12

Welding

12.1 Rivets
The most telling application of rivets is to attach aluminum plates (panels) to aircraft frames. Screws have
risks of self-loosening due to environmental factors which are severe for aircraft. Rivet is secured by plastically
deforming its end using a riveting machine. The rivet head produced will prevent a rivet from self-loosening and
thus will form a permanent fastener. It would require a determined effort for anyone who wants to remove a rivet;
a rivet can thus be considered tamper-proof. Its simple shape also makes rivet much cheaper and much faster to
install than a bolt.

Rivet is superior to bolt when shear loads are present because bolt has a smaller cross sectional area for
the same nominal diameter and also has sharp threads with high stress concentration factor. Recall that bolts are
preferred if the loads are axial (i.e., producing normal stress).

12.2 Welding
Welding is a permanent fastener. It is much stronger than rivet or bolt because it has no limitation on
either axial or shear load. Because welding fuses the fastening material into the two adjoined parts, it provides also
much more rigid connection than either rivet or bolt can. Welding is the chosen fastening method for building a
motorcycle or bicycle frame because of its rigidity.

Three main elements of welding are (i) heat source, (ii) electrode, and (iii) inert environment. Heat source
is required to melt the fastener, which in this case is the electrode. Each end of the parts to be joined will also
reach high temperature so that the melting electrode (weld pool) will fuse into the two ends and thus connect the
two parts. The heat source can be combustion of flammable gas or high voltage electrical arc.

The precise process of establishing the connection between the two parts and the resulting mechanical
strength is determined by the phase diagram of the melting electrode alloy material. Inert environment is required
to prevent oxygen to react with the alloy. Metallic oxides are hard and brittle; they promote crack formation. The
inert environment is obtained by shielding the heated electrode with either melted gel (flux) or inert gas (Ar, He, or
CO2).

Examples of welding methods are

1. Shielded metal arc welding (SMAW). It is commonly called stick welding and is performed manually. The heat
source is high voltage arc that is established between the electrode and the interface of the two plates being
welded. Distance between the electrode and the interface has to be close and steadily maintained. The task of
maintaining a constant distance requires training because the electrode is also consumed to become weld
pool. SMAW requires only one hand due to the electrode that acts both as a consumable and an electrical
wire. The electrode has a flux coating that melts to shield the weld pool.

2. Gas metal arc welding (GMAW). It is also called MIG (metal-inert gas) welding. It uses electric arc to generate
heat. Inert gas, such as Ar for aluminum electrode and CO 2 for steel electrode, is used. Electrode as a result
does not have flux core. For carbon steel electrode, quick welding does not require inert gas. Electrode is
typically extruded continuously from an electrode coil.

3. Gas tungsten arc welding (GTAW). It is also called TIG (tungsten-inert gas) welding. The electrode, i.e.,
electrical wire, serves to conduct the high voltage arc and to release inert gas close to the welded spot. The
filler wire that becomes weld pool is separate from the electrode, so both welder's hands are occupied for
welding.

4. Submerged arc welding (SAW). This welding is automated and is used, for example, to weld large pipe and
threaded connector. It uses granular flux that is sprinkled continuously and melts to submerge (shroud) the
fresh weld.

There are two welding types: bult weld and fillet weld. Butt weld connects two parallel plates as shown
below. Welding can be completed in one pass with a single V-groove shown in the second picture from the left. If
the plate thickness is h , while its length L, the weldment area for stress calculation is hL .

Fillet weld connects two plates that are perpendicular to each other. The weldment area for stress
calculation is 0.707 hL as shown in picture (a) below, where t=0.707 h , where t and h are throat length and leg
length, respectively. We shall use this area for determining shear stress in parallel loading in (c) and transverse
loading in (d) and (e). Notice that for (c) and (d), the total area is each 2 ×0.707 hL , while for (e) the total area is
4 ×0.707 hL .
12.3 Welding Fatigue
General fatigue calculation is applicable directly to a weld. The calculation starts with identifying stresses,
i.e., shear and normal stresses in the previous section, and is followed by determining the mean and alternating
components of these stresses. If the mean component is zero, we use the S-N curve and compute the endurance
'
limit Sn=C L CG C S CT C R Sn . If the mean component is not zero, we use the Goodman diagram and compute
the equivalent mean and alternating stresses using the following formulas when the stress state is two-
dimensional (2D):

1. Equivalent mean bending stress uses the larger principal stress, combining all mean stresses:

√
2
σm σ
σ em= + τ m2+( m ) ;
2 2
2. Equivalent alternating bending stress uses the maximum distortion energy theory, combining all alternating
stresses:
2 2 1 /2
σ ea=(σ a +3 τ a ) .
If the stress state is three-dimensional (3D), then we the equivalent mean bending stress is given by the
largest principal stress out of the three principal stresses. The equivalent alternating bending stress is given by the
3D version of the maximum distortion energy theory’s equivalent tress.

We have to account for fatigue stress concentration factor at a weld’s critical location as listed below.
Problem 24
Two 20-mm steel plates are butt-welded together. Both the plate and the welding electrode materials correspond
to strength properties of Su=500 MPa and S y =400 MPa. The imposed loading is perpendicular to the weld
length’s direction, along one of the plate’s axes, and fluctuates rapidly between –20 and +60 kN. Estimate the
length of weld required in order to provide a safety factor of 2.5: (a) If the weld reinforcement is not removed; (b)
If the excess weld material is carefully ground off to provide smooth, continuous surfaces.

The weldment can be one of the diagrams below. From the weld's strength perspectives, any one of them
is equivalent so long as the welding was properly performed as per welding code.

(a) The tensile stress area for the weld is thus 0.02 L m 2 . The load's mean stress is
6
( 60−20 ) 1000/2 10 SF × K f
σ m= × SF × K f = ,
0.02 L L
while the load's alternating stress is

( 60−(−20) ) 1000/2 2 ×106 SF × K f

σ a= × SF × K f = .
0.02 L L
Endurance limit (at 106 cycles) is given by
'
Sn=S n C L C G C S C T C R ,

where C L =1 for axial loading; C G is not known precisely because we only know one of the two weld’s
dimensions: 20-mm thickness, while the weld length L is unknown; C S=0.5 for the unremoved weld
reinforcement that resembles as-forged surface of 500 MPa tensile strength; and both C T and C R equal one.
Sn =0.5 Su =250 MPa. One way to obtain a first estimate of C G is to regard the smaller of the two weld’s
'

dimensions as diameter; thus, for a 20-mm diameter, we find

−0.107
C G=1.24 ∙ 20 =0.9 .
Thus,

Sn=250× 0.5 ×0.9=112.5 MPa.

The Goodman line for the endurance strength is thus given by the following straight-line equation

σ a=112.5 1−( σm
500 )
MPa .

The load line is given by

 σa
=2.
σm
The two lines intersect at

σ m=50.6 MPa , σ a=101.1 MPa.

The fatigue stress concentration factor for the unremoved weld reinforcement is K f =1.2 so that
6
10 × 2.5 ×1.2 6
=50.6 ×10 ,
L
yielding L=0.07 m.

Now we can verify whether the 0.9 size factor is consistent with the 0.07 weld length. The equivalent
diameter for a rectangular weld cross section is

0.808 √ 20× 70=30.2 mm ,

producing a size factor of 1.24 ∙ 30.2−0.107=0.86 . The change of size factor from the initial 0.9 to 0.86 is small,
and we should be led to conclude there is no need to repeat the calculation using this improved size factor.

(b) Given that the excess weld material (weld reinforcement) is ground off to give smooth surfaces, we now use
K f =1.0 for the absence of sharp profiles on the surface and C S=0.79 for machined surface of 500 MPa tensile
strength.

The endurance strength is now

Sn=250× 0.79 ×0.86=169.9 MPa.

Thus, there is a 51% improvement for the endurance strength from that in part (a) by grounding off the excess
weld material. The Goodman line is now

σ a=169.9 1−( σm
500 )
MPa ,

while the load line remains the same:

σa
=2.
σm
They intersect at

σ m=72.6 MPa , σ a=145.2 MPa.

Thus,
6
10 × 2.5 ×1.0 6
=72.6 ×10 ,
L
where 1.0 in the left hand side is the fatigue stress concentration factor, yielding
 L=0.034 m.
Grounding off the excess weld takes time, but the weld length is 50% shorter than that found in part (a).
The decision to ground off the excess weld should also save welding cost since the labour cost of an experienced
grinder is lower than that of an experienced welder.
Problem 25
The bracket is to support a 60 kN load. Using an E60 series welding rod and a safety factor of 3.0, what
size should the weld be specified?

Because the 60 kN is applied symmetrically to the two plates, we can regard one plate only with
a 30 kN load. Each plate has 3 welds: one 100-mm long vertical and two 75-mm horizontal. We regard
the 3 welds as one rigid body. The 30 kN load will create a torque to the rigid body in addition to the 30
kN load to the rigid body.

Each weld has a throat length t and is assumed identical for all 3 welds. Resulting torsional and
direct shear stresses from the torque and the load will have a shear area of tL, where L is either 100 or
75 mm.

We first have to determine the centroid of the 3 welds so that we can compute the magnitude
of the torque. The coordinate system used has its origin at the bottom left corner as shown below.

Thus,

( 100 t ) 0+ ( 75 t ) 37.5+ ( 75 t ) 37.5

x= =22.5 mm ,
100 t +75 t+75 t
( 100 t ) 50+ ( 75 t ) 0+ ( 75 t ) 100
x= =50 mm ,
100 t +75 t+75 t
and the external torque from the 30 kN load is therefore
 6
T =( 30000 N ) (55+ 22.5 mm )=2.325 ×10 N ∙ mm.

Our next task is to determine the polar area moment of the 3 welds as one rigid body. Polar area
moment formula for single rigid body is

J=∫ r 2 dA=∫ ( x 2 + y 2 ) dx dy

where the xy coordinate system has its origin at the single rigid body’s centroid. When this single rigid
body becomes a part of a composite rigid body, the centroid will move to a new location of the
composite rigid body. Erecting a new (global) x ' y ' coordinate system for the composite rigid body,
whose origin is now at the new centroid. The polar area moment is now defined with the new x ' y '
coordinate system

J ' =∫ r '2 d x ' d y ' =∫ ( x '2+ y '2 ) d x ' d y ' .

Let the position of the single-rigid-body centroid be denoted by a radius vector r 0 in the x ' y '
coordinate system, so that
'
r =r + r 0 ,
' '
implying that x =x+ x0 and y = y + y 0 . Substituting these into the last integral yields

J =∫ r d x d y =∫ ( ( x+ x 0 ) + ( y+ y 0 ) ) dx d y=∫ ( x + y ) dx dy+∫ ( 2 x x 0+2 y y 0 ) dx dy +∫ ( x 0 + y 0 ) dx dy .

' '2 ' ' 2 2 2 2 2 2

The first term is the polar area moment J for single rigid body. The second term is zero because it is
linear in x and y . The third term is the distance-square r 0 =( x 0 + y 0 ) term, which is called the “parallel
2 2 2

axis” term. Hence,

'
J =J + J 0 ,

where J 0=∫ ( x 0 + y 0 ) dx dy .
2 2

Thus, for the three-weld rigid body,

1 1
t 100 + 100 t + ( 100 t ) ( 22.5 )
3 3 2
J 1=
12 12
for the vertical weld, and
 1 1
J 2= t 753 + 75 t 3 + ( 75 t ) ( 502+ ( 37.5−22.5 )2 )
12 12
for the top weld, and for the bottom weld

J 3=J 2 .

The cubic terms in t can be ignored since they are much smaller than the linear terms in t . Thus,
' 4
J =J 1 + J 2 + J 3 ≈ 613021t mm .

Because the torsional shear stress is τ =Tr /J , the largest stress is located at the furthest point from the
centroid. There are four candidate locations: A, B, C, D.

Point B has the following stress directions, where τ B , x and τ B , y are horizontal and vertical (as in
the diagram) components of torsional shear stress. Direct shear stress τ shear and τ B , y act in opposite
directions so their net effect is diminished.

( 2.325× 106 ) ( 75−22.5 ) 199.1 N

τ B , y= = .
613021 t t 2
mm

( 2.325 ×106 ) ( 50 ) 189.6 N

τ B , x= = .
613021t t 2
mm
30000 120
τ shear = = .
100t +2 ×75 t t
We see that they have the same order of magnitude and thus cannot be ignored. The total stress at B is

√
2
( 199.1−120 ) 189.62 205.4
τ B= + 2 = .
t
2
t t

It can be verified that the total stress at D is the same as that at B.

 Points A and C have the same total stress. At A:

( 2.325 × 106 ) ( 22.5 ) 85.3 N

τ A , y= = .
613021 t t 2
mm

( 2.325 ×106 ) ( 50 ) 189.6

τ B , x= = .
613021t t
30000 120
τ shear = = .
100t +2 ×75 t t

Unlike in B and D, τ A , y and τ shear add up at A. The total stress is now

√
2
( 85.3+120 ) 189.62 279.4
τ B= + 2 = .
t2 t t

Thus, the largest stress locations are A and C.

E60 welding rod has a yield strength of ( 60−12 ) × 1000=48 ksi. Applying the safety factor
SF=3 we equate the stress limit
0.58 S y 0.58× 48000 Pa −6 m
2
τ A= = ×6895 × 10 =64
SF 3 psi mm 2
4.3
with 279.4 /t to give t =4.3 mm. Hence, the weld leg length is =6 mm.
0.707
Problem 26
Determine the maximum stress in the two-fillet-welds system due to the 23000 N load.

The two fillet welds are treated as one rigid body and we need to determine its centroid. The origin of the
coordinate system is at the bottom of the vertical weld. The x coordinate of the centroid is

6 ×150 ×0+ 6 ×150 × (−75 )

x= =−37.5 mm ,
6 ×150+ 6 ×150
while the y coordinate of the centroid is

6 ×150 × 75+6 ×150 ×150

y= =112.5 mm .
6 × 150+6 ×150
The 23,000 N downward force thus generates a torque of magnitude

( 0.125+0.0375 ) ×23,000=3737.5 N-m .

The polar area moment of the two-fillet-welds system is

6 × 1503 6 ×150 3
+6 × 150 [ (75−37.5 ) + ( 150−112.5 ) ]+ +6 ×150 [ ( 112.5−75 ) + 37.5 ]=8.44 × 10 mm ,
2 2 2 2 6 4
J≈
12 12
where the first three terms belong to the horizontal fillet weld and the last three terms the vertical fillet weld. The
polar area moment above has neglected the term 2 ×150 ×6 3 /12 as it is much smaller than the six terms above.

There are three candidate locations where the maximum stress could be found. The first is the left edge of
the horizontal weld. Its torsional shear stress is

3737.5× √ ( 0.150−0.0375 ) + ( 0.150−0.1125 )

2 2

−6
=52.5 MPa .
8.44 × 10
There is also a direct shear stress of magnitude

23000
−6
=12.8 MPa .
2× 6 ×150 ×10
The torque that causes the torsional stress has a roughly NE bearing from the centroid, while the 23,000 N force
that causes the direct shear stress has a S bearing. Thus, they will partially cancel each other at the left edge of the
horizontal weld. This first candidate location is therefore rejected.

The second candidate location is at the corner where the horizontal and the vertical welds meet. The
torque here has a SE bearing, while the direct shear stress will remain with the S bearing. Thus, they will add up
instead of cancelling each other.

The third candidate location is that the bottom edge of the vertical weld. The torque here has a SW
bearing, while the direct shear stress will remain with the S bearing. Thus, they will add up.

The radius from the centroid to the bottom edge location is much larger than the radius from the centroid
to the corner location. Therefore, the bottom edge is more prospective than the corner. The torsional shear stress
at the bottom edge is

3737.5× √ 0.03752+ 0.11252

−6
=52.5 MPa ,
8.44 ×10
while the direct shear stress remains at 25.6 MPa. The SW bearing of the torsional shear stress at the bottom edge
makes an angle of

arctan ( 0.1125
0.0375 )
=71.6 °

with the vertical line. Thus, the downward component of the sum of the two stresses is

12.8 MPa +52.5 MPa × cos 71.6 °=29.37 MPa ,

while the left-going component of the stress sum is

52.5 MPa × sin 71.6 °=49.82 MPa .

Their sum is thus

√ 29.372 +49.822=57.83 MPa .

The 57.83 MPa is the stress sum obtained by regarding the two stresses as vector-like quantities. Thus, the 57.83
MPa is not 100% correct.

If we regard the two stresses as tensors, then we have a 3D stress state

( )
0 0 −τ sin 71.6 °
0 0 −τ cos 71.6°−σ ,
−τ sin71.6 ° −τ cos 71.6 °−σ 0

where τ =52.5 MPa and σ =12.8 MPa. The three eigenvalues are

0 , √ σ +0.63 στ +τ ,−√ σ +0.63 στ +τ

2 2 2 2

so that the equivalent stress of the maximum distortion energy theory:

1/ 2
1
σ=
'

√2
[ ( σ 2−σ 1 ) + ( σ 3 −σ 1 ) + ( σ 3−σ 2 ) ]
2 2 2
is equal to

σ =1.73 √ σ + 0.63 στ+ τ =100.05 MPa .

' 2 2

The 100.05 MPa equivalent stress is the correct answer and is 1.73 times larger than the 57.83 MPa. The 1.73
factor, however, can be applied directly as a prefactor learned from previous problems. To obtain the 1.73 factor
from the 3D stress state, however, will require additional calculations.