F.3
F.4
R.3
R.4
R.5
R.6
R.7
Klueh,
Containment Designs
C.1 Regulations and Standards for Design of Nuclear Facilities
C.2 Nonlinear analysis of Reinforced Concrete Structures 0.Biiyiikoztiirk,
in Computers and Structures, V.7, 1977
C.3
P.3
P.4
Criteria of the ASME Boiler and Pressure Vessel Code for Design by
Analysis in Sections I11 and VIII, Division 2, ASME 1969
P.5
P.6
P.7
P.8
P.9
22.314/1.56/2.084/13.14
Fall 2006
Problem Set I
Due 09/19/06
28.5
50.2
12.2
51.2 (Max.)
10.4
43.6
Length (mm)
50.800
50.848
(yielding begins)
56.1
67.3
69.8 (fracture)
2. Calculate the maximum normal stress and the maximum shear stress in the cube shown
below.
3. This problem illustrates some important stress tensor concepts and some calculation methods
for treating a stress tensor.
Consider three stress tensors a , b , and c dened as follows:
55 5 30
a =
5 55 30 ;
30 30 20
10 0
0
b =
0 10 0 ;
0 10
c = a + b
where the stresses are given in MPa units.
(a) What are the three principal stresses that characterize each of these stress tensors ?
(b) Draw a 3D Mohrs circle to represent the three stress states.
(c) Also place points on your Mohrs circles that give the stresses on planes normal to each
of the original coordinate axes (x, y, z).
(d) What are the unit vectors (direction cosines) that dene principal directions for the
three stress tensors?
(e) What is the Tresca Stress for each stress tensor? What is the von Mises Stress for each
stress tensor?
22.314/1.56/2.084/13.14
Fall 2006
Strain:
xx 0
0
=
0 yy yz
0 zy zz
Note that the shear stresses on the plane normal to x direction are zero. Therefore, we
can derive the two principal stresses on the yz plane using the solution for a plane stress
condition.
yy + zz
yy zz 2
2
i,j =
) + yz
2
3. (a) The principal stresses are the eigenvalues of the stress tensor. Its solved by using
MATLAB (See the code in the end). For
55 5 30
a =
5
55
30
,
30
30 20
For
10 0
0
b = 0 10 0
,
0 10
1 = 2 = 3 = 10 MPa.
For
c = a + b ,
the principal stresses are found to be:
(b) See Figure1 and Figure 2. The Mohrs Circle for b is a single point (10, 0).
(c) The points on the 3D Mohrs circles that give the stresses on planes normal to each of
the original
coordinate
axes (x, y, z) are:
2 + 2 ), ( ,
2 + 2 ), and ( , 2 + 2 )
(x , xy
y
z
xz
yx
yz
zx
zy
as shown in Figure 1 and Figure 2.
u1 = 0.5774
; u2 =
0.7071
; u3 =
0.4082
.
0.5474
0.8165
1
0
0
u1 = 0 ; u2 =
1
; u3 =
0
.
0
0
1
Since c is the sum of a and hydrostatic pressure tensor b , the unit vectors should be
the same
as those
of a .
0.5774
0.7071
0.4082
u1 =
0.5774
; u2 =
0.7071
; u3 =
0.4082
.
0.5474
0.8165
(e)
T resca = max{1 2 , 2 3 , 3 1 }
vonM ises =
(1 2 )2 + (2 3 )2 + (3 1 )2
2
Therefore:
T resca
vonM ises
a
b
c
90 (MPa)
0 (MPa)
90 (MPa)
81.85 (MPa) 0 (MPa) 81.85 (MPa)
%
22.314 (Fall 06) Problem Set I3.
%
09/19/2006
%============================================================
Sigma a = [ 55 5 30
5 55 30
30 30 20];
Sigma b = [10 0 0
0 10 0
0 0 10];
Sigma c = Sigma a + Sigma b;
10
s2
= Eigval a(2,2);
s3
= Eigval a(1,1);
c1
= (s1 + s2)/2; r1
c2
= (s2 + s3)/2; r2
c3
= (s1 + s3)/2; r3
x1
y1p
x2
y2p
x3
y3p
=
=
=
=
=
=
20
= (s1 s2)/2;
= (s2 s3)/2;
= (s1 s3)/2;
axis equal;
plot(x1, y1p);
plot(x1, y1n);
plot(x2, y2p);
plot(x2, y2n);
plot(x3, y3p);
plot(x3, y3n);
= y1p;
30
= y2p;
= y3p;
40
tau x
= sqrt(Sigma a(1,2)2 + Sigma a(1,3)2);
sigma y = Sigma a(2,2);
tau y
= sqrt(Sigma a(2,1)2 + Sigma a(2,3)2);
sigma z = Sigma a(3,3);
tau z
= sqrt(Sigma a(3,1)2 + Sigma a(3,2)2);
50
c1
c2
c3
= (s1 + s2)/2; r1
= (s2 + s3)/2; r2
= (s1 + s3)/2; r3
x1
y1p
x2
y2p
x3
y3p
=
=
=
=
=
=
60
= (s1 s2)/2;
= (s2 s3)/2;
= (s1 s3)/2;
= y1p;
= y2p;
70
= y3p;
80
90
22.314/1.56/2.084/13.14
Fall 2006
Problem Set II
Due 09/26/06
1. Consider a cylindrical vessel of inner radius R and wall thickness t with at ends. The
pressure inside the vessel is Pi and the surrounding pressure Po . What is the relative error in
estimating the maximum value of the stress intensity in the cylinder based on the thin shell
approximation for values of:
t/R = 0.03
t/R = 0.10
t/R = 0.15
t/R = 0.30
Consider two cases:
Pi = 2Po
Pi = 20Po
2. A pressure vessel is constructed of a cylinder with a hemispherical head at each end. There
is no external restraint to either axial or radial displacement. Inside radius of both cylinder
and hemispheres is R. The wall thickness is uniform at a value t. The length of the cylinder
is L. No aws or stress concentrations are present. Dimensions are:
R = 110 cm
t= 11 cm
L = 433 cm
Material properties:
Youngs modulus = 200 GPa
Questions:
(a) What is the total (peak) stress as a function of radial position (z) at a junction between
cylinder and hemisphere.
(b) What is the maximum radial displacement of the vessel cylinder and sphere?
22.314/1.56/2.084/13.14
Fall 2006
Pi R2 Po (R + t)2
z =
(R + t)2 R2
Pi P o
t
=
(R + )
t
2
r =
(1)
(2)
(3)
Therefore:
Sthin = r =
Pi P o
t
Pi + Po
(R + ) +
t
2
2
(4)
dr r
+
=0
dr
r
(5)
Hooks law:
1
(r z )
E
1
= ( r z )
E
1
z = (z r )
E
r =
Since = u/r, r =
du
,
dr
(6)
(7)
(8)
we get:
d
1
= (r )
dr
r
(9)
For this close end cylinder far from the end, plane stress condition is assumed, i.e., z is
const.
Plug Eq 6 and Eq 5into Eq 9, we get
d
( + r ) = 0
dr
(10)
(11)
(12)
(13)
(14)
Sthin

Sthick
0.03
1.41%
1.31%
0.10
0.15
4.13% 5.67%
4.09% 5.88%
0.30
8.88%
10.46%
2. We use thin shell approximation to solve this problem. For a region of cylinder far from a
(15)
(16)
(17)
Radial displacement:
uc =
P R2
t
(2 + )
R
2Et
(18)
= =
(19)
(20)
Radial displacement:
us =
P R2
t
(1 + )
2Et
R
(21)
2
u0 = u c +
0 =
V0
M0
+
2 3 D 2 2 D
(22)
V0
M0
2 2 D D
(23)
2R
22
V0 +
M0
Et
Et
(24)
22
43
V0 +
M0
Et
REt
(25)
where, = s R, s = ( 3(1
)1/4 , u0 is the radial displacement at the junction, 0 is the
R2 t2
1
slope at the junction.
Due to the continuity of the displacement and slope, the four unknowns u0 , 0 , M0 , and V0
1
Note that the direction of the shear force Q in Note L4 is different from that in Note C. If you assume the
direction of Q0S is same as QOC , the continuity of shear force should give that: Q0S = Q0C .
Vo
Mo
Mo
z
Vo
M0 = 0
V0 = P R2 /(4R + Et/ 3 D)
V0
u = uc +
2D 3
P R
x =
2t
PR
EV0
=
+
t
2DR 3
r = P/2
(26)
(27)
(28)
(29)
(30)
(31)
With mean radius R = inner radius + t/2 = 1.155 m, t = 0.11 m, E = 200 GPa, = 0.3, we
get
(a) At the junction, from Eq 29Eq 31:
x = 81.38 MPa
= 122.06 MPa
r = 7.75 MPa
(b) Radial displacement as a function of radial postion z is: (R + z) . Thus, from Eq 18,
the radial displacement of cylinder is:
PR
t
(2 + )(R + z)
2Et
R
From Eq 21, the radial displacement of hemishpere is:
PR
t
(1 + )(R + z)
2Et
R
From Eq 28, the radial displacement of junction is:
(
PR
t
V0
(2 + ) +
)(R + z)
2Et
R
2DR 3
Therefore:
Cylinder Sphere
0.00089 0.00037
Junction
0.00063
22.314/1.56/2.084/13.14
Fall 2006
Due 10/05/06
1. For the same vessel described in Problem II2, nd out the maximum allowable pressure
of the coolant so as not to cause elastic failure under expected static loading and fatigue
conditions. State the criterion you wish to apply before proceeding to numerical evaluation.
Yield stress = 320 MPa, Ultimate tensile stress = 500 MPa.
SN curves for both high cycle fatigue and low cycle fatigue are attached.
Expected number of cycles between cold and hot zero power conditions = 500.
Expected number of cycles between hotzero power and hotfull power conditions =
105
Temperature prole is distributed parabolically at hotfull power conditions:
1
T = Tout T (1 2z/t)2
4
Where Tout is the outside wall temperature. T = 50 K is the temperature difference
between inside and outside wall.
22.314/1.56/2.084/13.14
Fall 2006
MPa
For cylinder:
Pm =  r  = 11P
For sphere:
Pm =  r  = 5.75P
Thus we have:
11P 166.7 MPa
P 15.6 MPa
For secondary stresses, we will consider three regions, the cylinder region far from junction,
the sphere region far from junction and junction region. The thermal stress is assumed
identical in three regions. Dene:
the stresses in cylinder region: ,c , x,c ,r,c
the stresses in sphere region: ,s ,x,s ,r,s
the stresses in junction region: ,j , x,j , r,j
It is easy to see that: ,s < ,j < ,c , since the shear force in junction is in the same
direction with the system pressure for sphere, and opposite direction for cylinder. Also,
x,s = x,j = x,c and r,s = r,j = r,c . Thus the maximum stress intensity will be in the
cylinder region. From ASME code:
2
1
Pm + Q 3Sm 3min( y , u ) = 500MPa
3
3
Pm + Q = max(,c r,c , x,c r,c , ,c x,c )
PR
E
P PR
E
P PR
= max(
+
(Tavg T (z)) + ,
+
(Tavg T (z)) + ,
)
t
1
2 2t
1
2 2t
Average temperature is Tavg :
t/2
Tavg =
T (z)2rdr
t/2
t/2
t/2
2rdr
= Tb 0.3T
1
MPa
MPa
n1
n2
+
1
N1 N2
Assuming from CZP to HZP, only the pressure oscillations, n1 =500.
Assuming from HZP to HFP, only the temperature oscillations, n2 = 105 .
From topic D, Salt = 12 max,T resca . For high cycle, the alternating stress is the thermal
stress, the amplitude is: Salt = 25.7MPa. Applying a safety factor of 2, and using the most
conservative curve 7 on the high cycle fatigue gure, it can be seen that at 51.4 MPa, the
cycle number is 6.5E5, i.e. N2 >6.5E5. Thus, N1 n1 /(1 n2 /N2 ) = 588.
From low cycle fatigue gure, it can be seen that Salt 65, 000 psi = 448 MPa , at low
cycle, the only stress variation is due to pressure variation and:
Salt = 12 max,T resca = 5.5P Again, applying a safety factor of 2:
5.5P 448/2 = 224 MPa
P 40.7
MPa
From above analysis results, it can be seen that the static primary stress condition in cylinder is the
limiting factor. Therefore, the system pressure:
P 15.6MPa
22.314/1.56/2.084/13.14
Fall 2006
Problem Set IV
Due 10/12/06
1. Consider a cantilever beam with one end xed into a wall. The dimensions are shown in the
gure below. A force F = bhy /2 is applied to the other end. y is the yield strength. Now a
bending moment is applied as shown in the gure. Determine the magnitude of the moment
M at which the the beam fails. Assume idealized material law: elastic perfect plastic.
M
F
h
b
2. The stress tensor can be written as the sum of a deviatoric stress tensor and a dilation stress
tensor and the strain can be written as the sum of volumetric strain and deviatoric strain as
follows:
1 0 0
S1 0 0
P 0
0
0 2 0 = 0 S2 0 + 0 P 0
0 0 3
0 0 S3
0
0 P
1 0 0
1 0 0
v /3 0
0
0 2 0 = 0 2 0 + 0 v /3 0
0 0 3
0 0 3
0
0 v /3
Where P = (1 + 2 + 3 )/3 is the hydrostatic pressure and v = 1 + 2 + 3 is the
volumetric strain.
1
UD =
Si i
2 i=1
3. Consider several ceramic fuel materials with properties listed in the table below:
Property
Thermal conductivity average (W/mo C )
Melting point (o C )
Linear coefcient of expansion (/o C )
Fracture strength (MPa)
1
UO2
3.6
2800
10.1E6
110
UC
23
2390
11.1E6
60
UN
21
2800
9.4E6
22.314/1.56/2.084/13.14
Fall 2006
h/2
h/2
h/2
h/2
y/2
(a)
(b)
h/2
h/2
h/2
z0
h/2
(c)
(d)
The moment for reaching the yield at a local space can be calculated. The moment shown in
the gure is counter clockwise, therefore:
1 y
h
2
h
M1 = 2 ( ) ( ) ( ) b
2
2
2
2
3
1
y bh2
12
When further increasing the magnitude of moment, the yielding zone in lower beam will
further extend but the maximum stress remains y as elastic perfect plastic material property
M1 =
is assumed. The upper beam become compressed and will also reach the yielding at a certain
moment. This situation continues until the beam gets to the yielding at the whole cross
section. The stress distribution is shown in Figure (d). To determine the moment, we rst
need to evaluate the point z0 which divide the compression and tension. Force balance in
axial direction will give:
F =
h
h
y bh
= ( z0 )(y )b + (z0 + )y b
2
2
2
h
h
h
= ( z0 ) + ( + z0 )
2
2
2
h
z0 =
4
Therefore, the moment M for actual yielding of the whole beam can be calculated:
h
h
h
h
M = y ( z0 ) ( + z0 )/2 b + y (z0 ( )) ( + z0 )/2 b
2
2
2
2
M=
3
bh2
16 y
is larger than M1
2. Total strain energy is the sum of distortion energy and dilation energy:
UT = UD + US
UD = UT US
3
UT =
1
i i
2 i=1
1
US = (P v )
2
Therefore:
3
1
1
UD =
i i (P v )
2 i=1
2
3
1
1
=
(Si P )(i + v /3) (P v )
2 i=1
2
3
(1)
3
1
1
1
1 1
=
Si i +
Si v /3 + (P )
i + (P )
v /3 (P )v
2 i=1
2 i=1
2
2
2
i=1
i=1
Since
3
Si =
i=1
3
i=1
(i + P ) = 1 + 2 + 3 + 3P = 0
(2)
(i v /3) = 1 + 2 + 3 v = 0
(3)
i=1
i =
i=1
1
UD =
Si i
2 i=1
3.
(a) From the class note, assuming a constant value of thermal conductivity the linear power
is proportional to the the temperature difference between the center line and fuel sur
face.
q = 4kT
The maximum linear power is limited by the melting of fuel, i.e. when melting rstly
occurs at the center line of the fuel. Thus assume Tf o = 700o C and plug melting points
and average thermal conductivities of UO2 , UC and UN into above equations, we can
get an estimation of the maximum achievable linear heat generation rates:
qmax,U
O2 = 95
kW/m
qmax,U
C = 489 kW/m
qmax,U
N = 554 kW/m
(b) Refer to note M12, thermal stress due to a parabolic temperature distribution would
give a maximum stress intensity at the outer surface of the fuel in elastic regime. Using
Trescas yield condition, when fracture rstly occurs at the outer surface of the fuel,
the linear heat generation rate can be related to the the fracture stress as follows:
q =
T 16k(1 )
E
From the class notes, E = 175 GPa, = 0.3 for UO2 , assume the same properties for
UC. Plug material properties into the above equation, we can get:
qU O2 = 7.7
kW/m
qU C = 24.3
kW/m
Thus the q to initiate cracking is typically less than 10% of that to initiate melting of the
fuel. Both UC and UN provide an opportunity to increase q signicantly without violating
the limits.
This problem illustrates some "plastic strain" concepts and some calculation methods for
treating plasticity.
Uniaxial StressStrain Information: Consider a material that has the uniaxial stressstrain curve
(actually two straight lines) shown below. The first portion of the curve, a line from the origin to
point A, has no plastic deformation. The second
STRESS
portion, a line from point A to point B, has both
elastic and plastic deformation.
The elastic stressstrain behavior is characterized
by a Young's Modulus of 195 GPa and a Poisson's
ratio of 0.3.
Derived Properties: Using the above uniaxial stressstrain information, what is the 0.2% offset
yield stress for this material? What is the plastic stress (cp)as a function of equivalent (strainhardening) strain (E,)?
Loadina Sequences and Questions:
A three dimensional stress is applied in a proportional manner to a solid made of the above
material until the following stress state (oa,J is reached (note that this state is 2.5 multiplied by
the oa tensor of problem set #A}:
Show that this equivalent stress is equal to the previously defined "von Mises stress."
2) What strain tensors (elastic, mechanical, and total) exist when aa is reached?
The stress on the same solid is next proportionally reduced until zero stress remains.
3) What strain tensors (elastic, mechanical, and total) exist when zero stress is reached?
The same solid is subjected to a different proportional loading to reach the stress state (abii),
where:
5) What strain tensors (elastic, mechanical, and total) exist when zero stress is reached?
22.314
Known
v =0.3
E =195000
A
= 150
oB :=260
sB =0.5410  2
Derived properties
EmA =0
on
rsmB =sB  E
smY :=0.002
ry :
aY
smY  smA
mA .(oB  aA) . aAd Linear interpolation between A anb B to get yield stress arY
emB  smA
204.098
aB
sB = B  i =0.. 100
e
= seB.so 8eA

Opi = teseB
.(oB 
A) oA
0.002
0.004
I/I ap as a function of so
sA
too
200
51
0
200
0.006
75
1)
akk:=o 0,0
Oa, I +
+l aa 2 , 2
ukk =325
i:=0..2
j =0..2
ac :=0
Sal1 :=oa,
S(ij,0)75
129.167 12.5
29.167
Sa =12.5
75
une = i
75
Sa)
75
58.333
(Sao,
)22+(Sao,22 (Sa
1 ,)2
oae = 204.634
Now calculate VonMises stress for comparision
eigena:= eigenvals(aa)
aVM
cVM =204.634
We can see cae=aVM
2.2 (Sa2,)]
1(Sa,2)
1
selaij
5(i
v)ai,j
.[(l
=2
S4.167.10 4 833310'
cela=
8.333.10 
[5104
5 104
510
4.16710
5.0
j,0)Okk]
4
1.66710
4
b) Mechanical strain
Since aae=204>aa=150, mechanical strain exists and has to be calculated
Since the stress was applied in a proportional manner, Saij/ce is constant, thus we have
cm(ij)=(3*Sa(i,j)2*ae)*se
Lae E
roota = roo
Ec
A.(oB 
A)+
A],ea
= roota
sae = 0.002
100sac
0.202
3Sa
mai 2.aa
4.31810
ma= 1.85110
4

0.001
Total stress
1.85110
4
0.001
4.31810
0.001
0.001
8.637"10
4
0.002
8.48510 4 2.684*.t0
 4 0.002
8.48510
Etola = 2.68410
. OOl
0.002
0.002
3) When stress is proportionally reduced to zero, sela vanishs while ema remains unchanged.
stola=tma when zero stress is reached
260 00
ab :=
0 00
000
akk := bo,0 t
bl. I
b2 ,2
akk = 260
i =0..2
j =0..2
Sbi, j
5(i
abi j 
173.333
Sb = 0
obe=
j,0)kk
0
86.667 0
86.667
0
3.[(Sbo 0) 2
2
(Sb, 1)22+S
(Sbb2)22+ (Sb
1 j 2+ (Sb2) 22+ (Sb22 ) ]
obe =260
4) Strain tensors for ab
a) Elastic strain
i =0..2
j =0..2
celbi j =
(Ij v)obi,
v5(ij, 0).kk
0.001 0
selb = 0
4 10
4
0
4104
b) Mechanical strain
During the second loading, when ae increases from 0 to oae, no mechanical strain is produced.
Since cbe=260 > aae=204, new mechanical strain will be produced when ae increases from aae to
abe.
bshe:=
roothb=ro
obe
a((aBA)
4eB 
aAse
zeA
she =rootb
abe = 0.004
100sbe= 0.407
smadd,, :=
3.Sb.i
.(sbe  sac)
0.002 0
cmadd
0
0
0.001 0
0
0.001
smadd is the addtional mechanical strain tensor that is produced when ve increases from ose to
abe. Adding smadd to Ema we can get the total mechanical strain when ob is reached.
smb := madd
Emb"
sma
0.002
1.85110
0.001
1.85110

5.916*10
0.001
c) Total stress
stolb = celb  smb
/ 0.004
Etolb=
11.85110
o0.001
1.851"10
9.916100.001
0.001
0.001
f002
Emb
remains unchanged.
Note: Some of the elements in the matrices may not be accurate because Mathcad sometimes
doesn't show enough significant digits.
22.31411.5612.084113.14Fall 2006
Problem Set VI
Due 10126106
his
Geometry (20C)
Young's Modulus
Poisson's Ratio
C o e f f i c i e n t o f Thermal
Expansion
;9 = c l a d creep r a t e (sl.) .
where a has the u n i t s MPa.
9
3.
Operating Conditions
= 76 GPa;
= 0.25;
= 6.7 m1m.K; and
= (1 x 109)ag;
 An uppower
 An uppower
4.
Calculational Basis
 Use a s i n g l e r i n g t o
represent t h e c l a d
 Consider
t h a t t h e r a d i a l d e f l e c t i o n a t the f u e l outer surface
and the c l a d average temperature vary l i n e a r l y w i t h time
during each uppower ramp.
 Use only
contact.
L e t the creep
5.
Questions
 c l a d stress;
 clad strain;
 radial
and
This problem set illustrates applications of beam theory to Zircaloy Follower in a BWR for
calculation of curvature caused by Zircaloy growth ; Consult Notes X on Beam Theory.
ZIRCALOY FOLLOWER
a)
Y
Z
W
T
C 1 N + C2 N2 ;
(1.1)
where :
the zdirection growth strain ( egz ) is given in percent;
the fast fluence (N) is given in the units of (1021 fast neutrons per cm2 ) with the fast flux
cutoff specified by E > 1 MeV; and
the constants are C1 = 0.013 and C2 = 0.0018.
b)
For points originally on the axial centerline (x = 0; y = 0 ), denote displacements in the xdirection, the ydirection, and the zdirection displacement, respectively, by u, v, and w.
At z = 0, the follower is supported so that u, v, and w are all zero and so that no moments are
applied. At z = L, the follower is supported so that the zdirection force is zero, so that u and v
are zero, and so that no moments are applied.
c) Fast Neutron Fluence:
After several refueling cycles, a follower has an accumulated fast fluence given by:
N = [ Nx (x)] [N z (z )] ;
(1.2)
( ( ))
z L
2
N z (z ) = 1.49cos p
Le
(1.3)
(1.4)
x
z
y
Beam model
Geometry and material properties:
L=2.4m; W=200mm; T=7mm
and
E=75 GPa and =0.25
Boundary conditions:
At z=0:
u=v=w=0
M0=0
Cross section
At z=L:
u=v=0
ML=0
Fz=0
z =
+ zo ; where zo = 0.01 gz
E
furthermore,
z = za k y x; where za =
d y d 2 u
dw
and k y =
= 2
dz
dz
dz
therefore,
dw d 2 u
x 0.01 gz )
dz dz 2
At any cross section,
M y = z xdA = M 0 = 0
z = E(
In the meanwhile,
M y = z xdA = E( za k y x 0.01 gz )xdA
A
Therefore,
0.1x
M y = E za (z) dA + Ek y (z) x 2 dA + 0.01EC1 N z (z) 151 +
xdA + 0.01EC2 N z (z) 2 N x (x) 2 xdA
0
.
2
A
A
A
A
where
dA = 0 for symmetry
T2 2
6
2
A x dA = 20 x wdx + T22 x Tdx = 4.6722 10
2
N
A
Then,
2z L
2z L
k y = 0.075C1 N z (z) 2.25C2 N z (z) = 0.001453cos
0.008991 cos
2
L
2
L
e
e
2z L
2z L
= 0.001453cos
0.004496 0.004496 cos
Le
2Le
z
z
2z L
2z L
y (z) y (0) = k y (z )dz = 0.001453cos
dz
+ 0.004496 + 0.004496cos
0
2 Le
Le
0.001453Le 2z L
0.004496Le 2z L
L
L
+ sin
+ sin
sin
0.004496z
sin
2Le
2
Le
Le
2Le
= 0.001175sin(1.2368z 1.4842 ) 0.004496z 0.001818sin(2.4736z 2.9684 ) 0.001484
Again,
z
u ( z ) u(0) = y ( z )dz
0
dw(z)
dz
We already know
Fz
= z dA = FzL = 0
A
In the meanwhile,
Fz = z dA = E( za k y x 0.01 gz )dA
A
where
2
dA = A = 2TW T = 0.0028
A
xdA = 0
A
Therefore,
2z L
2z L
= 0.002905cos
+ 0.008996 cos
2L
2L
e
e
2z L
2z L
= 0.002905cos
+ 0.004498 + 0.004498 cos
Le
2Le
0.002905Le 2z L
L
L 0.004498Le 2z L
+ sin
+ sin
sin
+
sin
+ 0.004498z
2Le
2
Le
Le
2Le
= 0.002349 sin(1.2368z 1.4842 ) + 0.001818sin(2.4736z 2.9684 ) + 0.004498z + 0.002653
=
Therefore:
22.314/1.56/2.084/13.14
Fall 2006
Problem S e t L. 54
This problem i l l u s t r a t e s some of t h e concepts i n r e i n f o r c e d /
p r e s t r e s s e d concrete containment s t r e s s c a l c u l a t i o n r .
1. Geometq
Directions
depth.= 165
+1
Steel
Tendons
Radial
Longitudinal
Hoop
p i t c h = 165
iameter = 57
Concrete
I n s i d e Radius = 20 600
Wall Thickness = 1380
Figure 1
plm
diameter s t e e l b a r s (one row near each surface) with a p i t c h of approximately L65 mm.
Harever t h e
Problem S e t L. 54
3.
Questions
(a)
What prest;ess
l e v e l s (one l e v e l f o r a l l l o n g i t u d i n a l tendons,
4.
C a l c u l a t i o n a l Basis
I l r s m e e l a s t i c behavior.
Consider t h a t t h e c y l i n d e r is "long" b u t consider t h e region near
t h e b a s e mat.
Consider t h e c y l i n d e r t o b e b u i l t  i n a t t h e p o a i t i o n i t j o i n s t h e
base mat (zero r a d i a l d e f l e c t i o n . zero s l o p e of r a d i a l d e f l e c t i o n ) .
   Lt'
Define t h e f i e x u r a l r i g i d i t y on t h e b a s i s of t h e s e c t i o n shown i n
L
C a l c u l a t e r i g i d i t y on the b a s i s t h a t
s =
4 As
tp
PR
= 3.708MPa m
2
P
r =
2
Ec
El = (1 ls )
1 2
Ec
E = (1 s )
1 2
E* = ls E s + El
Nl =
and then
=
=
2 El
t
(1
)
E
+
E
s s
E *
R
E
Nl +
E
t
r (1 s (1 ls ) )
E*
1
E*
Ec t 3
2
2
D=
( + ls s t) + E s ls s t , where s = thickness / 2 depth = 0.525m
2
1 12
According to notes for problem set L54, we get
w
R
c = p = (P )
R
R
Nl
r
El c + (1 ls )
1
t
lc =
E*
lc max =
t
for
2
Ec
( c max + lc max + bl ) +
r
2
1
1
where
bl (z) =
t d 2w
= t 2 w p e z [cos z sin z ]
2
2 dz
1
where in turn =
4DR
Since c max and lc max reach the maximum value at z , therefore, the strain due
to bending bl = 0 as z .
Now, the maximum longitudinal stress shall be offset by the tendon prestress to get
zero net concrete stress upon pressurization and therefore we get
1 ls
lsprestree =
lc max
ls
c max =
r
( c max + lc max ) +
2
1
1
Then, we get
1 s
sprestree =
c max
For the two cases where the longitudinal tendon pitch is equal to 165 mm and 200
mm, respectively, we just substitute corresponding numbers to the above equations, and
obtain the final results:
165 mm:
lsprestree = 99.5MPa
sprestree = 104.36MPa
200 mm:
lsprestree = 123.7MPa
sprestree = 104.41MPa
According to these result, when changing the longitudinal tendon pitch from 165
mm to 200mm, the required prestress in the longitudinal direction is increased by almost
25%, however the required prestress in the hoop direction almost remains unchanged.
This problem set illustrates some features of seismic analysis. A response spectrum
modal (lumped mass) method is to be used.
1)
The pipe has an outside diameter of 210mm, a wall thickness of 7mm, and a length
of 3m. The pipe is made of a material with Youngs modulus = 200 Gpa and density
= 8500 kg/m3 . Liquid water (density = 750 kg/m3 ) is present both inside and outside
of the pipe. The added mass coefficient for the outside water is 1.1. Neglect direct
treatment of fluid friction, fluid drag, and solid friction these effects are incorporated
in the damping matrix of item 2.
The support at z = 0 is fixed (zero displacement and zero slope). The support at the
other end (z=L=3 m) is a roller (zero displacement, zero moment, and zero axial
force).
2)
Massless springs should be chosen to give the same stiffness characteristics at the
point masses as for a beam with the same supports (the resulting stiffness matrix
has no zeros).
The damping matrix should have only two nonzero elements (located on the
diagonal). These elements should be equal and should give 2% critical damping for
vibraton at the system fundamental frequency.
3)
[S
Questions:
a)
What is the fundamental undamped frequency?
b)
What are the peak forces that act on the supports during the
earthquake?
Lumped mass
mB
PB
mA
PA
Pipe
m A = mB =
(R 2 Ri 2 )L + ( Ri 2 + 1.1R 2 )L water
2
= 133.724Kg
The governing motion eqution for dynamic response of this pipe can be expressed as:
[M ]{u&&} + [D]{u&} + [K ]{u} = {F}
where:
0
m
[M]=mass matrix, M = A
0 m B
[D]=damping matrix
[K]=stiffness matrix
First of all, we should calculate the stiffnes matrix of the pipe using beam theory:
Suppose the displacements of point A (L/3) and B (2L/3) are uA and uB, respectively.
Assuming that there is only a force PA acting on point A (at L/3), we can calculate the
corresponding displacements of points A and B, satisfying the boundary conditions:
u(0) = 0
u(L) = u(3) = 0
du(z)
=0
dz z=0
For L>z>L/3:
dM y
M y = Vx (3 z) =
(3 z)
dz
Solving this equation with the boundary condition My(L)=0, we get
M y = C1 (3 z)
For z<L/3:
dM y
(3 z)
M y = 2PA + V x (3 z) = 2PA
dz
Solving this equation:
M y = C2 (3 z) + 2PA
y (0) =
0.105
0.098
where I = x 2 dA = 2 cos 2 d
A
r 3dr = 2.3023 10 5
So that
C1
EI
Ky =
=
EI C 2 (3 z) + 2PA ,0 < z < 1
EI
EI
C1 (3 z) 2 9C2 4PA
+
+
EI
EI at z=0, y=0
2
2EI
y = K y dz =
2
C
2P
9C
(3
z)
2
+ A z + 2
EI
2
EI
2EI
3
C1 (3 z)
4P 13PA 9C 2
9C
+ 2 + A z
EI
6
2
EI
EI
3
EI
2EI
u(z) = y dz =
3
C
P
9
C
9C
(3
z)
+ A z2 + 2 z 2
EI
6
EI
2EI
2EI
At last, we have another boundary condtion, u(L)=0
So that
My
d d 2 u
=
dz dz 2
4P 13PA 9C2
9C
3 2 + A
=0
EI 3EI 2EI
2EI
23
C2 =
PA
27
C1 = C 2 + PA =
PA
27
Then, we obtain
C 2 8 PA 9C 2 9C 2
P
+
+
0.1358 A
u A
EI 6 EI 2EI 2EI
EI
u = C 1 9C
8P
13P
9C
P
A
B 1 + 2 + A
2 0.142 A
3EI 2EI
EI
EI
EI 6 EI
Similarly, assuming that there is only a force PB acting on point B (at 2L/3), we can
calculate the corresponding displacements of points A and B, satisfying the boundary
conditions:
For L>z>2L/3:
dM y
(3 z)
M y = Vx (3 z) =
dz
Solving this equation with the boundary condition My(L)=0, we get
M y = C1 (3 z)
For z<2L/3:
M y = PB + V x (3 z) = PB
dM y
dz
(3 z)
0.105
0.098
where I = x 2 dA = 2 cos 2 d
A
r 3dr = 2.3023 10 5
So that
C1
EI (3 z),2 < z < 3
=
Ky =
EI C2 (3 z) + PB , z < 2
EI
EI
C1 (3 z) 2 9C 2 5PB
+
+
2
2 EI 2 EI at z=0, y=0
y = K y dz = EI
2
C 2 (3 z) + PB z + 9C 2
EI
2
EI
2EI
My
d d 2 u
=
dz dz 2
C1 (3 z) 3 9C 2 5PB
9C 19PB
z 2
+
+
EI
6
2 EI 6EI
2 EI 2 EI
u(z) = y dz =
3
C 2 (3 z ) + PB z 2 + 9C 2 z 9C 2
EI
6
2EI
2EI
2EI
At last, we have another boundary condtion, u(L)=0
So that
5P 9C
19PB
9C
3 2 + B 2
=0
2EI 2EI 2EI 6EI
13
C2 =
PB
27
14
C1 = C2 + PB =
PB
27
Then, we obtain
9C
9C
C2 8 PB
P
+
+ 2 2
0.142 B
u A
EI 6 2EI 2EI 2EI
EI
=
u = C 1 9C
5P
9C
19P
P
B
B 1 + 2 + B 2
0.247 B
EI 2EI 6EI
EI
EI 6 EI
So that, according to superposition when PA and PB act on the system simultaneously,
u A 1 0.1358 0.142 PA
u = EI 0.142 0.247 P
B
B
In the meantime,
u A
PA
P = K u
B
B
We can get
1
0.1358 0.142
18.46232 10.614
18.46232 10.614
K = EI
= EI
= 4.6046 10 6
0.142 0.247
10.614 10.15054
10.614 10.15054
Secondly, we should compute the damping matrix:
The damping matrix has only two nonzero elements, located on the diagonal. These
elements are equal and give two percent critical damping for vibration at the system
fundmental frequency.
[M ]{u&&} + [K ]{u} = 0
0 u&&A
18.46232 10.614 u A
133.724
+ EI
=0
0
133.724 u&&B
10.614 10.15054 u B
133.724 2 + 18.46232EI
A
10.614EI
= 0
2
10.614EI
133.724 + 10.15054EI B
0.56364
0.826
Then, we should also calculate the critical damping matrix,
Suppose that the critical damping of lumped mass is considered seperately:
0
C
Dcr = 1
0 C2
[M ]{u&&} + [D]{u&} + [K ]{u} = 0
N
Multiply the above equation with ANT leftly, and also replace {u} with AN {u}, where
0.56364
0.826
AN = [u1 u2 ] =
0.56364 0.826
we get
133.724 2 + C1 + 25.7 EI
A1
0
= 0
2
0
133.724
2.9
C
EI
+
+
2
A2
If there are nontrivial solutions for A1 and A2, we obtain
133.724 2 + C1 + 25.7 EI = 0
133.724 2 + C2 + 2.9 EI = 0
So that critical damping matrix:
0
C 0 251592.2
=
Dcr = 1
0
84514.2
0 C2
198513 77788
T
Dcr = AN Dcr AN =
77788 137593.5
Damping matrix in this case
0
5031.844
D = 0.02Dcr =
0
1690.284
f1 =
According to the response spectrum of Fig 8 in note M32, extrapolating to the calculated
frequencies, we obtain the maximum displacements corresponding to Sa=0.33g are
S
0.33 9.8
S d 1 a2 =
= 3.6538 10 6 m
2
940.8
1
Sd 2
Sa
0.33 9.8
= 3.23 10 5 m
2
316.43
So
0
133.724
{F} =
u&&g
133.724
0
Finally, we obtain
0 u&&A 5031.844
0
0 u A
u& A 25.7EI
133.724
T 133.724
A
=
+
+
N
133.724u&&g
u
u& 0
0
u&&
133.724
0
1690.284
2.9EI
B
B
B
0 u& A
0 u A
u&&A 37.63
0.2624
25.7EI
1
=
+
u&& + 0
u&&g
0.5
2
2
= 0.02 940.8
5.4 10 7 + EI 10.614 10.15054 5.4 10 7
77788
137593.5
3.75 2 93.72 2
=
+
2.56 63.95
0.5
93.795
=
64
0.5
0.5
2
2
= 0.02 316.43
+ EI
13.52 2 337.6 2
=
+
19.85 497.4
0.5
337.87
=
497.8
Now, given the forces of PA and PB, we calculate the forces acted on supports:
FL
PB
PA
A
F0
Assuming the forces that act on the supports are F0 and FL, respectively, as in the above
figure.
dM y
dK y
4
14
FL =
= EI
=
PA +
PB
dz z=L
27
dz z=L 27
F0 = PA + PB FL =
23
13
PA +
PB
27
27
For P1,max:
F 1 23 13
110.71
F1 = 0 =
P1,max =
47.08
FL 27 4 14
Likewise, for P2,max:
F 1 23 13
527.5
F2 = 0 =
P2,max =
308.2
FL 27 4 14
539
=
Newton
311.8
So, the peak forces that act on the support z=0 and z=L are 539 and 311.8 Newton,
respectively.
2
Again, F = F1 + F2
2 0.5
0.5
What is the maximum linear hear generation rate (in kW/m) that the fuel
pin can operate at if the fuel is not to experience any fracture for r < 0.4R,
where R is the pellet radius? You may assume the pressure at the pelletclad gap to be 0.3 MPa.
1.2
Does the clad design satisfy the ASME criteria for structural integrity
under static loads? The fission gases will build up within the clad until the
gas plenum pressure reaches 5.0 MPa. You may assume the coolant
pressure to be 0.3 MPa and the maximum operating linear heat generation
rate is 45 kW/m.
Question #2 (45%)
A long, thinwalled cylindrical tank is used in transporting radioactive gases. While it is
being filled, the tank is subjected to radial constraint that can be approximated by a rigid
boundary that allows slip in the axial direction (see Figure 1).
2.1
What is the maximum pressure that should be allowed in the tank in order
to avoid plastic deformation of the membrane?
Material Properties:
Youngs Modulus, E = 30,000 ksi
(1ksi = 1000 psi)
Poissons Ratio,
= 0.3
Yield Stress,
y = 36 ksi
State any assumptions you make.
2.2
Is fatigue a factor in limiting the lifetime of the tank if the maximum gas
pressure is limited to 400 psi (see attached data in Figure 2)?
m a
+ 1
y N
UO2
200
0.32
150
3
10
UC
210
0.30
300
20
10
Steel
70
0.30
330
20
16
Question #3 (20%)
Consider a beam as shown in Figure 3, subjected to an axial force, F2 and a lateral force,
F1. Evaluate the value of F2 that will lead to a limit load condition for elastic behavior
of the beam when F1 = 8 MN.
Steel Data:
Youngs Modulus,
Poissons Ratio,
Yield Stress,
E = 191 GPa
= 0.28
Sy = 345 MPa
Contents
1 Introduction
1.1 Natural frequency of the harmonic oscillator . . . . . . . . . . . .
1.2 Forced excitation: earthquake . . . . . . . . . . . . . . . . . . . .
2 Decoupled dierential equations of a system
2.1 Lumped masses analysis method . . . . . . . . .
2.2 Modeling the pipe . . . . . . . . . . . . . . . . .
2.3 Dierential equation governing the beam motion
2.4 Natural frequencies of the undamped system . .
2.5 Uncoupled dierential equation . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
11
3 Earthquake data
11
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
15
16
16
17
18
20
21
. . . . . . . .
21
21
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
23
Abstract
The paper presents two methods to assess the cost eectiveness of a
steam generator pipe. The rst method is a time series analysis with real
earthquake data. The second method is a probabilistic assessment which
leads to the evaluation of a lifetime mean cost.
Introduction
1.1
d2 u
du
= F (t) k(u) d( )
dt2
dt
(1)
Here F(t) represents the force of the earthquake on the system. Actually
the earthquake is not really a force external to the system that is applied to
it. An earthquake is really an acceleration of the ground which was supporting
the structure. Thus equation (1) can be rewritten assuming that the ground is
d 2 u d2 u G
du duG
d2 uG
)
=
k(u
u
)
d(
m
G
dt2
dt2
dt
dt2
dt
(2)
Lets dene y(t) = u(t) uG (t). (2) becomes the simple equation :
d2 y
dy
d2 uG
+
d
+
ky
=
m
dt2
dt
dt2
k
Let us dene 0 = m
and 2Q0 =
can be written simply as :
d
m.
(3)
d2 y
dy
+ 2Q0
+ 0 2 y = 0
2
dt
dt
The characteristic equation is
(4)
r2 + 2Q0 r + 0 2 = 0
For our problem, well consider that 0 is xed whereas Q is the parameter
(actually Q is called the quality factor and it is directly related to the damping).
The delta of this equation is then = (2Q0 )2 40 . Thus = 0 when
Q = 1 this is called critical damping, > 0 if Q > 1 and < 0 if Q < 1. In
our case Q is always below 1, so < 0.
1.2
We shall now look at the forced oscillation of this system : an earthquake will
induce a ground acceleration term that will excite the oscillator at the frequency
.
To nd the response spectrum, one usually assume that the solution y(t) is
y(t) = Re(yejt)
Therefore, (4) with the forced term can be written as :
2 y + j2Q0 y + 02 y = 2 uG
(5)
(6)
Plotting this ratio with dierent values of damping on gure (2) suggests
that for undamped systems (Q < 1) the amplication is larger if Q is smaller.
More important, we can also plot the acceleration as a function of the accel
eration at the natural frequency :
Fdynamic
1
=
2 2
Fstatic
(1 ( 0 ) ) + 4Q2 ( 0 )2
3
(7)
Q
0.01
0.02
0.05
0.1
0.2
0.5
Fdynamic
Fstatic
50.0
25.0
10.0
5.0
2.6
1.2
2
2.1
The harmonic oscillator approach gives interesting results for the plant, but it is
somehow too simple to capture the complexity of the piping system in a nuclear
plant.
The approach broadly used in seismic analysis is the socalled lumped anal
ysis. The piping system of the nuclear plant is modeled as a succession of
harmonic oscillator with mass mi , coupled to one another by a spring and an
absorber. Of course we have mi = Mtotal . And the relation between the dier
ent oscillators is given by the beam theory. The strength of this approach is that
it enables the engineer to reduce the original continuous pipe to a succession of
linked masses which only have one degree of freedom (horizontal displacement
if the pipe is vertical). The fundamental principle of dynamic applied to the
mass mi yields
mi
d2 ui
dui
duG
= ki,i+1 (ui+1 ui ) ki1,i (ui1 ui ) ci (
)
2
dt
dt
dt
(8)
As we did before,
we
displacement. We
will dene
yi = ui uG the relative
y1
m1 0 . . . 0
y2
0 m2
0
also dene Y = . , M = .
..
.
..
..
..
.
y
0
0 . . . mn
m1
k1,1 k1,2 . . . k1,n
m2
k2,1 k2,2
k2,n
, m = . ,K = .
.. is the matrix dened as
..
..
..
.
.
mn
kn,1
kn,2
. . . kn,n
F1
F2
..
.
on each mass. So
Fn
F = KY . D is the socalled damping matrix and represents the ability of the
system to transform kinetic energy into heat and thus to dissipate some energy.
Equation (8) can then be written as :
M Y + DY + KY = muG
(9)
Equation (9) is not trivial to solve as is. But it can be reduced to a linearly
independent system of n equations by decomposing the homogeneous equation
in terms of harmonic solutions.
2.2
After this brief theoretical overview of the modeling method, we will in this
section eectively model a steam generator pipe. We will afterwards study the
eect of an earthquake on the pipe with dierent damping values.
Figure 5: Modeling the pipe: two lumped masses, xed at z=0 and z=L
The pipe will be considered to be xed between two walls. We will model
at a distance Ltotal
from
the pipe as two lumped masses, each of mass Mtotal
2
3
the wall. See gure (5).
Some usual gures for the pipes geometry and properties are summarized
below :
Pipe length: Ltotal = 3 m
Pipe outside diameter 21 mm, material thickness of 1.5 mm
Pipe composed of stainless steel grade 304: = 8000 kg/m3 , E = 193 GP a
Pipe lled and surrounded with water at 70 M P a (water = 750 kg/m3 )
Virtual mass coecient of 1.1
From these properties, we can directly compute some useful information on
w
w
the pipe. First its mass is Mtotal = Mpipe + Minside
+ Mmoved
. Where Mpipe =
2
2
w
2
w
2
Lwater .
(Rout Rin )Ltotal , Minside = Rin Lwater and Mmoved = Cvirtual Rout
w
w
This gives Mpipe = 2.545 kg, Minside = 0.779 kg, Mmoved = 1.120 kg. We end
up having a total mass Mtotal of 4.444 kg. Thus each lumped mass will have a
mass m = 2.222 kg.
7
2.3
locations.
y1
We introduce the matrix A dened as Y = AF where Y =
and F =
y2
F1
.
F2
Given the linearity of this equation, we can study the eect of F1 alone and
F2 alone.
For instance F1 will induce a displacement at z = L/3 and also at z = 2L/3
proportional to F1 .
To nd this relation, one has to solve the fundamental equation of the beam
dynamic motion. I is the moment of inertia of the pipe around the xaxis.
4 u Mtotal 2 u
f
+
=
z 4
EI t2
EI
4u
F1 (z = L/3)
=
(10)
4
z
EI
Given the geometry of the pipe, there is no displacement at the boundaries,
and there is no angular deviation as well. We have the following boundary con
ditions:
F1
y1
By solving equation (10) we get gure (6). So A
=
=
0
y2
F1
0.0988 EI
F1
0.0679 EI
Given the symmetry of the system we can predict that similar results will
be found for F2 . Thus putting all the results together we get A.
1
0.0988 0.0679
A=
0.0679 0.0988
EI
19.1806 13.1818
K = EI
(11)
13.1818 19.1806
2.4
(12)
To
the natural frequencies, we will solve equation (12) for Y =
obtain
y
1
. Equation (12) becomes
et
y2
2 M y + Ky = 0
0) shall verify:
In order to have a non trivial solution (i.e. y =
Det( 2 M + K) = 0
(13)
Y1 = 2
1
1
Y2 = 12
1
This two modes represents the two fundamentals oscillation modes of the
beam, and can be represented on gure (7)
10
2.5
Now that we have the two fundamental modes of the beam, we will rewrite our
initial problem in this coordinate system. To do so, we dene the modal matrix
associated with this base :
1 1
1
P = Y1 Y2 = 2
1 1
The new coordinate vector X is such that Y = P X. We also have a critical
damping matrix Dcr . So the fundamental equation M Y +Dcr Y +KY = muG
can be rewritten as
+ P t Dcr P X + P t KP X = P t muG
P tM P X
(14)
We nd that P t M P = M
a new critical damping matrix
. We dene
d
0
1
Dcr
= P t Dcr P as: Dcr
=
. Computing K = P t KP we get
0 d2
0.7802
0
3.1424
t
K = 10
. We also have m = P m =
0
4.2091
0
+ Dcr
MX
X + K X = m uG
x1
And if we take X =
, then we can write the new set of equation as
x2
the uncoupled following system of equations:
G
(15)
We now need
values, and then
critical damping.
lator we get dcr,1
263.3 kg/s.
Dcr
263.3
0
0
611.6
Earthquake data
The next step is to use our model and to compute the displacement when the
pipe is subject to an earthquake. To do this, we will use data obtained from
real earthquake (taken from [1]) and apply it to our model.
3.1
This earthquake occurred on October 16, 1999 in California. The data come
from the online database COSMOS Strong Motion Program. The seismograph
11
was located at 48.4 km from the epicenter (vertical projection of the focus), and
the earthquake was rated 7.1 on the Richter scale (which goes up to 9.0).
F (f )2
2
12
13
3.2
This earthquake occurred on January 19, 1982 in New Hampshire. It was rated
4.5 on the Richter scale, so it was much more modest than the precedent one,
but every cannot be sitting right on the San Anreas Fault zone! The accelero
graph we will use was recorded at Franklin Falls Dam, NH at 10.8 km from the
epicenter (note that the data were acquired much closer to the epicenter).
14
Now that we both have the uncoupled equation and some sample of ground
motion we can solve (15) with the data of the two dierent earthquakes and
study the evolution of the displacement of the lumped masses and the force on
the support caused by the ground motion as a function of the damping.
4.1
Methodology
15
4.2
4.2.1
First of all we note that the second equation of equation set (15) is an unexcited
harmonic oscillator. If we assume that the pipe was originally at rest (initial
displacement and velocity are zero) then x2 (t) = 0 all along the earthquake.
This means that y1 (t) = y2 (t). The only variable of interest is x1 (t).
A MatLab solve of this problem gives the following solution for x1 (t).
0.9555
Then Y = P X so Ymax = P Xmax . We nd Ymax =
103 cm.
0.9555
We
that F = KY , so Fmax = KYmax and therefore Fmax =
also recall
263.66
N.
263.66
Now given the symmetry of the problem, it is clear that the force on the
supports are the same. In addition we have just proved that F1 = F2 so by a
16
If we apply the same reasoning and calculations for dierent values of the damp
ing ratio we get the results summarized in table
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.20
0.50
1.00
ymax 103 in cm
0.9555
0.9032
0.8920
0.8872
0.8861
0.8847
0.8857
0.8926
0.8971
0.9022
0.9043
0.8177
0.7956
Fmax in N
263.66
249.24
246.13
244.82
244.53
244.12
244.40
246.30
247.55
248.94
249.53
225.65
219.53
Amplication ratio
Fmax
M ax(M uG )
17
3.000
2.284
1.904
1.616
1.437
1.247
1.161
1.070
1.016
0.980
0.740
0.536
0.432
4.3
18
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.20
0.50
1.00
ymax 103 in cm
1.2641
1.1089
1.0134
0.9909
0.9703
0.9179
0.9190
0.8908
0.8611
0.8368
0.6889
0.5876
0.5713
Fmax in N
348.81
305.99
279.64
273.43
267.74
253.28
253.58
245.80
237.62
230.91
190.08
162.14
157.63
Amplication ratio
Fmax
M ax(M uG )
1.680
1.475
1.264
1.196
1.104
1.100
0.994
0.960
0.927
0.971
0.874
0.524
0.340
19
4.4
The general idea here is to be able to determine the force that applies on the
support during an earthquake. From the results and the sketches showed before
one concludes that apart from a puzzling rise around 0.07 % for the JoshuaTree earthquake, the amplication of the ground motion is decreasing with the
damping ratio.
We also note that even though the Joshua Tree earthquake was much stronger
in terms of magnitude than the earthquake in New Hampshire, the eects are
relatively close. See gure 16 for a plot.
Figure 16: Maximum amplications on the support for the two earthquakes
We can make the following comments on the results:
The two earthquake have the same order of eect on the support even
though the Joshua Tree Earthquake is a 7.1 Richters scale earthquake
whereas the NHs one is only 4.5. But actually they were recorded at
dierent distance from the epicenter and on dierent types of rock. The
NH was recorded at 10.8 km, whereas the Joshua Tree at a distance of 48.4
km from the epicenter. If we consider that the seismic wave propagate
only at the surface, than the eect of the seismic wave are divided by
2
( 48.4
10.8 ) 20!
The NH has a much larger dependance in damping ratio than the Joshua
Tree. This can be explained by looking at the Fourier spectrum on gures
20
10 and 12 (One should not pay attention to the absolute values of the coef
cients which have not been normalized). We can see that the NH spectra
has most of its components around the rst natural frequency. Recalling
the simple HO on gure 4 we see that around the natural frequency the
amplication is very sensitive to the damping ratio. On the other hand,
the Joshua Tree spectra is mostly located well below the natural frequency,
where the amplication is less sensitive to the damping.
4.5
One can draw the following conclusion from the results obtained:
This kind of calculation enables the engineer to choose the right parameter
in order to avoid failure of the component over the life time of the plant.
Given the fact that we do not have data of the cost of damping, it is not
possible to go further and to assess the cost of the damping necessary to
meet the design criteria.
The choice of the appropriate damping ratio is really dependant on the
site where the plant is to sit.
This calculation does not introduce any probabilistic concepts, and is thus
conservative in the sense that a relatively larger than expected earthquake
has to be assumed in order to have a security margin
The following probabilistic model, still theoretical and hard to apply, intro
duces the concepts of probability of failure, cost of failure etc. which can allow
a more coherent and reliable costeectiveness assessment.
The lack of data, and also the lack of calculation power leads us to switch here
from a numerical case of a modeled pipe with real earthquake pipe to a more
theoretical approach where the purpose is to capture the cost of the deterioration
of the pipe as a function of the damping ratio, and to nd the best value of the
damping ratio that would come from a costbenet analysis.
This concept of costeectiveness design of a nuclear device is a new eld of
research. I am presenting here some results from some recent articles ([2] and
[3]).
5.1
The rst step is to get an idea of the probability at which the component that
we are studying (the pipe), will fail if an earthquake occurs.
21
Dening the component failure In our case the failure means breaking the
pipe. Most of the stresses are concentrated at the edge of the pipe (where the
junction is). At this place, the moment shall not exceed a critical moment Mlim
that is computed given the properties of the material. The condition of failure
is thus:
M (z = 0) orM (z = L) Mlim
(16)
It represents the content in power of the signal (i.e. the accelerograph) with
the frequency. From this PSD, one can easily determine the response spectra
by making a point by point calculation.
Probability of failure knowing an earthquake occurred The next step
is to determine, by using the appropriate PSD (and that is where I start to
lack some data), what is the probability that the ground motion will cause our
component to enter condition (16) (i.e. to fail). In [2], the authors assume
that uG is a zeromean stationary Gaussian process. SM () is dened as the
PSD of the bending moment in the support. It can be obtained by using the
transfer vector function of the moment hM () that transform the acceleration
of the ground into the moments on the supports (this function is easy to nd
using the dierential equation governing the motion of the beam). We nally
get SM () = hM ()2 Sg (). As long as the ground acceleration has zeromean,
than the moment is a Gaussian with zeromean, and thus its standard deviation
+
2
= SM ()d.
is M
What is of interest for us is to determine at which rate the moment will cross
Mlim . To obtain this value the authors are using the Rice Formula to compute
the cross rate. A classical result for a Gaussian process is the following:
M =
1 M
M2
exp( lim
2 )
2 M
2M
(17)
22
(M t)r
exp(M t)
(18)
r!
With this distribution, we verify that E(Xt ) = M t which is what we had
expected.
Now, we know that one failure is enough to fail the component, so the
probability of failure is 1 minus the probability of no failure over t i.e.
PfM (Xt =r) =
PfM
eq
= 1 eM t
(19)
E[C()] = C + Cf Pf () (1 exp(tlif e ))
(20)
Where tlif e is the total lifelength of the plant, is the annual discount rate,
C the marginal cost of and Cf the cost of failure.
Conclusions on the model This very simple model enables to get a simple
formula of the expected cost that can then be maximized in order to minimize
the expected cost of the pipe. Nevertheless, data such as PSD, and even C or
Cf are very hard to obtain and cannot be estimated by a backoftheenvelope
calculation.
Conclusion
23
References
[1] Database of accelerographs :
scripts/earthquakes.plx
COSMOS http://db.cosmoseq.org/
24
Abstract
Contact between the outer surface of a fuel pellet and the inner surface of the fuel rod
cladding can result in several kinds of undesirable behavior. As the pellet swells outward
and the cladding creeps down, large stresses are generated in both the pellet and the
cladding. This contact may exacerbate chemical degradation of the cladding and the
stress may enable rapid propagation of cracks. Because maintaining the cladding as a
barrier to fission gas release is necessary for reactor operation, it is important to have
accurate models of the pelletcladding mechanical interaction. This work analyzes the
predictions of a standard fuel rod modeling code utilizing a rigid pellet assumption, and
compares these results to a model allowing elastic deformation of both the pellet and the
cladding. These models are also used to compare two claddings with significantly
different mechanical behavior: a traditional Zircaloy metal cladding and a silicon carbide
ceramic cladding.
Introduction
As manufactured, a roughly axisymmetric gap exists between the outer radius of the
cylindrical UO2 fuel pellets and the inner surface of the cladding in current light water
reactor fuel rods. This gap, usually several tenths of a millimeter thick, forestalls
significant mechanical interaction between the fuel pellets and the cladding during
normal reactor operation. However, it is important to be able to analyze the onset and
consequences of pelletcladding mechanical interaction (PCMI), should it occur, as it
may severely challenge the integrity of the cladding.
The fuel pellet and cladding behavior is governed by time, the external reactor core
environment, and the feedback between the state of the pellet and the cladding.
Depending on the scale of system being considered, different models have been
developed to predict pellet and cladding behavior, and in particular the consequences of
PCMI. In the example studied here, the FRAPCON fuel rod modeling code is used to
predict the behavior of a UO2 fuelled fuel rod with either Zircaloy or silicon carbide
(SiC) cladding.
The FRAPCON code is primarily concerned with a conservative analysis of the fuel rod,
and therefore incorporates a simple model of pellet expansion and cladding response. In
this study, the effects of stress feedback on pellet behavior are explored, and the results
are compared to the FRAPCON predictions. The choice of cladding materials is
interesting as Zircaloy is a ductile metal and SiC is a brittle ceramic, therefore the
response to mechanical interaction with the cladding may be significantly different.
A particular point of interest is the change in calculated cladding strain when a more
elaborate pellet deformation model is considered. This relates directly to failure of the
cladding due to stress beyond the yield or fracture strength of the material. The state of
stress in the cladding is also an important input into more complex PCMI issues, such as
stress corrosion cracking of the cladding.
Causes of PCMI
The initial gap between the fuel pellets and the cladding is needed to allow loading of the
pellet stack into the cladding tube during fuel rod manufacturing. Typical radial clearance
between the oxide pellet and the cladding inner radius at this initial cold state is 200 to
400 microns. This gap space, connected to the plenum at the top of the fuel rod, is then
back filled with helium to an initial pressure of 1 to 3 MPa for pressurized water reactor
(PWR) fuel.
After insertion in the core and startup, both the pellet and cladding grow due to thermal
expansion. This addition of heat has significant implications for both the cladding and the
pellet. For a Zircaloy cladding with a typical service temperature of 600 K, creep
becomes an important consideration. The external coolant pressure is about 15 MPa,
therefore there is a significant stress driving cladding creepdown onto the pellet.
For the pellet, the initial heat up causes both thermal expansions leading to cracking as
well as densification due to additional sintering of the pellet. The cracking is a
consequence of the stresses induced in the pellet due to differential thermal expansion;
the temperature is highest in the center of the pellet and decreases in a logarithmic
manner towards the edges. Therefore, compared to some midradius cylinder of average
temperature, the central region will try to expand more and will be held in compression,
while the outer layers experience less thermal expansion and will be held in tension.
The thermal stress in the cylindrical pellet under elastic strain behavior can be related to
q, the linear heat rate, by,
t =
E q '
16k (1 )
(1)
where E is the elastic modulus of the fuel, is the thermal expansion coefficient, k is the
thermal conductivity, and is Poissons Ratio. This stress must be considered locally
within the pellet as it applies to hoop, radial, and axial stresses, but in general, the stress
intensity will be largest at the outer circumference. Typical beginning of life values for
the quantities in (1) are given in Table 1 based on formulas given in FRAPCON subcodes. Given these values and using the Tresca maximum shear strength theory of failure,
the predicted stress intensity at the pellet outer surface is 140 MPa, compared to a
fracture strength of 125 MPa predicted by the FRAPCON code. Therefore, pellet
cracking will occur almost immediately after reactor startup.
Table 1. Parameters for calculating pellet thermal stresses from FRAPCON.
9.9 m/mK
q
20 kW/m
k
5.4 W/mK
0.316
These fractures due to thermal expansion are predominantly radial cracks due to the
circumferential stress, and they will tend to propagate towards the center of the pellet.
These cracks also spread axially along the length of each pellet, eventually producing pieshaped pellet fragments. It can be shown through analysis of the stresses caused by this
cracking that the resultant pellet wedges will relocate radially outward. [1]
Simultaneously, due to the high temperatures at the center of the pellet, the pellet will
undergo additional sintering, and its density will approach the theoretical UO2 density
from an initial manufactured density of about 0.95TD. This densification, combined with
the high thermal stresses at the center of the pellet, will tend to create a central void.
These effects combine to relocate the pellet fragments outwards toward the cladding.
Early in life these processes may result in closing of the fuelcladding gap, however this
is not considered a cause of significant PCMI.
4
The contact pressure between the relocated pellet fragments and the cladding is low, and
in general as the cladding creeps down the pellet fragments will be compressed, closing
some of the initial cracking. Parts of these cracks may eventually heal as a result of the
high temperatures near the center of the pellet. PCMI will ultimately occur if cladding
creep down continues for a long period of time, however other mechanisms, such as
pellet expansion, may accelerate this process.
The main driver of pellet expansion after the startup period is fission product swelling.
Solid and gaseous fission products accumulate within the fuel grains as a function of
fluence. These initially constitute point defects in the crystal structure of the fuel, along
with irradiationinduced vacancies and interstitials. Due to thermal diffusion and
radiation assisted transport mechanisms, these defects tend to cluster. Large voids and
agglomerations of fission products will start to swell the pellet. In PWR fuel, this
swelling is typically manifest within 10 MWd/kg and continues at a rate roughly
proportional to burnup. Gaseous fission products may escape the fuel as the gas bubbles
grow and interconnect along the grain boundaries or due to thermal diffusion; however,
this release does not a have a significant impact on the rate of swelling. [2]
Another typical initiator of PCMI in PWRs is sudden increase in fuel rod temperature,
such as a power ramp during startup. The thermal expansion coefficient of Zircaloy is
less than that of the UO2 pellet, and expansion of accumulated solid and gaseous fission
products will add to the growth. Significant contact pressure may accumulate before the
cladding can creep outward to relieve the pressure. In addition, there is both an axial and
radial dependence to the temperature in a fuel pellet, even in one that is cracked, that will
cause special deformation. The central region will expand more than the edges, producing
an hourglass or wheatsheaf shape. The corners of the pellet then may impinge on the
cladding before the outside surface of the pellet, as shown in Figure 1. [3]
Figure 1. Mechanisms of stress concentration due to pellet radial cracking and hourglassing. [1]
cladding inner surface. Stress concentrations between the edges of these wedges are
favorable points for cladding crack initiation. [3]
A significant mechanism of cladding damage, which increases the problems of
mechanical stress discussed above, is chemical attack on the cladding inner surface. In
particular, the mechanism of stress corrosion cracking (SCC), which occurs due to a
combination of tensile stresses and chemical corrosion at the crack tip, is of significant
concern. A schematic of the SCC mechanism during PCMI is shown in Figure 2.
The fuelcladding gap is initially filled with inert helium gas, and although this becomes
contaminated with fission products such as krypton and xenon, it is not particularly
corrosive. However, once there is good contact between the pellet and the cladding, solid
fission products may migrate down the thermal gradient to the cladding. The most
corrosive of these is iodine, and it is a major contributor to SCC. [5] In addition, there is
oxygen released from the oxide pellet due to fission, and the cladding is particularly
susceptible to oxidation.
The formation of a zone of chemical bonding between the fuel pellet and the cladding,
facilitated by the mechanical interaction, tends to reduce ductility and increase local
stresses due to the dissimilar thermal expansion and swelling of the joined structures. The
corrosion and possibly irradiation assisted growth of cracks through the cladding,
combined with increased tensile stresses at the inner cladding surface and the transition
from compressive to tensile stress in the bulk cladding due to PCMI combine to increase
the probability of throughwall cladding failure.
mitigation of PCMI for boiling water reactors by adding a thin layer of zirconium on the
inside of the cladding. However, this is expensive, reduces the available fuel volume, and
increases the parasitic absorption of neutrons. Additionally, pellets can be manufactured
with chamfered edges to increase the distance between the pellet and the clad if the pellet
undergoes considerable hourglassing during operation. This also provides a larger contact
surface area if contact does occur. [4]
More desirable than these mitigation techniques is prevention of PCMI altogether. The
most obvious method of prevention is proper sizing of the initial pelletcladding gap.
Because of the low thermal conductance of the gap, and the fact that either the clad or
pellet must be thinned to achieve a larger gap size, it is important to make the gap wide
enough to prevent contact during the expected operating lifetime of the fuel rod. In this
sense, the major concerns are the relative thermal expansions of the pellet and cladding,
the creep down of the cladding during operation, and the irradiationinduced swelling of
the pellet. The initial helium fillgas pressure is also a relevant variable; higher gas
pressure reduces cladding creep down and increases gap conductance, however it also
significantly increases the internal pressure of the rod at end of life, therefore it is limited
by the expected fission gas release during operation.
To prevent contact due to pellet hourglassing, reactors operate under restrictions
governing the rate that the linear heat generation in the fuel can be increased. This allows
time for the relaxation of stresses induced by the thermal expansion of the pellet and
entrapped fission gases. [6]
together with a SiC matrix. In particular, a concept has been demonstrated that uses a SiC
duplex fuel rod clad to hold UO2 pellets in a standard PWR configuration. [8]
The duplex consists of a monolithic inner SiC tube surrounded by and bonded to a SiC
composite. It is difficult to model this duallayer cladding design in existing PWR fuel
rod modeling codes; however, it is possible to get a general idea of the performance by
approximating the construction as a single layer with the most limiting properties.
In addition to the mechanical stress and strain limitations, SCC mechanisms may need to
be considered in SiC materials; however, there is currently limited information available
on the susceptibility of SiC to SCC. SiC is vulnerable to oxidation under certain
conditions, especially if there is free silicon within the matrix, therefore it is important the
SiC materials be a stoichiometric as possible prior to irradiation. [9]
A conservative restriction for SiC is prevention of any PCMI during operation. This is
based on both limitations of the codes used in the analysis of SiC cladding and based on
experience with Zircaloy cladding. This analysis investigates this limitation in an attempt
to gauge the range of the true response of the fuel rod if PCMI does occur.
The fuel rod analysis code used in this work is FRAPCON, a steadystate fuel rod code
for light water reactors. It was developed for the U.S. Nuclear Regulatory Commission as
a conservative tool for modeling the behavior of fuel rods under realistic operating
conditions. FRAPCON tracks the thermomechanical behavior of a single fuel rod up to
high burnup. It models the fuel using a series of axisymmetric radial rings at specified
axial nodes, and uses a singlering model for the cladding. It also has detailed correlations
for calculating fission gas accumulation and release, fuel and cladding properties as a
function of time, temperature, and fluence, and corrosion due to hydriding and oxidation.
[2]
FRAPCON uses a rigid pellet model for PCMI calculations; after pelletcladding contact,
the pellet stresses and strains are calculated, then the cladding deformation is calculated.
However, there is no feedback to the pellet mechanical model concerning the cladding
behavior, unless it determines the pelletcladding gap is no longer closed. FRAPCON
also assumes that there is no axial slip of the pellet after the gap has closed, although in
general it does track the axial deformation of the pellet stack and the cladding.
As discussed above, pellet cracking, relocation, and chemical attack are all important
considerations when determining cladding integrity. FRAPCON does account for pellet
cracking and relocation by adjusting the thermal model based on assumption of uniform
soft closure of the gap. However, it does not account for the changes in circumferential
stresses based on this cracking. After relocation due to cracking has occurred, the pellet
swelling and densification continues, however special consideration is then given at the
first instance of hard pelletcladding contact. It is assumed that 50% of the relocation
strain is recovered before true PCMI can occur; this implicitly accounts for the partial reclosing of the radial cracks.
After the maximum amount of surface relocation strain has been recovered, the fuel rod
mechanical behavior will be dominated by the pellet swelling (assuming a constant or
decreasing linear heat generation rate and low fission gas release). The swelling will be
linear with burnup, and the cladding will expand outward to accommodate the pellets
radial strain. FRAPCON does not track the cladding for failure due to stresses or strains
beyond the yield or ultimate stresses, nor does it perform any fracture analysis.
In general, given the aversion to PCMI during normal operation, and the strength and
ductility of Zircaloy, the rigid pellet model may be an appropriate approximation.
However, for extended periods of PCMI, or for a ceramic SiC cladding with an elastic
modulus closer to that of the ceramic UO2 pellet, this model may fail to capture important
behavior, or misrepresent the available operational margin.
10
flux and temperature occurs just above the midplane of the fuel rod. The relation between
the various pressures and the cladding stresses in the r, , and z directions is,
rco
r rci + rco
= rco
r a
z co
rci
rci + rco
rci
rco rci
Pext
P
0
in
Fz
1
rco2 rci2
0
(2)
where rco and rci are the cladding outer and inner radii, respectively, Pext is the coolant
pressure external to the fuel rod, Pin is the pressure exerted on the inner cladding surface,
which includes gas and contact pressure, and Fz is the net axial force given by,
(3)
where Pg is the fuel rod internal gas pressure. The stressstrain relations for the cladding
are then given by,
1
r
= 1
E
r
1 z
(4)
z = (Pg + Ps )
(5)
r = Pin
(6)
where Ps is the pressure due to the pellet stack holddown spring. It is assumed that there
is no friction between the pellet and the cladding, and properties are taken to be
axisymmetric and uniform over the axial length under consideration. In addition,
relaxation of the cladding stress due to creep is not considered. These approximations
greatly simplify the physical situation, but should provide a bounding estimation of the
cladding behavior when compared to the results of the rigid pellet model.
Once the pelletcladding gap has closed, the FRAPCON code continues to calculate the
fuel strain outward due to irradiationinduced swelling without radial constraints. For this
model, that strain is treated as a virtual cladding strain, and the stress required to deform
the cladding is converted into the effective contact pressure between the pellet and the
cladding using,
11
c + c =
1
Ec
rco
r
Pext + ci Pin
rco rci
rco rci
rco
rci
c
Pext
Pin c
2
2
rco + rci
rco + rci
rco rci
(7)
where c is the virtual cladding strain. Then a relation for radial pellet strain,
rf =
1
( Pin f ( f (Pg + Ps )))
Ef
(8)
is substituted into (7). The FRAPCONpredicted swelling of the fuel pellet will therefore
result in an increase in the contact pressure between the pellet and the cladding. This
model calculates the new steady state geometry, based on radial strain of the pelletcladding interface and the resultant stresses in the pellet and cladding. The result will be a
smaller radial deformation in the cladding than predicted by the rigid pellet model;
however, the degree of difference may be small depending on the relative materials and
physical condition of the pellet and the cladding. Comparing the modeling assumptions
made in this model and those in the FRAPCON code, the true behavior of the fuel rod
likely lies in between.
A parameter of particular interest in this study is the stress intensity in the cladding. The
FRAPCON manual suggests use of the Von Mises Distortion Energy Theory to relate
stresses to yield and failure modes. According to this theory, failure occurs when (for a
cylindrical principal coordinate system),
2
( r )2 + ( z )2 + ( z r )2
2
(9)
(10)
where rfo is the pellet outer radius, which is equal to the cladding inner radius.
Furthermore, if it is determined that the gap remains closed (pellet swelling is the
predominate condition), then the deformation of the pellet outer surface equals that of the
cladding inner surface,
uci = u fo
(11)
and the relationship between cladding displacement and strain is given by,
12
r rco rci
= 1
r + r
z co ci
0
1
rco rci
1
rco + rci
0
0
uco
0 uci
1 z
(12)
The FRAPCON input case for this study is based on a 15x15 PWR with constant external
conditions (coolant and power). The input parameters for this case are given in Table 2.
This case is for UO2 fuel with Zircaloy cladding. The linear heat generation rate was
chosen to ensure pelletcladding contact during operation without excessive fission gas
release. Therefore, this case models PCMI initiated by prolonged operation, allowing
time for the cladding to creep down onto the pellet and for the pellet to swell outward
significantly due to fission product buildup.
Table 2. Parameters for FRAPCON input file.
Geometry and fill gas pressure is for cold, asmanufactured fuel rod.
Property
Value
LHGR
19.7 kW/m
Coolant Pressure
15.2 MPa
Inlet Temperature
561 K
Local Peaking Factor
1.12
Fuel OD
0.932 cm
Cladding ID
0.958 cm
Cladding OD
1.09 cm
Fill Gas Pressure
3.31 MPa
This input case was chosen to evaluate both the Zircaloy and the SiC cladding response
to PCMI both in order to ensure a fair comparison and because of limitations of the
FRAPCON numerical solver. Even with the aggressive linear heat generation rate chosen
for this study, maintaining the same cladding geometry and substituting the SiC cladding
leads to an open gap at end of life. This is primarily due to the lack of creep in the SiC.
This was compensated for by reducing the inner and outer radii of the SiC cladding until
hard fuelcladding contact occurred at the same time as in the Zircaloy cladding case
(0.939 cm ID and 1.071 cm OD). The average fuel temperature is about 10 K higher due
to the lower thermal conductivity of the SiC cladding, however it is not expected this will
alter the properties of the fuel significantly in this study.
Analytical Results
13
The results of the preliminary FRAPCON analysis (with Zircaloy cladding) show the
expected behavior. Figure 4 shows the fuel burnup proceeding linearly with fuel
residence time, and a small fission gas release less than 0.4% at discharge.
The initial heat up and relocation results in a slight closing of the gap, as shown in Figure
5, followed by fuel densification during the first 100 days (5 MWd/kg) of irradiation.
After 200 days, there is a generally linear rate of gap closure as the cladding creeps down
and the pellet swells outward. Between 700 and 1000 days there is soft contact between
the pellet and the cladding, and the fuel fragment relocation strain is being recovered.
Hard contact occurs soon thereafter, with the FRAPCON calculated contact pressure
increasing rapidly as the pellet pushes the cladding outward.
0.5
60
0.4
40
0.3
30
0.2
20
Burnup (MWd/kg)
50
0.1
10
0
0
200
400
600
800
1000
1200
1400
0
1600
Time (days)
Figure 4. Burnup and fission gas release as a function of time for the FRAPCON case.
80
30
70
60
20
50
40
15
30
10
20
25
5
10
0
0
200
400
600
800
1000
1200
1400
0
1600
Time (days)
Figure 5. Pelletcladding radial gap thickness and pelletcladding contact pressure as a function of
time for the FRAPCON case.
14
Comparing the results of the PCMI analysis for Zircaloy cladding between the
FRAPCON model and the new deformable pellet model, it is apparent there is not a
significant difference in deformation. As shown in Figure 6, there is only a 5% difference
in the final cladding hoop strain between the two models.
0.0045
0.004
Rigid Pellet
0.0035
Deformable
Hoop Strain
0.003
0.0025
0.002
0.0015
0.001
0.0005
0
1000
1100
1200
1300
1400
1500
1600
Time (days)
Figure 6. Zircaloy cladding hoop strain with and without a deformable pellet model.
There is a more significant difference, however, between the predicted cladding hoop
stresses, as shown in Figure 7. The FRAPCON rigid pellet model predicts that the stress
will approach a maximum near 70 MPa by discharge, while in the deformable pellet
model the stress increases to over 200 MPa at discharge. The maximum stress intensity in
the cladding with the deformable pellet model is 240 MPa, which is still below the
Zircaloy yield stress of 270 MPa calculated by FRAPCON.
15
250
Rigid Pellet
200
Deformable
150
100
50
0
50
100
1000
1100
1200
1300
1400
1500
1600
Time (days)
Figure 7. Zircaloy cladding hoop stress with and without a deformable pellet model.
Replacing the Zircaloy cladding with the average SiC layer, the primary difference in the
PCMI analysis is the increase in elastic modulus and decrease in the fracture strength. As
with the Zircaloy cladding, the difference in the cladding deformation predicted by the
two models is small. The difference in hoop strain between the models is about 8% at end
of life, as shown in Figure 8.
0.0045
0.004
Rigid Pellet
0.0035
Deformable
Hoop Strain
0.003
0.0025
0.002
0.0015
0.001
0.0005
0
1000
1100
1200
1300
1400
1500
1600
Time (days)
Figure 8. SiC cladding hoop strain with and without deformable pellet model.
Compared to the Zircaloy cases, the cladding stress accumulation for SiC cladding,
shown in Figure 9, is much more significant. The rigid pellet model predicts a rapid
increase in stress since there is no creep relief (up to 700 MPa at discharge), whereas the
deformable pellet model predicts a more modest but also linear increase up to 325 MPa at
discharge.
16
800
700
Rigid Pellet
600
Deformable
500
400
300
200
100
0
100
1000
1100
1200
1300
1400
1500
1600
Time (days)
Figure 9. SiC cladding hoop stress and without a deformable pellet model.
Analysis
The deformable pellet model consistently predicts less cladding deformation than the
rigid pellet model, as was expected. Due to the radial constraint, the fuel pellets
experience greater axial deformation and less radial expansion with time. Given the
model assumptions, the ability of the cladding to resist the pellet expansion is probably
overestimated; therefore, the true deformation state should lie between the two models.
As a consequence of the deformation of the pellet, the deformable pellet model also
generally predicts lower stresses in the cladding. Note that this is violated in the Zircaloy
analysis after creep becomes important (greater than 1200 days, shown in Figure 7). After
this point, the cladding creep rate is sufficient to maintain a nearly stable level of stress in
the cladding, however the deformable pellet model neglects creep and predicts an ever
increasing stress with imposed strain. For the Zircaloy cladding, even when neglecting
creep, this stress is not predicted to exceed the yield strength.
The fracture strain of SiC is around 0.15%. Both models are in general agreement that
this strain occurs around 1200 days. The fracture strength for the SiC composite is given
by,
3
(
1 x10 25 )20
(13)
where T is the cladding temperature in Kelvin and is the neutron fluence in n/m2. A
calculation of the Von Mises stress intensity shows that the hoop stress necessary to
17
surpass the fracture strength of the SiC is around 120 MPa. [8] The deformable pellet
model predicts exceeding this fracture strength around 1230 days, giving a failure strain
of 0.17%. This is a reasonable number and still in agreement with other studies of SiC
composites. FRAPCON predicts fracture around 1110 days and therefore a strain of
0.075%.
The FRAPCON rigid pellet model predicts cladding strain within reasonable agreement
of the deformable pellet model. The predictions of cladding stresses vary more
substantially depending on the material. For the Zircaloy cladding, the deformable pellet
model initially predicts lower stresses, as is realistically expected. When creep becomes
significant at high stresses this is taken into account in the rigid pellet model, and
therefore the limiting stresses in that case are more realistic. No failures are predicted in
either model due to excessive stress or strain, however the impact of the lower initial
stresses would reduce the predicted SCC and critical stress intensities for crack
propagation.
For the SiC cladding, the distinctions between the models are more straightforward. The
new deformable pellet model consistently predicts lower stresses in the cladding as a
function of time since creep is not an issue. Use of the FRAPCON model alone,
therefore, can provide reasonable prediction of failure when failure criteria are presented
in terms of limiting strains, but not of limiting stresses.
The deformable pellet model also gives insight into the aggravation of SCC phenomena
predicted the FRAPCON. Because the hoop stresses predicted by the deformable pellet
model are significantly lower than those calculated by FRAPCON after initial contact in
both claddings, stresses will be lower around the cracks. While the magnitude of this
effect is not calculated here, it is probable that the limiting parameter for the Zircaloy
cladding after PCMI will be SCC, since it is otherwise well below the yield strength.
Using stress information from FRAPCON would lead to overpredictions of the cladding
tensile stress, and therefore the rate of SCC damage. If SCC becomes an important
consideration for SiC cladding, this analysis indicates that the FRAPCON overestimation
of the cladding tensile stress would be even greater.
Conclusions
Pelletcladding contact is initiated due to either cladding creep down, prolonged fuel
pellet swelling, or sudden thermal expansion of the pellet due to power ramping.
Mechanical interactions are avoided in practice by providing sufficient initial margin
between the pellet and cladding surfaces, shaping the pellet to avoid protrusions
following deformation, and limiting reactor power ramp rates. A different choice of
cladding, such as SiC, reduces the parameter space by removing cladding creep as a
PCMI initiation mechanism, but also as a source of postcontact stress relief.
PCMI damage may occur due to stressing the cladding beyond the yield or fracture
strength at a point or over a region, producing bulges, ridges, or breaks in the cladding.
18
PCMI may also aggravate SCC by providing additional tensile stress to the cladding
inner surface and a convenient transport pathway for corrosive chemical species.
A complete treatment of PCMI, from onset to failure, requires threedimensional
treatment of the fuel rod and the thermomechanical, chemical, and nuclear phenomena
over large timescales. Simpler models, such as the rigid pellet model used in FRAPCON,
provide insight into the condition of the pellet and cladding during PCMI, but may be less
appropriate for different material choices.
An analysis of PCMI using the rigid pellet model, contrasted against a model of a simple
deformable pellet, highlights the limitations of each approach. The rigid pellet is
computationally simpler than a full implementation of the deformable pellet model, and it
provides reasonably precise estimates of the pellet and cladding strains in time for both a
ductile metal (Zircaloy) and a brittle ceramic (SiC) cladding. However, it likely
significantly overestimates the stresses induced in the cladding, namely the tensile hoop
stresses that may drive SCC failure mechanisms.
References
1. Helfer, T., et al., Modelling The Effect Of Oxide Fuel Fracturing On The Mechanical
Behaviour Of Fuel Rods, Pelletclad Interaction in Water Reactor Fuels, Seminar
Proceedings, AixenProvence, France, 911 March, 2004.
2. Berna, G. A., et. al., FRAPCON3: A Computer Code for the Calculation of SteadyState, ThermalMechanical Behavior of Oxide Fuel Rods for High Burnup, Vol. 2,
NUREG/CR6534, PNNL11513, prepared for the U.S. Nuclear Regulatory
Commission by Pacific Northwest Laboratory, Richland, Washington. 1997.
3. GuicheretRetel, V., Modelling 3D Mechanical Phenomena in a 1D Industrial Finite
Element Code: Results and Perspectives, Pelletclad Interaction in Water Reactor
Fuels, Seminar Proceedings, AixenProvence, France, 911 March, 2004.
4. Yagnik, S.K., Sunderland, D.J., and Cheng, B.C., Effect of PWR Restart Ramp Rate
on Pelletcladding Interactions, Pelletclad Interaction in Water Reactor Fuels,
Seminar Proceedings, AixenProvence, France, 911 March, 2004.
5. Le Boulch, D., Fournier, L., SainteCatherine, C., Testing and Modelling IodineInduced Stress Corrosion Cracking In StressRelieved Zircaloy4, Pelletclad
Interaction in Water Reactor Fuels, Seminar Proceedings, AixenProvence, France,
911 March, 2004.
6. Billaux, M. and Moon, H., PelletCladding Mechanical Interaction In Boiling Water
Reactors, Pelletclad Interaction in Water Reactor Fuels, Seminar Proceedings, AixenProvence, France, 911 March, 2004.
19
Other Resources
Garcia, Ph., Struzik, C., Agard, M., and Louche V., Monodimensional mechanical
modelling of fuel rods under normal and offnormal operating conditions, Nuclear
Engineering and Design, 216 (2002) 183201.
Suzuki, M., Uetsuka, H., and Saitou, H., Analysis of mechanical load on cladding
induced by fuel swelling during power ramp in high burnup rod by fuel performance
code FEMAXI6, Nuclear Engineering and Design, 229 (2004) 114.
20
December 7, 2006
Page 1
December 7, 2006
Abstract
The design of a nuclear power plant imposes to build an external barrier, nominally the
containment,madeofconcretereinforcedwithsteelbarswhichveryoftenareprestressed inorderto
augmenttheoverallcapabilitytoresist totensilestresses.Thestateoftheartofthesehugestructures
did not advance lastlybutthenewfeatures offuture plants(bigger thanthecurrent standardin size
and power or based on innovative technologies such as the Fast Gas Reactors with higher operating
pressures) are going to push the industry rethink the current standards in order to meet new
requirementsatleast incostterms.Thekeyvariablefor thecontainmentbuildingisthedesignofan
adequate margin which, from a structural point of view, corresponds to determine the strengthen
configuration of concrete and steel as a response to internal and external loads. In this paper we
analyzetheexistingandmorecommonconfigurations, astheywereadoptedby the USnuclearmarket,
fromtoSixtiestothesedays,and then,providesomecalculationsofthesafetymarginsforaclassical
largePWRdrycontainment.Theexampleprovided inthecalculationsisexecutedmainlybymeans of
thedataprovidedintheSafetyAnalysisReport,SAR,ofanexistingplant,theIndianPointUnit3,IP3.
The work is organized as follow: Sections 2 and 3 of this paper give an overview of the different
types of containment built in the US. In Section 4, the detailed model structure is provided and the
scenarios analyzed together with the results of the study are described. Finally, a discussion of the
obtainedresultsandsomefutureutilizationsofthemodelaregiven.
Inadditionto the calculationsperformedas anillustrativeexerciseforthe314classatMIT,thispaper
also analyzes the current regulatory framework and standards of the ASME and ACI codes and
emphasizes, in the conclusions, the excessive use of conservativism and the contradictions in the
adoptionofthesecodes.
This work has been prepared in partial fulfillment of the 314J Class.
Page 2
December 7, 2006
By
Edoardo Cavalieri d'Oro
Page 3
December 7, 2006
Index
1. Introduction: some Historical Notions about Containments
pp. 05
pp. 06
pp. 08
4. Calculations
4.1
4.2
pp. 08
pp. 09
pp. 10
pp. 11
pp. 13
pp. 15
pp. 16
pp. 16
pp. 16
pp. 19
pp. 25
pp. 26
6. References
pp. 28
Page 4
December 7, 2006
December 7, 2006
PWRcommondesignsarecategorizedaseither"largedry,""subatmospheric,"or"icecondenser."
ForBWRs,thecontainmentandmissileshieldfitclosetothereactorvessel.Thereactorbuilding
wallformsasecondarycontainmentduringrefuelingoperations.Thecontainmentdesignsarereferred
tobythenamesMarkI(oldest;drywell/torus),MarkII,andMarkIII(newest).Allthreetypes housea
largebodyofwaterusedtoquenchsteamreleasedfromthereactorsystemduringtransients.
DifferentpropertiesandfeaturesofthesecontainmenttypesarereportedindetailinTable1.
Page 6
December 7, 2006
Table 1: Main containment features and characteristics of the US fleet divided by design type (Pressure Water reactors and Boiling Water Reactors).
Characteristics of the US
Containments (109 units)
Mark I
Mark II
Mark III
Subatmospheric
Ice Condenser
LargeDry
Number of Units
24
57
Pressure Suppression
Yes
Yes
Yes
No
Yes
No
Volume, (10 *m )
12
15
48
5270
36 40
46100
1.7
1.3
1.3
0.528
0.48
0.20
0.41
0.30
0.420.52
0.4
0.4
0.16
0.34
Number of Barriers
3
1593 3293
3293 3323
2894 3833
24413411
3411
1500 3800
200.000 320.000
110.000 180.000
90.000 140.000
200.000310.000
140.000 190.000
340.000 500.000
1.440.000 1.800.000
250.000 280.000
1.165.000 1.550.000
1.800.000
1.200.000
2.600.000
15.86
25.63
49.43
54.93
36.62
79.34
7.533.64
4.514.56
12.5216.59
20.5221.30
11.14
26.3230.67
Containment strength
Containment design pressure (MPa)
Median containment failure press (MPa) in IPE
0.490.53
0.78 1.41
0.410.48
1.07 1.42
0.2
0.490.75
0.410.52
0.931.00
0.180.31
0.350.76
0.380.52
0.721.41
Containment construction
22 steel
2 concrete
1 steel
7 concrete
2 steel
2 concrete
7 concrete
7 steel
2 concrete
7 steel
50 concrete
Vertical vents
DW/WW vacuum
breakers
No
No
Containment spray(6)
and fan coolers
Containment
spray(6) fan coolers
Inerted by N2
Inerted by N2
Igniter System
Hydrogen igniters
Hydrogen
recombiner (for
designbasis)
0.5
0.5
0.4
0.1
0.25
0.1
0.91 (7)
1.07
0.52
1.03
0.45
0.92
(3)
Page 7
December 7, 2006
Page 8
December 7, 2006
December 7, 2006
building structure used to house, and to protect from designbasis hazards, nuclear safetyrelated
structures, mechanical and electrical components, and distribution systems associated with a reactor
coolant system; and (2) toserveaprimaryfunction asan engineeredsafeguardtocontainthepostulated
radiological consequencesofa lossofcoolant accident in the nuclearsteamsupplysystem.
In addition to the pure regulatory framework imposed by the NRC, we also found a historical
contraposition of the rules that of the ASME imposed to all boiling and vessel systems and the
specificationsgivenbytheACIincasethecontainmentisgoingtobebuiltwithconcrete.
In order to incorporate the earthquake ground acceleration load, which has a dynamic nature we
should use the model proposed by Wen [13] to combine them as showninfigure3(Table3offeran
ideaofthefinaloutputthatisusuallyprovidedbythesemodels).Becauseofallthepossiblesourcesof
uncertaintiesalreadymentioned, ASMEandACIcodesreducethem byfactorizingloadsintodifferent
well specified load factors and also as in Section III of the B&PV ASME Code factorize loads and take
intoaccountthestandarddeviationandmean valuesoftheconcreteproperties.Itisthusobviousthata
probabilisticframeworkwouldfitfineinthiscontextandlatersection willgofurtherexplainingit.
P a g e  10
December 7, 2006
CorrosionsoranyotherAgingMechanisms
ChemicalreactionswithCO2(carbonationoftheconcrete)
Fragilitycurveofthesteel(ASMEcodeforthetemperatureeffects)
Penetrationduetoimpactwithinternalorexternalmissiles
Waterreleasewithinthecontainment
The discussion provide her is actually given for completeness but actually none of these
mechanisms will be analyzed in the present analysis. But the hope is that, as emphasized in the
previoussection,thereaderwillbeawarethatnewregulations(bynewwe meanthelast15years or
so..)aretryingtoconsistentlyaddressmaterialdegradationmechanismsandthata completeanalysis,
asrecentpapersinthefieldareshowing,willincludetheeffectsofuncertainties astimechange.
Now weneed toconnecttheconceptof loadto the concept ofcapacity.Theloadistheevolutionof
pressure (which is varying in a more stringent time that is the time of the evolution of a possible
accident) Figure 4 provides a possible example of pressure load history during different scenarios of
evolutionof theaccidentandleadingtodifferent reliability endstates.Thecapacity(expectedfailure
pressure) is the demarcation zone for scenarios ending in a failure state. Note that from this figure
couldbebetterexplainedtroughthesetofobservationsthatfollow.
1
TheCapacityisgivenprimarilybytheexternalprestressedconcretebarriersstrength sointhiscaseitsdeduciblefromthemechanicalproperties
ofasinglecomponentandthisbecausetheinternalsteellinerplaysamarginalroleinsupportingthestructure.
P a g e  11
December 7, 2006
Theblacklineshowninfigure4represents thepressureevolutionduringanaccident.Theaccident
can follow different paths or scenarios. Until the peak, the accident can be thought as the DBA or
classicalLOCAwheretheenergyandamountofinventoryreversedfromprimarytroughcontainment
are essentially determining it. The peak pressure is used to determine the design basis accident
pressurebyincreasingitofafactorthatusuallyrangesfrom 10to25%2.Thebifurcationpointis,inthis
illustratorexample,thelastpointinwhichtoachieveornotthedesiredmitigationtroughi.e.theECCS
system. If we dont have the expected mitigation the pressure evolves following the path of a severe
scenario with meltdown where the pressure is going to pass the boundaries imposed by the design
pressureandthenultimatelyendsinafailurestateofthecontainment:thisistheregionwearegoing
to refer. The region between the red and the brown line constitute the safety margins in terms of
pressure for the reactor building. Passing the red line means going below the required margins for
safety, and,passingthebrownonemeans goingthroughafailuremode ofthecontainment.Thissecond
upperboundaryisalsoknownascapabilityofthesystemandwearegoingto refer toit later onin this
workwhen thephaseofconceptualizationofacalculationwillrequirethat.Finallynotethatthegraph
presentedbelowreferstopressureandnottostresses.Apropersafetymarginrepresentationwillbein
astresstime spaceandwouldtakeintoaccountnotjustpressurebut,asweanticipated intheprevious
Section, alsootherfactorspilingupwithitsuchastemperature,dynamicloads.
Figure 4: Pressure histories in the containment after a large LOCA leading to reactor meltdown.
2
Containment design pressures typically have been defined as 10%25% above maximum pressure. Initial design of containment typically used a 40% margin for
containment barriers and 20% for interior structures that did not serve a barrier function. These margins were reduced to half these values when all the geometry of
the interior compartments are known as part of the final design. An even lower margin can be used if rigorous modeling and dynamic load effects are considered.
P a g e  12
December 7, 2006
P ( PL > PC ) =
( f
( L)dL ) f C (C )dC
Figure 5: Capacity and load distributions within the containment and SM def.
Thequantity,safetymargin,SM,describesthereliabilityofa barrierorsysteminlightofloadstrength
orloadcapacityconsiderations.Itiscomputedfromthefollowingequation:
P a g e  13
December 7, 2006
SM =
CL
C2 + 2L
(1.1)
C ( f L ( L)dL ) fC (C )dC
+
P( L > C ) =
(1.2)
Under the strengthstress model (see Fig. 5), for failure mode i4, the reliability is defined as the
probability that the strength, C, is greater than the stress, L. The strength and stress are in general
sense. Any resistance, such as a yield strength, allowable force, or allowable deflection can be
considered as strength, and any loadingtype quantity, such as a bending stress, external force5, or
deflection,canbeconsideredasastress.
For normally distributed C, L, the probability of failure (i.e., 1reliability) can be expressed as a
function ofsafetymargin(Equation1.1)alone.Thisisonereasonwhysafetymarginemergedasthe
soleproxyforreliability inmany applications. Theotherreasonis thatthedesign goal(inthenuclear
industry,especiallyintherelatedfieldofthepressurevesselconstructionaspresentedhere)istobuild
components and systems that have negligible failure probabilities. This can be attained by having
sufficient safety margin (i.e., a large separation between mean capacity and load relative to their
combinedstandarddeviations).Thissolidifiedthegeneralizationthathavingadequatesafetymarginisa
sufficientconditionforhighreliability.Thus,ahighlyreliablesystem(i.e.,oneinwhichtheprobabilityof
failure is negligible) shows no overlap between the probability densities of capacity and load. On the
otherhand, refer tofigure5,a non reliablesystemshowsa configurationof thesystemthat isdefective
becauseitshowsacommonareaforthecapacityofthesystemandtheloadattheaccident.
With the definitions provided in the last two Sections, we can now define the Capacity and Load
condition in the case of the specific example of the Indian Point 3 unit for which data ( in terms of
distributions) are available. Beforedoingthat pleasenote onceagainthatthecapacityasreportedin
figures 4 and 5 is drown as a brown straight line but that in reality the strength of the concrete and
steel are subjected to several degradation mechanism (as reported in the list given in the previous
Section)notreportedhereandforwhichweassumetobeallnegligibleoverthelifetimeofthereactor.
3
The ratio in Eq. 1.1 means that for a given distance between C an L can be reduced if L or C present big uncertainties. This is a measure of the probability for the
two curves to overlap in presence of flat distribution curves and its aim is different from the (CL)/C ratio that we are going to use later which is typically showing
the variation of the margin with regard of the original C (mean) value. Note the two formulations can be both used and they are consistent with each other.
Or combinations of loads as it will be the case of the factored load that we have to deal with in the example provided in this paper.
P a g e  14
December 7, 2006
We conclude giving the assumptions we are going to use. Imagining not to consider labor costs,
specificcostsforconcreteare270 $/m3 [1]andforthesteelusedfor thebarsa priceestimateforsteel
equalto2.5$/kg(considerthatsteelhasadensityof7850kg/m3).
P a g e  15
December 7, 2006
December 7, 2006
Figure 6: Containment with spherical shell on the top of the PWR type as for the Indian Point 3 unit.
Maindimensionsare:
P a g e  17
December 7, 2006
Containment Capabilit
Containment capability is given from the probabilistic safety study of Indian Point Plant in the
appendix 4.4.1 [11]. It is an essential element of risk evaluation for degraded core events because it
defines the realistic lower bound ultimate containment building capability for withstanding internal
pressure loads. This had been determined to be 0.972 MPa (or 141 psia) for IP3 based on a failure
criteriondefinedasthestate ofhavingreachedgeneralyield inthereinforcedwalls.Atthatpressure
the strains start to significantly increase and radial deflections of the containment wall may become
unacceptablylarge.ThisvalueisconsistentwiththeaveragevalueforcapabilityweprovidedinTable
1,soitconfirmsthattheIP3isagoodrepresentativeexample oflargedrycontainmentbuilding.The
capabilitywasdefinedasthemaximumcombinationoftemperatureandpressuretoproduceageneral
yieldstate.
Load factors
LoadfactorsusedinIP3safetyreportwerecalculatedby followingtheACI31863:loadfactorsin
the designprimarilyprovideforasafety marginon the loadassumptions.Specificcombinationsusedin
thedesign arepresentedbelow.Thedesignincludedtheconsiderationofbothprimaryandsecondary
stresses. Theloadcapacity(thecapacityintheoldway todefineit) wasbasedontheultimatestrength
valuespresentedin partIV B ofACI318as reduced by acapacityfactor = 0.9 forflexureand0.85for
diagonal tension, bond and anchorage. For the liner steel it is equal to 0.95 for tension while for
compressionismaintainedbelow0.95yield.
The loads used to determine the required limiting capacity of any structural element on the
containmentstructuresare:
(1) L=D+/0.05D+1.5P+1.0(T+TL)
whereD=deadloadofthestructure,P=pressureassociatedtotheDBAaccidentasshownintransient
calculations and T = load due to the maximum temperature gradient trough the concrete shell upon
temperaturesassociatedwith1.5P whileTL=loadexertedbythelinerbasedon1.5P.
Three other loads were defined by inserting also other possible contributions such as
earthquakes, winds and tornado. We here report just the first condition which mean that the
containmenthasthecapacitytowithstandloadingsatleast50% greaterthanthosecalculatedforthe
LOCAalone6.
The loads we are going to consider are the DEAD LOAD, the ACCIDENTAL PRESSURE and
TEMPERATUREandthePRESTRESSINGLOAD.
6
Allthestructural components are designedtohave acapacity requiredbythemost severalloads combination.The loadsresulting fromtheuseofthese
equations willhereafterbe termedLoadFactors.
P a g e  18
December 7, 2006
Thepresentanalysisis conductedonlyonthemaincylindricalwallandsteellinerandupperdome
arenotconsideredsoanyanalysisofthedomeoverthecylinderisprovidedhere.
Aweldedsteelliner withaminimumthicknessofonequarter inchitisattachedtotheinsideface
of the concrete shell. The load carrying capacity of the liner is usually disregarded from calculations
becausethisisnotfunctionaltothescopeofthelinerwhichistoensureahighdegreeofleaktightness.
Note that the surface of the containment is imagined without penetrations, personnel locks and
equipmenthatches.
The containment wall is reinforced with hoop meridian and diagonal rebar. The typical rebar
arrangementwerefertoforthecylindricalwallisshowninFigure8.Thehoopandmeridionalrebars
are divided into two groups and each group is placed close to the wall surface. We consider that the
tendonsmoveinthelongitudinaldirectionthe sameastheconcrete.Sotherigidityiscalculatedonthe
basisthat"planesectionsremainplane"and thattransversePoisson'sratioof thebarscanbetreated
thesameastheconcrete.Thedetailsoftherebararrangementsforthecylindricalportionof the IP3
containment are shown byTable2[11]. Thesenseof thistableistoobservethatrebarischanging at
differentelevations(notethat apropercalculationwouldinspect thevariouselevationsby meansofa
P a g e  19
December 7, 2006
traditionalfiniteelementmashbutalsothesafetyreportdid notconducteditatthattime).Anaverage
value among all the elevations has been used to perform our calculations. Among the different
elevationsandpointinspaceweselectedtheregion nearthebasemat.Weconsideredthe cylinderto
bebuiltinatthepositionitjoinsthebasemat(zeroradialdeflection,zeroslopeofradialdeflection).
About prestressing conditions we assume there is no bending of the shell in the unpressurized
conditionandthatprestressing isgoingtobeachievedby posttensioning.
Finally we assumed elastic behavior and we define flexural rigidityon the basis of the section of
materialshowninFig.8.
December 7, 2006
Thesafetymarginsofthestructureasreportedinfigure6fromtheshelltheoryandthusreferred
tofigure7and8iscomputedasfollow:
1
1

25%
1
90%
85%
2
(1.3)
(1.4)
(1.5)
(1.6)
(1.7)
(1.8)
So,
according
toourassumptions,
theloadfactorLofEq.1.
4,or thedesignpressure,ismainly
definedfrom theaccidentsequencepressure Paandincreasedof themarginwepreviouslydescribed.
Here
t
he value i
s conservatively(
becauseherewedont knowall
thedetailsoftheload)settledequal
to1
.25.
Thed
esignpressurein
Eq.1.3isobtainedanalyzi
ng
t
heconcretewiththethinshelltheory.F
andDarethenobtainedbyassuming
thatthedeadl
oadismainly theweight of thestructureandits
uncerta
i
ntiesareovershadowed
bythelargevariabilityintroduced by
theothertwo,
soit has a minor
effectonthelimitstateprobabilities,
t
husitsvalueisgoingtob
edeterministicandequalto150lbs/ft3.
been
treated
as a Gaussian with mean
valueof47psi=0.32MPaandstddevof5.02Psi(0.0346MPa),accordingtothedataofref.[11]
ThecapacityCisincontrastdecreasedtotakeintoaccountpossibleimperfectionsofthematerial
strengthofthematerialasimposedtobeforaveragetensilestressesforboundedreinforcingsteel.
In addition to this, we have to avoid that the concrete faces creep deformations and
t
hus we
imposealimitontheprestressF,Eq.1.7,asgivenfromtheACI359.
Thusconstrictionimposedonthematerialtranslatesi
nto thetensionswithintheshellofreference:
MAX(Hs,preS =(1Xhs)/Xhs*Hc,max ,Ls,preS =(1Xls)/Xls*Lc,max)<0.85fc
(1.9)
P a g e  21
December 7, 2006
Equation 1.8 provides ultimately the safety margins in a deterministic formulation while the
probabilistic one,assumingthat both, LoadandCapacity,arenormallydistributedas was given inEq.
1.1, is at the moment not considered. Because the system of the equation identified revealed to be
linear,theformulationofEq1.2canbeusedandthusthemarginswefoundaregoingtobeexpressed
intermsoftheirrelativeinitialcapacity(seealsofootnote3atpage13):
SM % = (CL)/C
(1.10)
Note there are a set of different criteria we can establish and use to set up our limits or boundaries to the system which are related to the different
performancesdesiredfromthestructure (herewereferredtostrengthandthustomeasuresofperformancebasedontheyieldingstressbecauseamere
structuralcriteria wasadopted,but,thereare other performancesweshouldbeinterestedto,suchasserviceability,stability,plasticityandvibrations).
P a g e  22
December 7, 2006
Afinalclarificationaboutthewaywedeterminedcosts; assumingthereisnochangeoftechnology
orconstructionmethodscorrespondingtothicknessvariationsoftheconcrete,and,oftheareaofthe
steelbarsusedtopretension,costsaresimplycalculable bymultiplyingthecorrespondingvolumesof
materialinusefortheircostsforkg:
(1.11)
(1.12
)
Results provided by Table 5 gives an overall cost that correspond to 14.7 % of the initial
investmentcostwehypothesizedtobeforaPWRofthistype(1000MWe)around2billion$.
0.00%
4.73%
9.03%
12.95%
P a g e  23
December 7, 2006
0.9
0.8
0.7
0.6
0.5
0.4
P a g e  24
December 7, 2006
P a g e  25
December 7, 2006
Table 2: Spacing at different elevations and divided by layers for the IP3 PWR Cont [11]
elevation
hoop
meridional
diagonal
(m)
(dmnl,dmnl,cm)
primary
secondary
7.6
2 18 35.5
18
30.5
18
30.5
18
76.2
13.8
2 18 35.5
11
30.5
11
30.5
18
76.2
15.3
2 18 35.5
18
30.5
11 91.5
18
76.2
16.4
2 18 35.5
11
30.5
11
91.5
18
76.2
33.7
2 18 35.5
18
30.5
18
30.5
18
76.2
45.1
2 18 35.5
11
30.5
11
30.5
18
152.4
Table 3:
[11]
Load combo
Occurrence rate
[combined occurrences/year]
Cumulative occur.
[# occurrences]
mean duration
[sec]
D+P
D+E
D+P+E
2.16E03
1.64E02
1.36E09
8.64E02
6.56E01
5.46E08
1200
15
14.81
Table 4:
The Indian Point 3 Unit. Features of the Containment and of the Plant [11]
LOADS
D=dead load=150 lbs/ft3= 2226.586 kg/m3
P=accidental pressure=(m,s)=N(0.288, 0.034)
E=earthquake loads
T=accidental temperature corresponding to P
GEOMETRY
R+L=H m
R= 20.574 m
MATERIAL
Concrete:
Ec=21.373 GPa
H= 66.7512 m
L= 46.1772 m
vc=0.2
fc=fc(m,s)=N(33.75, 4.32) [MPa]
Ss= 1.0668 m
Sc=1.3716 m
fc,min=20.68 MPa
Reinforced bars:
Es= 200 GPa
vs=0.2
fy=fy(m,s)=lognormal(495, 35.71)
REACTOR FEATURES
Reactor type: PWR
24 MI N of New York City, NY
Docket Number: 05000286
Operating License: Issued 04/05/1976, Expires
12/15/2015
Operator: Entergy Nuclear Operations, Inc.
Electrical Output: 979 MWe
Reactor Vendor/Type: Westinghouse FourLoop
Containment Type: Dry, Ambient Pressure
P a g e  26
December 7, 2006
Table 5: Final Results of the IP3 Calculation for Safety Margins of a Concrete Containment
GEOMETRY INPUT
STRUCTURAL CALCULATIONS
t=Sc
1.371
Lc,max
6.69286E05
20.6
Wp=W,max
0.004671207
0.057
de
0.165
hc,max
0.000226758
PL
0.165
bL(z)
PH
0.2
bL,max(z=0)
0.6855
Lc,max
2431003.13
Pa
Xs
0.5205
Hc,max
5277691.233
Pa
46.1772
Ls,preS
105323971.4
Pa
R+L=H
66.7512
Hs,preS
136501287.9
Pa
Ss
1.0668
nominal/
distribution
LOADS AN CAP
mean
[MPa]
1.25
dmnl
MATERIAL INPUT
Es
2E+11
Pa
Pa*
142.1888416
MPa
Ec
2.137E+10
Pa
21.8205428
MPa
Pi
450000
Pa
136.5012879
MPa
vc=vs
0.2
dmnl
127.9699574
MPa
fc=fc(m,s)
33750000
Pa
L=tot
428.4806298
MPa
fy=fy(m,s)
495000000
Pa
0.9
dmnl
Pa
C=*fy
445.5
MPa
SM
8.509685116
MPa
fuc
OTHER QUANTITIES
nominal/
distribution
275000
Pa
VOLUMES
mean
[ m^3]
As
0.0025518
m2
Volume (free)
61406.58409
m^3
Xls
0.0225605
dmnl
Vcil
8456.642372
m^3
6.644E+09
Pa*m3
Vs/bar
0.117833069
m^3
Xhs
0.0372248
dmnl
# bars
484
dmnl
NL
3605000
Pa*m
Vs=Vs/bar*# bars
114.06
m^3
EL
2.176E+10
Pa
Vc=VcilVs
8342.579961
m^3
mean
value
Eh
2.143E+10
Pa
COSTS
E'
2.627E+10
Pa
268.5646061
$/m^3
1.875E+09
Pa
2.5
$/kg
512503.67
s (steel density)
7850
kg/m^3
0.241911
Unit Costs
0.000318471
$/m^3
Tot Costc=Vc*Ucc
2.240521701
Million $
Tot Costs=Vs*Ucs
285.1560264
Million $
287.3965481
Million $
OPTIMIZATION ON COSTS
max F, max P, max LS
given f'c, fy
Total invest costs
Cost fraction
2 Billion $
14.37%
P a g e  27
December 7, 2006
6. References
[1] ABWRCost/Schedule/COLProjectatTVA'sBellefonteSite,August 2005TennesseeValleyAuthorityNew
Nuclear Power Plant Licensing Demonstration Project, DEAI0704ID14620. Prepared by: Toshiba
CorporationGECompanyUSECBechtel PowerCorporationGlobal NuclearFuelUS,August2005.
[2] Ellingwood, Bruce et al. R.:Nuclear Power Plant Containment Pressure Boundary Aging Research (D8
A1US) ORNL/CPl02338. Department of Civil Engineering at the Johns Hopkins University Baltimore,
MD212182682(USA),May1999.
[3] NUREG0800 (Formerly NUREG75/087): U.S. Nuclear Regulatory Commission Standard Review Plan
OfficeofNuclearReactorRegulation.Section3.8.1:Concrete Containment,draftreview 2,April1996.
[4] NUREG0800: U.S. Nuclear Regulatory Commission Standard Review Plan Office Of Nuclear Reactor
Regulation.Section3.8.2:SteelContainment,draftreview2,April1996.
[5] NUREG0800: U.S. Nuclear Regulatory Commission Standard Review Plan Office of Nuclear Reactor
Regulation.Section3.8.3:Concreteand SteelInternalStructuresofSteelor ConcreteContainments,draft
review2,April1996.
[6] NRC Information Notice 200409: Corrosion of Steel Containment and Containment Liner. Office of
NuclearReactorRegulation,Washington,D.C.20555000April27,2004.
[7] NUREG1150,Volume 1: ReactorRiskReferenceDocument.MITNuclearEng.Dept.InternalCopy,1996.
[8] D.C. Williams, J.J. Gregory, S.R. Tieszen, J.L. Tillis:Containment Loads Scoping Calculations for Severe
Nuclear Reactor Accidents. Sandia Report SAND922774 *UC940 NPRWSA905 Unlimited Release
PrintedFebruary2003.
[9] Shinozuka, M.H. Hwang and M. Reich Reliability assessment of reinforced concrete containment
structuresNucleEng& Des.80(1984)pp.247267.
[10] Kawakami,HHwangM.T.ChangandMReichReliabilityassessment ofIndianpointunit3containment
structuresBNLNUREG/CR3641January 1984.
[11] PowerAuthorityofthestateofNewYork:IndianPointNPPunit3: finalsafetyanalysis report.Docket
250286,1994.
[12] Mirela Gavrilas: A framework for Integrating Risk and Safety Margins, RES/DSARP/NRCA, August
2006.
[13] Wen.Y.K.Methodsforreliabilityin structuresundermultiple timevaryingloads.Nucl.Eng&Des.60.
[14] John D. Stevenson, Companion Guide Book. A guide to the ASME CODE: Code for Concrete Reactor
vesselsandContainments,EditedbyK.R.Raoandpublishedin2006.
[15] Levin,HowardAlan.Prestressedconcretecontainmentsfornuclearpowerplants.InstituteArchives
NoncirculatingCollection3 M.SThesis,1976.
[16] NicholasJ.andCarino:StructuresDivisionPredictionof CrackinginReinforcedConcreteStructures,
NISTIreport#5634. Gaithersburg,MD20899 April1995.
[17] Crystall Ball, a risk analysis, simulation and optimization Software distributed from Decision
Engineering.Ver.7.2,August2006.
P a g e  28
Abstract
Introduction
One of the difficulties facing nuclear fuel designers today is ensuring that the
design stays below the melting point of the fuel or 2840C in the case of UO2. Recent
work has shown that a significant thermal operating benefit can be obtained from the
2400
2200
Assumptions:
Hot spot linear powers
Same core peaking of 2.5
Same core power for
45kW/m and 74kW/m cases
2000
Temperature ( oC)
1800
1600
1400
1200
1000
P=150%
800
600
400
200
0
Radius (mm)
Figure 1: Solid/Annular Radial Fuel Temperature Profile (Kazimi)
These annular pellets obtain this additional thermal margin due to the unique internally
and externally cooled geometry of the individual fuel pins shown in Figure 2 below.
Coolant
Fuel
Cladding
Gap
Besides the increased surface area for heat transfer, the annular pellets also benefit from
not having fuel in the center of the pin which effectively spreads out the heat source. Due
to fabricability concerns, additional schemes such as Duplex pellets have also been
proposed in order to decrease the centerline temperature. Duplex pellets differ from
annular fuel in that the inner hole of the annular pellet is filled with a solid material such
as ZrO2 or Zircaloy4 as shown in Figure 3.
Like nearly all ideas within the nuclear field, the duplex pellet concept is not new.
Duplex pellets were originally proposed during the late 1970s light water reactor breeder
program in hopes of increasing the breeding potential of the fissile material. However, its
use has started to recently resurrect due to increased interest in proliferation resistance
(Shwageraus) and waste management (Bays). Shwageraus utilized this geometry as one
possible avenue in order to help develop a thoriumbased fuel cycle for light water
reactors which reduces the plutonium generation rate and enhance the proliferation
resistance of the spent fuel. However, due to the significantly large thermal resistance
path from the UO2 in the center to the ThO2 on the periphery, it was found that the design
led to higher than acceptable centerline temperatures. Bays, on the other hand, took
advantage of this characteristic by putting spent fuel in a zirconium metal matrix on the
periphery with thoria/zirconium CERMET in the center. This design allowed for a deeper
burn of the transuranic waste because of the higher burnup rim effect on the periphery
from the high thermal flux coming in from the moderator. Additionally, since the
CERMET shares a common metallic phase with the outer pellet, these two can be co
extruded together with the zircaloy cladding in a heated die extrusion process.
However, in order to further develop this duplex fuel pin design, there are several
areas, including fission gas release, which need to be better characterized. Therefore, this
paper seeks to ascertain the relative difference in fission gas release between a standard
solid UO2 reference design and a duplex design with Zircaloy and UO2. Although the
central area in the duplex doesnt produce any fission gas, it is not well known if the
higher temperatures in the outer annulus (when compared to the equivalent area in the
solid design) will cause overall more or less fission gas release.
Method
For purposes of this study, an equivalently sized solid UO2 reference pellet will be
compared with a UO2/Zircaloy duplex pellet. In order to make a constant comparison, the
linear heat generation rate for the two designs will be matched. This necessitates that the
UO2 in the outer ring of the duplex pellet will have to be enriched above that of the 5
weight % UO2 in the solid pellet. The particular calculations showing this determination
follow in the next section.
The next step will be to determine the temperature profiles for the two respective
designs. For simplicity in the calculation, the thermal conductivity of the UO2 will be
assumed constant. This will allow the establishment of average temperature values for
both the inner and outer zones. These temperature averaged zones will then be fed into a
fission gas release correlation based on Booth type diffusion with grain boundary gas
accumulation and resolution (Weisman).
In order to perform the fission gas estimation, this correlation also requires
several other input parameters including, discharge burnup of the fuel, total incore
residence time and plutonium content. A CASMO simulation of both the solid reference
fuel case as well as the duplex fuel case allowed these parameters to be determined. The
pellet design parameters used to construct the CASMO input deck are shown below in
Table 1.
Dco (mm)
Dci (mm)
Dfo (mm)
Dfi (mm)
Solid Pellet
9.52
8.37
8.25

Duplex Pellet
9.52
8.37
8.25
2.74
The solid reference fuel was ran with 5 weight percent enrichment however in
order to fairly compare the duplex fuel, the total number of U235 atoms needs to be
equivalent on a per unit length basis. Therefore the following calculation was performed
to determine the enrichment of the duplex fuel.
10.4gU 1molUO2
gU
1molU
238gU
*
*
*
= 9.1674 3
3
cm
270gUO2 1molUO2 1molU
cm
gU
gU 235
9.1674 3 * 0.05 = 0.45837
cm
cm 3
The area ratio between the duplex and solid fuel is given by
AR =
(4.125mm) 2
((4.125mm) 2 (1.370mm) 2 )
= 1.10834
Thus multiplying the U235 atom density by the area ratio gives
gU 235
gU 235
0.45837
*1.10834 = 0.50803
cm 3
cm 3
Dividing this by the total original U atom density gives the new enrichment
gU 235
cm 3 = 5.54%
gU 235
9.1674
cm 3
0.50803
duplex = (
Figure 4 shows a plot of the Eigenvalue versus burnup of the two designs. The
simulation assumed a 3 batch refueling scheme with 3% leakage. The calculated
discharge burnups are indicated on the graph by the arrows from the xaxis as well as in
Table 2.
Keff versus Incore Life for Solid and Duplex Fuel Pins
1.5
1.4
1.3
1.2
Keff
Solid Pin
Duplex Pin
1.1
0.9
0.8
0
10
20
30
40
50
56
60
63
70
80
Burnup (MWd/kg)
Figure 4: Keff versus InCore Life for Solid and Duplex Fuel Pins
Discharge BU (MWd/kg)
Discharge BU (MWs/kg)
Specific Power (W/g)
EFPD
EFPs
Solid Fuel
Duplex Fuel
56.25
63.75
4860000
5508000
0.0343346
0.0395727
1638.289073 1610.95907
141548175.9 139186863.7
Calculations
First the linear power ratings will be matched between the two designs.
'
'
qsolid
= qduplex
'
'''
qsolid
= (R 2fo R 2fi )qduplex
'''
qduplex
= 1.1228*106
W
m3
Then the volumetric heat generation rate for the solid fuel is
'''
=
qsolid
'
solid
2
fo
kW
W
m
=
= 1.03855*106 3
2
m
(0.0040456m)
53.4
In order to determine the temperature distribution in the fuel pins we start with the
general heat conduction equation as below
V * q '' + q ''' = 0
which at steady state and assuming we are only considering one dimension in the radial
direction, the heat conduction equation reduces to
1 d dT '''
kr
+q =0
r dr dr
k
dT
r C
+ q ''' + 1 = 0
dr
2
r
Now we write the general heat flux condition knowing that for the duplex pellet
no heat flux exists at Rfi and for the solid pellet Rfi=0
q
which gives
''
r=R fi
dT
= k
dr
= 0 C1 =
r=R fi
q ''' R 2fi
2
T max
kdT =
r
q ''' 2
r
R
2fi +
C1 ln
fi
Since Rfi=C1=0 for the solid fuel, the above expression simplifies to
T
max
q ''' r 2
kdT =
Tfo
kdT =
q'
4
W
3
(Tmax
mC
3W
53.4 *10
400
C ) =
T
max =
1816.48
C
4
However, for the duplex pellet we substitute in the expression for C1 and solve
T max
T max
kdT =
q ''' R 2fi r
q ''' 2
(
r
R
2fi )
ln
4
2
fi
2
2
2
q ''' r 2
R
fi
R
fi
r
kdT =
ln
r
r
R
fi
q'
R 2
1
fi
fo
R
fo
ln
'
R
fi
T max
q
Tfo kdT =
4 1
R 2
fo 1
fi
we find
4.125mm
3 W
ln
53.4 *10
m
1.37
mm
W
3
(Tmax
400
C )
=
2
1
T
max =
1603.11
C
4
mC
4.125mm
1.37mm
1
Figure 5 below shows the plotted fuel pin temperature profile and the breakdown
between Zone 1 and 2. The temperature averaged values for Zones 1 and 2 are shown in
Table 3.
Fuel Pin Temperature Profiles
1900
Zone 1
1800
Zone 2
1700
1600
y = 440.7x + 2206.9
1500
1400
y = 345.48x + 1816.5
Temp (C)
1300
Solid Pin
Duplex Pin
1200
1100
1000
900
800
700
600
500
400
0
0.5
1.5
2.5
3.5
4.5
This zone averaged temperature profile for both the solid and duplex fuel is
depicted below in Figure 6.
Now that the relevant data has been determined, we can start to calculate the
fission gas release. Provided b=burnup in MWs/kg and t=cycle length in s, we calculate
the high burnup factor as
1.25*1014
bunkel =
b 3
86.4
ak1 =
9 *10
e ftemp + 0.0005
ak2
=
0.00005*
ak1*
1
ftemp1900
40 +1 (bunkel )
+
e
(1 e ak 2*t )
f = (1 (1 ak1)) *
ak 2 * t
comp comp
* 1
*10.96
100
100
fp = 1 0.001*
p=
fdens
roth
1
1
105
1 + 1 *
3
(
)
fp
fgrn
1
(1 + 0.47047 * x )
+ 0.0958798*
1 + 0.47047 * x
0.7478556 *
1
+
0.47047
*
x
+ 0.000025
Finally by defining cc1 as the output concentration of grain trapped fission gas,
gbgout as the output concentration of grain boundary trapped fission gas and fmgrot as
cumulative fission gas released we can calculate
Results
The calculations showed that the outer fueled annulus did indeed exhibit a
significantly higher fission gas release that the equivalent region for the reference solid
pellet. Since the fuel is being compared on a per unit length basis, the fission gas release
results are organized into area weighted ratios as shown below in Table 4.
Solid Pellet
Weighted Fission Gas Release Zone 1
Weighted Fission Gas Release Zone 2
Duplex Pellet
Weighted Fission Gas Release Zone 1
Weighted Fission Gas Release Zone 2
3.23%
21.06%
0.00%
25.37%
Thus the final fission gas release ratios for the solid reference and duplex fuel can
be easily calculated and are given below in Table 5.
As shown above, the 20.4% higher fission gas release in the duplex fuel designs outer
annulus more than compensates for the larger fission gas release which occurs in Zone 1
of the solid reference fuel. Overall, the 4.4% higher fission gas release for the duplex fuel
is not significantly larger than that of the solid reference. However, this result is
surprising in that it goes against the conventional wisdom of a lower centerline
temperature always yielding a lower cumulative fission gas release.
Conclusions
Since this was a scoping study, the treatment of several variables was fairly
rough. Aside from treating the thermal conductivity as non constant, the analysis could
also be further refined by considering a larger number of temperature averaged cylinder
zones. The 4.4% difference in fission gas release determined in this scoping study is
likely to be overshadowed by the uncertainties in the analysis. Thus it would be safe to
conclude that while the duplex design does have lower average temperatures, the fission
gas release is comparable to that of the solid fuel design.
References
Kazimi, M., Hejzlar, P., High Performance Fuel Design for Next Generation PWRs:
Final Report, MIT Center for Advanced Nuclear Energy Systems, MITNFCPR082,
January 2006.
Shwageraus, E., et. al, Microheterogenous ThoriaUrania Fuels for Pressurized Water
Reactors, Nuclear Technology, Volume 147, July 2004.
Bays, S., Evaluation of Recent Transmutation Scenarios for Partitioning/Transmutation
of Actinides: A Once Recycle Commercial Waste Disposition Fuel Cycle Utilizing
Thorium, Idaho National Laboratory, p 7256.
Weisman, J., MacDonald, P., Fission Gas Release Correlation, ANS Transactions,
Volume 12, Number 2, November 1969.
Pacific Northwest National Laboratory, FRAPCON3: A Computer Code for the
Calculation of SteadyState, ThermalMechanical Behavior of Oxide Fuel Rods for High
Burnup, NUREG/CR6543, Volume 2, PNNL11513, December 1997.
Abstract
Because of the irradiation coming from the core, the beltline region of LWR vessels
experiences a slow degradation of its mechanical properties, which is known as Reactor
Pressure Vessel (RPV) embrittlement. An accurate prediction of the severity of this
phenomenon, at each time during the operating life of the RPV, is fundamental in order
to prevent the vessel from experiencing stress states that may lead to a nonductile
propagation of a critical crack that may be present in the vessel walls, thus yielding vessel
failure.
Most of the literature, as well as current NRC regulations, relate the severity of
embrittlement to fast neutron fluence only, neglecting the irradiation time rate as well as
the effect of gammarays. In some circumstances, however, this procedure leads to an
underestimation of the severity of embrittlement. Irradiation rate plays an important role
when RPV walls experience a low fast neutron flux, like in BWRs. Gammarays give a
significant contribution to embrittlement when a thick water gap separates the core from
RPV walls, like in the ABWR. The current design tendency of decreasing fast neutron
flux at the RPV wall, coupled with the introduction of new and innovative reactor
features, is making the mentioned circumstances more and more frequent, requiring a
revision of the current fastfluencebased regulations.
the limitations of the current regulations and their proven inadequacy in some
circumstances.
The paper has a practical nature, in that it was written with the purpose of highlighting
readytouse information and rules of thumb, whose importance is often underrated.
1. Introduction
With respect to the vast majority of the mechanical components used throughout the
industry, those aimed at invessel nuclear applications must be provided with an
additional feature consisting of a satisfactory behavior under the effect of nuclear
irradiation. This satisfactory behavior essentially consists of preserving, at some extent,
the preirradiation characteristics, such as mechanical properties and geometry. The
component representing the boundary between the high and the low radiation regions, i.e.
the RPV, does not constitute an exception and it must satisfy this requirement too. In
spite of absorbing a lower dose per unit time than most of the internals, the RPV is a nonreplaceable component which is designed for a 4060 year life. The cumulative damage
experienced in this long time interval causes a slow modification of the mechanical
properties of the RPV in the beltline region, which mainly materializes in the form of
toughness reduction (embrittlement) and increase of the Ductile Brittle Transition
Temperature (DBTT).
Based on Figure 6, Mansur and Farrell ([1]) state that thermal neutron induced
displacements become significant (10%) with respect to those from fast neutrons when:
nt
10
nf
(1)
where nt and nf are the thermal and fast neutron flux respectively. This situation is
however quite unlikely since, even in the core of LWRs, the thermal flux is lower or
comparable to the fast flux.
Neutrons cause vessel embrittlement also by inducing the 10B(n, )7Li reaction, which
has a larger cross section for thermal neutrons than for fast neutrons:
n + 10B 4He + 7Li
Since boron is often present in RPV steels as impurity, neutron irradiation can yield
helium production. Then, the bubbles so formed tend to coalesce at the grain boundaries,
initiating the socalled grainboundary cracking, which causes structure embrittlement.
Again, by using the cross section as comparison tool, Mansur and Farrell ([1]) conclude
that in order that induced displacements contribute more than 10% compared to fast
neutron induced displacements the following condition must be satisfied:
(2)
100
nf
where is the ray flux. As discussed in Section 7.2, this inequality, not satisfied at the
RPV walls of LWRs, can become true in case of wide water gaps interposed between
core and vessel walls. This fact significantly affects the applicability of the current
embrittlementrelated regulations, which are neutronfluence based, to reactors having
the mentioned characteristic. The ABWR is an example of such reactors.
More precisely, once transition from brittle to ductile behavior has occurred, most materials experience a
very slow toughness reduction as the temperature is further increased.
Since the brittleductile transition does not occur at a single temperature, but over a
temperature interval, Appendix G of 10 CFR Part 50 ([4]) recommends to refer the
increase of the temperature transition from the unirradiated condition, DBTT, to the
measurement performed at an impact energy of 30 ftlb (41 J), as shown in Figure 4.
3.3 Quantitative prediction of embrittlement
Should experimental data not be available, Section 50.61 of 10 CFR Part 50 ([4])
recommends, with some limitations discussed later, the use of the following relation for
DBTT (C):
DBTT =
5
CF f
9
+M
(3)
where:

As mentioned above, equation (3) is not universally applicable, and its use is
recommended as long as ([5]):
Chemistry Factors for several CuNi content combinations are listed in Table 1 and Table 2 of Section
50.61 of 10 CFR Part 50 ([4]).
the material under investigation is a ferritic steel of type SA302, 336, 533, 508
(typical RPV steels);
Figure 5 shows the DBTT increase obtained by applying equation (3) to two RPV steels:
a modified SA302B type (0.23 wt% Cu, 0.49 wt% Ni), and a low Cu/Pcontent steel,
i.e. SA533B1 (0.07 wt% Cu, 0.60 wt% Ni) (compositions from [3]). It is evident how
significantly alloy elements can affect the increase of DBTT 3 . Figure 5 also shows the
asymptotic nature of the irradiationinduced embrittlement: the rate of DBTT increase
tends to decrease as the neutron fluence increases.
Calculated increase in DBTT (C)
140
120
100
80
60
40
SA302B modified
SA533B1 (low Cu/P content)
20
0
0.01
0.1
10
100
Figure 5: DBTT increase as a function of fast neutron fluence (En>1 MeV) for two types
of ferritic steel. Curves obtained from equation (3).
The user of equation (3) must be aware of the significant dependence of the postirradiation DBTT on the temperature at which irradiation takes place. Regulatory Guide
1.99, Rev.2, ([5]), states that, with respect to the values predicted by equation (3):
irradiation below 525F (274C) should be considered to produce greater
embrittlement, and irradiation above 590F (310C) may be considered to produce less
embrittlement. Figure 6 shows that irradiation experienced below 400F (204C) yields a
postirradiation DBTT that can be up to 100C higher than that resulting from an
irradiation at 550F (288C).
It is important to point out, however, that Cu and Ni content can not be indefinitely reduced since the
corrosionresistance properties of the steel would be negatively affected.
If together with the DBTT increase, the actual toughness reduction was also of interest,
Regulatory Guide 1.99, Rev.2, ([5]), provides an useful graph to conservatively estimate
the reduction percentage in Upper Shelf Energy (USE) as a function of neutron fast
fluence and Cu content. This graph is represented in Figure 7 together with the fluence
ranges for typical BWRs and PWRs.
wet annealing: the annealing temperature is typically below 345C since water is
used as heating medium. Although the relatively low temperature prevents
achieving a complete toughness recovery, the process is not complicated from an
engineering viewpoint since the primary water temperature is controlled by pump
heat up, and 345C is a typical value for the RPV design temperature, at which the
RPV is tested during preservice testing.
Concerning the duration of the annealing treatment, Mager et al. (referenced by [7])
concluded that an excellent recovery of toughnessrelated properties could be achieved
by annealing at 450C for 168 hours. After the treatment, the reembrittlement caused by
irradiation follows the same rate as that experienced before annealing. For this reason, the
increase of DBTT with fluence is often referred as lateral shift, since the effect of
annealing is simply to laterally shift the DBTT vs fluence curve towards higher fluence
levels, as illustrated in Figure 8.
Regulatory Guide 1.162 states that: Although thermal annealing has not yet been
applied to a U.S. commercial power reactor, it has been successfully applied to other
reactors. Two reactor vessels that have been successfully annealed are the Army's SM1A
in 1967, and the BR3 in Mol, Belgium, in 1984. Both of these reactors operated at
temperatures low enough to permit "wet annealing" at a temperature of 650F using the
reactor coolant pumps as the heat source. In addition, at least 12 Russiandesigned
VVER440 PWRs, which operate at conditions similar to U.S. PWRs, have been annealed
in Russia and Eastern Europe at temperatures of approximately 850F, using dry air and
radiant heaters as the heat source ([7]).
minimum allowed toughness of the RPV in the beltline region (102 J before
irradiation and 68 J throughout the entire RPV life);
pressure and temperature limits for the various operating phases that a reactor
experiences during its design life, including unanticipated operating events.
without hazard of brittle fracture. Instead, it explicitly refers to the ASME Code, whose
most important points are discussed below.
5.2.1 ASME Code: pressure and temperature limits during normal operation
From the operating condition viewpoint, the effect of the embrittlement is to constrain the
number of possible combinations pressuretemperature (pT) that each point of the RPV
can tolerate without experiencing nonductile propagation of a preexisting crack assumed
to exist at that point.
It is well known that the brittle crack propagation is governed by the stress intensity
factor at the crack edges. When such a factor exceeds a critical value KIc, referred as Plain
Strain Fracture Toughness 4 , the propagation occurs. KIc depends on the type of crack and
in particular on the actual toughness of the material, i.e. the toughness calculated
accounting for the irradiation effect. At any time during the RPV operating life, i.e. at any
given level of fluence, the verification of whether a pT combination can be tolerated at a
certain location in the vessel is performed by comparing an adhoc defined stress
intensity factor K I* , which is function of p and T, to the lower bound of experimental
measures of fracture toughness performed on specimens made of the same material of the
RPV and irradiated to the same level of fluence. The following inequality must be
satisfied:
K I* < K Ic
(4)
Figure 9 shows the curve used to perform this comparison. The parameter plotted is the
experimental Plain Strain Fracture Toughness, which was measured at different
temperatures for various specimens, each irradiated to a certain level of fluence, and
therefore having a certain DBTT. It is important to note that, no available data points
for static tests fall below the curve (Appendix G of [8]), i.e. it has never been observed a
4
Fracture toughness, Kc, is a quantitative way of expressing a material's resistance to brittle fracture
when a crack is present ([9]). It represents the minimum stress intensity factor that, once originated in a
specimen at the edges of a preexisting crack, causes the crack to propagate, yielding brittle fracture. This
parameter depends on several factors, among which the dependences on the material toughness (via TDBTT) and on strain rate & are usually graphically (and not analytically) represented. For a given
temperature T and & , Kc is defined as: K c = Y S a , where S is the stress intensity, a is a function of
the size and type (internal, edge, etc.) of the preexisting crack, while Y is a dimensionless factor depending
on the type of crack and on the ratio of crack size to specimen width (measured in the direction of crack
propagation). It is important to note that, for very large values of this ratio, i.e. for large specimen width b,
Y looses its dependence on the cited ratio, tending to a constant value which depends only on the type of
crack. Because of the relation above, such independence of Y from the specimen dimensions translates into
an independence of Kc from the specimen dimensions. In fact, Kc decreases as b increases, but remains
constant and equal to KIc, referred as Plain Strain Fracture Toughness, for widths larger than a critical width
2
usually indicated as B. It has been experimentally verified that, as a general rule, B 2.5 K Ic . Since for
y
0.5
a typical ferritic steel ( y = 350 MPa ), KIc varies from about 35 MPa m (at T=DBTT50C) to about 200
MPa m0.5 (at T=DBTT+50C), the resulting values for B ranges between few cm to ~80 cm. Being the
typical RPV wall thickness well below 80 cm, the RPV resistance to brittle crack propagation is actually
larger than KIc. Therefore, the use of the Plain Strain Fracture Toughness KIc to characterize the RPV
resistance to brittle fracture can be defined as a conservative estimate of its actual ability to withstand
nonductile fracture.
brittle crack propagation occurring for stress intensity factors lower than those
represented by the curve.
K I* = 2 K Im + K It
(5)
where KIm is the membrane tension stress intensity factor while KIt is the thermal stress
intensity factor, both referred to the location of the RPV under investigation. The former
is calculated as:
K Im = M m
5
pRi
t
(6)
where t is the thickness of the RPV section under examination, p is the internal pressure,
Ri is the vessel inner radius and Mm is a function of t and of whether the crack is on the
inside or on the outside surface (Appendix G of [8]). The thermal stress intensity factor is
instead the maximum stress intensity factor that generates in the RPV because of the
throughwall thermal gradient produced during heatup and cooldown phases. Being K I*
dependent on the heatup/cooldown rate, the coolant temperature variation with time can
not take any value, but is typically limited at 56 C/hr.
By introducing (5) and (6) into (4), the maximum allowed pressure at a given time during
the RPV operating life can be calculated as:
K K It
p = min Ic
2 M m
Ri
(7)
where the subscript j indicates the jth geometric location in the RPV.
It is important to note that, besides the conservativism implicit in the use of the
maximum postulated cracks, additional conservativisms characterize the procedure
recommended for the verification of the allowable pressure. The use of a factor 2 in
equation (5) together with the conservative choice previously described for the
development of the KIccurve are good examples of such an extraconservativism. For
this reason, while discussing the concept of maximum postulated cracks Appendix G of
[8] states that: the prevention of nonductile fracture is ensured () even if the defects
were to be about twice as large.
5.2.2 ASME Code: p and T limits during unanticipated operating events
Appendix E of the ASME Code ([8]), provides acceptance criteria on the maximum
allowed pressure during two types of unanticipated operating transients: pressurized
thermal transients, in which the coolant temperature variation rate is larger than 5.6 C/hr,
and isothermal pressure transients in which this rate is less than 5.6 C/hr. During the
former, the pressure must not exceed the design pressure, while the coolant temperature
must be at least 31 C above the RPV DBTT. Instead, during the more frequent
isothermal pressure transients, the maximum pressure must not exceed the allowable
values listed in Table 1 (from Table E2 of Appendix E of [8]):
Table 1: Maximum allowable pressure as a function of TDBTT
(for design pressure grater than 2400 psig)
Type of unanticipated
operating event
Pressurized thermal transients
TDBTT, C
31
13.9
8.3
5.5
0
5.5
13.9
27.8
41.6
58.3
92.2
111.1
Design pressure
1.1design pressure
2400 (16.548)
2250 (15.513)
2000 (13.790)
1750 (12.066)
1500 (10.342)
1200 (8.274)
1000 (6.895)
850 (5.860)
800 (5.516)
750 (5.171)
surveillance specimen capsules must be located in the beltline region so that they
can duplicate the neutron spectrum, the temperature history and the maximum
neutron fluence experienced by the RPV;
well beyond the original 40year design life. In fact, the simultaneous occurrence of
crtiticalsize flaws, embrittled vessel, and a severe PTS transient is a very low probability
event (NUREG1808, [11]).
Figure 10: Pressure and temperature variations during the 1978 accident at Rancho Seco
([3])
6.1 PTS prevention regulations
Section 50.61 of 10 CFR Part 50 ([4]) is aimed at monitoring the level of RPV
embrittlement in order to specifically prevent PTS occurrence. The monitoring process is
extremely important bacause, only for PTSrelated purposes, the neutron fluence of
interest is that corresponding to the expiration date of the operating license. Based on the
maximum allowed fluence for PTS prevention, the continuous monitoring of the level of
fluence allows to plan the future neutron flux history on the RPV walls, so that a
certain safety margin is ensured at the end of the operating license and, at the same time,
the fuel loaded in the peripheral assemblies can be completely burnt. In other words, the
fluence monitoring allows satisfying safety and economic requirements simultaneously.
The above mentioned maximum allowed fluence for PTS prevention is expressed, in
Section 50.61 of [4], in the form of the DBTT that is expected to characterize the RPV
beltline region at the end of the operating license, DBTTPTS. The PTS Screening
Criterion establishes that the maximum allowed DBTTPTS is 132C for plates, forgings
and axial weld materials, and 149C for circumferential weld materials ([4]). The
verification of whether these limits are exceeded or not is performed using the following
relation for DBTTPTS:
DBTTPTS = DBTTU + M + DBTTPTS
(8)
where DBTTU is the Ductile Brittle Transition Temperature of the unirradiated material,
M is a positive factor accounting for uncertainties in the value of DBTTU, in the material
composition, fluence and on the calculation procedures. Finally, DBTTPTS, which can be
calculated using equation (1), is the increase in Ductile Brittle Transition Temperature
experienced by the material at the end of RPV operating license.
It is interesting to note that the value of DBTTPTS calculated using equation (8) will
cause about five US plants to exceed the PTS screening criteria limits before the
expiration of their 40year operating licenses, and four more will be within 1F of the
screening criteria limits ([3]). In these cases, in order to avoid premature retirement of
the RPV, Section 50.61 of [4] allows the RPV beltline to be given a thermal annealing
treatment so that toughness and DBTT can be recovered and the PTS screening criterion
be satisfied.
The reason of this apparent inconsistency lie just in neglecting, when the fast neutron flux
is relatively low like in BWRs, the flux history that has led to the observed final value of
fast neutron fluence. A possible explanation was given by H. Ino, professor in the
Department of Mechanical Engineering at Hosei University, Japan ([12]). By means of a
computer simulation based on a chemical rate equation theory, he investigated the effect
of different values of fast neutron flux on the increase of yield stress, which is one of the
consequences of neutron irradiation (see Figure 3). He also performed a sensitivity
analysis by changing the copper content, based on the well known embrittlementenhancement effect due to this alloy element. It is crucial to note that the simulation was
performed with a constant fluence, but varying the neutron flux. The results are
summarized in Figure 12 where, for the purposes of this discussion, only the continuous
lines need to be considered. The parameter on the xaxis is the damage rate expressed in
displacement per atom per second (dpa/s), which can be interpreted as the fast neutron
flux (not fluence!) in view of the approximate correspondence: 1 dpa ~ 71020 n/cm2
([12]). It can be noticed that the level of embrittlement, in the form of change of yield
stress with fixed fluence, is higher at lower fluxes. In particular, for a low fast flux
reactor like a BWR, the level of embrittlement that would be predicted by the current
methodologies and regulations is that corresponding to harder fluxes, i.e. PWRlike
fluxes and even harder (from test reactors), resulting in an evident underestimate of the
embrittlement. Although this difference increases as the copper content increase, the
underestimate corresponding to typical RPV Cu contents (below 0.3%) is not negligible.
The importance of the neutron flux in determining the level of embrittlement has been
also observed by comparing two types of specimens extracted from Japanese BWRs:
those located very close to the core and therefore subjected to a neutron flux larger by an
order of magnitude, and those located on the RPV walls. In Figure 13 they are referred as
BWR accelerated and BWR respectively. Figure 13 also shows the calculated profile
for very embrittlementsensitive steels (worst case), embrittlementsensitive steels, and
embrittlementinsensitive steels. It can be noticed that, in spite of absorbing a lower
fluence, the specimens located at the BWR vessel walls (full dots) show the same
DBTT as those located closer to the core and thus absorbing higher fluence (other than
higher flux). This is again the lowflux embrittlement enhancement phenomenon
investigated by Ino.
Oak Ridge National Laboratory ([1]) proved that, in presence of large water gaps, the use
of data coming from neutrondominatedembrittlement tests to predict RPV damage
yields a significant overestimate of the time needed to reach a certain level of
embrittlement. In fact, the term accelerated embrittlement was coined to indicate the
early embrittlement of the coupons analyzed during the High Flux Isotope Reactor
(HFIR) surveillance program ([14]). Like the ABWR, also the HFIR has a large water
gap (~600 mm wide) interposed between the core and the RPV. The analysis of the
surveillance data showed that HFIR specimens irradiated at fast neutron fluences of the
order of 1017 n/cm2 showed a DBTT increase of tens of degree C, when instead
experimental measures performed in test reactors not provided with large water gaps
indicated that no embrittlement should occur at fast fluences below about 61017 n/cm2.
In the HFIR specimens a significant ray effect was in fact superposed to that of fast
neutrons. Figure 15 shows the HFIR data compared to those coming from typical test
reactors.
interposing, in some cases, large water gaps between core and RPV
References
[1]
[2]
Singh, B.N., Horsewell, A., Toft, P., Edwards, D.J., Temperature and dose
dependencies of microstructure and hardness of neutron irradiated OFHC
copper, Journal of Nuclear Materials 224 (1995), 131140.
[3]
Shah, V.N., MacDonald, P.E., Aging and Life Extension of Major Light water
Reactor Components, Elsevier, 1993.
[4]
[5]
[6]
[7]
Regulatory Guide 1.162, Format and Content of Report for Thermal Annealing
Reactor Pressure Vessels, February 1996. USNRC.
[8]
ASME Boiler and Pressure Vessel Code, Section XI, Division 1, Rules for Inservice Inspection of Nuclear Power Plant Components, American Society of
Mechanical Engineers, New York, 1986.
[9]
[10]
[11]
[12]
[13]
[14]
Abstract
Stability has been the main concern in the development stage of BWRs in history.
Although stability studies of BWRs have a long history and there is extensive literature
available, the literature on fuel performance under unstable power oscillations is very
limited. It is very desirable to investigate the fuel integrity in BWR power oscillation
conditions.
This goal of this work is to examine the cladding performance under the power
oscillations. The calculations of fuel deformation were first conducted by using the
FRAPTRAN code to simulate the Japanese FK11 test and power oscillation under normal
BWR operating conditions. Then, based on the results of FRAPTRAN, the cladding
fatigue analysis was conducted.
Based on the analysis, it could be concluded that: FRAPTRAN code could be used to
analyze fuel performance under BWR power oscillations. The FRAPTRAN code results
were in good agreement with FK11 test in Japan. The fuel deformation was mainly
caused by PCMI and was roughly proportional to the fuel enthalpy. Enhanced cladding
deformation due to ratcheting was not found. The cladding could satisfy the design
criterion in ASME Code under power oscillation conditions, which means it could
maintain the fuel integrity. Cladding thermal fatigue is not an issue under power
oscillations, unless dryout takes place.
1. Introduction
1.1 BWR Stability and its Effects to Safety
In historical perspective, stability has been the main concern in the development stage of
BWRs. It is of practical importance for designing and operating BWRs.
1
The boiling twophase flow in the core may become less stable because of the time lag
between vapor generation and pressure loss perturbation. Furthermore, in BWRs, the
reactivity depends strongly on the core void fraction. Thus, when a void fraction
oscillation is established in a BWR, the power oscillates according to the neutronic
feedback. This feedback mechanism may under certain conditions lead to poorly damped
or even limitcycle power oscillations. Their frequency lies around 0.5 Hz (about twice
the transport time of the coolant through the core). Amplitudes from nearly 0% to more
than 100% in power have been observed. The oscillations are mostly global, i.e. inphase. [Hanggi P., 2001]
Higher mode power oscillations are also possible; these divide the core into two regions
oscillating in opposite directions at constant overall power. These regional oscillations,
also called outofphase oscillation, are cumbersome for the operators since their
detection is not directly possible with standard instruments that display only coreaverage
data.
Generally, the BWR stability issue is not a major industry safety problem from a
technical point of view. Given appropriate instrumentation, power oscillations are easy to
detect and there exist simple, as well as effective, counter measures. A scram will
normally solve the problem. But the concerns of outofphase oscillation mode and the
unavailability of reactor shutdown system (the case of ATWS originated oscillations) keep
the stability issue a safety concern.
In terms of safety, the concerned variables in an instability occurrence with high
oscillation amplitudes are the neutron flux and the rod surface temperature: the control of
the first of the above quantities may prevent any undesired excursion of the second one.
Additional problems might arise due to thermal cycling that may affect the fuel rod
integrity, making pelletcladding interaction more probable; thermal cycling may also
induce greater than normal fission product release from pellets. [F. DAuria, 1997]
The numerous modifications in reactor size, reactor power, fuel design, power density,
discharge burnup and loading strategies changed the core stability behavior of the BWR
reactor to a significant extent. Since, for economic reasons, the trend towards smallerdiameter fuel rods and different loading strategies will persist, the stability problem has to
be taken care of. It is, therefore, important to understand the underlying mechanism of
powervoid instability as thoroughly as possible, as well as to be able to understand its
consequence to fuel integrity.
2
In literature, there are numerous works on BWR stability. But most of them are focused
on knowing the mechanism of different of oscillations, and on how to predict the
oscillations by developing all kinds of codes. Very limited work was done on the effects
of oscillation.
This work is focused on if the fuel integrity (failure of cladding) could be conserved
under the power oscillations.
1.2 Fuel Failure Modes
An LWR fuel rod typically consists of UO2 fuel pellets enclosed in Zircaloy cladding.
The primary function of the cladding is to contain the fuel column and the radioactive
fission products. If the cladding does not crack, rupture, or melt during a reactor transient,
the radioactive fission products are contained within the fuel rod. During some reactor
transients and hypothetical accidents, however, the cladding may be weakened by a
temperature increase, embrittled by oxidation, or over stressed by mechanical interaction
with the fuel. These events alone or in combination can cause cracking or rupture of the
cladding and release of the radioactive products to the coolant. Furthermore, the rupture
or melting of the cladding of one fuel rod can alter the flow of reactor coolant and reduce
the cooling of neighboring fuel rods. This event can lead to the loss of a coolable
reactor core geometry.
The primary means of cladding failure result from mechanical loading of the internal clad
surface by fuelclad interactions or from the buildup fission product gasses in the plenum
to critical values. Failure in both cases is due to rupture from plastic deformation and
creep. Analysis of the mechanical behavior is a very complicated task due to the
complicated relationships which exist between the mechanical, thermal, chemical, and
structural properties. Additionally, not all of the individual mechanisms involved are
completely understood. [MIT 22.314 Lecture Notes F.1]
The most prevalent failure mode for fuel rods is due to the failure of cladding from
PelletCladding Mechanical Interactions (PCMI). PCMI failures are the result of
mechanical interaction between the fuel and cladding in a corrosive environment.
Thermal expansion mismatches between the fuel pellets and cladding material result in
pellet expansion exceeding that of the clad. This internal stress is opposed by the external
stress of the pressurized cooling system such that a tight interface between the two
materials is maintained. Volatile fission products of cesium and iodine released to the gap
3
4
5
Then, based on the results of FRAPTRAN, the cladding fatigue analysis was conducted.
The work includes:
(1) Simulating the power oscillation tests of the NSRR (FK11) by using FRAPTRAN
code, and make a comparison with experiment results.
(2) Simulating the power oscillations without scram in normal BWR operating
conditions.
(3) Analyzing the influence on cladding fatigue during power oscillations.
cladding in particular. The code has a heat conduction model to calculate the transfer of
heat from the fuel to the cladding, and a cooling model to calculate the transfer of heat
from the cladding to the coolant. The code has an oxidation model to calculate the extent
of cladding embrittlement and the amount of heat generated by cladding oxidation. A
mechanical response model is included to calculate the stress applied to the cladding by
the mechanical interaction of the fuel and cladding, by the pressure of the gases inside the
rod, and by the pressure of the external coolant.
2.2 Cladding Deformation Model
The detailed description of cladding deformation model used in FRAPTRAN code could
be found in Report NUREG/CR6739. Here the author only discussed the different
models in the strainstress calculation in case whether the claddingfuel gap was open or
closed, which resulted in different mechanisms of deformation.
Deformation and stresses in the cladding in the open gap regime are calculated using a
model which considers the cladding to be a thin cylindrical shell with specified internal
and external pressures and a prescribed uniform temperature.
Calculations for the closed gap regime are made using a model which assumes that the
cladding is a thin cylindrical shell with prescribed external pressure and a prescribed
radial displacement of its inside surface. The prescribed displacement is obtained from
the fuel thermal expansion model. Furthermore, because no slippage is assumed to take
place when the fuel and cladding are in contact, the axial expansion of the fuel is
transmitted directly to the cladding. Hence, the change in axial strain in the shell is also
prescribed.
The decision as to whether or not the fuel is in contact with the cladding is made by
comparing the radial displacement of the fuel with the radial displacement that would
occur in the cladding due to the prescribed external (coolant) pressure and the prescribed
internal (fission and fill gas) pressure. The decision is expressed by the equation:
..
urfuel urclad + a
..
7
(2.1)
If the above equation is satisfied, the fuel is determined to be in contact with the cladding.
The loading history enters into this decision by virtue of the permanent plastic cladding
strains imposed in the cladding by the cladding loads.
If the fuel and cladding displacements are such that Equation 2.1 is not satisfied, the fuelcladding gap has closed during the current loading step and the open gap solution is used.
If Equation 2.1 is satisfied, however, the fuel and cladding have come into contact during
the current loading increment. At the contact interface, radial continuity requires that:
..
u rclad = u rfuel a
(2.2)
while in the axial direction the assumption is made that no slippage occurs between the
fuel and cladding. This state is referred to as PCMI.
Note that only the additional strain which occurs in the fuel after PCMI has occurred is
transferred to the cladding. Thus, if z,clad
o is the axial strain in the cladding just prior to
contact and z,fuel
is the corresponding axial strain in the fuel, then the noslippage
o
condition in the axial direction becomes:
fuel
zclad zclad
zfuel
,o = z
,o
(2.3)
After u rclad and rclad have been calculated, a solution is made of the stresses and strains
in a thin cylindrical shell with prescribed axial strain, external pressure, and prescribed
radial displacement of the inside surface. The solution also gives the interface pressure
between the fuel and cladding.
2.3 Fuel Rod Internal Gas Pressure Response Model
The pressure of the gas in the fuel rod must be known in order to calculate the
deformation of the cladding and the transfer of heat across the fuelcladding gap. The
pressure is a function of the temperature, volume, and quantity of gas. The detailed
description of cladding deformation model used in FRAPTRAN code could be found in
Report NUREG/CR6739.
It should be mentioned here that FRAPTRAN does not have a model to calculate the
transient release of fission gases as a function of temperature. The fill gas composition
and pressure at the time of the transient, which is dependent on fission gas release prior to
the transient, is either manually entered by the user or read from a FRAPCON3 burnup
initialization file.
8
The released fission gas does affect the gas pressure and composition, which in turn
impacts the transient thermal and mechanical calculations. It has also been proven that
some amount of fission gas was released during power oscillation transients. Therefore, a
user input option (MODEL data block in the input file) is used to specify the fission gas
release to the fuelcladding gap and rod plenum during power oscillation transients.
Although the rodaverage fractional fission gas release could be specified as a function of
time during the transient, we dont know the detailed information about the fission gas
release in a power oscillation test. For simplicity, the internal gas pressure, rather than
fission gas release fraction, was specified as a function of time. In our analysis, different
internal gas pressure history was used to identify what is the effect of this simplification.
9
It should be noticed that typical fuel performance codes (like FRAPCON) calculate Pm
and Q, but not Pl and Pb.
3.2 Cladding Strain Criterion
The total permanent uniform strain shall not exceed:
1 % membrane strain (limiting)
2 % bending strain
5 % local strain
The intent of this requirement is to limit cladding damage due to slow rate strain
accumulation at which the stress does not reach the stress limit (yield stress). The clad
loading mechanism is the rod internal differential pressure with the system pressure and
clad straining by the pellet expansion and PCMI. A bending strain and local strain are not
calculated by FRAPCON and the limits are not applied at this time.
 10 
Power history in test FK11 is illustrated in Fig. 4.1. At the beginning of the test, rod
power was kept at 27 kW/m for about 5 s to develop a center peaked temperature profile
before initiating the power oscillation transient. Then, the test rod was subjected to four
power oscillations to the peaks at 38, 43, 47, 50 kW/m at intervals of 2 s. Fuel enthalpy
during the test relative to room temperature was estimated with FRAPT6 and is
illustrated in Fig. 4.1.
 11 
However, the FRAPCON steadystate output file, which is used as an input for
FRAPTRAN, is prepared based on Fukushima Diana 2 Reactor, also known as base
irradiation file for FK1. There are some differences between FK1 and FK11 pretest
conditions, which are shown in the table below.
Table 4.2 Difference between FK11 pretest and FRAPTRAN input condition
Design
Irradiation
Parameter
Fill Gas Pressure
Fuel Density
Fuel Enrichment
Peak Linear Power
Irradiation Time
Burnup
EOL Gas Pressure
Unit
MPa
%
%
kW/m
day
MWd/kg
MPa
FK11 pretest
FRAPTRAN input
0.5
0.3
97
95
4.5
3.9
35
17.3
4 cycles
2500
56
65
1.4
0.6
The major differences between the pretest condition and the FRAPCON simulation
results may be the burnup and the internal gas pressure. As in our FRAPTRAN input file,
the internal gas pressure was specified as a user input, we could changed the initial gas
pressure to 1.4 MPa, the same as FK11 pretest condition. The difference of burnup may
affect the results in some degree, so our comparison between the test and simulation
could only be qualitative. It should be also mentioned that the FK11 test is operating
under cold zero conditions before power oscillation starts. Thus, we need to further
conduct analysis in normal BWR operating conditions to investigate if the fuel integrity
could be maintained under power oscillations.
4.2 Results of tests and FRAPTRAN simulation
In test FK11, synchronized axial elongation of fuel stack and cladding was observed as
shown in Fig. 4.2. This deformation behavior suggests that strong PCMI controlled the
cladding deformation during the test. Posttest diameter measurement showed that both
radial and axial deformation was elastic. The cladding surface temperature remained
about 100C throughout the power transient. Rod internal pressure increased gradually
from 1.4 to 2.2 MPa during the power transient, a result of heat up of the gap/plenum gas
and fission gas release from the pellets. Posttest gas analysis indicated that fission gas
release during the test was 0.4% of the total produced in the pellets. Test results are
summarized in Table 4.3
 12 
After changing the FRAPTRAN input file to satisfy the FK11 test conditions,
FRAPTRAN 1.3 Code was ran, and the results was shown in the Fig. 4.3 and Fig. 4.4.
 13 
2.5
50
60
40
30
20
1.5
0.5
10
10
15
20
25
30
10
Time (sec )
20
25
30
40
40
35
35
15
Time (sec )
30
25
20
15
10
30
25
20
15
10
10
15
20
25
30
10
Time (sec )
15
20
25
30
Time (sec )
Fig. 4.3 Results of elongation of pellet stack and cladding, rod internal pressure and
reactor power of FRAPTRAN simulation for test FK11
Cladding Outside Temperature (K )
1.60E02
4.50E+02
4.50E+01
1.40E02
4.00E+02
4.00E+01
3.50E+02
1.20E02
3.50E+01
3.00E+02
1.00E02
3.00E+01
2.50E+02
8.00E03
2.50E+01
2.00E+02
6.00E03
2.00E+01
1.50E+02
4.00E03
1.50E+01
1.00E+02
2.00E03
1.00E+01
5.00E+01
0.00E+00
5.00E+00
0.00E+00
10
15
Time (sec)
20
25
30
0.00E+00
0
10
15
Time (sec)
20
25
30
15
20
Time (sec)
25
30
7.00E+01
10
1.20E02
5.00E+01
4.00E+01
3.00E+01
2.00E+01
1.00E+01
1.00E02
4.00E03
4.50E03
6.00E+01
8.00E03
6.00E03
4.00E03
3.50E03
3.00E03
2.50E03
2.00E03
1.50E03
1.00E03
2.00E03
5.00E04
0.00E+00
0.00E+00
0
10
15
Time (sec)
20
25
30
0.00E+00
0
10
15
Time (sec)
20
25
30
10
15
Time (sec)
Fig. 4.4 Results of other parameters of FRAPTRAN simulation for test FK11
 14 
20
25
30
4.3 Discussion
From FRAPTRAN simulation results, we can find that the axial elongation of fuel stack
and cladding was nearly the same in Fig. 4.3.c and 4.3.d; and also that the structural
radial gap is zero in power oscillation periods (Fig. 4.4.a) and enormous amount of gap
interface pressure (cladpellet interaction, Fig. 4.4.d). These both suggest that strong
PCMI controlled the cladding deformation in the oscillations, which is consistent with the
test results.
However, the test results shows that very small plastic cladding deformation was left after
the power oscillation, while the FRAPTRAN simulation results indicated fairly large
plastic cladding deformation after the oscillations. Also the amplitude of the peak clad
axial elongations between test and simulation are a little different. These differences may
arise from the difference in initial burnups (56 MWd/kg in test, 65 MWd/kg in
simulation).
Furthermore, from Fig.4.4.e and Fig.4.4.f, the cladding deformation was not accumulated
through the power cycling, indicating that ratcheting deformation of the cladding by the
pellets did not occur, which agree with the test results. It could also be shown that the
fuel axial elongation is proportional to the fuel enthalpy increase from Fig. 4.3.c and Fig.
4.4.c., which also is consistent with the test results.
In FK11 test, the cladding axial elongation is around 47mm, the corresponding axial
strain is exceeding 1% membrane strain limits. However, the posttest examination
showed that the cladding is not failed. In FRAPTRAN simulation, the cladding axial
strain is around 0.8%, below the 1% membrane strain limit.
Although the existence of the small difference in initial conditions between FK11 test
and FRAPTRAN simulation, the results of them are in good agreement. This gives us
some confidence to use FRAPTRAN code to simulate power oscillations in normal BWR
condition, rather than cold zero condition in FK11 test. It could be concluded that the
fuel deformation in power oscillations is mainly caused by PCMI, and is roughly
proportional to the fuel enthalpy, and enhanced cladding deformation due to ratcheting by
the cyclic load was not observed (Also due to the assumption of non axial slippage at
fuelcladding interface in the model).
 15 
 16 
Table 5.1 Needed BWR/4 information for Power and flow rate calculation
Parameter
Core
Thermal power
Core flow rate
Power density
Typical BWR/4
3293 MWt
12915 kg/s
51 kW/liter
4.75 m
6.375 m
7.2 MPa
0.15 MPa
275 C
1.4
764
Square 8x8
63
Coolant Mass Flux
30
2.00E+02
1.80E+02
35
25
1.60E+02
20
1.40E+02
1.20E+02
15
1.00E+02
10
8.00E+01
6.00E+01
4.00E+01
2.00E+01
10
20
30
40
50
60
0.00E+00
Time (sec )
10
20
30
Time (sec)
40
50
60
Fig. 5.1 History of linear heat generation rate and flow rate in power oscillations
5.2 Results of FRAPTRAN Simulation
The major results of deformation of fuel and cladding under oscillations is shown in the
Fig. 5.2, Fig. 5.3, and Fig. 5.4. It should be mentioned here that the fission gas release is
not considered in FRAPTRAN code. Thus, as discussed in Section 2.3, the internal gas
pressure was specified as a function of time in our analysis. The internal gas pressure
history could be found in Fig. 5.2.a.
 17 
2.5
1.20E+03
2
1.00E+03
1.5
8.00E+02
6.00E+02
4.00E+02
0.5
2.00E+02
10
20
30
40
50
60
0.00E+00
Time (sec )
20
30
Time (sec)
40
50
60
35
40
30
10
25
20
15
10
35
30
25
20
15
10
10
20
30
40
50
60
10
20
Time (sec )
30
40
50
60
Time (sec )
Fig. 5.2 Results of FRAPTRAN simulation for BWR power oscillation (1)
Structural Radial Gap
2.50E+01
5.75E+02
1.40E03
1.60E03
5.70E+02
1.20E03
2.00E+01
5.65E+02
1.00E03
8.00E04
1.50E+01
5.60E+02
6.00E04
1.00E+01
5.55E+02
4.00E04
5.00E+00
5.50E+02
2.00E04
0.00E+00
0.00E+00
5.45E+02
10
20
30
Time (sec)
40
50
60
10
20
30
Time (sec)
40
50
60
10
30
Time (sec)
40
50
60
50
60
7.00E+01
20
5.00E03
1.20E02
4.50E03
6.00E+01
1.00E02
4.00E+01
3.00E+01
2.00E+01
4.00E03
5.00E+01
8.00E03
6.00E03
4.00E03
3.50E03
3.00E03
2.50E03
2.00E03
1.50E03
1.00E03
2.00E03
1.00E+01
5.00E04
0.00E+00
0.00E+00
0.00E+00
10
20
30
Time (sec)
40
50
60
10
20
30
Time (sec)
40
50
60
10
20
30
Time (sec)
Fig. 5.3 Results of FRAPTRAN simulation for BWR power oscillation (2)
 18 
40
5.00E03
9.00E04
4.50E03
8.00E04
4.00E03
3.50E03
3.00E03
2.50E03
2.00E03
1.50E03
1.00E03
7.00E04
6.00E04
5.00E04
4.00E04
3.00E04
2.00E04
1.00E04
5.00E04
0.00E+00
0.00E+00
10
20
30
Time (sec)
40
50
60
10
30
40
50
60
Time (sec)
6.00E+02
3.50E+02
3.00E+02
5.00E+02
20
1.00E04
4.00E+02
3.00E+02
2.00E+02
1.00E+02
2.50E+02
2.00E+02
1.50E+02
1.00E+02
5.00E+01
0.00E+00
10
20
30
40
50
60
0.00E+00
10
1.00E+02
20
30
40
50
5.00E+01
60
1.00E+02
Time (sec)
Time (sec)
Fig. 5.4 Results of FRAPTRAN simulation for BWR power oscillation (3)
In Fig. 5.2.c and Fig. 5.2.d, cladding axial elongation follows quite well with the fuel
pellet axial elongation; also Fig. 5.3.a and Fig. 5.3.d indicate the gap is closed under
power oscillations. These facts suggest the strong PCMI controlled cladding deformation
during oscillations.
5.3 Discussion
In this section, we followed the limiting criteria discussed in Section 3 to examine if the
cladding is failed under power oscillations. The results were shown in Table 5.2.
Table 5.2 Simulation results applying to the design criterion
Simulation Results
Criteria Values
If Satisfy
Design Criterion
Yes
Pm< 1.0 Sm
Pm = tresca = r  , Sy 500 MPa ,
Pm+Q<3.0 Sm
Sm = 2 / 3Sy = 333MPa
3Sm = 1000MPa
Yes
1%
No
Pm + Q = 460 MPa
1% membrane strain
Overall, z = 0.008
 19 
It should be mentioned that the thermal stress Q achieved its maximum value when the
fuel rod increased its power form zero to rated power before power oscillation started,
which may not happen in the real plants. Although the 1% membrane strain criteria is
not satisfied, we still have enough confidence that the fuel rod would not fail as we can
see the cladding strain and permanent axial and hoop strain did not changed with power
oscillations from Fig. 5.4.a and 5.4.b. But it raised some concern about the cladding creep
behavior which is above the content of current work.
We should be careful for the discussion above because we are taking the internal gas
pressure as an input, which may be totally different with the real case. Thus, we discard
the input for the internal gas pressure history and conducted analysis again. Its result was
shown in the Fig. 5.5. The curve shape of the each parameter was basically the same,
while the amplitude was a little different. This proves that the internal gas pressure only
has very limited effects on fuel deformation in power operation conditions, when the
coolant pressure is high enough.
1.16
1.155
1.15
1.145
1.14
1.135
35
35
30
30
1.17
1.165
25
20
15
10
10
20
30
40
50
60
70
20
15
10
25
10
20
30
Time (sec )
40
50
60
70
10
20
30
4.50E+02
9.00E03
4.00E+02
8.00E03
50
60
70
5.00E+02
40
Time (sec )
Time (sec )
3.50E+02
3.00E+02
2.50E+02
2.00E+02
1.50E+02
1.00E+02
6.00E+01
7.00E03
6.00E03
5.00E03
4.00E03
3.00E03
5.00E+01
4.00E+01
3.00E+01
2.00E+01
2.00E03
1.00E+01
5.00E+01
1.00E03
0.00E+00
0.00E+00
10
20
30
40
50
Time (sec)
60
0.00E+00
10
20
30
40
50
Time (sec)
60
10
20
30
40
Time (sec)
Fig. 5.5 Results of FRAPTRAN simulation for BWR power oscillation (4), different
internal gas pressure
 20 
50
60
It could be concluded that under a certain amount of undamped power oscillation cycles,
the cladding would not fail and the fuel integrity is conserved.
(6.1)
 21 
shown here in Fig. 6.1. In the Fig., the average cladding temperature oscillation was only
around 4C.
Cladding Average Temperature
5.85E+02
5.80E+02
5.75E+02
5.70E+02
5.65E+02
5.60E+02
5.55E+02
5.50E+02
5.45E+02
0
10
20
30
Time (sec)
40
50
60
7. Summary
In summary, the fuel integrity under power oscillations with ATWS was examined in this
work by using FRAPTRAN code. Based on the analysis, the following conclusions could
be drawn:
(1) FRAPTRAN code could be used to analyze fuel performance under BWR power
oscillations. The FRAPTRAN code results were in good agreement with FK11 test
in Japan.
(2) The fuel deformation was mainly caused by PCMI and was roughly proportional to
the fuel enthalpy. Enhanced cladding deformation due to ratcheting was not found.
(3) The cladding could satisfy the design criterion (stressstrain criterion) in ASME Code
under power oscillation conditions, which means it could maintain the fuel integrity.
(4) The fission gas release (internal gas pressure) was of importance to tests in cold zero
power conditions, but not the normal power operating conditions.
(5) Cladding thermal fatigue is not an issue under power oscillations, unless dryout takes
place.
 22 
It should be pointed out there that the analysis in current study is based on consideration
of stressstain criterion and thermal fatigue. Cladding creep analysis, fission gas release
model, boiling transition (dryout), and stress corrosion cracking are not considered in the
analysis. These may weak the conclusion drawn above.
Reference:
1. F. DAuria et al., State of the Art Report on Boiling Water Reactor Stability,
OCDE/GD(97)13, 1997
2. Hanggi, P., "Investigating BWR Stability with a New Linear FrequencyDomain
Method and Detailed 3D Neutronics," a Ph.D. dissertation submitted to the Swiss
Federal Institute of Technology, Zurich, 2001.
3. Zhao, Jiyun. Stability Analysis of Supercritical Water Cooled Reactors. Ph.D. Thesis,
Department of Nuclear Science and Engineering, Massachusetts Institute of
Technology, Cambridge, Massachusetts, September 2005.
4. T. NakaMura, et al., Irradiated Fuel Behavior under Power Oscillation Conditions,
Journal of Nuclear Science & Technology, Vol. 40, No. 5, p. 325333, 2003
5. T. Fukahori, et. al., Development of an analysis system for fuel integrity evaluation
and application to BWR core instabilities, 2005 Water Reactor Fuel Performance
Meeting, Paper umber: 1091, Track No. #53.
6. Frost, Brian R. T., Nuclear Fuel Elements, Pergamon Press, New York (1982) pp 5770, 7778, 184190.
7. Ygnik, S. K., Secondary Hydriding of Defected ZircaloyClad Fuel Rods, Electric
Power Research Institute, Inc., Palo Alto, CA (1993) pp 212.
8. NRC NUREG/CR6739, FRAPTRAN: A Computer Code for the Transient Analysis
of Oxide Fuel Rods, August 2001
9. Gialdi, E., et al, Prog. Nucl. Energy 15, 447, 1985.
10. ASME, 1998 ASME Boiler & Pressure Vessel Code III Devision 1, 1998
11. NRC, 10 CFR Part 50 General Design Criterion,
http://www.nrc.gov/readingrm/doccollections/cfr/part050/part050appa.html
12. H. Garkisch, Design Limits Input for Performance Evaluations, Westinghouse
Electric Co., Talked given at Univ. of Wisconsin, July 2002
13. Michael A. Speer, LWR Fuel Pin Performance During Burnup,
http://www.nuc.berkeley.edu/thyd/ne161/mspeer/uo2fuel.html
 23 
I.
Nomenclature
a)
ei
i
gij
b)
u, v, w
c)
si
t ij
d)
E
G
n
e)
II.
StrainDisplacement Relationships
a)
Cartesian Coordinates
u
x
ex =
ey =
ez =
v
y
w v
+
y z
g xz =
u w
+
z x
er =
u
r
g rQ =
v 1 u v
+
r r Q r
eQ =
u 1 v
+
r r Q
gQ z =
1 w v
+
r Q z
g zr =
u w
+
z r
ez =
III.
g yz =
Cylindrical Coordinates
v u
+
x y
w
z
b)
g xy =
w
z
StrainStress Relationships
ex =
1
[s x  v(s y + s z )]
E
ey =
1
[s y  v(s x + s z )]
E
ez =
1
[s z  v(s x + s y )]
E
t xy = Gg xy , t yz = Gg yz , t xz = Gg xz
b) Cylindrical Coordinates
er =
1
[s r  v(s Q + s z )]
E
eQ =
1
[s Q  v(s r + s z )]
E
ez =
1
[s z  v(s r + s Q )]
E
IV.
t rQ = Gg rQ , t rz = Gg rz , t Qz = Gg Qz
StressStrain Relationships
a)
Cartesian Coordinates
sx =
E
[(v 1)ex  v(ey + ez )]
(v + 1)(2v 1)
sy =
E
[(v 1)ey  v(ex + ez )]
(v + 1)(2v 1)
sz =
E
[(v 1)ez  v(ez + ey )]
(v + 1)(2v 1)
b)
Cylindrical Coordinates
sr =
E
[(v 1)er  v(eQ + ez )]
(v + 1)(2v 1)
sQ =
E
[(v 1)eQ  v(er + ez )]
(v + 1)(2v 1)
sz =
E
[(v 1)ez  v(er + eQ )]
(v + 1)(2v 1)
3
V.
Equilibrium Equations
a)
Cartesian Coordinates
s x t xy t xz
+
+
+X =0
x
y
z
s y t xy t yz
+
+
+Y = 0
y
x
z
t yz
t z t xz
+
+
+Z =0
z
x
y
b)
Cylidrical Coordinates
2t rQ t zQ
1 s Q t rQ
+
+
+
+Q=0
r Q
r
r
z
s r 1 t rQ s r  s Q t r z
+
+
+
+R=0
r r Q
r
z
s z 1 t zQ
t rz
t
+
+
+ rz + Z = 0
z
r Q
r
r
VI.
6 strains
3 displacements
6 stresses
15 unknowns
eQ
eQ
v
DQ]  [rDQ + v]
Q
rDQ
lim
=
DQ 0
lim
[(r + u)DQ + v +
lim
DQ 0
u 1 v
u 1 v
[ +
]= +
r r Q
r r Q
eQ =
er =
lim
Dr 0
u+
u
Dr  u
u
r
=
Dr
r
5
NOTE L.4
25
45
100
Max theory
DE theory
Limiting
Mohr theory Points
Max theory
57.7
60
Load Line
100
Shear Diagonal
2
22.312
NOTE L.4
L. Wolf
Revised by
M.S. Kazimi and N.E. Todreas
12/23/03
TABLE OF CONTENTS
1.
Page
Definition of Concepts.......................................................................................................... 1
2.
3.
4.
5.
Thermal Stress.......................................................................................................................31
6.
6.4 Maximum Shear Stress Theory (The Coulomb, later Tresca Theory)...........................44
Pressure Vessels.........................................................................................................58
References ...................................................................................................................................61
ii
1.
DEFINITION OF CONCEPTS
Structural mechanics is the body of knowledge describing the relations between external
forces, internal forces and deformation of structural materials. It is therefore necessary to clarify
the various terms that are commonly used to describe these quantities. In large part, structural
mechanics refers to solid mechanics because a solid is the only form of matter that can sustain
loads parallel to the surface. However, some considerations of fluidlike behavior (creep) are also
part of structural mechanics.
Forces are vector quantities, thus having direction and magnitude. They have special names
(Fig. 1) depending upon their relationship to a reference plane:
a)
b)
c)
z
Compressive
z
Tensile
Shear
Note L.4
Page 2
Fig. 2
F1
Fn
F2
Figure 3.
Note L.4
Page 3
F1 + F2 + ... + Fn =
n Fn
= 0
(1.1a)
The total moment of all the forces in Fig. 3 about an arbitrary point 0 must be zero since the vector
sum of forces acting on each particle is zero. Again, observe that internal forces occur in selfcanceling pairs along the same line of action. This leads to the second condition for equilibrium:
the total moment of all the external forces about an arbitrary point 0 must be zero.
r1 F1 + r2 F2 + ... + rn Fn =
rn F n = 0
(1.1b)
where rn extends from point 0 to an arbitrary point on the line of action of force Fn .
In Fig. 4A, an arbitrary plane, aa, divides a body in equilibrium into regions I and II. Since the
force acting upon the entire body is in equilibrium, the forces acting on part I alone must be in
equilibrium.
In general, the equilibrium of part I will require the presence of forces acting on plane aa.
These internal forces applied to part I by part II are distributed continuously over the cut surface,
but, in general, will vary over the surface in both direction and intensity.
n
II
dF
a
a
dA
a
Point 0
Point 0
Note L.4
Page 4
r
T = stress at point 0 on plane aa whose normal is n passing through point 0
dF where dF is a force acting on area dA.
= lim
dA
dA0
n
[Reference 1 uses the notation T to introduce the concept that T is a stress vector]
NOTE: Stress is a point value.
The stress acting at a point on a specific plane is a vector. Its direction is the limiting direction
of force dF as area dA approaches zero. It is customary to resolve the stress vector into two
components whose scalar magnitudes are:
normal stress component : acting perpendicular to the plane
shear stress component : acting in the plane.
1.1
The selection of different cutting planes through point 0 would, in general, result in stresses
differing in both direction and magnitude. Stress is thus a secondorder tensor quantity, because
not only are magnitude and direction involved but also the orientation of the plane on which the
stress acts is involved.
NOTE: A complete description of the magnitudes and directions of stresses on all possible
planes through point 0 constitutes the state of stress at point 0.
NOTE: A knowledge of maximum stresses alone is not always sufficient to provide the best
evaluation of the strength of a member. The orientation of these stresses is also
important.
In general, the overall stress state must be determined first and the maximum stress values
derived from this information. The state of stress at a point can normally be determined by
computing the stresses acting on certain conveniently oriented planes passing through the point of
interest. Stresses acting on any other planes can then be determined by means of simple,
standardized analytical or graphical methods. Therefore, it is convenient to consider the three
mutually perpendicular planes as faces of a cube of infinitesimal size which surround the point at
which the stress state is to be determined.
Figure 5 illustrates the general state of 3D stress at an arbitrary point by illustrating the stress
components on the faces of an infinitesimal cubic element around the point.
Note L.4
Page 5
y
0
yz
y
yx
zy
zx xz
xy
Figure 5.
Stress Element Showing General State of 3D Stress at a Point Located Away from
the Origin.
+ yx
 yx
x
Figure 6. Definition of Positive and Negative yx.
Note L.4
Page 6
xz
yz
z
9components
(1.2)
(1.3)
Therefore, six quantities are sufficient to describe the stresses acting on the coordinate planes
through a point, i.e., the triaxial state of stress at a point. If these six stresses are known at a point, it
is possible to compute from simple equilibrium concepts the stresses on any plane passing through
the point [Ref. 2, p. 79].
1.2
The tensor S becomes a symmetric tensor if Eq. 1.3 is introduced into Eq. 1.2. A fundamental
property of a symmetrical tensor (symmetrical about its principal diagonal) is that there exists an
orthogonal set of axes 1, 2, 3 (called principal axes) with respect to which the tensor elements are all
zero except for those on the principal diagonal:
S' =
1 0 0
0 2 0
0 0 3
(1.4)
Hence, when the tensor represents the state of stress at a point, there always exists a set of mutually
perpendicular planes on which only normal stress acts. These planes of zero shear stress are called
principal planes, the directions of their outer normals are called principal directions, and the stresses
acting on these planes are called principal stresses. An element whose faces are principal planes is
called a principal element.
For the general case, the principal axes are usually numbered so that:
1 2 3
1.3
The concept of strain is of fundamental importance to the engineer with respect to the
consideration of deflections and deformation.
A component may prove unsatisfactory in service as a result of excessive deformations,
although the associated stresses are well within the allowable limits from the standpoint of fracture
or yielding.
Strain is a directly measurable quantity, stress is not.
Note L.4
Page 7
dy
dx
y
x
x = lim dx
x
x0
dy
y = lim y
y0
y
dx
dx
yx = lim dx
y
y0
= tan
x
Figure 8.
Subscript notation for strains corresponds to that used with stresses. Specifically,
yx: shear strain resulting from taking adjacent planes perpendicular to the yaxis and
displacing them relative to each other in the xdirection (Fig. 9).
x, y: normal strains in x and ydirections, respectively.
yx > 0
initial
deformed
rotated clockwise
xy < 0
initial
deformed
rotated counterclockwise
Note L.4
Page 8
Sign conventions for strain also follow directly from those for stress: positive normal stress
produces positive normal strain and vice versa. In the above example (Fig. 7), x > 0, whereas
y < 0. Adopting the positive clockwise convention for shear components, xy < 0, yx > 0. In
Fig. 8, the shear is yx and the rotation is clockwise.
NOTE: Half of xy, xz, yz is analogous to xy , xz and yz , whereas x is analogous to x.
T =
1 xy
2
1 yx
2
1 zx
2
1 xz
2
1 yz
2
1 zy
2
It may be helpful in appreciating the physical significance of the fact that is analogous to /2
rather than with itself to consider Fig. 10. Here it is seen that each side of an element changes in
slope by an angle /2.
2
Figure 10. State of Pure Shear Strain
1.4
Plane Stress
In Fig. 11, all stresses on a stress element act on only two pairs of faces. Further, these
stresses are practically constant along the zaxis. This twodimensional case is called biaxial stress
or plane stress.
y
y
yx
0
xy
z
A
Note L.4
Page 9
The stress tensor for stresses in horizontal (x) and vertical (y) directions in the plane xy is:
S = x xy
yx y
When the angle of the cutting plane through point 0 is varied from horizontal to vertical, the
shear stresses acting on the cutting plane vary from positive to negative. Since shear stress is a
continuous function of the cutting plane angle, there must be some intermediate plane in which the
shear stress is zero. The normal stresses in the intermediate plane are the principle stresses. The
stress tensor is:
S' = 1 0
0 2
where the zero principal stress is always 3. These principal stresses occur along particular axes in
a particular plane. By convention, 1 is taken as the larger of the two principal stresses. Hence
1 2
3 = 0
In general we wish to know the stress components in a set of orthogonal axes in the xy plane,
but rotated at an arbitrary angle, , with respect to the x, y axes. We wish to express these stress
components in terms of the stress terms referred to the x, y axes, i.e., in terms of x, y, xy and the
angle of rotation . Fig. 12 indicates this new set of axes, x1 and y1, rotated by from the original
set of axes x and y. The determination of the new stress components x1 , y1 , x1 y1 is
accomplished by considering the equilibrium of the small wedge centered on the point of interest
whose faces are along the xaxis, yaxis and perpendicular to the x1 axis. For the general case of
expressing stresses x1 and x1 y1 at point 0 in plane x1, y1 in terms of stresses x, y, xy and the
angle in plane x, y by force balances in the x1 and y1 directions, we obtain (see Ref. 1, p. 217):
x1 = x cos2 + y sin2 + 2xy sin cos
x1 y1 = (y  x) sin cos + xy (cos2  sin2 )
y1 = x sin2 + y cos2  2xy sin cos
where the angle is defined in Fig. 12.
y
y1
y
x1
x1
x
x
(A)
(B)
(1.5a)
(1.5b)
(1.5c)
Note L.4
Page 10
These are the transformation equations of stress for plane stress. Their significance is that
stress components x1 , y1 and x1 y1 at point 0 in a plane at an arbitrary angle to the plane xy are
uniquely determined by the stress components x, y and xy at point 0.
Eq. 1.5 is commonly written in terms of 2. Using the identities:
sin2 =
1  cos 2
;
2
we get
cos2 =
1 + cos 2
;
2
x + y
 y
+ x
cos 2 + xy sin 2
2
2
 y
x1 y1 =  x
sin 2 + xy cos 2
2
+ y x  y
y1 = x
cos 2  xy sin 2
2
2
x1 =
(1.6a)
(1.6b)
(1.6c)*
The orientation of the principal planes, in this two dimensional system is found by equating x1 y1 to
zero and solving for the angle .
1.5
Mohr's Circle
Equations 1.6a,b,c, taken together, can be represented by a circle, called Mohrs circle of stress.
To illustrate the representation of these relations, eliminate the function of the angle 2 from
Eq. 1.6a and 1.6b by squaring both sides of both equations and adding them. Before Eq. 1.6a is
squared, the term (x + y)/2 is transposed to the left side. The overall result is (where plane y1x1
is an arbitrarily located plane, so that x1 is now written as and x1 y1 as ):
[  1/2 (x + y)]2 + 2 = 1/4 (x  y)2 + 2xy
(1.7)
Now construct Mohrs circle on the and plane. The principal stresses are those on the axis
when = 0. Hence, Eq. 1.7 yields the principal normal stresses 1 and 2 as:
1,2 =
x + y
center of circle
2xy +
x  y
2
(1.8)
radius
The maximum shear stress is the radius of the circle and occurs in the plane represented by a
vertical orientation of the radius.
* y1 can also be obtained from x1 by substituting + 90 for .
Note L.4
Page 11
max =
2xy +
x  y
2
= 1 1  2
2
(1.9)
The angle between the principal axes x1 y1 and the arbitrary axis, xy can be determined from
Eq. 1.6b by taking x1 y1 as principal axes. Hence, from Eq. 1.6b with x1 y1 = 0 we obtain:
2xy
2 = tan1
(1.10)
x  y
From our wedge we know an arbitrary plane is located an angle from the general x, yaxis set
and ranges from zero to 180. Hence, for the circle with its 360 range of rotation, planes
separated by in the wedge are separated by 2 on Mohrs circle. Taking the separation between
the principal axis and the xaxis where the shear stress is xy, we see that Fig. 13A also illustrates
the relation of Eq. 1.10. Hence, Mohrs circle is constructed as illustrated in Fig. 13A.
x  y
x + y
( y , yx)y
( 2 , 0)
locus of yx
( 1 , 0)
x( x , xy)
2xy +
locus of xy
x  y
2
Figure 13. (A) Mohrs Circle of Stress; (B) Shear Stress Sign Convection for Mohrs Circle.*
When the principal stresses are known and it is desired to find the stresses acting on a plane
oriented at angle from the principal plane numbered 1 in Fig. 14, Eq. 1.6 becomes:
and
= 1 + 2 1  2 cos 2
2
2
(1.11)
=  1  2 sin 2
2
(1.12)
For shear stresses, the sign convection is complicated since yx and xy are equal and of the same sign (see Figs 5
and 6, and Eq. 1.3), yet one must be plotted in the convention positive ydirection and the other in the negative ydirection (note the axis replaces the yaxis in Fig. 13A). The convention adopted is to use the four quadrants of the
circle as follows: positive xy is plotted in the fourth quadrant and positive yx in the second quadrant.
Analogously, negative xy lies in the first quadrant. Further, the shear stress at 90 on the circle is positive while
that at 270 is negative. [Ref. 1, p. 220]
Note that the negative shear stress of Eq. 1.12 is plotted in the first and third quadrants of Fig. 14, consistent with
this sign convention.
*
Note L.4
Page 12
1 + 2
2
2
1  2 cos 2
2
1  2 sin 2
2
1
Figure 14. Mohrs Circle Representation of Eqs. 1.11 & 1.12.
Example 1:
Consider a cylindrical element of large radius with axial (x direction) and torque loadings
which produce shear and normal stress components as illustrated in Fig. 15. Construct Mohrs
circle for this element and determine the magnitude and orientation of the principal stresses.
From this problem statement, we know the coordinates x, xy are 60, 40 ksi and the
coordinates y, yx are 0, 40 ksi. Further, from Eq. 1.10, 2 = tan1 2(40)/(60) = 53. From these
values and the layout of Fig. 13, we can construct Mohrs circle for this example as shown in
Fig. 16. Points 1 and 2 are stresses on principal planes, i.e., principal stresses are 80 and 20 ksi
and points S and S' are maximum and minimum shear stress, i.e., 50 and 50 ksi. These results
follow from Eqs. 1.8 and 1.9 as:
From Eq. 1.8
From Eq. 1.9
From Fig. 13A
2
402 + 60  0 = 80, 20 ksi
2
2
max = 402 + 60  0 = 50 ksi
2
at plane of max = (60 + 0)/2 = 30 ksi, i.e., at S and S'
1 , 2 = 60 + 0
2
S (30, 50)
y (0, 40)
F
xy = T = 40 ksi
A
x = F = 60 ksi
A
y = 0
37
(20, 0) 2
0 3
53= 2
(80, 0)
x (60, 40)
S' (30, 50)
Note L.4
Page 13
The principal element, i.e., its faces are principal planes, and the maximum shear element are
shown in Fig. 17.
NOTE: The angle between points y and S is the complement of 53 or 37, and plane y is rotated
37/2 counterclockwise from plane S.
NOTE: The procedure for determining the orientation of these elements: point 1 is 53 counter
clockwise from point x on the circle; hence the plane represented by 1 is half of 53
counterclockwise from plane x. Similarly, the S plane is 143/2 counterclockwise of
plane x, or 37/2 clockwise of plane y.
y axis
26.5
Plane perpendicular
to x axis
Plane y
1 = 80 ksi
S axis
37
2 = 30 ksi
Plane y
max= 50 ksi
(2) (1)
(1)
y axis
143
2
x axis
x axis
37
2 = 30 ksi
(2)
S axis
Plane x
2 =  20 ksi
Principal Element
Plane x
max= 50 ksi
Figure 18 illustrates the orientation of one of the eight octahedral planes which are associated
with a given stress state. Each of the octahedral planes cut across one of the corners of a principal
element, so that the eight planes together form an octahedron.
The stresses acting on these planes have interesting and significant characteristics. First of all,
identical normal stresses act on all eight planes. By themselves the normal stresses are therefore
said to be hydrostatic and tend to compress or enlarge the octahedron but not distort it.
Note L.4
Page 14
oct = 1 + 2 + 3
3
(1.13a)
Shear stresses are also identical. These distort the octahedron without changing its volume.
Although the octahedral shear stress is smaller than the highest principal shear stress, it constitutes
a single value that is influenced by all three principal shear stresses. Thus, it is important as a
criterion for predicting yielding of a stressed material.
oct = 1 1  2 2 + 2  3 2 + 3  1 2 1/2
3
In cases in which x, y, z, xy, xz and yz are known:
+ y + z
oct = x
3
oct = 1
x  y 2 + y  z 2 + z  x 2 + 6 2xy + 2xz + 2yz
3
(1.13b)
(1.14a)
1/2
(1.14b)
oct
1
3
oct
Having observed the correspondence between strain and stress, it is evident that with suitable
axis transformation one obtains an expression for the strain tensor T ' which is identical to that of
stress tensor S' except that the terms in the principal diagonal are 1, 2, 3. Hence, recalling from
Section 1.3 that /2 corresponds to , strain relations analogous to Eqs. 1.8, 1.9 and 1.10 can be
written as follows:
1 , 2 =
x + y
max = 2
1 xy
2
1 xy
2
2 = tan1
xy
x  y
x  y
2
x  y
2
(1.15)
(1.16)
(1.17)
Note L.4
Page 15
Conclusion
The preceding sections dealt separately with the concepts of stress and strain at a point. These
considerations involving stresses and strains separately are general in character and applicable to
bodies composed of any continuous distribution of matter.
NOTE: No material properties were involved in the relationships, hence they are applicable to
water, oil, as well as materials like steel and aluminum.
Note L.4
Page 16
Note L.4
Page 17
2.1
Let us consider the various components of stress one at a time and add all their strain effects.
For a uniaxial normal stress in the x direction, x, the resulting normal strain is
x = x
E
(2.1)
(2.2)
This fixed fraction is called Poissons ratio, . Analogous results are obtained from strains due to
y and z.
The shearstress components produce only their corresponding shearstrain components that
are expressed as:
zx = zx ,
xy = xy ,
yz = yz
(2.3a,b,c)
G
G
G
where the constant of proportionality, G, is called the shear modulus.
Note L.4
Page 18
For a linearelastic isotropic material with all components of stress present:
x = 1
x  y + z
E
y = 1 y  z + x
E
z = 1 z  x + y
E
xy = xy
G
yz = yz
G
zx = zx
G
(2.4a)
(2.4b)
(2.4c)
(2.5a) same as (2.3)
(2.5b)
(2.5c)
E
.
2 (1 + )
(2.6)
Hence,
2 (1 + )
xy
E
2 (1 + )
=
yz
E
2 (1 + )
=
zx .
E
xy =
(2.7a)
yz
(2.7b)
zx
(2.7c)
Equations 2.4 and 2.5 may be solved to obtain stress components as a function of strains:
E
1  x + y + z
(1 + ) (1  2)
E
y =
1  y + z + x
(1 + ) (1  2)
z =
1  z + x + y
(1 + ) (1  2)
E
xy =
= Gxy
2 (1 + ) xy
E
yz =
= Gyz
2 (1 + ) yz
E
zx =
= Gzx .
2 (1 + ) zx
x =
(2.8a)
(2.8b)
(2.8c)
(2.9a)
(2.9b)
(2.9c)
Note L.4
Page 19
x =
+
+ y + z
(1 + ) x
(1  2) x
E
y =
+
+ y + z
(1 + ) y
(1  2) x
E
z =
+
+ y + z .
(1 + ) z
(1  2) x
(2.10a)
(2.10b)
(2.10c)
For special case in which the x, y, z axes coincide with principal axes 1, 2, 3, we can simplify the
strain set, Eqs. 2.4 and 2.5, and the stress set Eqs. 2.8 and 2.9, by virtue of all shear strains and
shear stresses being equal to zero.
1 = 1 1  2 + 3
(2.11a)
E
2 = 1 2  3 + 1
(2.11b)
E
3 = 1 3  1 + 2
(2.11c)
E
1 =
1  1 + 2 + 3
(2.12a)
(1 + ) (1  2)
2 =
1  2 + 3 + 1
(2.12b)
(1 + ) (1  2)
3 =
1  3 + 1 + 2 .
(2.12c)
(1 + ) (1  2)
For biaxialstress state, one of the principal stresses (say 3) = 0, Eqs. 2.11a,b,c become:
1 = 1 1  2
E
2 = 1 2  1
E
3 =  1 + 2 .
E
(2.13a)
(2.13b)
(2.13c)
In simplifying Eqs. 2.12a,b,c for the case of 3 = 0, we note from Eq. 2.12c that for 3 to be zero,
3 = 
+ 2 .
1  1
Substituting this expression into the first two of Eqs. 2.12a,b,c gives:
E
1 =
1 + 2
1  2
E
2 =
2 + 1
1  2
3 = 0 .
(2.14)
(2.15a)
(2.15b)
(2.15c)
Note L.4
Page 20
In case of uniaxial stress Eqs. 2.13 and 2.15 must, of course reduce to:
1 = 1 1
E
2 = 3 =  1
E
1 = E1
2 = 3 = 0 .
2.2
(2.16a)
(2.16b)
(2.17a)
(2.17b)
(2.18)
NOTE: Hydrostatic compressive stress exists within the fluid of a pressurized hydraulic cylinder,
in a rock at the bottom of the ocean or far under the earths surface, etc.
Hydrostatic tension can be created at the interior of a solid sphere by the sudden
application of uniform heat to the surface, the expansion of the surface layer subjecting
the interior material to triaxial tension. For 1 = 2 = 3 = , Eqs. 2.11a,b,c show that:
1 = 2 = 3 = = 1  2 .
E
This state of uniform triaxial strain is characterized by the absence of shearing deformation; an
elemental cube, for example, would change in size but remain a cube. The size of an elemental cube
initially of unit dimension would change from 13 to (1 + )3 or to 1 + 3 + 3 2 + 3. If we
consider normal structural materials, is a quantity sufficiently small, so that 2 and 3 are
completely negligible, and the volumetric change is from 1 to 1 + 3. The volumetric strain, V/V,
is thus equal to 3 or to:
V = 3 = 3 (1  2) .
(2.19)
V
E
Hence,
E
k =
.
(2.20)
3
3 (1  2)
Now 0.5, so that k cannot become negative. A simple physical model of a representative
atomic crystalline structure gives:
= 1/3 ,
(2.21)
k = E .
(2.22)
so that
Note L.4
Page 21
3.
3.1
Stresses
Stresses in a ThinWalled Cylinder:
Consider a portion of the cylinder sufficiently remote from the ends to avoid end effects.
The equation of equilibrium of radial forces (Fig. 19) acting on the element is:
pi ri ddL = 2t,av tdL sin
yielding
t,av =
d
2
(3.1)
pi ri
.
t
(3.2)
when sin (d/2) is approximated by d/2, which is valid for small d. This is a correct expression
for the average tangential stress (hoop stress) regardless of cylinder thickness.
NOTE: It is only when this equation is used as an approximation for the maximum tangential
stress, t,max , that the thickness t must be very small in comparison with the cylinder
diameter.
Equation 3.1 can be modified to give a good quick estimate of the maximum tangential stress due to
internal pressure for cylinders of moderate thickness by replacing ri with rav, which we label R:
R = ri + t .
2
Thus,
t,max
pi
t
(3.3)
pi R
t
(3.4)
t,avtdL
d
dL
p i r i ddL
d/2
2ri
2ro
Figure 19. Radial Forces in an Internally Pressurized ThinWalled Cylinder.
Note L.4
Page 22
The following are errors in estimating t,max from Eqs. 3.2 or 3.4 compared to the thickwalled
cylinder results:
t/ri
0.1
0.2
0.5
1.0
% Error Using
Eq. 3.2
5 (low)
10 (low)
23 (low)
40 (low)
% Error Using
Eq. 3.4
0.5 (low)
1.0 (low)
3.8 (low)
10.2 (low)
If the ends are closed, the cylinder is also subjected to an axial force of magnitude pi r2i . This is
distributed over an area of cross section that can be expressed as:
A = r2o  r2i = 2Rt .
(3.5)
pi r2i
p r2
= ii .
2rav t
r2o  r2i
(3.6)
For the thinwalled case, ri rav R since ri = R(1  0.5t/R). Hence, Eq. 3.6 reduces to:
a,av =
pi R
.
2t
(3.7)
3.2
(3.8)
The deformation of the diameter of a thinwalled cylinder or sphere caused by the internal
pressure is to be determined. Note that the state of stress at the inner surface is triaxial and is not
plane stress. The principal stresses of the inner surface are 1 = , 2 = z and 3 = p.
However, 3 = p is so small compared to 1 and 2 that it can be neglected when considering
strains. Thus, the state of stress in thinwalled pressure vessels is usually considered to be plane
stress.
The deformation of such pressure vessels is influenced by the tangential and axial (transverse
for the sphere) stresses, and hence use must be made of the following relations obtained from Eqs.
2.4, 2.5 and 2.8, respectively, with y = 0.
Note L.4
Page 23
x = x  z
E
E
y =  x + z
E
z = z  x
E
E
= xy
G
E
x =
x + z
1  2
E
z =
z + x
1  2
(3.9a)
(3.9b)
(3.9c)
(3.9d)
(3.10a)
(3.10b)
to express Hookes law for plane stress. Equations 3.10a,b are obtained from Eqs. 2.8a and 2.8c
upon inspection of y evaluated from Eq. 2.8b with y taken as zero.
Let , z, and z represent the tangential and axial stress and strain, respectively, in the wall.
The substitution of these symbols in Eqs. 3.9a,b,c, i.e., x and z = z, gives:
 z
E
E
z = z 
.
E
E
(3.11)
(3.12)
ClosedEnd Cylinder:
For the strains in the closedend cylinder, the values of and z as derived in Eqs. 3.4
and 3.8, respectively, are substituted into Eqs. 3.11 and 3.12 to give:
pR
pR
pR
= 1

=
(2  )
E t
2t
2Et
(3.13)
pR
pR
pR
z = 1

=
(1  2) .
E 2t
t
2Et
(3.14)
Let the change in length of radius R be r when the internal pressure is applied. Then the change in
length of the circumference is 2r. But the circumferential strain, , is, by definition, given by the
following equation:
= 2r = r .
2R
R
(3.15)
pR2
(2  ) .
2Et
(3.16)
Note L.4
Page 24
The change in length, l, for a closedend cylinder is equal to:
l = zl
or
l =
pRl
(1  2)
2Et
(3.17)
(3.18)
Sphere:
z = =
pR
2t
(3.19)
(3.20)
3.3
pR2
(1  ) .
2Et
(3.21)
Figure 20 illustrates a cylinder closed by thinwalled hemispherical shells. They are joined
together at AA and BB by rivets or welds. The dashed lines show the displacements due to internal
pressure, p. These displacements are given by Eqs. 3.16 and 3.21 as:
pR2
(2  )
2Et
pR2
Rs =
(1  )
2Et
Rc =
(3.22a)
(3.22b)
The value of Rc is more than twice that of Rs for the same thickness, t, and as a result the
deformed cylinder and hemisphere do not match boundaries. To match boundaries, rather large
shearing forces, V, and moments, M, must develop at the joints as Fig. 20 depicts (only those on the
cylinder are shown; equal but opposite shears and moments exist on the ends of the hemispheres).
This shear force is considerably minimized in most reactor pressure vessels by sizing the
hemisphere thickness much smaller than the cylinder thickness. For example, if the hemisphere
thickness is 120 mm and the cylinder thickness is 220 mm, for = 0.3, then the ratio Rc to Rs is
Rc = ts 2  = 1.32
tc 1 
Rs
Note L.4
Page 25
Rs
Rs
V
V Shearing Forces
Rc
V
Figure 20. Discontinuities in Strains for the Cylinder (Rc) and the Sphere (Rs).
The general solution of the discontinuity stresses induced by matching cylindrical and
hemisphere shapes is not covered in these notes. However, a fundamental input to this solution is
the matching of both displacements and deflections of the two cylinders of different wall
thicknesses at the plane they join. Thus, the relationship between load, stress, displacement and
deflection for cylinders and hemispheres are derived. These relations are presented in Tables 1
and 2, respectively.
Table 1. Relationship Between Load, Stresses, Displaced and Deflection for the Cylinder
(equations cited are from Ref. 2)
Stresses
t , l, r
, z, r
pR pR
p
,
, t
2t
2
Load
p
Mo
Edge
Moment
per Unit
Perimeter
Length
6Mo , 6Mo , 0
t2
t2
2 DR
EMo
2
(Eq. 7.3.22,
p. 7.312)
(Eq. 7.3.21,
p. 7.311)
Note: u o =
Mo
; so first term
2 2 D
Eu o
.
R
Displacement
Deflection
uo
pR2
2  + t
2tE
R
Mo
2
2 D
(Eq. 7.3.18, p. 7.3.7)
units
F L/L
= L
1 F L3
L2 L2
0
 Mo
D
(Eq. 7.3.19, p. 7.3.8)
dimensionless
Note L.4
Page 26
Table 1. Relationship Between Load, Stresses, Displaced and Deflection for the Cylinder (contd)
EQo
Qo
Qo
, 0, 0
R2 D
2 D
(Eq. 7.3.22,
(Eq. 7.3.18, p. 7.3.7)
p. 7.312)
units F/L = L
Qo
1 FL
Note: u o = 3 ; so first term
2 D
L3
u E
= o .
R
where Plate Flexural Rigidity:
Edge Shear
Force
per Unit
Perimeter
Length
D =
4
Et3
,
12 1  2
units
2
Et = 3 1 R2 t2
4R2D
F L3
L2
Qo
2
2 D
(Eq. 7.3.19, p. 7.3.8)
dimensionless
1 1
L2 L2
Table 2. Relationship Between Load, Stresses, Displaced and Deflection for the Hemisphere
Load
Stresses
t , l, r
Displacement
o
Deflection
o
pR pR
p
,
, 2t
2t
2
t
pR2
1 +
2tE
R
Mo
2 Mo 6Mo , 6Mo , 0
tR
t2
t2
2 Mo
Et
4 Mo
REt
2RQ0
Et
2 Qo
Et
Edge
Moment
per Unit
Perimeter
Length
Qo
Edge Shear
Force
per Unit
Perimeter
Length
Note: 2 Mo = o ; so first
t
E
E
o
term =
.
R
2Q0
, 0, 0
t
2Q0
= Eo .
Note:
t
R
where
= R ; 4 =
3 1 2
R2t2
Note L.4
Page 27
4.
The cylinder is axially restrained (z = 0) at either end and subjected to uniform radial pressure,
not body forces. The cylinder cross section and stresses on an element are illustrated in Fig. 21.
d
2
pi
r + dr
po
ro
ri
r
r + dr
d
2
= ur ,
r = dv  vr
dr
(4.1a,b,c)
where u is the outward radial displacement and v is the direction displacement. Since the effect of
internal pressure is to move the material in the radial direction without any rotation, both v and r
should be zero. This will be demonstrated formally from the boundary conditions.
StressEquilibrium Relations (for forces in the r and directions)
dr
2r
dr + r  = 0 ,
+
r
r = 0.
dr
dr
StressStrain Relations
)]
]
)]
1
+ z
E r
1
= ( z + r )
E
1
z = 0 = z r +
E
r =
(4.2a,b)
(4.3a)
(4.3b)
(4.3c)
Note L.4
Page 28
r
r
=0
(4.4)
(4.5a)
(4.5b)
Displacement Approach
When displacements are known, it is preferable to use this solution approach. In this case we
solve for the displacements. Proceeding, solve Eq. 4.3 for stresses and substitute them into the
stressequilibrium relations of Eq. 4.2 to obtain:
r 
d
 r
(1  ) dr +
+
+
= 0
r
r
dr
dr
dr
2r
+ r = 0
dr
(4.6a)
(4.6b)
Substituting Eq. 4.1 in Eq. 4.6 and simplification show that the displacements u and v are
governed by the equations:
d2 u + 1 du  u = d 1 d (ru) = 0
(4.7a)
r dr
dr r dr
r2
dr2
d2 v + 1 dv  v = d 1 d (rv) = 0 .
r dr
dr r dr
r2
dr2
Successive integration of these equations yields the displacement solutions
u = C1 r + Cr2 ;
v = C3 r + Cr4
2
2
(4.7b)
(4.8a,b)
where the constants of integration are to be determined by the stress boundary conditions
(Eqs. 4.5a,b). With the use of Eqs. 4.8a,b, the strain and stress solutions are obtained from Eq. 4.1
and
E
r =
(1  )r +
(4.9a)
(1 + ) (1  2)
E
=
(1  ) + r
(4.9b)
(1 + ) (1  2)
r = Gr
(4.9c)
as
r = C1  C2 ,
= C1 + C2 ,
r =  2C4
(4.10a,b,c)
2
2
2
2
r2
r
r
Note L.4
Page 29
C1  (1  2) C2
E
(1 + ) (1  2) 2
r2
C1 + (1  2) C2
E
=
(1 + ) (1  2) 2
r2
EC1
z = r +
(1 + ) (1  2)
r =  2GC4 .
r2
r =
(4.11a)
(4.11b)
(4.11c)
(4.11d)
By virtue of the boundary conditions, Eqs. 4.5a,b, the first and last of Eq. 4.11 yield the
following equations for the determination of the constants C1 and C2.
C1  (1  2) C2 =  (1 + ) (1  2) p
i
2
E
r2i
C1  (1  2) C2 =  (1 + ) (1  2) p
o
2
E
r2o
C4 = 0
The solutions for C1 and C2 are
C1 = (1 + ) (1  2) pi r2i  po r2o
2
E r2o  r2i
(1 + ) pi  po r2i r2o
C2 =
.
E r2o  r2i
(4.12a)
(4.12b)
(4.12c)
(4.13a)
(4.13b)
As to the constant C3, it remains undetermined. However, with C4 = 0, Eq. 4.8 shows that v = C3r/2
which corresponds to a rigidbody rotation about the axis of the cylinder. Since this rotation does
not contribute to the strains, C3 is taken as zero. Hence Eqs. 4.11a,b,c and Eq. 4.8a become:
r =
=
z =
u =
4.2
1
ro /ri 2  1
1
ro /ri 2  1
2
ro /ri 2
1 1 +
ro /ri 2
ro 2 p
pi  1  1
ri o
2
2
r/ri
r/ri
ro /ri 2
ro 2 p
pi  1 + 1
ri o
2
2
r/ri
r/ri
2
pi  rro po
i
 1
(1 + ) r/ri ri
ro /ri 2  1
(1  2) +
(4.14a)
(4.14b)
Plane Strain!
(4.14c)
ro /ri 2 pi
ro 2 po
 (1  2) + 1
r/ri 2 E
r/ri 2 ri E
(4.14d)
Stress Approach
In this case we solve directly for the stresses and apply the boundary conditions as before.
Thus, reexpressing the stressstrain relations (Eq. 4.3) directly as strain when z = 0 we get:
Note L.4
Page 30
r = 1 + (1  ) r 
E
1
+ (1  )  r
=
E
r
r = ,
G
(4.15a)
(4.15b)
(4.15c)
and then substituting them into Eq. 4.4 upon simplification yields
d
r
d
(1 ) r +
= 0 .
dr
dr
r
(4.16)
By the first stressequilibrium relation in Eq. 4.2, this equation is reduced to the simple form
d + = 0.
(4.17)
dr r
Integration of this equation and the second of Eq. 4.2 yields
r + = C1 ;
r =
C2
.
r2
(4.18a,b)
pi r2i  po r2o
,
r2o  r2
i
C2 = 0 ,
C3 =
pi  po r2i r2o
.
r2o  r2i
(4.20a,b,c)
With these values for the constants, Eqs. 4.18, 4.19 and the relation z = (r + ), the complete
stress solution is explicitly given. This solution can then be substituted in Eqs. 4.15a and 4.15b to
determine r and and, finally, u is determined by the relation u = r .
NOTE: Since the prescribed boundary conditions of the problem are stresses, it is evident that the
stress approach in obtaining the solution involves simpler algebra than the displacement
approach.
Note L.4
Page 31
5.
THERMAL STRESS
Let us consider the additional stresses if the cylinder wall is subjected to a temperature gradient
due to an imposed internal wall temperature, T1, and an external wall temperature, To. For this case
the following stresses and stress gradients are zero,
r ;
and
rz = 0
and
z
z
= 0
= 0
(5.1a,b,c)
(5.1d, e)
 r
= 0
r
(5.2)
= r   z + ET
r
E = E ur =  r  z + ET
Er = E
Ez = E
w
z
= z  r  + ET
(5.3a)
(5.3b)
(5.3c)
Compatibility
u
1 u  1 u
=
r r
r r
r r
(5.4)
 r  z + ET = 1r r   z + ET  1r  r  z + ET
r
= 1r r  + r r  = 1 +r r 
and therefore,
 r =  r
 r  z + ET = r r
1 + r
r
from the
Equilibrium Condition
(5.5)
(5.6)
Note L.4
Page 32
or,
or,
 E
z
r
+ E
T
r
(5.7)
r
= 0.
(5.8)
z
r
= 0
or
rz
= 0
(5.9)
r
r
r
Eliminating z/r from Eqs. 5.8 and 5.10 we get
r
r
 2
 2
(1  )
r
r
+ E
+ E
+ r + E
T
r
T
r
T
r
= 0.
+ E
(5.10)
T
r
= 0
= 0
(5.11a)
(5.11b)
By integration we get
E T = Z
(independent of r).
(1  )
Let us solve the energy equation to get the wall radial temperature distribution:
+ r +
T
r
T
+ 1r
= 0
r
1 r T = 0
r r r
T
5.1
(5.11c)
(5.12a)
(5.12b)
= br
r
(5.12c)
T = a + b lnr .
(5.13)
Stress Distribution
Radial Stress
From Eq. 5.11c
+ r +
E T = Z .
(1  )
(5.14)
Note L.4
Page 33
 r  r
= 0
(5.15)
E T = Z
(1  )
(5.16)
or by subtraction
2r + r
or
r
r
1
r2 r + E (a + blnr) = Z
r r
(1  )
(5.17)
Ear2 + Eb r2 lnr  r2
= Zr2 + B
2 (1  )
(1  ) 2
4
2
(5.18)
By integration we get
r2 r +
or
E T
r = A + B r2 2 (1  )
(5.19)
(5.20)
E T E b
= A  B 2
2
(1
)
2
(1
 )
r
(5.21)
Axial Stress
1)
w
z
= 0
ro
2)
zrdr = 0 .
z = + r  ET + Ez
E T E b
 ET + Ez
(1  ) 2 (1  )
= 2A  Eb  ET + Ez
2 (1  ) (1  )
= 2A  Eb  Ea  Eblnr + Ez
2 (1  ) (1  ) (1  )
(5.22a)
z = 2A 
(5.22b)
Note L.4
Page 34
1)
Eb  ET
2 (1  ) (1  )
(5.23)
NOTE: T = a + blnr
ro
2)
zrdr = 0 ;
or
ro
Dr ri
where
Ea  Eblnr dr = 0
(1  ) (1  )
(5.24)
D 2A  Eb/2(1  ), and
= D  Ea  Eblnr
z =
D  ET
(1  )
(1  ) (1  )
(5.25)
or
ro
Dr ri
(5.26)
or
2
Dr2 E
ar2 + br2 lnr  br
2
2 (1  )
2
2
Dr2 E
r2 T  br
2
2 (1  )
2
5.2
z =
= 0
(5.27a)
ri
= 0
(5.27b)
r2o To  r21 T1
E
 b
(1  )
2
r2o  r12
(5.27c)
r2o To  r21 T1
E
 b  T
2
(1  )
2
2
ro  r 1
(5.28)
D =
and
ro
ro
ri
Boundary Conditions
r = ro,
r = 0,
T = To
r = r1,
r = 0,
T = T1
Therefore, from Eq. 5.19 at the inner and outer boundaries, respectively:
0 = A + B  ET1
r21 2 (1  )
(5.29a)
Note L.4
Page 35
0 = A + B  ETo .
r2o 2 (1  )
(5.29b)
B =
Er2o r2
1 T1  To
2 (1  ) r2o  r21
(5.30a)
A =
E r2o To  r21 T1
.
2 (1  ) r2o  r21
(5.30b)
With regard to temperature, from Eq. 5.13, at the inner and outer boundaries, respectively:
T1 = a + blnr1,
To = 1 + blnro .
(5.31a,b)
b = T1 
r To
ln r1
(5.32a)
a = To lnr1 r T1 lnro
ln r1
(5.32b)
5.3
Final Results
r =
E
2 (1  )
E
2 (1  )
r2o r21
T1  To
r2
 T
r2o  r21
(5.33a)
r2o r21
T1  To
r2
 T1 r To  T
r2o  r21
ln r1
o
(5.33b)
r2o To  r21 T1 +
r2o To  r21 T1 +
1)
z =
2)
No axial load
z =
r2o To  r21 T1
T  T
E
 1 r o  T
2
(1  )
2
ro  r 1
2ln r1
o
(5.34)
(5.35)
Note L.4
Page 36
Note L.4
Page 37
6. DESIGN PROCEDURES
There are four main steps in a rational design procedure:
Step 1:
Determine the mode of failure of the member that would most likely take place if the loads
acting on the member should become large enough to cause it to fail.
The choice of material is involved in this first step because the type of material may
significantly influence the mode of failure that will occur.
NOTE: Choice of materials may often be controlled largely by general factors such as:
availability
cost
weight limitations
ease of fabrication
rather than primarily by the requirements of design for resisting loads.
Step 2:
The mode of failure can be expressed in terms of some quantity, for instance, the maximum
normal stress.
Independent of what the mode of failure might be, it is generally possible to associate the
failure of the member with a particular cross section location.
For the linearly elastic problem, failure can be interpreted in terms of the state of stress at the
point in the cross section where the stresses are maximum.
Therefore, in this step, relations are derived between the loads acting on the member, the
dimensions of the member, and the distributions of the various components of the state of stress
within the cross section of the member.
Step 3:
By appropriate tests of the material, determine the maximum value of the quantity associated
with failure of the member. An appropriate or suitable test is one that will produce the same action
in the test specimen that results in failure of the actual member.
Note L.4
Page 38
NOTE: This is difficult or even impossible. Therefore, theories of failure are formulated such
that results of simple tests (tension and compression) are made to apply to the more
complex conditions.
Step 4:
By use of experimental observations, analysis, experience with actual structures and machines,
judgment, and commercial and legal considerations, select for use in the relation derived in Step 2 a
working (allowable or safe) value for the quantity associated with failure. This working value is
considerably less than the limiting value determined in Step 3.
The need for selecting a working value less than that found in Step 3 arises mainly from the
following uncertainties:
1.
2.
3.
These considerations clearly indicate a need for applying a socalled safety factor in the design of a
given loadcarrying member. Since the function of the member is to carry loads, the safety factor
should be applied to the loads. Using the theory relating the loads to the quantity associated with
failure desired in Step 2 and the maximum value of the quantity associated with failure in Step 3,
determine the failure loads which we will designate Pf. The safety factor, N, is the ratio:
N = Pf = failure load
Pw
working load
NOTE: If Pf and Pw are each directly proportional to stress, then
N = f
w
The magnitude of N may be as low as 1.4 in aircraft and space vehicle applications, whereas in
other applications where the weight of the structure is not a critical constraint, N will range from 2.0
to 2.5.
6.1
This section will treat the problem of predicting states of stress that will cause a particular
material to faila subject which is obviously of fundamental importance to engineers.
Note L.4
Page 39
Materials considered are crystalline or granular in nature. This includes metals, ceramics
(except glasses) and highstrength polymers.
The reason for the importance of crystalline materials is their inherent resistance to
deformation. This characteristic is due to the fact that the atoms are compactly arranged into a
simple crystal lattice of relatively low internal energy.
In this section we neglect the following types of failures:
creep failures which occur normally only at elevated temperature,
buckling and excessive elastic deflection, and
fatigue failure which is dynamic in nature.
Thus we limit ourselves to failures which are functions of the applied loads.
Definition:
Failure of a member subjected to load can be regarded as any behavior of the member
which renders it unsuitable for its intended function.
Eliminating creep, buckling and excessive elastic deflection, and fatigue, we are left with the
following two basic categories of static failure:
1. Distortion, or plastic strainfailure by distortion is defined as having occurred when the
plastic deformation reaches an arbitrary limit. The standard 0.2% offset yield point is
usually taken as this limit.
2. Fracturewhich is the separation or fragmentation of the member into two or more parts.
I.
II. Fracture can be either brittle or ductile in nature (or a portion of both).
As tensile loading acts on an atomic structure and is increased, one of two events must
eventually happen:
Either the shear stress acting in the slip planes will cause slip (plastic deformation), or
The strained cohesive bonds between the elastically separated atoms will break down (brittle
fracture) with little if any distortion. The fractured surfaces would be normal to the applied
load and would correspond to simple crystallographic planes or to grain boundaries.
NOTE: The stress required for fracture ranges from about 1/5 to as little as 1/1000 of the
theoretical cohesive strength of the lattice structure because of submicroscopic flaws or
dislocations.
Note L.4
Page 40
Many fractures are appropriately described as being partially brittle and partially ductile,
meaning that certain portions of the fractured surface are approximately aligned with planes of
maximum shear stress and exhibit a characteristic fibrous appearance, while other portions of the
fractured surface appear granular as in the case of brittle fracture and are oriented more toward
planes of maximum tensile stress.
NOTE: Tensile fractures accompanied by less than 5% elongation are often classed as brittle. If
the elongation is > 5% elongation, then the fracture is classed as ductile.
Brittle fractures often occur suddenly and without warning. They are associated with a release
of a substantial amount of elastic energy (integral of force times deflection) which for instance may
cause a loud noise. Brittle fractures have been known to propagate great distances at velocities as
high as 5,000 fps.
Primary factors promoting brittle fracture are:
a. low temperature increases the resistance of the material to slip but not to cleavage,
b. relatively large tensile stresses in comparison with the shear stresses,
c. rapid impact rapid rates of shear deformation require greater shear stresses, and these
may be accompanied by normal stresses which exceed the cleavage strength of the
material,
d. thick sections this "size effect" has the important practical implication that tests made
with small test samples appear more ductile than thick sections such as those used in
pressure vessels. This is because of extremely minute cracks which are presumably
inherent in all actual crystalline materials.
6.2
Engineers concerned with the design and development of structural or machine parts are
generally confronted with problems involving biaxial (occasionally triaxial) stresses covering an
infinite range or ratios of principal stresses.
However, the available strength data usually pertain to uniaxial stress, and often only to uniaxial
tension.
As a result, the following question arises: If a material can withstand a known stress in
uniaxial tension, how highly can it be safety stressed in a specific case involving biaxial (or triaxial)
loading?
Note L.4
Page 41
The answer must be given by a failure theory. The philosophy that has been used in
formulating and applying failure theories consists of two parts:
1. Postulated theory to explain failure of a standard specimen. Consider the case involving a
tensile specimen, with failure being regarded as initial yielding. We might theorize that
tensile yielding occurred as a result of exceeding the capacity of the materials in one or
more respects, such as:
a)
b)
c)
2. The results of the standard test are used to establish the magnitude of the capacity chosen
sufficient to cause initial yielding. Thus, if the standard tensile test indicates a yield
strength of 100 ksi, we might assume that yielding will always occur with this material
under any combination of static loads which results in one of the following:
a) a maximum normal stress greater than that of the test specimen (100 ksi),
b) a maximum shear stress greater than that of the test specimen (50 ksi),
cf) are defined analogously to a and b.
Hence, in the simple classical theories of failure, it is assumed that the same amount of whatever
caused the selected tensile specimen to fail will also cause any part made of the materials to fail
regardless of the state of stress involved.
When used with judgment, such simple theories are quite usable in modern engineering
practice.
6.3
In a generalize form, this simplest of the various theories states merely that a material subjected
to any combination of loads will:
Note L.4
Page 42
1.
Yield whenever the greatest positive principal stress exceeds the tensile yield strength in a
simple uniaxial tensile test of the same material or whenever the greatest negative principal
stress exceeds the compressive yield strength.
2.
Fracture whenever the greatest positive (or negative) principal stress exceeds the tensile
(or compressive) ultimate strength in a simple uniaxial tensile (or compressive) test of the
same material.
NOTE: Following this theory, the strength of the material depends upon only one of the principal
stresses (the largest tension or the largest compression) and is entirely independent of the
other two.
Uniaxial
Tension Hydrostatic
Tension
Hydrostatic
Compression
Pure Shear
= S yc
Figure 22.
= S yt
Principal Mohrs Circles for Several Stress States Representing Incipient Yielding
According to Maximum Normal Stress Theory (Note, for Pure Shear 1 = 2 = ).
NOTE: Each of the circles is a principal circle for the state of stress which it represents. This
theory implies that failure (in this case yielding) occurs when and only when the principal
Mohrs circle extends outside the dashed vertical lines given by Syt and Syc.
The failure locus for the biaxial stress state for yield according to the maximum normalstress
theory is to be illustrated. 1,2 are the principal stresses. Yield will occur if either the compressive
yield strength, Syc, or the tensile yield strength, S yt, is exceeded by either of the principle stresses
1 or 2. Hence, the maximum value of + 1 is Syt, + 2 is Syt, 1 is Syc, and  2 is S yc. These
maxima are plotted in Fig. 23, thus defining the failure locus as a rectangle. Failure is predicted for
those states that are represented by points falling outside the rectangle.
Note L.4
Page 43
NOTE: If use is made of Sut and Suc instead of Syt and S yc, the theory would have predicted
failure by fracture.
NOTE: In the 3D case we have to deal with a cube.
2
Syc
Syt
Syt
1
Syc
Failure Locus
2
Figure 23. Failure Locus for the Biaxial Stress State for the Maximum NormalStress Theory.
Failure for Brittle Material
Suc
S ut
Sut
1
Suc
Note L.4
Page 44
It is often convenient to refer to an equivalent stress, Se (e), as calculated by some
particular theory.
NOTE: The equivalent stress may or may not be equal to the yield strength.
Mathematically, the equivalent stress based on the maximum stress theory is given by:
Se = i max
i = 1, 2, 3
(6.1)
Applicability of Method
Reasonably accurate for materials which produce brittle fracture both in the test specimen
and in actual service such as: Cast iron, concrete, hardened tool steel, glass [Ref. 3, Fig. 6.8].
It cannot predict failure under hydrostatic compression (the state of stress in which all three
principle stresses are equal). Structural materials, including those listed above, can withstand
hydrostatic stresses many times Suc.
It cannot accurately predict strengths where a ductile failure occurs.
6.4
The theory states that a material subjected to any combination of loads will fail (by yielding or
fracturing) whenever the maximum shear stress exceeds the shear strength (yield or ultimate) in a
simple uniaxial stress test of the same material.
The shear strength, in turn, is usually assumed to be determined from the standard uniaxial
tension test. Principle Mohrs circles for several stress states representing incipient yielding
according to maximum shear stress theory are shown in Fig. 25.
= Sys =
Sy
2
Uniaxial
Tension
Uniaxial
Compression
Pure Shear
=  S ys =
Sy
2
Figure 25. Stress States for Yielding According to the Maximum Shear Stress Theory
Note L.4
Page 45
(6.2a)
where max occurs on faces inclined at 45 to faces on which the maximum and minimum principle
stresses act. Hence, in this failure theory, it is important to recognize 1 and 2 are the maximum
and minimum principle stresses, or
max = 1 max  min .
2
In the tensile test specimen, 1 = S y, 2 = 3 = 0, and thus:
max = 1 Sy .
2
(6.2b)
(6.3)
The assumption is then made that this will likewise be the limiting shear stress for more
complicated combined stress loadings, i.e.,
max = 1 Sy = 1 max  min
2
2
(6.4)
Sy = max  min
(6.5)
or
The failure locus for the biaxial stress state for the maximum shear stress theory is shown in
Fig. 26. This locus is determined as follows. There are three principal stresses involved, 1, 2 and
3 (3 is always equal to zero). In the first quadrant, along the vertical line, 1 > 2 > 3, which
means that 1 = max and 3.= min . Thus, the value of 2 is free to be any value between 1 and
3, yielding the vertical line. Similarly, along the horizontal line in the first quadrant, 2 > 1 > 3,
which means that 2.= max.and 3.= min . Thus, in this situation, 1 is free to be any value
between 2 and 3, yielding the horizontal line. In Quadrant II, 1.is a compressive stress. Hence,
this stress is now min . Thus, one now has the situation:
2 = max > 3 = 0 > 1 = min
(6.6)
(6.7)
Note L.4
Page 46
Syt
Sy
A
E
Syc
Syt
Syc
Figure 26.
45
Sy
1
Yield strength is
Sy in tension and
compression
B
Sy
D
F
H
Figure 27.
Locus of
principal
axes
1
Pure shear
C S
y
Pure Shear State Representation on
the Failure Locus for the Maximum
ShearStress Theory
NOTE: When 1 and 2 have like signs, the failure locus is identical to that of the maximum
stress theory.
NOTE: The boundaries of all principal Mohr circles not representing failure are the two
horizontal lines Sys (or S us). This theory predicts that failure cannot be produced
by pure hydrostatic stress.
The failure locus for the biaxial stress state is shown in Fig. 27. EF represents the shear
diagonal of the 1  2 plot, since it corresponds to the equation 1 = 2 which yields Mohrs
circle with 1 = 2 = which represents pure shear in the 12 plane. GH corresponds to
1 = 2, which yields Mohrs circle as a point with = 0. Hence, GH represents the locus of
principal axes.
Applicability of Method
For ductile failure (usually yielding) steel, aluminum, brass.
conservative side.
6.5
This theory suggests that Mohrs circles be drawn representing every available test condition
and that the envelope of these circles be taken as the envelope of any and all principal Mohr circles
representing stress states on the verge of failure.
Note L.4
Page 47
Figure 28 represents what might be called Mohrs theory in its simplest form where only
uniaxial tension and compression data are available, and where the envelope is assumed to be
represented by the two tangent straight lines.
In this form, the Mohr theory is seen to be a modification of the maximum shearstress theory.
If both compression and tension data are available, the Mohr theory is obviously the better of the
two.
Uniaxial
Compression
Syc
Uniaxial
Tension
S yt
Figure 28. Mohrs Failure Theory for Uniaxial Tension and Compression
6.6
Failure will occur whenever a principal normal strain reaches the maximum normal strain in a
simple uniaxial stress test of the same material.
The principal normal strains have been written as follows in Eq. 2.4:
i = 1 i  j + k
E
which for a biaxial stress state are
1 = 1 1  2
E
2 = 1 2  1
E
For failure in a simple tensile test, Eq. 6.9 reduces to
f = f
E
where f and f are taken in the uniaxial loading direction.
(6.8)
(6.9a)
(6.9b)
(6.10)
Note L.4
Page 48
Hence, taking f as Sy, the failure criteria are
and
1  2 Sy
2  1 Sy
(6.11b)
1  2  Sy
(6.11c)
2  1  Sy
(6.11d)
(6.11a)
where failure is predicted if any one of the relations of Eq. 6.11 are satisfied.
NOTE: Unlike the previously considered theories, the value of the intermediate principal stress
influences the predicted strength.
The graphical representation of this failure theory is presented in Fig. 29.
This theory predicts failure in hydrostatic states of stress, i.e., ductile, which is not in agreement
with experimental evidence plus does not work well for brittle material failures. It is of historical
but not current importance.
2
Sy
Sy
Sy
1
Yield strength is Sy
in tension and
compression.
Sy
Figure 29. The Failure Locus for the Maximum Normal Strain Theory (for fixed ).
6.7
The total amount of elastic energy absorbed by an element of material is the proper criterion
for its yielding. It is a forerunner to the important maximum distortionenergy theory discussed
next.
Note L.4
Page 49
6.8
Given a knowledge of only the tensile yield strength of a material, this theory predicts ductile
yielding under combined loading with greater accuracy than any other recognized theory. Where
the stress involved is triaxial, this theory takes into account the influence of the third principal
stress.
NOTE: Its validity is limited to materials having similar strength in tension and compression.
Equations can be developed from at least five different hypotheses! The most important of these
relate to octahedral shear stress and distortion energy. [see Ref. 3, p. 139 for a derivation based on
direct evaluation of distortion energy.]
We consider this theory as the maximum octahedralshearstress theory, i.e., yielding will
occur whenever the shear stress acting on octahedral planes exceed a critical value. This value is
taken as the octahedral shear existing in the standard tensile bar at incipient yielding.
The maximum octahedralshearstress theory is closely related to the maximum shearstress
theory but may be thought of as a refinement in that it considers the influence of all three principal
stresses.
From
1/2
oct = 1 1  2 2 + 2  3 2 + 3  1 2
3
(1.13b)
(6.12)
According to the theory, yielding always occurs at a value of octahedral shear stress established by
the tension test as
oct (limiting value) = 2 Sy .
(6.13)
3
Thus, the octahedral shearing stress theory of failure can be expressed as follows by utilizing
Eqs. 6.13 and 1.13b:
Sy = 2 1  2 2 + 2  3 2 + 3  1 2 1/2 .
2
(6.14)
Equation 6.14 implies that any combination of principal stresses will cause yielding if the right side
of this equation exceeds the tensile test value of Sy. This may be written alternatively as
2S2y = 1  2 2 + 2  3 2 + 3  1 2 .
(6.15)
Note L.4
Page 50
A variation of Eq. 6.14 which is sometimes useful involves the concept of an equivalent
uniaxial tensile stress, e, where e is the value of uniaxial tensile stress which produces the same
level of octahedral shear stress as does the actual combination of existing principal stresses, thus
e = 2 1  2 2 + 2  3 2 + 3  1 2 1/2
2
(6.16)
Obviously, if the loads are such that e > Sy, yielding would be predicted. For design purposes, e
should be made equal to the allowable working uniaxial stress.
Case of Pure Biaxial ShearStress
Recalling Mohrs circle for this case, we have the principal stresses
1 = , 2 = , 3 = 0.
Sy
or e = 3 ,
(6.17)
S
This means that if > y = 0.577 Sy, the material will yield. Hence, according to the maximum
3
octahedralshearstress theory, a material is 57.7% as strong in shear as it is in tension.
General Case of Biaxial Stress (3 = 0)
Equation 6.16 reduces to
e =
21 + 22  1 2
1/2
(6.18)
In many biaxialstress problems it is more convenient to work directly with stress x, y and
xy, because these can be determined more readily than principal stresses. Equation 6.18 can be
modified for this purpose by application of Eq. 1.8 to yield Eq. 6.19:
1 , 2 =
x + y
2xy +
x  y
2
e = 2x + 2y  x y + 32xy
1/2
(1.8)
(6.19)
Equation 6.19 can also be derived by superposition of Eqs. 6.17 and 6.18.
The locus of failure conditions for this failure theory is illustrated by the ellipse in Fig. 30.
NOTE: The theory can be independently developed from the maximum distortionenergy theory,
which postulates that failure (yielding) is caused by the elastic energy associated with this
distortion.
Note L.4
Page 51
2
S yt
0.577 S yt
Syt
0.577 Syt
45
0.577 Syt
Shear Diagonal
0.577 Syt
Figure 30.
6.9
Failure Locus for the Biaxial Stress State for the Maximum Distortion Energy Theory
The failure theories are compared graphically for a biaxial state of stress in Fig. 31. From this
figure, it can be seen that:
The distortion energy and maximum shear stress theories predict similar results with the
shear stress theory being more conservative.
The maximum normal stress and maximum shear stress theories agree in the first and third
quadrants where the signs of the principal stresses are the same but not in the second and
fourth quadrants.
Biaxial strength data for a variety of ductile and brittle materials are shown in Fig. 32 with
several failure theory limits. From this figure it can be seen that experimental data supports:
 Maximum normal stress theory is appropriate for brittle behavior.
 Distortion energy or maximum shear stress theories is appropriate for ductile failure.
Note L.4
Page 52
2
f
1.6
1.0
1.0
1.2
0.8
0.4
0.4
0.8
1.2
1.6
1
f
2
yp
1.0
2 Brittle materials
ult
Ductile materials
Aluminum
Copper
Nickel
Steel
Mild Steel
Carburized Steel
1.0
1
yp
1.0
Distortion
energy theory
Maximum shear
theory
1.0
Brass
Cast iron
Cast iron
1.0
1
ult
Maximum normal
stress theory
1.0
Figure 32. Comparison of Biaxial Strength Data with Theories of Failure for a Variety of Ductile
and Brittle Materials. (From Ref. 3, p. 144.)
Note L.4
Page 53
stress (plane state of stress), the critical inner surfaces are subjected to uniaxial stress. For these
cases, the failure theories are, of course, in complete agreement as to the load intensity causing
failure.
Example 1:
Internally pressurized cylinder. Determine the internal pressure required to yield the inner
surface of a cylinder, where ri = 1 in., r o = 2 in., (t = 1 in.), and the material is steel with the
properties Sy = 100 ksi and = 0.3. Assume plane stress.
Maximum stresses located at the inner surface are
r2o + r2i
= pi 4 + 1 = 5 pi
2
2
4  1
3
ro  r i
r =  pi .
t = pi
The predictions of the various failure theories are displayed as numbered curves on Fig. 33 for
a spectrum of geometries of internally pressurized cylinders. The specific results for several failure
theories for this example are as follows:
1.6
1.4
1.2
1.0
pi
= r
ri
Sy
(6)
0.8
1.0 at r/r i =
(5)
0.6
0.4
(4)
Max
theory
Max
theory
for = 0.3
Max DE
theory
Max
theory
1/(1 + ) at r/r i =
(3)
(2) 0.577 at r/r i =
0.5 at r/r i =
(1)
0.2
0
0.5
1.0
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Cylinder Proportion ro /ri = (r i + t)/r i
1.5
2.0
2.5
3.0
Cylinder Proportion t/ri
3.5
4.0
4.5
5.0
Curves
based on
initial
yielding
Note L.4
Page 54
Curve (1)
According to the maximum shearstress theory, yielding will begin when the highest shear
stress in the cylinder reaches that in the standard tensile test at initial yielding, which is Sy/2. Thus,
S
max = y
2
or
t  r = 50 ksi ,
2
4 pi = 50 ksi
3
or
pi = 37.5 ksi .
Hence, for this case where t/ri = 1.0, pi/Sy = 0.375. This value appears on Fig. 33.
Curve (2)
According to the maximum distortionenergy theory, yielding will begin when the equivalent
tensile stress at the inner cylinder wall reaches the tensile stress in the standard tensile test at initial
yielding, which is Sy. Hence, using Eq. 6.18 and the known principal stresses:
e = S y
2t + 2r  t r
1/2
= 100 ksi
pi = 43 ksi
Curve (4)
According to the maximum normalstress theory, yielding will begin when the highest normal
stress in the cylinder reaches the highest normal stress in the standard tensile test at initial yielding,
which is Sy. Hence,
t = S y
5/3 pi = 100 ksi
or the internal pressure required to yield the inner surface is
pi = 60 ksi for t/ri = 1.0.
For the full range of cylindrical geometries, pi can be determined for each theory and is presented
on Fig. 33.
Note L.4
Page 55
Example 2:
Straingage tests on the surface of a steel part indicate the stress state to be biaxial with
principal stresses of 35 ksi tension and 25 ksi compression. The steel has been carefully tested in
tension, compression and shear, with the results that Syt = S yc = 100 ksi and Sys = 60 ksi.
Estimate the safety factor with respect to initial yielding using the following failure theories:
Maximum normalstress,
Maximum shearstress,
Distortionenergy, and
Mohrs theory.
Solution:
We assume that the safety factor should be computed on the basis of all stresses increasing
proportionally as the load is increased to failure. On this basis the load line has been extended
outward from the nominal load point until it intersects the limiting lines corresponding to each
failure theory.
NOTE: By simple proportion, we obtain the safety factor as the ratio by which the nominal
stresses can be increased before yielding is predicted.
a)
2
S yc
Syt
Syt = 100 ksi
1
35, 25
Note L.4
Page 56
Syt Syc
,
1 2
= min 100 , 100
35
25
= min 2.86 , 4.00
= 2.86 2.9
b)
2
Sy
Sy
Sy
(35, 25)
Sy
m m =  1 m = m1 =  1 = 1
1
Thus,
1 =
y
x
 25  2
35  1
At the point of intersection between the failure locus line and the minimum distance to failure line
(i.e., at 1, 2), 2 = 1  Sy. So:
1 =
 25  1  Sy
35  1
35 + 1 = 25  1 + S y
21 = 35  25 + Sy
Sy = 100
1 = 35  25 + 100 = 55
Note L.4
Page 57
2 = 1  Sy = 55  100 = 45
max =
max  min
55   45
=
= 100 = 50
2
2
2
max
30
Solutions for the distortion energy theory and Mohr theory are shown together with the
foregoing solutions in Fig. 36. Satisfactory accuracy can be obtained by drawing the partial ellipse
representing the distortionenergy theory as a careful freehand curve through the three known
points. The curve representing the Mohr theory is drawn similar to the distortion energy curve
except that it passes through a value of 60 on the shear diagonal instead of through 57.7.
25
45
100
Max theory
DE theory
Limiting
Mohr theory Points
Max theory
57.7
60
Load Line
100
Shear Diagonal
2
Figure 36. Failure Theories for Example 2 Conditions
NOTE: Only the lower right quadrant is needed, because of one tensile and one compressive
stress.
Note L.4
Page 58
Hence, the resulting safety factors for the failure theories examined are:
THEORY
SAFETY FACTOR
50/30 = 1.7
DistortionEnergy
65/35 = 1.9
Mohr
69/35 = 2.0
Maximum NormalStress
100/35 = 2.9
conservative
preferable
no basis for believing this factor
pR/t = f
Rearranging yields
p
= 1 .
f R/t
(6.20)
2
2
pR/t + p/2 = f, or
p
1
=
f
R/t + 1/2
(6.21)
oct = 1
3
Now
1  2 2 + 2  3 2 + 3  1 2 = 2 f
3
1 = pR/t,
2 = pR/2t,
3 = p/2
(6.22)
Note L.4
Page 59
Hence
1
3
pR pR
t
2t
pR
p
+
2t
2
+ 
p pR
2
t
= 2 f
3
(6.23)
When the terms on the left side of Eq. 6.23 are squared and like terms collected, the following
equation is obtained:
p
2
=
(6.24)
2
f
3 R + 3 R + 1
2 t
2 t
2
Figure 37 compares the foregoing results for p/f for various thinwalled cylinders, and shows that
the maximum shearstress theory predicts the lowest value of pressure to cause failure and the
maximum distortionenergy theory the largest. The maximum normalstress theory predicts
intermediate values.
NOTE: The maximum difference between the lowest and largest values of the failure pressure is
about 15%.
NOTE: For cylinders of brittle materials, the maximum normalstress theory (curve (4) of Fig. 33
with S y replaced by Su) may be applied. It is obvious that brittle cylinders must be
designed with much larger safety factors than ductile cylinders, as failure involves
complete fracture, whereas ductile cylinders have a substantial plastic reserve between
initial (or small) yielding and fracture.
0.12
0.10
0.08
Max. shearing stress theory
0.06
0.04
0.02
0
10
20
30
40
R = Intermediate Radius
Wall Thickness
t
Maximum Internal Pressure in ClosedEnded ThinWalled Circular Cylindrical
Pressure Vessel Predicted by Various Theories of Failure
Figure 37. Failure of ClosedEnd Circular Cylinder Pressure Vessels
Note L.4
Page 60
6.12Examples for the Calculation of Safety Factors in ThinWalled Cylinders
Example:
The most stressed volume element of a certain loadcarrying member is located at the free
surface. With the xy plane tangent to the free surface, the stress components were found to be
x = 15,000 psi
y
= 2,000 psi
xy
7,000 psi
The loadstress relations for the member are linear so that the safety factor, N, can be applied either
to loads or stresses. Determine N if the member is made of a ductile metal with a tensile yield
stress of e = 44,000 psi.
NOTE: In general N Yield Stress . Hence, we can write N(Load Stress) Yield Stress.
Load Stress
1) Maximum DistortionEnergy Theory From Eq. 6.12 together with the definition of N we
obtain:
N oct = 2 Sy
3
For a uniaxial tensile test e = S y. Also, oct is given by Eq. 1.14b for stresses in the general x,
y, z directions. Hence, the above equation becomes:
N
3
2 + 2 + 2 2
x  y 2 + y  z 2 + z  x 2 + 6 xy
xz
yz
1/2
= 2 (44,000)
3
17,000
+  2,000
+  15,000
812 10
+ 6 7,000
= 2 (44,000)
3
= 2 (44,000)
N = 2.18
NOTE: This result can also be obtained directly from the result for a biaxial stress state,
Eq. 6.19 as:
2 1/2 =
N x2 + y2  x y + 3xy
e
N
15,000
+  2,000
N
+ 30 106 + 3 7,000
2 1/2
= 44,000
2)
Maximum ShearStress Theory In order to use this theory it is necessary to determine the
three principal stresses. From Mohrs circle:
Note L.4
Page 61
x + y
x  y 2
+
+ 2xy
2
2
1 (17,000) 2 + (7,000) 2
= 1 (13,000) +
2
4
= 6,500 + 11,010
= 17,510 psi
1 =
2 = 6,500  11,010
=  4,510 psi
3 = 0
At failure the three principal stresses are:
1 = 17,510 N psi
2 =  4,510 N psi
3 = 0
Now, from Eq. 6.5, which expresses the failure criterion for the maximum shear stress theory:
17,520 N + 4,510 N = 44,000
N 2
If the maximum distortionenergy theory is used, the loads can be increased by a factor of 2.18
before failure by general yielding. Assuming this is correct, the maximum shearstress theory
of failure is conservative since it predicts smaller failure loads.
REFERENCES
[1]
S.H. Crandall, N.C. Dahl and T.J. Lardner, An Introduction to the Mechanics of Solids,
McGrawHill, 2nd ed., 1972.
[2]
[3]
[4]
NOTE L.4
25
45
100
Max theory
DE theory
Limiting
Mohr theory Points
Max theory
57.7
60
Load Line
100
Shear Diagonal
2
22.312
NOTE L.4
L. Wolf
Revised by
M.S. Kazimi and N.E. Todreas
12/23/03
TABLE OF CONTENTS
1.
Page
Definition of Concepts.......................................................................................................... 1
2.
3.
4.
5.
Thermal Stress.......................................................................................................................31
6.
6.4 Maximum Shear Stress Theory (The Coulomb, later Tresca Theory)...........................44
Pressure Vessels.........................................................................................................58
References ...................................................................................................................................61
ii
1.
DEFINITION OF CONCEPTS
Structural mechanics is the body of knowledge describing the relations between external
forces, internal forces and deformation of structural materials. It is therefore necessary to clarify
the various terms that are commonly used to describe these quantities. In large part, structural
mechanics refers to solid mechanics because a solid is the only form of matter that can sustain
loads parallel to the surface. However, some considerations of fluidlike behavior (creep) are also
part of structural mechanics.
Forces are vector quantities, thus having direction and magnitude. They have special names
(Fig. 1) depending upon their relationship to a reference plane:
a)
b)
c)
z
Compressive
z
Tensile
Shear
Note L.4
Page 2
Fig. 2
F1
Fn
F2
Figure 3.
Note L.4
Page 3
F1 + F2 + ... + Fn =
n Fn
= 0
(1.1a)
The total moment of all the forces in Fig. 3 about an arbitrary point 0 must be zero since the vector
sum of forces acting on each particle is zero. Again, observe that internal forces occur in selfcanceling pairs along the same line of action. This leads to the second condition for equilibrium:
the total moment of all the external forces about an arbitrary point 0 must be zero.
r1 F1 + r2 F2 + ... + rn Fn =
rn F n = 0
(1.1b)
where rn extends from point 0 to an arbitrary point on the line of action of force Fn .
In Fig. 4A, an arbitrary plane, aa, divides a body in equilibrium into regions I and II. Since the
force acting upon the entire body is in equilibrium, the forces acting on part I alone must be in
equilibrium.
In general, the equilibrium of part I will require the presence of forces acting on plane aa.
These internal forces applied to part I by part II are distributed continuously over the cut surface,
but, in general, will vary over the surface in both direction and intensity.
n
II
dF
a
a
dA
a
Point 0
Point 0
Note L.4
Page 4
r
T = stress at point 0 on plane aa whose normal is n passing through point 0
dF where dF is a force acting on area dA.
= lim
dA
dA0
n
[Reference 1 uses the notation T to introduce the concept that T is a stress vector]
NOTE: Stress is a point value.
The stress acting at a point on a specific plane is a vector. Its direction is the limiting direction
of force dF as area dA approaches zero. It is customary to resolve the stress vector into two
components whose scalar magnitudes are:
normal stress component : acting perpendicular to the plane
shear stress component : acting in the plane.
1.1
The selection of different cutting planes through point 0 would, in general, result in stresses
differing in both direction and magnitude. Stress is thus a secondorder tensor quantity, because
not only are magnitude and direction involved but also the orientation of the plane on which the
stress acts is involved.
NOTE: A complete description of the magnitudes and directions of stresses on all possible
planes through point 0 constitutes the state of stress at point 0.
NOTE: A knowledge of maximum stresses alone is not always sufficient to provide the best
evaluation of the strength of a member. The orientation of these stresses is also
important.
In general, the overall stress state must be determined first and the maximum stress values
derived from this information. The state of stress at a point can normally be determined by
computing the stresses acting on certain conveniently oriented planes passing through the point of
interest. Stresses acting on any other planes can then be determined by means of simple,
standardized analytical or graphical methods. Therefore, it is convenient to consider the three
mutually perpendicular planes as faces of a cube of infinitesimal size which surround the point at
which the stress state is to be determined.
Figure 5 illustrates the general state of 3D stress at an arbitrary point by illustrating the stress
components on the faces of an infinitesimal cubic element around the point.
Note L.4
Page 5
y
0
yz
y
yx
zy
zx xz
xy
Figure 5.
Stress Element Showing General State of 3D Stress at a Point Located Away from
the Origin.
+ yx
 yx
x
Figure 6. Definition of Positive and Negative yx.
Note L.4
Page 6
xz
yz
z
9components
(1.2)
(1.3)
Therefore, six quantities are sufficient to describe the stresses acting on the coordinate planes
through a point, i.e., the triaxial state of stress at a point. If these six stresses are known at a point, it
is possible to compute from simple equilibrium concepts the stresses on any plane passing through
the point [Ref. 2, p. 79].
1.2
The tensor S becomes a symmetric tensor if Eq. 1.3 is introduced into Eq. 1.2. A fundamental
property of a symmetrical tensor (symmetrical about its principal diagonal) is that there exists an
orthogonal set of axes 1, 2, 3 (called principal axes) with respect to which the tensor elements are all
zero except for those on the principal diagonal:
S' =
1 0 0
0 2 0
0 0 3
(1.4)
Hence, when the tensor represents the state of stress at a point, there always exists a set of mutually
perpendicular planes on which only normal stress acts. These planes of zero shear stress are called
principal planes, the directions of their outer normals are called principal directions, and the stresses
acting on these planes are called principal stresses. An element whose faces are principal planes is
called a principal element.
For the general case, the principal axes are usually numbered so that:
1 2 3
1.3
The concept of strain is of fundamental importance to the engineer with respect to the
consideration of deflections and deformation.
A component may prove unsatisfactory in service as a result of excessive deformations,
although the associated stresses are well within the allowable limits from the standpoint of fracture
or yielding.
Strain is a directly measurable quantity, stress is not.
Note L.4
Page 7
dy
dx
y
x
x = lim dx
x
x0
dy
y = lim y
y0
y
dx
dx
yx = lim dx
y
y0
= tan
x
Figure 8.
Subscript notation for strains corresponds to that used with stresses. Specifically,
yx: shear strain resulting from taking adjacent planes perpendicular to the yaxis and
displacing them relative to each other in the xdirection (Fig. 9).
x, y: normal strains in x and ydirections, respectively.
yx > 0
initial
deformed
rotated clockwise
xy < 0
initial
deformed
rotated counterclockwise
Note L.4
Page 8
Sign conventions for strain also follow directly from those for stress: positive normal stress
produces positive normal strain and vice versa. In the above example (Fig. 7), x > 0, whereas
y < 0. Adopting the positive clockwise convention for shear components, xy < 0, yx > 0. In
Fig. 8, the shear is yx and the rotation is clockwise.
NOTE: Half of xy, xz, yz is analogous to xy , xz and yz , whereas x is analogous to x.
T =
1 xy
2
1 yx
2
1 zx
2
1 xz
2
1 yz
2
1 zy
2
It may be helpful in appreciating the physical significance of the fact that is analogous to /2
rather than with itself to consider Fig. 10. Here it is seen that each side of an element changes in
slope by an angle /2.
2
Figure 10. State of Pure Shear Strain
1.4
Plane Stress
In Fig. 11, all stresses on a stress element act on only two pairs of faces. Further, these
stresses are practically constant along the zaxis. This twodimensional case is called biaxial stress
or plane stress.
y
y
yx
0
xy
z
A
Note L.4
Page 9
The stress tensor for stresses in horizontal (x) and vertical (y) directions in the plane xy is:
S = x xy
yx y
When the angle of the cutting plane through point 0 is varied from horizontal to vertical, the
shear stresses acting on the cutting plane vary from positive to negative. Since shear stress is a
continuous function of the cutting plane angle, there must be some intermediate plane in which the
shear stress is zero. The normal stresses in the intermediate plane are the principle stresses. The
stress tensor is:
S' = 1 0
0 2
where the zero principal stress is always 3. These principal stresses occur along particular axes in
a particular plane. By convention, 1 is taken as the larger of the two principal stresses. Hence
1 2
3 = 0
In general we wish to know the stress components in a set of orthogonal axes in the xy plane,
but rotated at an arbitrary angle, , with respect to the x, y axes. We wish to express these stress
components in terms of the stress terms referred to the x, y axes, i.e., in terms of x, y, xy and the
angle of rotation . Fig. 12 indicates this new set of axes, x1 and y1, rotated by from the original
set of axes x and y. The determination of the new stress components x1 , y1 , x1 y1 is
accomplished by considering the equilibrium of the small wedge centered on the point of interest
whose faces are along the xaxis, yaxis and perpendicular to the x1 axis. For the general case of
expressing stresses x1 and x1 y1 at point 0 in plane x1, y1 in terms of stresses x, y, xy and the
angle in plane x, y by force balances in the x1 and y1 directions, we obtain (see Ref. 1, p. 217):
x1 = x cos2 + y sin2 + 2xy sin cos
x1 y1 = (y  x) sin cos + xy (cos2  sin2 )
y1 = x sin2 + y cos2  2xy sin cos
where the angle is defined in Fig. 12.
y
y1
y
x1
x1
x
x
(A)
(B)
(1.5a)
(1.5b)
(1.5c)
Note L.4
Page 10
These are the transformation equations of stress for plane stress. Their significance is that
stress components x1 , y1 and x1 y1 at point 0 in a plane at an arbitrary angle to the plane xy are
uniquely determined by the stress components x, y and xy at point 0.
Eq. 1.5 is commonly written in terms of 2. Using the identities:
sin2 =
1  cos 2
;
2
we get
cos2 =
1 + cos 2
;
2
x + y
 y
+ x
cos 2 + xy sin 2
2
2
 y
x1 y1 =  x
sin 2 + xy cos 2
2
+ y x  y
y1 = x
cos 2  xy sin 2
2
2
x1 =
(1.6a)
(1.6b)
(1.6c)*
The orientation of the principal planes, in this two dimensional system is found by equating x1 y1 to
zero and solving for the angle .
1.5
Mohr's Circle
Equations 1.6a,b,c, taken together, can be represented by a circle, called Mohrs circle of stress.
To illustrate the representation of these relations, eliminate the function of the angle 2 from
Eq. 1.6a and 1.6b by squaring both sides of both equations and adding them. Before Eq. 1.6a is
squared, the term (x + y)/2 is transposed to the left side. The overall result is (where plane y1x1
is an arbitrarily located plane, so that x1 is now written as and x1 y1 as ):
[  1/2 (x + y)]2 + 2 = 1/4 (x  y)2 + 2xy
(1.7)
Now construct Mohrs circle on the and plane. The principal stresses are those on the axis
when = 0. Hence, Eq. 1.7 yields the principal normal stresses 1 and 2 as:
1,2 =
x + y
center of circle
2xy +
x  y
2
(1.8)
radius
The maximum shear stress is the radius of the circle and occurs in the plane represented by a
vertical orientation of the radius.
* y1 can also be obtained from x1 by substituting + 90 for .
Note L.4
Page 11
max =
2xy +
x  y
2
= 1 1  2
2
(1.9)
The angle between the principal axes x1 y1 and the arbitrary axis, xy can be determined from
Eq. 1.6b by taking x1 y1 as principal axes. Hence, from Eq. 1.6b with x1 y1 = 0 we obtain:
2xy
2 = tan1
(1.10)
x  y
From our wedge we know an arbitrary plane is located an angle from the general x, yaxis set
and ranges from zero to 180. Hence, for the circle with its 360 range of rotation, planes
separated by in the wedge are separated by 2 on Mohrs circle. Taking the separation between
the principal axis and the xaxis where the shear stress is xy, we see that Fig. 13A also illustrates
the relation of Eq. 1.10. Hence, Mohrs circle is constructed as illustrated in Fig. 13A.
x  y
x + y
( y , yx)y
( 2 , 0)
locus of yx
( 1 , 0)
x( x , xy)
2xy +
locus of xy
x  y
2
Figure 13. (A) Mohrs Circle of Stress; (B) Shear Stress Sign Convection for Mohrs Circle.*
When the principal stresses are known and it is desired to find the stresses acting on a plane
oriented at angle from the principal plane numbered 1 in Fig. 14, Eq. 1.6 becomes:
and
= 1 + 2 1  2 cos 2
2
2
(1.11)
=  1  2 sin 2
2
(1.12)
For shear stresses, the sign convection is complicated since yx and xy are equal and of the same sign (see Figs 5
and 6, and Eq. 1.3), yet one must be plotted in the convention positive ydirection and the other in the negative ydirection (note the axis replaces the yaxis in Fig. 13A). The convention adopted is to use the four quadrants of the
circle as follows: positive xy is plotted in the fourth quadrant and positive yx in the second quadrant.
Analogously, negative xy lies in the first quadrant. Further, the shear stress at 90 on the circle is positive while
that at 270 is negative. [Ref. 1, p. 220]
Note that the negative shear stress of Eq. 1.12 is plotted in the first and third quadrants of Fig. 14, consistent with
this sign convention.
*
Note L.4
Page 12
1 + 2
2
2
1  2 cos 2
2
1  2 sin 2
2
1
Figure 14. Mohrs Circle Representation of Eqs. 1.11 & 1.12.
Example 1:
Consider a cylindrical element of large radius with axial (x direction) and torque loadings
which produce shear and normal stress components as illustrated in Fig. 15. Construct Mohrs
circle for this element and determine the magnitude and orientation of the principal stresses.
From this problem statement, we know the coordinates x, xy are 60, 40 ksi and the
coordinates y, yx are 0, 40 ksi. Further, from Eq. 1.10, 2 = tan1 2(40)/(60) = 53. From these
values and the layout of Fig. 13, we can construct Mohrs circle for this example as shown in
Fig. 16. Points 1 and 2 are stresses on principal planes, i.e., principal stresses are 80 and 20 ksi
and points S and S' are maximum and minimum shear stress, i.e., 50 and 50 ksi. These results
follow from Eqs. 1.8 and 1.9 as:
From Eq. 1.8
From Eq. 1.9
From Fig. 13A
2
402 + 60  0 = 80, 20 ksi
2
2
max = 402 + 60  0 = 50 ksi
2
at plane of max = (60 + 0)/2 = 30 ksi, i.e., at S and S'
1 , 2 = 60 + 0
2
S (30, 50)
y (0, 40)
F
xy = T = 40 ksi
A
x = F = 60 ksi
A
y = 0
37
(20, 0) 2
0 3
53= 2
(80, 0)
x (60, 40)
S' (30, 50)
Note L.4
Page 13
The principal element, i.e., its faces are principal planes, and the maximum shear element are
shown in Fig. 17.
NOTE: The angle between points y and S is the complement of 53 or 37, and plane y is rotated
37/2 counterclockwise from plane S.
NOTE: The procedure for determining the orientation of these elements: point 1 is 53 counter
clockwise from point x on the circle; hence the plane represented by 1 is half of 53
counterclockwise from plane x. Similarly, the S plane is 143/2 counterclockwise of
plane x, or 37/2 clockwise of plane y.
y axis
26.5
Plane perpendicular
to x axis
Plane y
1 = 80 ksi
S axis
37
2 = 30 ksi
Plane y
max= 50 ksi
(2) (1)
(1)
y axis
143
2
x axis
x axis
37
2 = 30 ksi
(2)
S axis
Plane x
2 =  20 ksi
Principal Element
Plane x
max= 50 ksi
Figure 18 illustrates the orientation of one of the eight octahedral planes which are associated
with a given stress state. Each of the octahedral planes cut across one of the corners of a principal
element, so that the eight planes together form an octahedron.
The stresses acting on these planes have interesting and significant characteristics. First of all,
identical normal stresses act on all eight planes. By themselves the normal stresses are therefore
said to be hydrostatic and tend to compress or enlarge the octahedron but not distort it.
Note L.4
Page 14
oct = 1 + 2 + 3
3
(1.13a)
Shear stresses are also identical. These distort the octahedron without changing its volume.
Although the octahedral shear stress is smaller than the highest principal shear stress, it constitutes
a single value that is influenced by all three principal shear stresses. Thus, it is important as a
criterion for predicting yielding of a stressed material.
oct = 1 1  2 2 + 2  3 2 + 3  1 2 1/2
3
In cases in which x, y, z, xy, xz and yz are known:
+ y + z
oct = x
3
oct = 1
x  y 2 + y  z 2 + z  x 2 + 6 2xy + 2xz + 2yz
3
(1.13b)
(1.14a)
1/2
(1.14b)
oct
1
3
oct
Having observed the correspondence between strain and stress, it is evident that with suitable
axis transformation one obtains an expression for the strain tensor T ' which is identical to that of
stress tensor S' except that the terms in the principal diagonal are 1, 2, 3. Hence, recalling from
Section 1.3 that /2 corresponds to , strain relations analogous to Eqs. 1.8, 1.9 and 1.10 can be
written as follows:
1 , 2 =
x + y
max = 2
1 xy
2
1 xy
2
2 = tan1
xy
x  y
x  y
2
x  y
2
(1.15)
(1.16)
(1.17)
Note L.4
Page 15
Conclusion
The preceding sections dealt separately with the concepts of stress and strain at a point. These
considerations involving stresses and strains separately are general in character and applicable to
bodies composed of any continuous distribution of matter.
NOTE: No material properties were involved in the relationships, hence they are applicable to
water, oil, as well as materials like steel and aluminum.
Note L.4
Page 16
Note L.4
Page 17
2.1
Let us consider the various components of stress one at a time and add all their strain effects.
For a uniaxial normal stress in the x direction, x, the resulting normal strain is
x = x
E
(2.1)
(2.2)
This fixed fraction is called Poissons ratio, . Analogous results are obtained from strains due to
y and z.
The shearstress components produce only their corresponding shearstrain components that
are expressed as:
zx = zx ,
xy = xy ,
yz = yz
(2.3a,b,c)
G
G
G
where the constant of proportionality, G, is called the shear modulus.
Note L.4
Page 18
For a linearelastic isotropic material with all components of stress present:
x = 1
x  y + z
E
y = 1 y  z + x
E
z = 1 z  x + y
E
xy = xy
G
yz = yz
G
zx = zx
G
(2.4a)
(2.4b)
(2.4c)
(2.5a) same as (2.3)
(2.5b)
(2.5c)
E
.
2 (1 + )
(2.6)
Hence,
2 (1 + )
xy
E
2 (1 + )
=
yz
E
2 (1 + )
=
zx .
E
xy =
(2.7a)
yz
(2.7b)
zx
(2.7c)
Equations 2.4 and 2.5 may be solved to obtain stress components as a function of strains:
E
1  x + y + z
(1 + ) (1  2)
E
y =
1  y + z + x
(1 + ) (1  2)
z =
1  z + x + y
(1 + ) (1  2)
E
xy =
= Gxy
2 (1 + ) xy
E
yz =
= Gyz
2 (1 + ) yz
E
zx =
= Gzx .
2 (1 + ) zx
x =
(2.8a)
(2.8b)
(2.8c)
(2.9a)
(2.9b)
(2.9c)
Note L.4
Page 19
x =
+
+ y + z
(1 + ) x
(1  2) x
E
y =
+
+ y + z
(1 + ) y
(1  2) x
E
z =
+
+ y + z .
(1 + ) z
(1  2) x
(2.10a)
(2.10b)
(2.10c)
For special case in which the x, y, z axes coincide with principal axes 1, 2, 3, we can simplify the
strain set, Eqs. 2.4 and 2.5, and the stress set Eqs. 2.8 and 2.9, by virtue of all shear strains and
shear stresses being equal to zero.
1 = 1 1  2 + 3
(2.11a)
E
2 = 1 2  3 + 1
(2.11b)
E
3 = 1 3  1 + 2
(2.11c)
E
1 =
1  1 + 2 + 3
(2.12a)
(1 + ) (1  2)
2 =
1  2 + 3 + 1
(2.12b)
(1 + ) (1  2)
3 =
1  3 + 1 + 2 .
(2.12c)
(1 + ) (1  2)
For biaxialstress state, one of the principal stresses (say 3) = 0, Eqs. 2.11a,b,c become:
1 = 1 1  2
E
2 = 1 2  1
E
3 =  1 + 2 .
E
(2.13a)
(2.13b)
(2.13c)
In simplifying Eqs. 2.12a,b,c for the case of 3 = 0, we note from Eq. 2.12c that for 3 to be zero,
3 = 
+ 2 .
1  1
Substituting this expression into the first two of Eqs. 2.12a,b,c gives:
E
1 =
1 + 2
1  2
E
2 =
2 + 1
1  2
3 = 0 .
(2.14)
(2.15a)
(2.15b)
(2.15c)
Note L.4
Page 20
In case of uniaxial stress Eqs. 2.13 and 2.15 must, of course reduce to:
1 = 1 1
E
2 = 3 =  1
E
1 = E1
2 = 3 = 0 .
2.2
(2.16a)
(2.16b)
(2.17a)
(2.17b)
(2.18)
NOTE: Hydrostatic compressive stress exists within the fluid of a pressurized hydraulic cylinder,
in a rock at the bottom of the ocean or far under the earths surface, etc.
Hydrostatic tension can be created at the interior of a solid sphere by the sudden
application of uniform heat to the surface, the expansion of the surface layer subjecting
the interior material to triaxial tension. For 1 = 2 = 3 = , Eqs. 2.11a,b,c show that:
1 = 2 = 3 = = 1  2 .
E
This state of uniform triaxial strain is characterized by the absence of shearing deformation; an
elemental cube, for example, would change in size but remain a cube. The size of an elemental cube
initially of unit dimension would change from 13 to (1 + )3 or to 1 + 3 + 3 2 + 3. If we
consider normal structural materials, is a quantity sufficiently small, so that 2 and 3 are
completely negligible, and the volumetric change is from 1 to 1 + 3. The volumetric strain, V/V,
is thus equal to 3 or to:
V = 3 = 3 (1  2) .
(2.19)
V
E
Hence,
E
k =
.
(2.20)
3
3 (1  2)
Now 0.5, so that k cannot become negative. A simple physical model of a representative
atomic crystalline structure gives:
= 1/3 ,
(2.21)
k = E .
(2.22)
so that
Note L.4
Page 21
3.
3.1
Stresses
Stresses in a ThinWalled Cylinder:
Consider a portion of the cylinder sufficiently remote from the ends to avoid end effects.
The equation of equilibrium of radial forces (Fig. 19) acting on the element is:
pi ri ddL = 2t,av tdL sin
yielding
t,av =
d
2
(3.1)
pi ri
.
t
(3.2)
when sin (d/2) is approximated by d/2, which is valid for small d. This is a correct expression
for the average tangential stress (hoop stress) regardless of cylinder thickness.
NOTE: It is only when this equation is used as an approximation for the maximum tangential
stress, t,max , that the thickness t must be very small in comparison with the cylinder
diameter.
Equation 3.1 can be modified to give a good quick estimate of the maximum tangential stress due to
internal pressure for cylinders of moderate thickness by replacing ri with rav, which we label R:
R = ri + t .
2
Thus,
t,max
pi
t
(3.3)
pi R
t
(3.4)
t,avtdL
d
dL
p i r i ddL
d/2
2ri
2ro
Figure 19. Radial Forces in an Internally Pressurized ThinWalled Cylinder.
Note L.4
Page 22
The following are errors in estimating t,max from Eqs. 3.2 or 3.4 compared to the thickwalled
cylinder results:
t/ri
0.1
0.2
0.5
1.0
% Error Using
Eq. 3.2
5 (low)
10 (low)
23 (low)
40 (low)
% Error Using
Eq. 3.4
0.5 (low)
1.0 (low)
3.8 (low)
10.2 (low)
If the ends are closed, the cylinder is also subjected to an axial force of magnitude pi r2i . This is
distributed over an area of cross section that can be expressed as:
A = r2o  r2i = 2Rt .
(3.5)
pi r2i
p r2
= ii .
2rav t
r2o  r2i
(3.6)
For the thinwalled case, ri rav R since ri = R(1  0.5t/R). Hence, Eq. 3.6 reduces to:
a,av =
pi R
.
2t
(3.7)
3.2
(3.8)
The deformation of the diameter of a thinwalled cylinder or sphere caused by the internal
pressure is to be determined. Note that the state of stress at the inner surface is triaxial and is not
plane stress. The principal stresses of the inner surface are 1 = , 2 = z and 3 = p.
However, 3 = p is so small compared to 1 and 2 that it can be neglected when considering
strains. Thus, the state of stress in thinwalled pressure vessels is usually considered to be plane
stress.
The deformation of such pressure vessels is influenced by the tangential and axial (transverse
for the sphere) stresses, and hence use must be made of the following relations obtained from Eqs.
2.4, 2.5 and 2.8, respectively, with y = 0.
Note L.4
Page 23
x = x  z
E
E
y =  x + z
E
z = z  x
E
E
= xy
G
E
x =
x + z
1  2
E
z =
z + x
1  2
(3.9a)
(3.9b)
(3.9c)
(3.9d)
(3.10a)
(3.10b)
to express Hookes law for plane stress. Equations 3.10a,b are obtained from Eqs. 2.8a and 2.8c
upon inspection of y evaluated from Eq. 2.8b with y taken as zero.
Let , z, and z represent the tangential and axial stress and strain, respectively, in the wall.
The substitution of these symbols in Eqs. 3.9a,b,c, i.e., x and z = z, gives:
 z
E
E
z = z 
.
E
E
(3.11)
(3.12)
ClosedEnd Cylinder:
For the strains in the closedend cylinder, the values of and z as derived in Eqs. 3.4
and 3.8, respectively, are substituted into Eqs. 3.11 and 3.12 to give:
pR
pR
pR
= 1

=
(2  )
E t
2t
2Et
(3.13)
pR
pR
pR
z = 1

=
(1  2) .
E 2t
t
2Et
(3.14)
Let the change in length of radius R be r when the internal pressure is applied. Then the change in
length of the circumference is 2r. But the circumferential strain, , is, by definition, given by the
following equation:
= 2r = r .
2R
R
(3.15)
pR2
(2  ) .
2Et
(3.16)
Note L.4
Page 24
The change in length, l, for a closedend cylinder is equal to:
l = zl
or
l =
pRl
(1  2)
2Et
(3.17)
(3.18)
Sphere:
z = =
pR
2t
(3.19)
(3.20)
3.3
pR2
(1  ) .
2Et
(3.21)
Figure 20 illustrates a cylinder closed by thinwalled hemispherical shells. They are joined
together at AA and BB by rivets or welds. The dashed lines show the displacements due to internal
pressure, p. These displacements are given by Eqs. 3.16 and 3.21 as:
pR2
(2  )
2Et
pR2
Rs =
(1  )
2Et
Rc =
(3.22a)
(3.22b)
The value of Rc is more than twice that of Rs for the same thickness, t, and as a result the
deformed cylinder and hemisphere do not match boundaries. To match boundaries, rather large
shearing forces, V, and moments, M, must develop at the joints as Fig. 20 depicts (only those on the
cylinder are shown; equal but opposite shears and moments exist on the ends of the hemispheres).
This shear force is considerably minimized in most reactor pressure vessels by sizing the
hemisphere thickness much smaller than the cylinder thickness. For example, if the hemisphere
thickness is 120 mm and the cylinder thickness is 220 mm, for = 0.3, then the ratio Rc to Rs is
Rc = ts 2  = 1.32
tc 1 
Rs
Note L.4
Page 25
Rs
Rs
V
V Shearing Forces
Rc
V
Figure 20. Discontinuities in Strains for the Cylinder (Rc) and the Sphere (Rs).
The general solution of the discontinuity stresses induced by matching cylindrical and
hemisphere shapes is not covered in these notes. However, a fundamental input to this solution is
the matching of both displacements and deflections of the two cylinders of different wall
thicknesses at the plane they join. Thus, the relationship between load, stress, displacement and
deflection for cylinders and hemispheres are derived. These relations are presented in Tables 1
and 2, respectively.
Table 1. Relationship Between Load, Stresses, Displaced and Deflection for the Cylinder
(equations cited are from Ref. 2)
Stresses
t , l, r
, z, r
pR pR
p
,
, t
2t
2
Load
p
Mo
Edge
Moment
per Unit
Perimeter
Length
6Mo , 6Mo , 0
t2
t2
2 DR
EMo
2
(Eq. 7.3.22,
p. 7.312)
(Eq. 7.3.21,
p. 7.311)
Note: u o =
Mo
; so first term
2 2 D
Eu o
.
R
Displacement
Deflection
uo
pR2
2  + t
2tE
R
Mo
2
2 D
(Eq. 7.3.18, p. 7.3.7)
units
F L/L
= L
1 F L3
L2 L2
0
 Mo
D
(Eq. 7.3.19, p. 7.3.8)
dimensionless
Note L.4
Page 26
Table 1. Relationship Between Load, Stresses, Displaced and Deflection for the Cylinder (contd)
EQo
Qo
Qo
, 0, 0
R2 D
2 D
(Eq. 7.3.22,
(Eq. 7.3.18, p. 7.3.7)
p. 7.312)
units F/L = L
Qo
1 FL
Note: u o = 3 ; so first term
2 D
L3
u E
= o .
R
where Plate Flexural Rigidity:
Edge Shear
Force
per Unit
Perimeter
Length
D =
4
Et3
,
12 1  2
units
2
Et = 3 1 R2 t2
4R2D
F L3
L2
Qo
2
2 D
(Eq. 7.3.19, p. 7.3.8)
dimensionless
1 1
L2 L2
Table 2. Relationship Between Load, Stresses, Displaced and Deflection for the Hemisphere
Load
Stresses
t , l, r
Displacement
o
Deflection
o
pR pR
p
,
, 2t
2t
2
t
pR2
1 +
2tE
R
Mo
2 Mo 6Mo , 6Mo , 0
tR
t2
t2
2 Mo
Et
4 Mo
REt
2RQ0
Et
2 Qo
Et
Edge
Moment
per Unit
Perimeter
Length
Qo
Edge Shear
Force
per Unit
Perimeter
Length
Note: 2 Mo = o ; so first
t
E
E
o
term =
.
R
2Q0
, 0, 0
t
2Q0
= Eo .
Note:
t
R
where
= R ; 4 =
3 1 2
R2t2
Note L.4
Page 27
4.
The cylinder is axially restrained (z = 0) at either end and subjected to uniform radial pressure,
not body forces. The cylinder cross section and stresses on an element are illustrated in Fig. 21.
d
2
pi
r + dr
po
ro
ri
r
r + dr
d
2
= ur ,
r = dv  vr
dr
(4.1a,b,c)
where u is the outward radial displacement and v is the direction displacement. Since the effect of
internal pressure is to move the material in the radial direction without any rotation, both v and r
should be zero. This will be demonstrated formally from the boundary conditions.
StressEquilibrium Relations (for forces in the r and directions)
dr
2r
dr + r  = 0 ,
+
r
r = 0.
dr
dr
StressStrain Relations
)]
]
)]
1
+ z
E r
1
= ( z + r )
E
1
z = 0 = z r +
E
r =
(4.2a,b)
(4.3a)
(4.3b)
(4.3c)
Note L.4
Page 28
r
r
=0
(4.4)
(4.5a)
(4.5b)
Displacement Approach
When displacements are known, it is preferable to use this solution approach. In this case we
solve for the displacements. Proceeding, solve Eq. 4.3 for stresses and substitute them into the
stressequilibrium relations of Eq. 4.2 to obtain:
r 
d
 r
(1  ) dr +
+
+
= 0
r
r
dr
dr
dr
2r
+ r = 0
dr
(4.6a)
(4.6b)
Substituting Eq. 4.1 in Eq. 4.6 and simplification show that the displacements u and v are
governed by the equations:
d2 u + 1 du  u = d 1 d (ru) = 0
(4.7a)
r dr
dr r dr
r2
dr2
d2 v + 1 dv  v = d 1 d (rv) = 0 .
r dr
dr r dr
r2
dr2
Successive integration of these equations yields the displacement solutions
u = C1 r + Cr2 ;
v = C3 r + Cr4
2
2
(4.7b)
(4.8a,b)
where the constants of integration are to be determined by the stress boundary conditions
(Eqs. 4.5a,b). With the use of Eqs. 4.8a,b, the strain and stress solutions are obtained from Eq. 4.1
and
E
r =
(1  )r +
(4.9a)
(1 + ) (1  2)
E
=
(1  ) + r
(4.9b)
(1 + ) (1  2)
r = Gr
(4.9c)
as
r = C1  C2 ,
= C1 + C2 ,
r =  2C4
(4.10a,b,c)
2
2
2
2
r2
r
r
Note L.4
Page 29
C1  (1  2) C2
E
(1 + ) (1  2) 2
r2
C1 + (1  2) C2
E
=
(1 + ) (1  2) 2
r2
EC1
z = r +
(1 + ) (1  2)
r =  2GC4 .
r2
r =
(4.11a)
(4.11b)
(4.11c)
(4.11d)
By virtue of the boundary conditions, Eqs. 4.5a,b, the first and last of Eq. 4.11 yield the
following equations for the determination of the constants C1 and C2.
C1  (1  2) C2 =  (1 + ) (1  2) p
i
2
E
r2i
C1  (1  2) C2 =  (1 + ) (1  2) p
o
2
E
r2o
C4 = 0
The solutions for C1 and C2 are
C1 = (1 + ) (1  2) pi r2i  po r2o
2
E r2o  r2i
(1 + ) pi  po r2i r2o
C2 =
.
E r2o  r2i
(4.12a)
(4.12b)
(4.12c)
(4.13a)
(4.13b)
As to the constant C3, it remains undetermined. However, with C4 = 0, Eq. 4.8 shows that v = C3r/2
which corresponds to a rigidbody rotation about the axis of the cylinder. Since this rotation does
not contribute to the strains, C3 is taken as zero. Hence Eqs. 4.11a,b,c and Eq. 4.8a become:
r =
=
z =
u =
4.2
1
ro /ri 2  1
1
ro /ri 2  1
2
ro /ri 2
1 1 +
ro /ri 2
ro 2 p
pi  1  1
ri o
2
2
r/ri
r/ri
ro /ri 2
ro 2 p
pi  1 + 1
ri o
2
2
r/ri
r/ri
2
pi  rro po
i
 1
(1 + ) r/ri ri
ro /ri 2  1
(1  2) +
(4.14a)
(4.14b)
Plane Strain!
(4.14c)
ro /ri 2 pi
ro 2 po
 (1  2) + 1
r/ri 2 E
r/ri 2 ri E
(4.14d)
Stress Approach
In this case we solve directly for the stresses and apply the boundary conditions as before.
Thus, reexpressing the stressstrain relations (Eq. 4.3) directly as strain when z = 0 we get:
Note L.4
Page 30
r = 1 + (1  ) r 
E
1
+ (1  )  r
=
E
r
r = ,
G
(4.15a)
(4.15b)
(4.15c)
and then substituting them into Eq. 4.4 upon simplification yields
d
r
d
(1 ) r +
= 0 .
dr
dr
r
(4.16)
By the first stressequilibrium relation in Eq. 4.2, this equation is reduced to the simple form
d + = 0.
(4.17)
dr r
Integration of this equation and the second of Eq. 4.2 yields
r + = C1 ;
r =
C2
.
r2
(4.18a,b)
pi r2i  po r2o
,
r2o  r2
i
C2 = 0 ,
C3 =
pi  po r2i r2o
.
r2o  r2i
(4.20a,b,c)
With these values for the constants, Eqs. 4.18, 4.19 and the relation z = (r + ), the complete
stress solution is explicitly given. This solution can then be substituted in Eqs. 4.15a and 4.15b to
determine r and and, finally, u is determined by the relation u = r .
NOTE: Since the prescribed boundary conditions of the problem are stresses, it is evident that the
stress approach in obtaining the solution involves simpler algebra than the displacement
approach.