Sei sulla pagina 1di 11

Efficient seismic analysis of

piping systems with joint


deformations
Dong-Guen Lee and Yoon-Hwan Song
Department of Cirri Engmeering, Korea Advanced Institute of Science and Technology,
Ku,~ong<hmg, Yusong-gu, Taelon 305-701, Koren
(Rccetvcd November 1990, revtsed verston accepted August 1991)

A piping system is composed of p~pes with various thmkness,


dmmeter and length. Accurate analysis of a piping system requires a
comphcated three-dimensional finite element model and a computer
system with large memory size, while a simplified beam model may
result in response prediction with deteriorated accuracy. An efficient
model for seismic analysis of piping systems is proposed. The proposed model as developed by introducing the pipe joint element which
accounts for the local deformations of a pipe joint. Pipes are
represented by beam elements and the effect of local deformations of
pipe jo)nts is modelled by the pipe joint element deformations. The
proposed model which is as s~mple and efficient as a beam model can
be used to obtain the seismic response of piping systems with an
accuracy very close to that obtained by a complicated finite element
model.

Keywords: piping

systems,
analys~s, pipe joint element

A piping system is generally considered to include the


complete lnterconnectlon of pipes, including in-line
components such as pipe fittings and flanges. The contributions of piping systems are essential in an
industrlahzed society. Nuclear power piping systems are
very complicated systems which have various types of
connections and supports In particular, the stress concentration around a pipe joint results from the vibration
which power plant operation, abrupt failure and earthquake induce t_ Thus, a piping system demands a high
level of structural safety during a hmlted lifetime Also,
the behaviour of nuclear power piping systems must be
estimated with accuracy in both static and dynamic
analyses against an accident.
There are two major types of analysis model used for
the analysis of piping systems the one which uses shell
elements and the other which uses beam elements. In the
former case, even if the behaviour of piping systems can
be accurately estimated, it is inappropriate and uneconomical to analyse entire piping systems because of the
difficulty m analysis model generation and long computation time. In the latter case, the saving in computation time is significant and the analysis model is simple
But it neglects the flexibility effect such as the ovahzatlon of a pipe section Thus, we can merely obtain
approximate analysis result by this method.
To esumate the behavour of a piping system with
simplicity and accuracy, it is essential to develop a new
0 1 4 1 0 2 9 6 9 3 0 1 0 0 0 2 11
,c, 1 9 9 3 B u t t e r w o r t h - H e m e m a n n Ltd

Eng. Struct. 1993, Vol. 15, No 1

local

deformation

effect,

seismic

element which can consider the local flexibility effect


This paper proposes a pipe joint element which consists
of one radial spring and two rotational springs.
Generally, the stlffnesses of these springs depend on
geometric properties such as diameter, length and
thickness of run pipe and diameter and thickness of
branch pipe.
The Boiler and pressure vessel code (NB-3686) provides empirical formulae for the stiffness of two rotational springs-'. However, the radial spring constants
are not provided in the code
Murad and Sun 3 derived these spring constants based
on the theory of shells in terms of shell geometries and
internal pressure. The expressions, In closed forms, are
simple and very efficient for the estimation of spring
constants since the proposed formulations can be applied
to a rectangular attachment as well as a circular attachment But their research is not on a real hollow nozzle
as Murad and Sun 3 mentioned.
Batra and Sun 4 studied spring constants for the juncture of a hollow nozzle and a spherical shell. Their paper
dealt only with radial spring constants using an
analytical method. They specified the relationships of
spring constants in terms of shell and nozzle parameters
(diameter and thickness) except the essential parameter,
the length of shell
The aforementioned studies and code provisions are
limited to the estimation of the spring constants only for

Setsmtc analysts of pJpmg systems D -G. Lee and Y.-H. Song


the relatively regular shape of branch connections or
attachments. Further research for p~pe joints with
various geometries and local ~rregulantles such as the
insert plate reinforcement in a run pipe needs to be performed
The objectwe of this paper is to propose an efficient
analysis model for the dynamic analysis of piping
systems with local flexibdlty effect of the branch connection rather than to introduce an efficient method for
the determination of the stlffnesses of the springs. Time
h~storles of stress and displacement around the pipe joint
connecuon can easily be obtained by the linear combmauon of static analysis results for a ptpe joint model The
time histories of 6 components of forces and moments
obtained using the proposed model are used as the scaling factors for the combination of stresses and
displacements caused by the 6 unit loads applied to the
pipe joint model.

m Figure 3 Subelement I is used to represent pipe joint


deformations shown in Figure 2b and subelement II is
employed to account for those shown in Ftgures 2c and
2d. The length of subelement I is the same as the radius
of the run p~pe and the length of subelement II is zero
The pipe joint shown m Ftgure 1 can be modelled as
shown in Ftgure 4 using the proposed pipe joint element

Formulation of stiffness matrtx for the pipe joint


element
The element il shown in Figure 4 consists of a beam element q, a rotational deformation element jk and an axial
deformation element kl In the first stage of the stiffness
matrix formulation, the elements/./and jk are combined
to form an element tk as shown in Figure 5. The rotational spring constant m either longitudinal (Kt) or
c~rcumferential direcuon (K() ,s defined as the moment
KR

Development of the pipe joint element


A pipe joint, as shown in Figure 1 Is an mterconnection
of a run pipe and a branch pipe In general, a run pipe
has a relatwely large dmmeter compared to the thickness
of the pipe and the local deformations of the pipe joint
are as shown In Figure 2 When the conventional beam
model is used for the analys~s of a piping system, ~t is
assumed that p~pes are interconnected at the centrehne of
a run pipe and the effect of local deformations of the
pipe joint is ignored.
In this study, a pipe joint element is proposed to take
the pipe joint flexibility into consideraUon The pipe
Iolnt element is composed of two subelements as shown

K C , KI

Subelernent I

Figure 3

Subelement II

Subelements of a pipe joint e l e m e n t

Run pipe
A
w

0
Branch pipe
n

F~gure 1

Figure 4

Typical p=pe joint

~
C

P=pe joint model using a pipe joint e l e m e n t

%Z-3%%%%~'~jUc
d

Loeal d e f o r m a t m n s of a p~pe joint (b), due to radial force, (c), due to c~rcumferentlal
m o m e n t , (d), due to longitudinal m o m e n t

F~gure 2

Eng. Struct. 1993, Vol. 15, No 1 3

Setsmtc analysts of p~pmg systems D G Lee and Y H Song


the force that would produce unit d,splacemcnt Ill Its
direction, several terms related to axial deforlnatlon on
the right-hand side of equanon (7) must be moddmd
That ,s easdy dem'ed from the inouon ol ~er~e~ ol
sprmgs
To be compatible with the stiffness inatrix at the ten
trelme of the run p,pe, the equdibnuln equations ruth
respect to forces at nodes t and k, are presented as

't
/4

~/ ~

~9

10

[P,I =

Ftguro 5 Member w~th elastm spnngs at one end

[K,t] ]

[K4,]

[K~41

(1)

(2)

ILl =

-L

(lO)

[Uzl = [ T , z ] ' I U 4 I

(11)

The transformanon matrix [Tu] is formed a~ ~


F

[L~] =

(5)

(6)

As stated above, the member stiffness matrix which IS


considered to include the rotational springs at one end of
the beam element is entirely derived from the flexibility
matrices.

[T'~][K~k][T'~]T[K~] [T,~] z

- [T,~]
[K~]][K~]

--'u

3'~/

0
0

:'u

-x u

-Yu

-~u

(7)

Considering the elastic spring of the subelement 1 in


its spring constant (Kn) is defined as

Eng. Struct

1993, Vol. 15, No 1

(131

IP, i = [K,,] IU, i + [K,~] [L,]

'lull

(14)

[ & l = [K,,]IU, I + [K,~] [L,]

'lUll

(15)

When equation (10) ,s substituted into equation (15). we


can obtain equation (16) with respect to [P~I-

[P,] = [T~/]

I[K~,]IU, I

+ [Tu] - t [ K u ] [Tu] -rl Uzj

(16)

Thus equilibrium equations with respect to nodes t and


l can be written as

IP~I

[T~j]

'[&,]

xr' 'l
I UII

Figure 3, in which

(12)

Substituting equation (13) into equations (8) and (9), the


following equations are derived

And the submatrlx [K,,] can be obtained from [K~,] as

[KM] =[

l u l l = [T~,] rlU/I

Because the member stiffness matrix [KM] IS symmetric, the submatrix [K~,] can be obtained as

[K,,] = - [T,~] [K~,] = [T,~] [ K ~ ] [T,,] T

(3)

(4)

[&,] = [K,~]~= - [&~] [E~]7

where h/, Y,/and . . are the distance between two nodes


/, and / with rcspect to the ~, x and z axes. respecnvely.
By premultiplying equation (11) by [Tu] r, It can be
rewritten, as shown in equation (13)

Then. terms in the submatrlx [K,~] can be computed as


the stanc eqmhbrants of those m [K~]

[K,~] = - [ T , ~ ] [ K ~ ]

(9)

I & l = [L~] [P~]

Terms in the submatrlx [K~] are defined as the reactions at node k of the member due to unit displacements
at the same node Statically equivalent actions at node t
may be computed using the transformation matrix 5
1

(8)

And, using the transformation matrix [ Tt~] with respect


to nodes k and l, the load and displacement relations
between these nodes are given by

The submatnx [K~] can bc obtained as the reverse


matrix of the tlex,b,l,ty matrix [F~]

[ K ~ ] = [F~]

[K,~] l U l l

IP, I = [&,] [U,I + [Kaa] IU, I

would produce one radlan of rotation in ItS respective


direction and is Installed at node j of the beam element
Thus, the member stiffness matrix of the element tk is
expressed as

[KM] = [ [K.]

[K,,]IU,} +

[T~,] ' [ & , I [ T ~ / ]

'

(17)

UIt,mately. the member stiffness matrix which includes


the characteristics of the proposed pipe joint elements is

Seismic analysts of plpmg systems. D.-G. Lee and Y.-H Song


derwed as

[K*] = [

[T,~] [K~] [T,,]T


-- [T~I]

-'

[K.]

__ [Ti/. ] [KI.~]

[T,~] r

[Ttd ] -T~
r]

(18)

Formulation of mass matrix for the pipe jomt element


In the Ibrmulatlon of the mass mamx with a pipe joint,
using the concept of a ng~d body moUon as stated above,
mass matrices corresponding to the pipe joint element
can be acted on the centrehne of the run pipe. The
lumped mass mamx [M] of the element ik, shown m
Figure 5, should be obtained as

[E I [M~] 1
0

(19)

The mema force with respect to node k Is


[Pk] = [M~] lUll

(20)

Differentiating the displacement vector [Ut] of equation (11) twice with respect to ume, the acceleration vector can be obtained as

[U~} =

[T.]rlgJ~l

(21)

Substituting equations (10) and (21) into equation (20),


the following relationship ~s obtained

IP~I = [T~,]

'[M~][T.]

r[O~l

(22)

system since these springs account for the flexibility of


a pipe joint. The spring constants depend on the
parameters such as diameter, thickness and length of the
run pipe and diameter and thickness of the branch pipe,
etc_ Much research has been performed on these spring
constants using empirical and theoretical methods.
In this study, a s~mple procedure for the determination
of the stlffnesses of the springs is used. Rectangular and
triangular (near the joint connection) plate/shell
elements are used for analysis of the ptpe joint model
shown in Figure 6a. A radml load and moments are
applied at the end of the branch pipe and the spring constants (KR, K o KL) have been obtained from the static
load-displacement relationships at the pipe connection
as

Mc

ML

where
fR
0c
0L
P
Mc
ML

radial d~splacement at pipe connection


clrcumferentml rotation at p~pe connection
longitudinal rotation at pipe connection
radial load
circumferential moment
longitudinal moment

The spring constants obtained from the proposed procedure and the Boiler and pressure vessel code z and
results by Murad and Sun 3 are presented in Table 1
The dimensionless parameters c~, /3 and 3' and a thickness ratio p are
L~
Rr
Rr
tr
c~=-- B=-7=P=
Rr
Rb
tr
tt,

(26)

Thus, the mass matrix with respect to node l is obtained


as

[MA = [T.] '[Mk][T.]

(23)

Consequently, the mass matrix for the element d can be


expressed as
[M*]=[

[M']0 [Mr]0]

(24)

Determination of the spring constants


The proposed pipe joint element with three springs plays
an important role in the efficient analysis of a piping

where Rr and tr are the radius and the thickness of the


run pipe, respectwely, Rb and tb are the radius and the
thickness of the branch pipe, respectwely and Lr is the
length of the run pipe.
In the code, the radial spring constant was not
specified and the rotational spring constants are independent of the length of a run pipe However, the results by
Murad and Sun and by the proposed procedure indicate
that ot is an important factor for the radml spring constant as shown in Table 1
Insert plate reinforcements (see Figure 7) are frequently used in nuclear piping systems. The spring constants are increased when an insert plate reinforcement
is installed The effects of the thickness and radms of the
insert plate reinforcement on the spring constants are
represented in Figure 8.

Table 1 Spring c o n s t a n t s (fl = 5,-v = 50, p = 1)


Proposed model

KR/Rb (N/cm 2)
Kc/R 3 (N/cm2-rad)
KL/R,a (N/cm2-rad)

Code

a=4

a=8

73656 8
521 8
1732_3

40003.6
445 9
1672 7

-533 5
4274 1

Murad and Sun 3


c~=4

(x=8

27768 5
326 6
1273 7

16490 1
277.1
1245 3

Eng. Struct. 1993, VoI. 15, No 1 5

Seismic analysJ~ of piping systems

D G Lee and Y H Song


P

ML

Figure 6 Pipe joint model (a), with branch pipe, (b) with flcttttous plate

Eng. S t r u c t .

1 9 9 3 , V o l . 15, N o 1

Se/sm/c analysts of ptpmg systems. D -G. Lee and Y.-H. Song

Branch

pipe

Insert

plate

Run

Fmgure 7

pipe

Definition of geometric parameters

/'
6

/
/,

5 ~

5-

_2
ji
f

j J

3-

.----"

,jf'f'"

//

7
/

/
/
/

//

/
/

11/ I /

il//" / I
7/ ,11"

; p,P"

1 -

II

2-

4-

o)
c-

/I

I -

1
1

I
2
Thickness

1
3
ratto

ol

I
q

(tt/tr)

I
1

I
2
Thtckness

1
3

I
4

ratto ( t t / t r )

Figure 8 Effects of insert plate r e i n f o r c e m e n t on spnng constants w h e n c~ = 8, /3 = 5, ~ = 5 0 and p = 1 (Stiffness rat=o is defmed as
spring constant divided by spring constant w i t h o u t plate reinforcement) (a), R, = 2Rb, (b), R, = 3R~, ( ), K R, (---), Kc, ( - - ) , K L

When the stress m the branch pipe ~s not of major


interest the branch pipe can be modelled by a beam element which is connected to the run pipe by the fictitious
plate as shown m Figure 6b. Spring constants with the
ficutlOUS plate instead of the branch pipe are plotted in
Figure 9 When the thickness ratio is about four, the
stiffness ratio is very close to unity. Therefore, the
branch pipe can be replaced by a beam element when the

fictitious plate thickness (tj) is about four times the run


pipe thickness (tr) without any sigmficant deviation of
the spring constants.
In general, a piping system may have many p~pe
joints, while the number of pipe joint types are limited
Therefore, the results of the stauc analysis for the
limited number of pipe joints will be required for the
prediction of local stress in the run pipe.

Eng. Struct. 1993, Vol. 15, No 1

Selsmtc analysts of piping systems D G Lee and Y -H Song

1.0

- " , , I I.,-"

0
rQ

, /
/' //

/
L/3

0-1 F

t~

10

11

Thickness

ratio

12

(tf/t v)

Ftgure 9

Effects of thmkness of flctmus plate on spring con


stants (Stiffness ratm ~s defined as spring c o n s t a n t (Figure 6b)
d~vpded by spring c o n s t a n t w i t h branch pipe (Figure 6a)
(---),
KR, (- --), Kc, ( - ) , KL

Numerical example

shown m Ftgure 10 The material tor all ol the plpe~ is


assumed to be structural steel.
Three types of analysis models are used m this study,.
model C is the conventional beam model that cannol
account for the effect of the local deformation of the pipe
joint, while model P ts the proposed model which
utilizes the proposed pipe joint element m addition to the
beam elements A three-dimensional flmte element
model. (model F), consl,~ts of plate~qlell elements
representing the run pipe and beam elements employed
Io model the branch p~pes Finite elements mesh used l]~r
anal)s],, ol the run pipe m ]nodel F is shm~n m Fleulc
11. The number of nodes and elements used for these
models are listed m Table 2 For a comphcated piping
system, the number of nodes and elements to be used for
model F will be increased significantly while those R)r
the lOmt model used lor model P wdl not be increased
and will reqmre anal,,sls ol individual model~ which arc
,ndependent

Example structure
One of several example structures used to verify the performance of the proposed pipe joint element and analysis
model ~s shown m F~gure 10 The length, diameter and
thickness of the run pipe are 550 cm, 50 cm and 0.5 cm,
respectively, and two branch pipes are joined to the run
pipe at the pipe joint 150 cm and 350 cm apart from the
left end of the run p]pe, respectively Both ends of the
run pipe are fixed to a rigtd base. The diameter and
thickness of the branch pipes connected to the run pipe
are 10 cm and 0 5 cm, respectively, and lengths are

Frequencies and mode shapes


Analyses for model C and model P were p e d b r m e d
using the computer program (PISAP) developed for this
study, and the SAP IV computer code was used to
analy,,es model F The mare purpose of deternunmg the
frequencies and mode shapes of the example structure is
to verify the equivalence of structural properties of two
beam models and the finite element model, since the
seismic response of a structure mainly depends on the
elgenvalues and eigenvectors of the system Vibration

12

11

/~+

Zso%

Run p i p e

ZSOc'~_... 6

Ftgure 10 Examplestructure

Eng. Struct. 1993, Vol. 15, No 1

Seismic analysts of ptptng systems' D -G. Lee and Y -H Song

'A'

Details of tA~

Element S

Figure 11 Details of run p~pe for model F

Table 2 The number of nodes and elements of models

Model C
Model F
Model P
Pipe joint model used for model P

Number of nodes

Number of beam elements

Number of shell elements

26
1092
26
428

25
12
25
0

0
1280
0
520

frequencies for models C, F and P are lasted m Table 3


and corresponding mode shapes are shown in Table 4 for
the first five modes. The frequencaes of model F and P
are very close to each other and somewhat lower than
those of model C which does not account for the flexabihty of the pipe joint.
As a result, we can expect that the seasm~c response
prediction of a piping system using model P may
~mprove the accuracy significantly without markedly
increasing computational effort. The mode shapes of the
example structure demonstrate that the seismic
behaviour of two branch pipes is almost independent and
the run pipe can be considered as a flexible support to
a branch pipe when the run pipe is very stiff.

Table 3 Frequencies (Hz)


Mode

Model P

Model F

Model C

1
2
3
4
5

4
4
10
13
14

4694
4 734
10 360
13 760
14 290

5
5
14
14
16

711
742
651
778
281

378
919
956
987
180

Seismic responses
Taking the N-S component of the El Centro 1940 earthquake for the first 6 s as the input ground acceleration

Eng. Struct.

1993,

Vol.

15, No 1

SeJsmtc analysis of pJptng systems


Table 4

D G Lee and Y -H Song

Mode shapes of u~,omph q[rLlgtLIrt'

Model

t'

NIodc'l ('

//

/
/

/
Y

//

/
in the x-direction, the selsmac response t,me history
analyses are performed using three types of analysas
models and the results are shown m Ftgures 12 and 13.
Figure 12 shows the displacement ume history of node
C of the example structure shown in Ftgure 10 in the xdirection when models F, P and C are used as the
respective analysis models Good agreement is observed
between responses predicted using models F and P,
whale the result obtained using model C .s
underestimated. S~mdarly, the member end moment
ttme histories of element 1 at node A are shown in Ftgure
13 The responses obtained using model P and model F
are almost identical and that obtained using model C zs
somewhat different. Since model C neglects the local
deformation of joints, ~t gives reduced displacements
and increased member end moments compared to those
of model F.

10

Eng. Struct

1993, Vol. 15, No 1

/
//
/

Stmple evaluation of stress or dtsplacement near the


pipe connectton
Nodal displacements and member end acUons obtained
using model P may provide enough reformation for the
safety assessment of branch ptpes However, more
detailed reformation will be required for run pipes for
which local deformation and stress concentrauon can be
of major interest. Such information can easily be
obtained by the linear combination of displacements
and/or stresses, at the locatmns of interest, obtained m
the previous stage for the determination of spring constants through the finite element analysis of the p i p e j o m t
model shown in Figure 6 For this purpose, the time
histories of six components of member end actions of
branch pipe attached to the run pipe can be used as scaling factors.

Seismic analysts of piping systems. D.-G Lee and Y.-H. Song

~u

,0]

"~

0.5

AAA
vVVV Ivvrvvvvvvvv

I00

__A~A

50

Q.

_~ -0.5

E3

~^~AAA~^AA_A AAAAAAA
v VvvVvVVvvv ~vvvv~v,

-50

-1.0

-100

100-

II A All

50"

0.5

....

-0.5

I
Z

vVVV vvv vVvvvvv

~^AAAAA~^A~

^AA^AAA

~ vvvVVVVvvv ~vvvvv

E
-50

~5
-1.0

-100

b
100

1t

50

.... A AA// hh/h^^ A~A~AAAA

0
o

v vv iViVvvvvv
vvvvvvv

-50

-100
0

Ftgure 12

T,me (s)

Nodal dLsplacement of node C in x direction


model P, (b), model F, (c), model C

(a),

As an example, the major component of stress m element S(Figure 11) obtained using model P is compared
to that of the model F in Figure 14. The stress time
history obtained by the proposed model is very close to
that obtained by the model F Therefore, cumulative
fatigue damage analysis for pipe joints can be performed
very efficiently
When a piping system has more pipe joints of fewer
types, the proposed method will be more efficient
because of shorter computation time and a reduction in
the required memory size Consequently, it is essential
to account for the flexibility of pipe joints using the proposed model in the ,~elsmlc analysis of piping systems

Conclusions
An efficient analysis model with the pipe joint element
is introduced herein to account for the local deformation
of a pipe joint where a branch pipe is connected to a run
pipe with a relatively larger diameter. Seismic analysis
of an example piping system has been performed and the
following conclusions have been drawn from a comparison of the analysis results obtained using three types
of analysis models

Time (s)

Ftgure 13

M e m b e r end m o m e n t in element 1 at node A , (a),


model P, (b), model F, (c), model C

10"

E
u
Z

5-

o.

..J
un

-5

-10-

"

"

Time (s)

Ftgure 14
P, ( - -

(1)

(2)

Bending stress at e l e m e n t S m r u n p . p e
), model F

), model

Beam model with the proposed pipe joint element


leads to an accurate prediction of frequencies and
mode shapes which is essenual for the accurate
estlmauon of seismic response of a piping system
subjected to a severe earthquake
Time histories of stresses near a pipe joint can be
obtained efficiently using the proposed model and
fatigue analysis of the pipe joint of a complicated
piping system can be performed on a personal computer.

Eng. Struct. 1993, Vol. 15, No 1

11

Setsm~c analysts of ptpmg systems D -G. Lee and Y-H. Song


(3)

(4)

The proposed analysis model can provide the


dynamic response predlcUon of a piping system
with an accuracy comparable to that obtained using
a compl,cated three-dimensional finite element
model whde the slmphcity of the analysis model
and requirements for the memory size and computational time are s~mdar to those of the conventional beam model.
The proposed model will be most efficient when a
p~plng system has more p~pe .lomts of fewer types

References
I

12

Smith, P R and Van Laan, T


McGraw-Hill, 1987

~Ptptng andplpe ~upport ~ t e m ~ ' ,

Eng. Struct. 1993, Vol. 15, No 1

ASME Boder and pressure vessel code (Section II1. Subsection NBI.
ASME Boder and Pressure Vessel Committee, 1989
3 Murad. F P and Sun. B C "On radial and rotahonal spring constants
tit pipmg-miHIc~ _ PHi( 5th hit ( . n / on PJc~toc l'('~e/ TeChmdog~, ASML. Vol I. pp 85 236-251) (1984)
4 B Batra and B C Sun. 'On radial spring constant,, at the juncture
o[ a radial nozzle and a spherical shell' Pro Fifth hit Cmt] on
Pre~iure Ve,~el Te(hnologv. ASME, Vol I, pp 4 9 0 - 5 0 4 (1984)
5 Wea'~er, W Jr and Gere. J M ~Matrt~ anal~t~ oJ ]lamed strutui-e~'. Van Nostrand 1972
6 Ashwcll. D G and Gallagher. R H "Finite element~ jor thin shell~
and (urved member~ . John Wiley. 1976
7 Roark, J and Young. W C "Formula~ /or ~tte~ and ~tram
McGraw-Hill. 1975
8 Le~ck R D and Potvm A B 'Automated mesh gcnc~,tl..1 lor
tubular joint stress analysts', Comput & Stru(t 1975. 7, 7 3 - 9 1
9 Lo, S H "Finite element mesh generation over curved surfaces',
Comput & Struct , 1988. 29, 7 3 1 - 7 4 2
10 Cheung, Y K and Yeo. M F 'A practwal introduction tofimte element anal~'~t~" Pitman. 1980

Potrebbero piacerti anche