You are on page 1of 8

Torsion of a Thin Rectangular Section

z ∂ ψ d 2 2 = y + z ) n = 0, n
z
ψ
d
2
2
=
y
+
z
)
n =
0,
n =
1
∂ n
1 2 (
ds
ds = − dy
y
z
∂ ψ 1
d
(
2
2
=
yz +
)
=−
y
∂ z
2
dy
ds = dz
n
=−
1,
n =
0
n =
1,
n =
0
y
z
y
z
ds = − dz
ψ
1
d
ψ
1
d
2
2
=
(
y
+
z
)
(
2
2
=
y
+
z
)
y
∂ n
2
ds
∂ n
2
ds
ψ
1
d
ψ
1
d
(
2
2
)
(
=
yz +
=
z
2
2
=
y
+
z
)
=−
z
∂ y
2
dz
−∂ y
2
dz
ψ
= z
∂ y
ds = dy
ψ
1
d
n
=
0,
n
= −
1
2
2
=
(
y
+
z
)
y
z
∂ n
2
ds
∂ ψ d 2 2 = yz += y −∂ z 1 2 ( )
ψ
d
2
2
=
yz +=
y
−∂ z
1
2 (
)
dy

J

eff

= 4

z

T = GJ φ

eff

ψ =

y

z

t
t
2 2 ∂ ψ ∂ ψ ∇ 2 ψ =+= 0 2 2 ∂ y
2
2
ψ
ψ
∇ 2
ψ
=+= 0
2
2
y
z
ψ = yz
∂ψ =
z
∂ y
ψ
= G
φ
− z
σ xy
y
⎝ ⎠
y
= G
ψ
φ
+ y
σ xz
∂ z
⎠ ⎟
2 T y
σ
= 2 G y =
φ
xz
J eff
= 0
σ xy
2 2 ⎫ ⎪⎛ ⎧ ∂ ψ ⎞ ⎛ ∂ ψ ⎞ ⎪ = ∫
2
2
⎪⎛ ⎧
ψ
ψ
=
∫ ⎨⎜
z
++ ⎜ y
dA
∂ y
∂ z
⎟ ⎠
A ⎩ ⎪
⎪ ⎭
3
bt
2
3
y dA
== 4 I
bt
z
3
J
=
A
eff
3
σ xz Tt τ = max J eff
σ xz
Tt
τ
=
max
J eff

Out of plane warping

u

x

= φyz

>> z= linspace(-5,5, 20); >> y =linspace(-0.5,0.5,10); >> [zz, yy] = meshgrid(z,y); >> ux=zz.*yy; >> mesh(zz,yy,ux) >> axis equal >> view (50,20)

no warping no warping
no
warping
no warping
>> ux=zz.*yy; >> mesh(zz,yy,ux) >> axis equal >> view (50,20) no warping no warping

The results we obtained for the torsion of a thin rectangle can also be used, with some qualifications, for other thin open sections such as shown in the figure below

other th in open sections such as shown in the figure below For example, the effective
For example, the effective area moments for the cross sections shown can be calculated as
For example, the effective area moments for the cross sections shown can be calculated as
1
(a
)
bt 3
J eff =
3
1
1
3
(b)
J eff = 3 b 1 t 1 3 +
b 2 t 2
3
1
3
(c
)
J eff = 1 3 b 1 t 1 3 + 1 3 b 2 t 2 3 +
b 3 t 3
3

Also, the maximum shear stress formula can still be applied as

where

t max

τ max = T t max

τ max = T t max J eff

J

eff

is the largest thickness of the cross section. However, this maximum shear stress occurs on the outer edges of the thickest section and does not account for the stress concentrations that occur at re- entrant corners such as those marked with a C in Fig. 1. At such locations, the stresses depend on the local radius of curvature of the corner and may be considerably larger than the value predicted from Eq. (1). Such stress concentrations can be taken into account by finding either numerically or experimentally a stress concentration factor, K, for each re-entrant corner and then examining all high stress points and choosing the one with the highest stress, i.e.

⎤ ⎥ τ max = ⎡ K Tt ⎢ ⎣ ⎦ J eff max
τ max =
⎡ K Tt
J eff
max

K =

3

2.5

2

τ max

τ 1.5

1

0.5

0

t Stress Concentrations open section r τ τ max t r τ τ max closed
t
Stress Concentrations
open section
r
τ
τ
max
t
r
τ
τ
max
closed section
0.5
1.0
1.5

r / t

Centerline warping of thin open sections

u x = −φ′ω ω = ∫ r ds ω + sectorial area function ⊥
u
x = −φ′ω
ω
=
r ds ω
+
sectorial area function
0
ds r ⊥ d Ω T O center of twist d ω = r ds
ds
r ⊥
d Ω
T
O
center of twist
d
ω
=
r ds
1
d
Ω =
r ds
2
ω
=
2
d
Ω=
2
Ω+
ω
ω +
0

since we are taking T as positive counterclockwise, ω is positive if the area is swept out in a counterclockwise manner

ω

To locate the center of twist, O, we must have

To fix

ω

0

y

z

ω

dA = 0 dA = 0

ω

we can specify

ω dA = o

y and z are measured from the centroid

of the cross section

An ω satisfying all three of the above conditions is called a principal sectorial area function, ω p

u x

= −φω

p