT φ ′′ − EIωφ = −mz
x
y

87
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion

The foregoing differential equations may be re-expressed in an equivalent form

facilitating the identification of a solution:

Mz
Equal and opposite concentrated end torques
φ ′′′ − λ2φ ′ = −
EIω

mz
Distributed torque along the member z – axis
φ IV − λ2φ ′′ =
EIω

Where,

GKT
λ2 =
EIω

88
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion

The solution for the case of equal and opposite concentrated end torques is given as:

Mz z
φ = A + Be λz + Ce − λz +
λ2EIω
or,

Mz z
φ = A + B cosh(λz ) + C sinh(λz ) +
λ2EIω

For the case of a distributed torque along the member z – axis, the solution is:

mz z 2
φ = A + Bz + C cosh(λz ) + D sinh(λz ) −
2GKT

where A, B, C, and D are arbitrary constants of integration; the values of which can
Be discerned from the boundary conditions of the specific problem being solved.

89
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Differential equations governing torsion

Expressions for physically meaningful boundary conditions:

Twisting restrained φ=0

Warping restrained φ’ = 0

Warping free φ’’ = 0

Free end φ’’’ = 0

(shear flow due to warping is zero)

90
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Summary of cross-sectional properties

Properties that depend on the cross-sectional curvilinear coordinate “s”:

∫
ω = ρds
0
“Double Sectorial Area” / “Unit Warping wrt Centroid”; useful in computing the warping
product of inertia

∫
ωo = ρods = ω + y o x − y o x1 − xo y + xo y1
0
Relative warping of one point at “s” wrt another point
with warping deflection wo

1 E
ωn =
A 0 ∫
ωotds − ωo Normalized unit warping; useful in computing warping normal and shear stress

∫
Sω = ωntds
0
Warping statical moment; useful in computing warping shear stress

91
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Summary of cross-sectional properties
Overall cross-sectional properties:
E E E

∫ xtds = ∫ ytds = ∫ xytds = 0

0 0 0
Properties of centroid

E E E

∫ ∫ ∫
2
tds = A; y tds = I x ; x 2tds = I y Area moments for use with elementary beam theory
0 0 0

∫ ωytds = I
0
ωy
“Warping Product of Inertia”

Useful in computing shear center location

E

∫ ωxtds = I
0
ωx “Warping Product of Inertia”

∫0
ωn2tds “Warping Moment of Inertia”; used in warping term within the governing DE for torsion response

Iωy I ωx
xo = ; yo = − Shear Center coordinates in centroidal reference frame
Ix Iy
92
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Summary of cross-sectional properties

Units of previous quantities:

Length : x, y, s, xo, yo

Length2 : ω, ωo, ωn, A

Length4 : Sω, Ix, Iy

Length5 : Iωx, Iωy

Length6 : Iω

Mapping of our terms into those employed in the AISC Manual:

KT
J
This class… Iω
Cw …AISC Manual
ωn
ωn
Sω
Sw

93
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Summary of torsional stresses

(τ SV )MAX =
(M SV )t Intensity varies linearly across thickness of
KT cross-sectional component

σ w = Eωnφ ′′
Warping induced; uniform across thickness of cross-
sectional component.
ESωφ ′′′
τw =−
t

94
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Combined stresses

We may now incorporate torsion into our existing theories from structural mechanics:

AXIAL BENDING TORSION

P Mx y My x
σ = + − + Eωnφ ′′
A Ix Iy

s s

τ=
∫
− Vx xtds
0
−
∫
Vy ytds
0
−
ESωφ ′′′
+ Gtφ ′
tI y tI x t

BENDING TORSION

95
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Combined stresses

…and what about the signs?

(remember: tensile stresses positive)

Signs arise out of positive axial

load eccentricity.

Tensile warping assumed.

P Mx y My x
σ = + − + Eωnφ ′′ z
A Ix Iy ex
The internal shear stress equilibrating ey
the external shear arises out of the an x
equal and opposite internal shear reaction. y
s s

τ=
∫
− Vx xtds
0
−
∫
Vy ytds
0
−
ESωφ ′′′
+ Gtφ ′ Vx P
tI y tI x t
Vy

Tensile warping strain results in a

negative warping shear stress. Always positive since St. Venant’s shear flows both ways in a
a given plate element – there is always a positive choice.
96
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Combined stresses
…and now to the design problem!!

P Mx y My x dictates longitudinal stress variation

σ = + − + Eωnφ ′′ P, M x , M y ,φ ′′,Vx ,Vy , φ ′′′,φ ′ *
A Ix Iy
defines

response 1 s 1 s
∫ ∫
s s
x, y , ωn , ytds, xtds, Sω , t
τ=
∫
− Vx xtds
0
−
∫
Vy ytds
0
−
ESωφ ′′′
+ Gtφ ′
t 0 t 0
dictates cross-sectional stress variation
tI y tI x t

…thus, in design, consideration must be given to both the longitudinal position as

well as the cross-sectional location; as critical stress conditions are searched for.

*Helpful charts depicting variations in φ’, φ’’, φ’’’ may found in:
Heins, C.P., Seaburg, P.A. “Torsion Analysis of Rolled Sections,”
Bethlehem Steel Company, Steel Design File No. 13-A-1, 1964.

97
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
For sections built up from thin flat plates, the cross-sectional warping properties
vary linearly between points of intersection ☺!

…as an example, consider the variation in unit warping with respect to the centroid
in the general flat cross-sectional plate element drawn below:
ω
Consider the equation of a line:
x y = mx + b
x
dx
ωj ω ωi ω (x – xi) ωi
ω j − ωi
t xi,yi αij y
t x j − xi
cos α ij
αij
ds
ω = ωi +
(ω j − ωi )(x − x i )
(x j − x i )
xj,yj tij bij
i =n
t ij xj  (ω j − ωi )(x − x i )xdx
I ωx = ∑ cos α ∫
i =0 ij xi

ω
 i +
x j − xi


y
98
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design ω

Formulas for sections with straight plates: x

x
dx
i
ωj = ∑ρ b
i =0
ij ij
ωj ω ωi
xi,yi αij y
i
ωoj = ∑ρ
i =0
oij bij the “j” denotes the value after
integration along the element
1 1 
i =n
ωnj = 
A  2 ∑ (ωoi + ωoj )t ij bij  − ωoj
i =0 
ds

xj,yj tij
1
i =n i =n bij
I ωx = ∑ (ωi x i + ω j x j )t ij bij + 1 ∑ (ω x i j + ω j x i )t ij bij
3 i =0
6 i =0
i =n i =n
1
I ωy = ∑ (ωi y i + ω j y j )t ij bij + 1 ∑ (ω y i j + ω j y i )t ij bij y
3 i =0
6 i =0
i =n
1
Iω =
3 ∑ (ωi =0
2
ni )
+ ωni ωnj + ωnj2 t ij bij

i =n
1
KT =
3 ∑
i =0
t ij3 bij

99
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Consider the following example involving an I-shaped beam:

z - axis

L/2 L/2

eccentricity, e
tf
Q
tw
x - axis
Shear Center d’ = d - tf
& Centroid

y - axis
bf
100
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Compute cross-sectional properties:
Given the regular geometry of the I-shaped cross-section, the integrations
are trivial and may be carried out in tabular form

3 6
4
+ +

Shear Center
& Centroid d’

+ 2 -
1
5
bf / 2 bf / 2
101
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design Sign applies to term at the head of the arrow.
(negative means opposite to the assumed integration direction)

NOTE: The integrations may be performed

ρ = ρo b ρb ω = ωo ωn in any order; the same ωn will be obtained.
However, the order of integration will impact
1 +
0 d’ bf / 4 * the values of ω and ωo ; as intermediate steps.

d’ / 2 bf / 2 d’ bf / 4 =
2 +
d’ bf / 4 0
No warping shear and normal stress in
0 d’ 0 = the web…
3 + d’ bf / 4 0
d’ / 2 bf / 2 d’ bf / 4 =
4 d’ bf / 2 -d’ bf / 4* 3 6
5 = d’ bf / 2 -d’ b / 4 *
f
4
+ +
-d’ / 2 bf / 2 -d’ bf / 4 -
2 d’ bf / 4 0 Here we back
into the results
6 = 0 d’ bf / 4 * from joints 2 & 3
Shear Center
& Centroid d’

d’ / 2 bf / 2 d’ bf / 4
-
3 d’ bf / 4 0
+ 2 -
* The warping normal stresses are distributed in a manner that is consistent 1
with minor axis bending of each flange, in the opposite sense from each other. 5
This observation leads to the notion of a “bi-moment” – described on the next slide. bf / 2 bf / 2

102
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
ASIDE…
When surveying the literature on non-uniform torsion, be aware of some
terminology that arises from time to time.

Consider the differential expression for warping normal stress in a thin-walled

open cross-section:

σ w = E ωn ( s )  φ ′′ ( z )

A new quantity may be defined as an aggregation of the foregoing:

d ' bf The “Bi-moment”
Bw
−
4
-
+ Bw = EIωφ ′′ Units of Bw are: force(length)2
d ' bf
4
ωn
d ' bf −
d ' bf Bw ωn
4
4 σw = Reminiscent of (Mc / I)
here we have the
-
Iω raison d’être of this notion…(sigh)
+

103
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Back to the example…
Earlier results may now be used to fill in the last column of the table:

i =n
1
ωnj =
2A ∑ (ω
i =0
oi + ωoj )t ij bij − ωoj

1  d ' bf  bf t f   d ' bf d ' bf   d ' bf d ' bf  bf t f   d ' bf d ' bf  bf t f   d ' bf  bf t f 

= 0 +  + + (d ' t w ) +  +  + +   + 0 +   − ωoj
2 A  4  2   4 4   4 2  2   2 4  2   4  2 

Recognizing the fact that A = 2bf t f + d ' tw, it is observed that:

d ' bf
ω nj = − ωoj
4

the “j” denotes the value after

integration along the element

104
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
The diagram for Sω is obtained from an integration of the ωn diagram:
d ' bf
−
4 d ' t f bf2
- 16
s

∫
+ 4 6
d ' bf
Sω = ωn tds 3
0
4 x
ωn Sω
d ' bf
d ' bf −
4 2
4
- 1 5
+ d ' t f bf2
16
area under ωn diagram y
multiplied by thickness

In constructing the Sω diagram, we arbitrarily begin at pt. 1. Considering this starting point, it can be seen
that the combination of positive ωn and a positive direction for integration results in a a positive Warping
Statical Moment, Sω. A positive sense to the Warping Statical Moment implies a counter-clockwise warping
induced shear flow moment about the shear center (as depicted by the black arrow – twist positive with
RHR). NOTE: this not necessarily the direction of warping shear stresses since they are also proportional
to –Eφ’’’.
Proceeding now to pt. 2, it is observed that the Warping Statical Moment flowing into pt. 2 must branch out
and balance at this junction of plate elements. It is noted that in integrating from pt. 2 to pt. 5, a negative
direction of integrations is employed with a negative ωn; once again yielding a positive Sω (see arrow). The
value of Sω from pt. 2 to pt. 5 is the same as from pt. 1 to pt.2 and thus it is noted that there is no contribution
to the Warping Statical Moment from Pt. 2 to Pt. 3.
105
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design

The warping moment of inertia may now be computed as:

i =n
1
Iω =
3 ∑ (ω
i =0
2
ni )
+ ωni ωnj + ωnj2 t ij bij

Substituting the quantities from the example cross-section yields

1   d ′bf   d ′bf  − d ′bf   − d ′bf    b 3t (d ′)2

2 2
Iω = 2  +  +  t f bf  = f f
3   4   4  4   4    24


106
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Consider now the longitudinal twist in a member subjected to a concentrated
torque at mid-span…

The earlier solution developed for the case of a concentrated end torque may be
Used if proper boundary conditions are formulated at mid-span:

General Solution: Q·e

MzZ GKT
φ (z ) = A + B cosh(λz ) + C sinh(λz ) + ; λ =
λ2EIω EIω
Boundary Conditions:

φ (0 ) = 0
twisting restrained and warping unrestrained at support

φ ′′(0 ) = 0

L warping restrained at mid-span as a result of member continuity

φ ′  = 0
2
107
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Begin by taking derivatives of the general solution with respect to the longitudinal
position, z:
Mz z
φ (z ) = A + B cosh(λz ) + C sinh(λz ) +
λ2EIω
Mz
φ ′(z ) = Bλ sinh(λz ) + Cλ cosh(λz ) +
λ2EIω
φ ′′(z ) = Bλ2 cosh(λz ) + Cλ2 sinh(λz )

φ ′′′(z ) = Bλ3 sinh(λz ) + Cλ3 cosh(λz )

Substitute boundary conditions into derivatives and solve for integration constants:

φ (0 ) = A + B + 0 + 0 = 0 ⇒ A = −B

φ ′′(0 ) = Bλ2 + 0 = 0 ⇒ B = 0 ⇒ A = 0

L  L Mz − Mz
φ ′  = Cλ cosh λ  + =0 ⇒ C =
2  2
2
λ EIω  L
λ3 EIω cosh λ 
 2
108
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
The solution (considering the length L/2), and subsequent useful derivatives,
take the form:
 
Mz  sinh ( λ z ) 

φ (z) = 3 λz −
2λ EIω   L 
 cosh λ 2 
  

 
M z  cosh ( λ z ) 

φ′(z) = 2 1−
2λ EIω   L 
 cosh λ 2 
  

−Mz sinh ( λ z )
φ ′′ ( z ) =
 L
2λ EIω cosh  λ 
 2

−Mz cosh ( λ z )
φ ′′′ ( z ) =
 L
2EIω cosh  λ 
 2
109
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Assign a W12x50 as the section for this example:

d’ = 11.5” bf = 8.08” tf = 0.640” tw = 0.370” Ix-x = 391 in4 Iy-y = 56.3 in4

Sx-x = 64.2 in3 KT = J = 1.71 in4 Iω = Cw = 1880 in6 (Sω)max = 30.18 in4

(ωn)max = 23.32 in2 G = 11,500 ksi E = 29,000 ksi L = 240”

Q = 10 kips e = 4”
Mz = 40 k-in

Consider shear stresses:

TORSION INDUCED:
St. Venant shear stresses vary with plate thickness at a given cross-sectional location, z:
M GKT φ ′(z )
τ SV (t , z ) = SV (t ) = (t ) = (t )Gφ ′( z ) rads / in
KT KT
( )
Flange: (max value at z = 0)
(τ SV )Flange −Max = 0.640 " (11,500ksi ) 5.76 x10 = 0.424ksi
−5

( )
Web: (max value at z = 0)
(τ SV )Web −Max = 0.370" (11,500ksi ) 5.76 x10 = 0.245ksi
−5

110
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Warping shear stresses vary with the cross-sectional longitudinal position as well as the value of Sω:
Recall: max occurs at flange-web junction

− ESωφ ′′′(z )
τ w (t , z ) =
t

At the supported end,

(
− ( 29,000ksi ) 30.18in 4 )
τw =
0.640in
( −5.22x10 ) = 0.00714ksi
−9

At the mid-span,

(
− ( 29,000ksi ) 30.18in 4 )
τw =
0.640in
( −2.658x10 ) = 0.03635ksi
−8

111
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
FLEXURE INDUCED:
Vy s
In general, flexural shear stresses are computed as: τ flexure = −
Ix − xt ∫ ytds
0

sometimes called “Q”

d ′bf t f
4

d ′bf t f
x 2

2
d ′bf t f t w (d ′)
+
y 2 8

(τflexure)Flange = 0.295 ksi (τflexure)Web = 1.233 ksi

112
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design

Consider normal stresses:

TORSION INDUCED:

Warping normal stresses are proportional to φ’’(z)
max at mid-span, thus:

L
(σ w )max (
= φ ′′   E (ωn )max = −1.354 x10−6
2
) ( 29,000ksi ) 23.32in 2
= −0.915ksi

FLEXURE INDUCED:

Flexural normal stresses are maximum at mid-span (as dictated by the moment diagram)

M QL 10kips (240" )
σ flexure = = = = 9.346ksi
S x − x 4S x − x ( )
4 64.2in 3

113
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
10 kips
Stresses may now be superposed:
4”

Normal stresses at mid-span:

σtotal = -10.261 ksi

σtotal = -8.431 ksi
σflexure = -9.346 ksi
σwarping = -0.915 ksi

σwarping = 0.915 ksi

σwarping = -0.915 ksi

σwarping = 0.915 ksi

σflexure = 9.346 ksi
σtotal = 8.431 ksi
σtotal = 10.261 ksi
σflexure σwarping σtotal

114
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
Shear stresses at mid-span:
τflexure = 0.295 ksi τwarping = 0.0364 ksi

τflexure = 1.233 ksi τflange = 0.331 ksi

τweb = 1.233 ksi

τflexure = 0.295 ksi

τwarping = 0.0364 ksi
τflexure τSV τwarping

Shear stresses at support:

τflexure = 0.295 ksi τSV = 0.424 ksi τwarping = 0.00714 ksi

τflexure = 1.233 ksi τSV = 0.245 ksi τflange = 0.726 ksi

τweb = 1.478 ksi

τflexure = 0.295 ksi τSV = 0.424 ksi τwarping = 0.00714 ksi
τflexure τSV τwarping

115
Advanced Behavior of Metal Structures – Dr. Earls
TORSION: Analysis and Design
The AISC specification (2005) may be used to assess the adequacy of the
the member in this example (see section H3.3):

fnormal f
+ shear ≤ 1.0
Fnormal Fshear

fnormal ≡ sum of all coincident normal stresses; flange
11.18 ksi; web
9.35 ksi

fshear ≡ sum of all coincident shear stresses; flange
0.368 ksi; web
1.23 ksi

Fnormal = φT Fy = 0.9(50 ksi) = 45 ksi

Fshear = φT (0.6Fy) = 0.9(0.6)(50 ksi) = 27 ksi

10.261ksi 0.331ksi
Mid-span flange: + = 0.24 ≤ 1.0 ∴ safe !
45ksi 27ksi

9.35ksi 1.23ksi
Mid-span web: + = 0.253 ≤ 1.0 ∴ safe !
45ksi 27ksi

116
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear

“Eccentric Shear” is characterized as an in-plane loading condition in which the

line of action of the applied loading resultant does not coincide with the center
of gravity for the connection bolt group.

Examples: e P
P
e

The Specification does not specify an analysis method to arrive at required fastener strength; a “rational”
means for arriving at a required strength is all that is required…

117
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear

Two commonly employed methods of rational analysis are:

• Elastic Vector Analysis [based on elastic, small strain theory; has a closed
form solution that can be solved for directly (as guaranteed by Kirchhoff’s
Uniqueness Theorem from the Theory of Elasticity); conservative, but
unpredictably so.]

• Ultimate Strength Analysis [more accurate than Elastic Vector Analysis;

yields more uniform results; based on a limit states approach, and thus
philosophically more consistent with the design approach espoused in this
course; iterative solution procedure required.

118
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Elastic Vector Analysis:

Assumptions

•Connection rotation occurs about the centroid of the bolt group

•All fasteners are elastic
•All fasteners have the same cross-sectional area
•The force, R, on each fastener is proportional to its distance “d” from the centroid of the
bolt group.

In light of all of the foregoing assumptions, we note that the Principle of Superposition applies here
and thus…

e P P
M = Pe

eccentric shear moment shear

119
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Elastic Vector Analysis:

Consider the following pure moment condition

M M ≡ resultant moment due to eccentric load

R1 di ≡ radial distance from bolt group centroid

to bolt “i”
R2
d1 d2
R6 Ri ≡ Force contribution on bolt “i”, as a result
d6 d3 of moment “M”

d4 R3
R5 d5
R4

∑
6
M = R1d1 + R2d 2 + … + R6 d 6 = Ri d i
i =1

120
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Elastic Vector Analysis:

Based on the assumption that the force, R, on each fastener is proportional to its distance “d” from
the centroid of the bolt group, we note that

R1 R2 R3 R Rd Rd Rd Rd
= = = … = 6 ⇒ R1 = 1 1 ; R2 = 1 2 ; R3 = 1 3 ;…; 1 6
d1 d 2 d 3 d6 d1 d1 d1 d1

thus resulting in the following moment equation

6
R1d12 R1d 22 R1d 32 R1d 62 R1 2 R
M=
d1
+
d1
+
d1
+…+
d1
=
d1
[ ]
d1 + d 22 + d 32 + … + d 62 = 1
d1 ∑d
i =1
i
2

While the forgoing development focused on bolt 1, any of the bolts within the group would suffice for a
similar normalization. Thus, we may generalize this last result to be

Md n
Rn = 6
; where " n" denotes the fastener of interest
∑d
i =1
i
2

121
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Elastic Vector Analysis:
Since we would like to use the Principle of Superposition to combine the previous pure moment case
with the pure shear case, we should decompose Ri into its x-y components:

R Ry

Rx

d
y y x
Rx = R and Ry = R
d d

We may use this result to restate our pure moment result as:

M (y n )
(R x )n = 6
and (Ry )n = M6 (x n )
∑d
i =1
i
2
∑d
i =1
i
2

122
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Elastic Vector Analysis:

Since d2 = x2 + y2, we note that

M (y n ) M (x n )
(R x )n = 6 6
and (R y )n = 6 6

∑
i =1
x i2 + ∑
i =1
y i2 ∑
i =1
x i2 + ∑y
i =1
2
i

and then,

Rn = (R x )2n + (R y )2n

Considering now the pure shear condition,

P
RV =
nmax

Combining loading conditions gives,

(Rn )required − capacity = [(R ) y n + RV ]

2
+ (R x )n
2
 This is what we design for

123
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Ultimate Strength Analysis: From Equilibrium:

P ∑F = 0 H
n
V
xo ∑ R sinθ
i =1
i i − P sin δ = 0

e
∑F = 0

,
ity
Ri

c
V

tri
n
n

ce
∑ R cos θ
Ec
i i − P cos δ = 0
i =1

θi di
∑M = 0
n

yo
ro
α
H
∑ R d − P (e + x
i =1
i i o cos δ + y o sin δ ) = 0

δ For the case of δ = 0˚ or 90˚, α = δ, hence

Instantaneous
∑M = 0
n
Center of Rotation
∑ R d − P (e + r ) = 0
i =1
i i o

124
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Ultimate Strength Analysis:

For a bearing connection, the deformation of each fastener may be assumed

to be proportional to its distance from the instantaneous center and thus the
resistance of each fastener may be described by an expression that is based
upon a empirically derived load-deformation relationship such as:

(
Ri = Rult 1 − e −10 ∆ )
0.55

Since the most common bolt type is A325, we may specialize this expression
for that case:

Rult = τu Ab ; τu
≈70% of tensile ultimate strength (120ksi)

Ab is the cross-sectional area of the bolt

∆ = 0.34 in (experimentally determined deformation at failure)

e = 2.718281828…. (NOT loading eccentricity!)

125
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Ultimate Strength Analysis Example:
The individual fastener capacity can be expressed using the
P previously obtained expression:
ro = 3” e = 5” 0.55
( )(
Ri = 0.7 (120ksi ) 0.6013in 2 1 − e −10 ∆ )
1 4 0.55
(
= 50.5 1 − e −10 ∆ )
Loading coincides with y-axis
δ =0.
2 @ 3”
2 5
yi
sin θ i =
di
3 6 xi
cos θ i =
di
4” permits
yi
7/8” A325 bolt ∑R i
di
=0

xi
∑ Ri
di
= Pn Solution must satisfy equilibrium…

∑R d i i = Pn (e + ro )

di d
We now state our fundamental assumption regarding bolt deformations: ∆i = ∆ max = i (0.34 )
d max d max
126
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Ultimate Strength Analysis Example:

Since it is not at once obvious where the instantaneous center of rotation is, we will adopt an iterative
solution approach, beginning with ro = 3:

ro = 3”
Fastener xi yi di ∆i Ri (Rixi) / di Ridi

1 1 3 3.162 0.184 45.9 14.53 145.3

2 1 0 1 0.058 32.2 32.23 32.2

3 1 -3 3.162 0.184 45.9 14.53 145.3

4 5 3 5.831 0.340 49.6 42.51 289.1

5 5 0 5 0.292 49.0 48.99 244.9

6 5 -3 5.831 0.340 49.6 42.51 289.1

These summations do not identically satisfy the equilibrium statements
Σ = 195.3 Σ = 1145.9
Thus we must iterate…

127
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Eccentric Shear
Ultimate Strength Analysis Example:

Converged solution…

ro = 2.06”
Fastener xi yi di ∆i Ri (Rixi) / di Ridi

1 0.06 3 3 0.202 46.7 0.93 140.2

2 0.06 0 0.06 0.004 8.6 8.55 0.5

3 0.06 -3 3 0.202 46.7 0.93 140.2

4 4.06 3 5.05 0.340 49.6 39.87 250.3

5 4.06 0 4.06 0.273 48.7 48.68 197.6

6 4.06 -3 5.05 0.340 49.6 39.87 250.3

These summations do identically satisfy the equilibrium statements
Σ = 138.9 Σ = 979

128
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Balanced Welded Connection

In certain applications, such as joints in a truss, it is desirable to proportion a

welded connection to ensure that the connection resultant is collinear with the
line of action of the applied member force; thus reducing the likelihood that
connection moments are generated.

L1
F1

F2
d L2 centroidal line T
y
o
F3
L3

Assume that resultant weld forces act at plate edges and not at fillet weld throat

129
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Balanced Welded Connection

ΣMo = 0; F1d + F2(d/2) – T(y) = 0
F1 = -0.5F2 + T(y/d) [eqn. 1]

The force, F2, is equal to the weld strength per unit length , Rw, times the
weld length, L2 (see section J.4 of the Specification for Rw)

F2 = Rw2 L2 [eqn. 1a]

ΣFy = 0; F1 + F2 + F3 – T = 0
F1 = T – F2 – F3 [eqn. 2]

Substituting [eqn. 2] into [eqn. 1] gives

T – F2 – F3 = -0.5F2 + T(y/d)
F3 = T{1-(y/d)} – 0.5F2 [eqn. 3]

130
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Balanced Welded Connection

A four step procedure may now be used to balance the weld:

1) Proportion the end weld using the weld size limits imposed by Specification
Chapter J and then compute the capacity of the end weld, F2, using [eqn. 1a]
(if no end weld, then F2 = 0)

2) Compute F1 using [eqn. 1]

3) Compute F3 using [eqn. 3]

4) Compute weld lengths of side welds using:

F1 F3
L1 = L3 =
Rw 1 Rw 3

and satisfying the weld size limits imposed by Specification Chapter J

131
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Balanced Welded Connection
Example

L1
F1
L6 x 4 x 3/8, A992

F2
6” L2 centroidal line 150k
1.94”
o
F3
L3 Determine weld size limitations using Chapter J of the Specification:

Minimum size of fillet weld = 3/16 in. (T-J2.4)

Maximum weld size = 5/16 in. (Sec. J.2.2(b))

Use 3/16 in. fillet weld with E70 electrodes – the design strength of which is:

φRnw = φte(0.60FEXX) = 0.75(3/16”)0.707(42ksi) = 4.18 k/in.

Make sure that weld controls over shear rupture of the base metal:

Max φRnw = φte(0.60Fu) = 0.75(0.375”)0.60(65ksi) = 11.0 k/in. OK

132
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Balanced Welded Connection
Example

Now we compute the weld lengths need to balance the connection:

F2 = φRnwLw = 4.18 k/in. (6”) = 25.1 kips

Summing the moments about point “o” (at F3), we obtain:

150k (1.94" ) − 25.1k (3" )

F1 = = 36.0k
6"

Whereupon summation of forces yields:

F3 = Tu – F1 – F2 = 150kips - 36.0 kips -25.1 kips = 89.9 kips

F1 36.0k
Lw 1 = = = 8.6" → 9"
φRnw 4.18 k
in

F3 88.9k
Lw 3 = = = 21.2" → 22"
φRnw 4.18 k
in

133
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Bolt tension in clip angle connection

What is the tension in this bolt as a result of moment, M?

db

M d

b
Each of the bolts (shown in red) are pre-tensioned with a force, Tb

134
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Bolt tension in clip angle connection

Assume that the connection contact area on the column flange is d x b. The
contact stress distribution on such an assumed surface can be obtained using
superposition:
fbi ftb = -fcb fbi - ftb

fcb ftb + fbi

Initial compressive stress Stress on contact area (b x d)
on contact area (b x d) as resulting from the action of M
It is pointed out that due to Newton’s Third Law,
a result of bolt pre-tension alone.
these contact stresses, developed from the
d  bending, are subsequently resisted equally by
fbi =
∑T b
M 
ftb =   =
2 6M fasteners in a given connection line parallel to
the “b” direction, and having a tributary width, db.
bd I bd 2
Thus:
ΣTb ≡ Sum of all bolt pre-tensioning ftb bd b
T =
forces n
n ≡ number of fasteners in a line, in the “b” direction
135
Advanced Behavior of Metal Structures – Dr. Earls
CONNECTIONS: Bolt tension in clip angle connection
Using this last result for the bolt-line force due to bending, we may solve for
stress and equate it with the flexural contact stress expression obtained earlier:

Tn 6M
ftb = =
bd b bd 2

6M bd b 6Md b
T = =
bd 2 n nd 2

Assuming further, that the top-most bolt is a distance db from the top edge of
the clip angle results in a modification of this last expression (similar triangles):

6Md b  d − d b 
Tmod =  
nd 2  d 

NOTE: db used here is not the diameter of the bolt (denoted as db in

the Specification). Rather it is the bolt spacing.

136
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

“There is nothing more practical

than a good theory” - Hilbert
• Q: What is stability notionally?

• A: If a given physical system exists in some

initial state, and an external agency provides a
perturbation, or disturbance, to this state, then a
stable system remains close to the initial state
and exhibits a bounded response. Unstable
systems exhibit the converse behavior.

137
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Simple Physical Examples:

−δ δ

STABLE −δ δ
138
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Simple Physical Examples:

−δ δ −δ δ

UNSTABLE
139
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Simple Physical Examples:

−δ δ
−δ δ

NEUTRAL
140
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Definitions:

Equilibrium State – Force components balance (in any

direction) and moments balance (about any point). Any
free body is in equilibrium.

Stability – A equilibrium state is unstable if a small

disturbance in displacement or velocity can lead to
large motion.

141
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Systems of Linear Equations:
A finite set of equations of the form:

a1x1 + a2x2 + … + anxn = b

where a1, a2, … , an , b are constants

142
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Solution of Systems of Linear Equations:
A sequence of numbers s1, s2, … sn is called a
solution of the system if x1 = s1, x2 = s2, … , xn = sn
is a solution of every equation in the system.

A system that has no solution is called “inconsistent”

and the converse system is called “consistent.”

143
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Consider a system of two linear equations:
“x” and “y” are the unknowns,

a1x + b1y = c1
a2x + b2y = c2

where a1, a2, b1, b2 are all non-zero.

each of these equations may be conceived of as a

line in x-y space ( ) called L1 and L2, respectively
2

144
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Considering the graphs of L1 and L2, we immediately
observe three possibilities for the solution of the L1, L2
system:
L2
L2

L1 L1 & L2
L1

No Solutions One Solution Infinity of Solutions

145
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:

Theorem:

Every system of linear equations has either no solution,

exactly one solution, or infinitely many solutions.
(proof omitted)

146
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Homogeneous Systems of Linear Equations:

A finite set of equations of the form:

a11x1 + a12x2 + … + a1nxn = 0
a21x1 + a22x2 + … + a2nxn = 0
. . . .
. . . .
. . . .
am1x1 + am2x2 + … + amnxn = 0
where a1, a2, … , an are constants & m = n
147
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:

Observation:
All such systems are consistent since
x1 = x2 = x3 = …= xn = 0 is always a solution
(the trivial solution). However, other solutions
may exist (non-trivial solutions).

148
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Observation:

As a result of this consistency, and since the trivial

solution always exists, we can make the following
statement:
For a given system of homogenous linear equations,
exactly one of the following is true:
1) The system has only the trivial solution.
2) The system has infinitely many non-trivial
solutions in addition to the trivial solution.
149
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
if [A] exists, then the matrix [B] leading to the
statement:

[A] [B] = [I] ; where [I] is the identity matrix

implies that [B] is the so-called “inverse” of [A]

which is typically denoted as [A]-1. Thus,

[A] [A]-1 = [I]

150
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
Theorem: If [A] is an n-by-n matrix, then the following
statements are equivalent:

1) [A] is invertible
2) [A] {x} = {0} has only the trivial solution
(proof omitted)

151
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
If [A] is an invertible matrix, then
adj [A]
[A]−1
=
det[A]
where adj [A] is the transpose of the matrix
of cofactors from [A]
Theorem: A square matrix [A] is invertible iff
det [A] = 0
152
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
Eigenvalues and eigenvectors:
One may view a coefficient matrix, [A], as a
means for transforming one vector, {x}, into
another vector, {y} (i.e. [A] is a linear
operator).
[A]{x} = {y}
y
or graphically,
x

153
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
Eigenvalues and eigenvectors:
From the foregoing, it is clear that it is possible
that we may find a particular vector {x} whose
image under multiplication by [A] is simply a
scalar multiple of the original vector {x}.

Ax
x
154
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction

Mathematics Review:
Fundamentals:
Definition: If A is an n-by-n matrix, then a non-
zero vector x in Rn is called an eigenvector of A if
Ax is a scalar multiple of x; that is Ax = λx for
some scalar, λ. The scalar, λ, is called an
eigenvalue of A, and x is said to be an
eigenvector corresponding to λ.

155
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Introduction
Mathematics Review:
Fundamentals:
To find the eigenvalues of an n-by-n matrix, A, we
rewrite Ax = λx as
Ax = λIx
or
(λI – A)x = 0
For λ to be an eigenvalue, there needs to be a non-
trivial solution to this. This can be true only if
det (λI – A) = 0
which is called the characteristic equation of A
156
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

x P

k Assumptions:

•Bar of length “L” is rigid

•Collar is frictionless
a
L
θ •Spring remains horizontal
during motion

157
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Free Body Diagram (FBD) of Rigid Bar: FBD of collar:

P
ΣMpin = 0; R (a / cos(θ))-P L sin(θ)
+ kx=k a tan(θ)
θ
R R

ΣFx = 0; R cos(θ) =k a tan(θ)

Equilibrium Equation
A / cos θ Thus, R= k a (sin(θ) / cos2(θ))

158
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Substituting the results from the consideration of the collar into the equilibrium equation

sin (θ ) a
ka − PL sin (θ ) = 0
cos (θ ) cos (θ )
2

or,

 ka 2 
 − PL  sin (θ ) = 0; Equilibrium Equation for the System
cos 3
(θ ) 
 

159
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Considering small θ (i.e. “linear” analysis)
1 1
1 1

θ
cos( θ ) 0.5 sin( θ )
2 1 0 1 2

0 −1 1
2 1 0 1 2
− 1.571 θ 1.571 − 1.571 θ 1.571

Taylor Series:
θ2 θ4 θ3 θ5
cos(θ ) = 1 − + +… sin(θ ) = θ − + +…
2! 4! 3! 5!

Small θ ⇒ cos (θ) ≈ 1 and sin (θ) ≈ θ

160
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Using the small angle approximations, we re-state our equilibrium equation as:

(ka 2
)
− PL θ = 0 This is a linear, homogeneous equation

Solutions,
1) θ = 0 with arbitrary “P”; vertical configuration – trivial solution
2) P = (ka2) / L for arbitrary small θ

161
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

P
Single Degree of Freedom Systems:
Plot equilibrium states in θ – P plane → ka 2
L
If we non-dimensionalize “P” by multiplying
by (L / ka2), then at the bifurcation point we
experience a unit non-dimensional load:
Bifurcation Point
ka 2 ka 2 L
At P = we get 2
=1 -θ p θ
L L ka

Plot equilibrium states in θ – p plane →

1
PL
Define p=
ka 2
Bifurcation Point

-θ θ
162
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Now consider a non-linear analysis (i.e. finite θ):

We retain our original system equilibrium equation

 ka 2 
 cos 3 (θ ) − PL  sin(θ ) = 0
 
 
Solutions:
1) sin (θ) = 0; θ = ±π, ±2π, …, ±n; θ = 0 → trivial solution

ka 2 1
2) P = or, in non-dimensional form p=
L cos3 (θ ) cos3 (θ )

163
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

1.0
1
cos(θ ) ≤ 1.0 ⇒ cos 3 (θ ) ≤ 1.0
1

1 π π
cos( θ ) 0.5 ∴ ≥ 1.0 if - < θ <
cos (θ )
3
2 2

0
2 1 0 1 2 P

− 1.571 θ 1.571 x P

a
Ring falls off when L
θ
θ = cos-1 (a / L)

164
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Rigid Bar and Collar Stability Model Response: θ := −1, −0.99.. 1

1
p ( θ ) :=
3
cos ( θ )
6

x P
4
k
p ( θ)

2
a
L
θ

1 0.5 0 0.5 1
θ

165
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Intuitive Summary of previous method:

Destabilizing moment due to “P”; PL sin (θ)

ka 2 sin(θ )
Restoring moment due to spring;
cos 3 (θ )

For small θ, de-stabilizing moment is PLθ and restoring moment is ka2θ

166
Advanced Behavior of Metal Structures – Dr. Earls
STABILITY: Rigid bar models

Single Degree of Freedom Systems:

Intuitive Summary of previous method:

PL sin (θ) If θ > 0, then which dominates? Destabilizing or restoring?

If P < ka2 / L, then ka2 θ > PLθ is true and thus restoring
moment dominates and hence system is stable at θ = 0
ka 2 sin(θ )
cos 3 (θ )
If P > ka2 / L, then ka2 θ < PLθ is true and thus destabilizing
moment dominates and hence system is unstable at θ = 0

167
Advanced Behavior of Metal Structures – Dr. Earls