F 1/2

PIPE ELBOW STIFFNESS COEFFICIENTS INCLUDING

SHEAR AND BEND FLEXIBILITY FACTORS
FOR USE IN DIRECT STIFFNESS CODES

R. F. PERRY
Mechanical Analysis Group. United Engineers & Cons/ruc/ors. Inc..
30 South 17th Street. Philadelphia. Pennsylvania 19101. U.S.A.

SLMMAR\
Historically, developments of computer codes used for piping analysis were based upon
the flexibility method of structural analysis. Because of the specialized techniques em-
ployed in this method, the codes handled systems composed of only piping elements. Over
the past ten years, however, the direct stiffness method has gained great popularity be-
cause of its systematic solution procedure regardless of the type of structural elements
composing the system.
A great advantage is realized with a direct stiffness code that combines piping elements
along with other structural elements such as beams, plates, and shells, in a single mode!.
One common problem, however, has been the lack of an accurate pipe elbow element that
would adequately represent the effects of transverse shear and bend flexibility factors.
The purpose of the present paper is to present a systematic derivation of the required
12 x 12 stiffness matrix and load vectors for a three dimensional pipe elbow element which
includes the effects of transverse shear and pipe bend flexibility according to the ASME
Boiler and Pressure Vessel Code, Section III. The results are presented amllytically and
as FORTRAN subroutines to be directly incorporated into existing direct stiffness codes.
J. E. Srock, in the Piping Handbook, presents the 6 x 6 stiffness matrix for the straight
cantilevered pipe element and points out that the difficulties of determining and exhibiting
the stiffness matrices in literai form for more complicated elements become overwhelming
when obtaining them from the compliance (Flexibility) matrix. He further states that in-
stead, each individu al problem must be treated from a numerical point of view. In the
present paper these difficulties are overcome without the necessity of treating the problem
from the numerical view point. By selecting appropriate coordinate axes a diagonally par-
titioned form of both the force trànsfer and flexibility matrices is derived and efficiently
used to develop the complete 12 x 12 stiffness matrix. Subsequent transformations are ap-
plied to obtain the final form in the same local coordinate system used for the straight
members. The subroutines presented perform ail the required computat,ions inc1uding the
transformation of the elbow input parameters of end point and tangent intersection point
coordinates.
Aiso discussed and often overlooked are the stiffness contributions arising from the in-
teraction between bending moment and axial force as a result of the bend radius. lt is
also shown how certain stiffness coefficients may change sign depending on the values
of the bend flexibility factors.
In summary, this paper presents results useful to the analyst in modifying available
direct stiffness codes to allow analysis of pipe elbow elements in accordance with Section
III of the ASME Code.
 -2-
F 1/2

1. Introduction

The present paper presents a systematic derivation of the required 12 x 12 stiffness

matrix for a three-dimensional pipe elbow element which includes the effects of trans-
verse shear and pipe bend flexibility according to the ASME Boiler and Pressure Vessel
Code [IJ. More importantly, a computer subroutine is presented in FORTRAN IV language
which computes the stiffness matrix for irnmediate application in direct stiffness codes
such as that developed by Weaver, [2J ' entitled, "Analysis of Framed Structures - FR2".
With the elbow stiffness routine incorporated, a frame program such as this can be uti-
lized, with Sorne minor changes, to perform piping calculations in accordance with the
elbow flexibility and stress intensification'requirements of [lJ I t can th us
provide a useful tool for the design of piping systems. In addition, the possibility
of using frame members together with pipe members in a single model gives a practical
advantage over the standard piping codes which, generally,accept only pipe members.

The computation of a 12 x 12 stiffness matrix has been obtained by Morris, L3J ' for
a curved beam element neglecting the contributions of transverse shear, and f.1exibllity
factors due to cross-section ovalization in a pipe elbow. In addition, the results of
Morris, are not in a form suitable for immediate use in existing direct stiffness com-
puter codes.

Section 2 develops the derivation of the 6 x 6 Elbow Flexibility Matrix using the
atandard assumption of small displacements and a linear elastic material. The ef-
fects of transverse shear and cross-section ovalization are covered in this section.
In Section 3, the 6 x 6 rnatrix i8 inverted and the necessary Force Transfer Matrices
expressing ,static equilibrium are developed. These transfer matrices are then used
to formulate the complete 12 x 12 stiffness matrix referenced to the radial, tan-
gential, normal elbow coordinate system. In Section 4, the member degrees of freedom
typically used for straight beam pipe elements, such as in Weaver's [2J ' FR-2
progrsm are defined. Rotation matrices are spplied to the 12 x 12 stiffness matrix
previously obtained resulting in the final 12 x 12 elbow e1ement stiffness matrix
in terms of the member degrees of freedom for straight members. In this form, the
12 x 12 stiffness mstrix is suitable for direct use in the FR-2 program,simply by
using it to replace the straight beam (pipe) stiffness matrix whenever an elbow
e1ement is desired.

Section 5 presents the FORTRAN IV aubroutine used to generate the 12 x 12 elbow

stiffness matrix for direct use in the FR-2 program.

2. Derivation of the 6 x 6 Elbow Flexibi1ity Matrix

2.1 Radial-Tangential-Normal, Elbow Degrees of Freedom

Figure 1 shows a cantileveredpipe e1bow fixed at point Band arbitrari1y 10aded

at point A with three components of force and moment defined with respect to the
 -3-
F 1/2

radial, tangential and normal directions. The tangential y-axis is directed from
the A end towards the B end and the plane of curvature ia taken as the x-y plane.
The radial x-axis is directed toward the center of curvature and the normal z-axis
is chosen such that the x, y, z axes form a right handed system. The forces
acting on the elbow at end A are denoted as Fxl' Fyl' FZl snd the moments as
~l' MYl' Mzl ' The reactions acting on the elbow at end B are denoted in a similar
fashion with the subscript 2 instead of 1. The elbow subtends an angle, e, in
the x-y plane with a centerline bend radius of R. The bending moments of inertia
about both the x and z axes are denoted by, l, which is taken to be, for the
present, the same as if the elbow were straight. In section 2.3, l, will be
modified to account for the ovalization of the cross-section in accordance with
the flexibility factor specified in [lJ The cross-sectional are a of the pipe
is denoted by A, the modulus of elasticity, E, and the effective transverse shear
are a in either the x or z direction as As. The torsional constant about the y-
axis is denoted by J, and the shear modulus as G.

Let the angle 0( , measured from the A end, locate an arbitrary cross-section
and let V, P, Vn and Hn, T, M denote the three internaI forces and internaI
moments respectively acting at the cross-section on the portion of elbow con-
nected to A.

The six degrees of freedom, three displacements and three rotations, corresponding
to the three forces and three moments at sny point along the elbow, aredefined in
Figure 1 and denoted by dx ' dy ' d z ' ex' By, Bz. The subscripts 1 and 2 are used
when referring to the A end and B end respectively.

2.2 Flexibility Matrix Using the Complementary Strain Energy Theorem

When the pipe elbow of Figure 1 is loaded at A and fixed at B, the complementary
strain energy, U*, stored in the elbow is given by [4J as

U*
oc: =6
1
"2
f(V2 p2 Vn 2
GAs + AE + GAs +
Hn2
El +
T2 M2 MP )
GJ + El - 2 AER Rdo<
(1)

<><=0
The first and third terms of eq. (1) account for the in-plane and normal trans-
verse shear contributions. The second, fourth, fifth, and sixth terms represent
the axial, normal bending, torsional, and in-plane bending contributions to the
complementary strain energy respectively.

The last term in eq.(l) relates to the additional complementary strain energy
atored from the interaction of the axial force and in-plane bending moment as
a result of the elbow bend radius, R. lt is considered negligible [4J ' if
R/D> 10; where D is the pipe diameter. For a detailed discussion of this
term, the reader is referred to [5] where solid section curved beams are
treated. In the present paper, it is assumed that its effect is negligible
 -4-
F 1/2

although its inclusion presents no theoreticsl or practical difficulty.

Application of the complementary strain energy theorem [4J to the expression
U* of eq. (1) yields the following matrix equation:

In order to evaluate the above equation, it is first necessary to express the

internaI forces and moments appearing in eq. (1) in terms of the applied forces
and moments at end A. These relations follow directly from static equilibrium
of the elbow section spanned by 0< in Figure 1. They are given in matrix form
by:

~
= [U( 0<. ~
FXl]
Fyl (3)
[: Mzl

[vj~ = [v(~~ :;~

[FZl]
(4)

wnere the 3 x 3 matrices [Y(~~ and [y (oC ~ are defined as:

lOO'~
sino<..
~(oqJ -s 10 0<.

-ll.s1.noc:,
-cosO(.

R(l-cos 0<.) U (5)

~' :.<n~J
a
[v(o<.~ R8ill"'- -cos <X. ( 6)
R(l-coscx) -a in 0<. -coso<

Equation (3) expresses equilibrium within the plane of the elbow and eq. (4)
expresses equilibrium normal to the plane of the elbow. The fact that the
six equations of equilibrium can be uncoupled in this manner greatly sim-
plifies the calculations to follow. Substitution of eqs. (3) and (4) into
eq. (1), performing the necessary integrations, and then substituting these
results into eq . (2) with the required partial differentiat10n,yields the
following matr1x relations :

(7)

(8)
 -5-
F 1/2

where the 3 x 3 matrices [A] and [AN] are given by

[~ ~]
A2
CA] - A4 (9)
AS A6

AA~
AN2
[ AAI
~J = AN2 AN4 ANS (10)
AN3 ANS AN6

The A'S and AN's are presented in Table land computed in lines 1006 to 1017
of the computer subroutine in Table II.

Equations (7) and (8) may be written in partitioned matrix form as

dxl l'x
1

dyl l'y
[A] 1 [0]
Gzl 1 M~
---r --- ( 11)
d zl Fz
1
Gxl
[OJ 1 [AN] ~Jx1

Gyl 1
Hyl

wnere the partitioned coefficient matrix in eq. (11) is known as the elbow
flexibility matrix.

As will be seen in the following sections, the fact that the elbow flexi-
bility matrix is diagonally partitioned provides the simplification necessary
to easily obtain the elbow stiffness matrix.

, 3 Modification of the Flexibility Coefficients to Include the Effects

of Cross-Section Ovalization

In order to properly account for the cross-section ovalization in pipe elbows

f~r Class 2 piping, the ASME Code [lJ requires thst the flexibility factor, FF,
be used to modify the corresponding straight pipe bending moment of inertia, l,
for both in~plane and out-of-plsne bending. The computation of FF according to
Fig. NC-3673.2(b)-1, [lJ ' is given in line 227 of the subroutine of Table II.

The modification of, l, by division of the flexibility factor is performed on

line 990 of the subroutine. This modified value , XI, ie subsequently used in
aIl the computations required for eqs. (9) and (10).
 -6-
f 1/2

3. Construction of the 12 x 12 E1bo" Stiffness Matrix

3.1 Inversion of the 6 x E1bow F1exibl1ity Matrix

Equation (11) i9 now solved for the forces and moments in terms of the dis-
placements and rotations. Because of the diagona11y partitioned form of the
flexibi1ity matrix, the solution i8 ea8i1y obtained as fo11ows:

-[[:~~~l
Fxl dxl
Fy1 dyl
Mz1 Ozl
(12)
Fa]. dzl
Mxl ra] : [c] Oxl
Myl 9yl

~here,

(13)
[B] [Ar'
and,

[c] [~-I (14)

Cramers rule is then used to de termine the coefficients in [B] and [c]. The
computations are carried out in lines 1018 through 1038 of the subroutine. The
coefficient matrix thus obtained in eq. (12) is the 6 x 6 stiffness matrix.

3.2 Force Transfer Matrices

In order to achieve the complete 12 x 12 stiffness matrix, the force transfer

matrices which express equilibrium of the forces and moments at end A of the
elbow in terms of the reactions at end B are used to expand the 6 x 6 stiff-
ness matrix of eq. (12). The force transfer matrices [U(OB and [V(9)] are
obtained from eqs. (5) and (6) by setting 0(= 9.

The equilibrium conditions expressed by eqs. (3) and (4) with 0(= 9 are th en

l
stated in partitioned form as

Fx2 Fxl

l""~:
Fy2 Fyl
Mz2 [OJ Mz1
(15)
Fz2
Mx2
-[:]--i;,:; FZl
Mxl
My2 My1

The computation of the coefficients in [U(9TI and [V(9TI are performed in

lines 1039 through 1056 of the subroutine.
 - 7-
F 1/2

3.3 Construction of the 12 x 12 Elbow Stiffness Matrix

Let the 6 x 6 stiffness matrix of eq. (12) be denoted as

[k] = [c~_iJ~lJ (16)

[roJ :[cJ]
and the force transfer matrix of eq. (15) be denoted as

["rJ ~ fJul1 lOI] (17)

LL01 ! [v]
It can be shown, using the symmetry property of the stiffness matrix a10ng
with the equilibrium condition of eq. (15),that the complete 12 x 12 stiff-
ness matrix, [K] , may be expressed as fo11ows :

(18)

Becsuse of the diagonal1y partitioned form of aIl the matrices forming

eq. (18), the final form of [K] 12 x 12 is shown to be:

Fx 1 1 .
1
dx1
Fy 1 ~lJ : [OJ ,1 ~2J [OJ d)'l
Mz 1 1 1 9z1
- - - - - - 1 - - - -1- - - -
Fz 1 , 1 1 dz 1
Mx 1 [0 J 1LH~ 1 [0] : [H~ 9x 1
1 1
My l _ _ L __ L __ --' ___ _ 9y 1
(19)
, 1 1
Fx 2
T. 1 1
dx 2
Fy 2 r21 1[0] : @31 : [0] dy 2
Mz 2 1 1 1 9z2

Fz 2
--1- ~- - - ,- - - -
dû
MxZ [01 1 [Hf : [oJ : [H~ 9x 2
My 2
l , 9yz

where,

(20)

The computation of the matrices shown in eq. (20) are performed in 1ines 1171
through 1190 of the subroutine. Lines 1062 through 1134 th en rearrange the
terms of the stiffness matrix [K] to arrive at the stiffness matrix ~~ given
by:
 -8-
F 1/2

Fxl dxl
Fyl dyl
Fzl dzl

Mxl 8x l
Myl 8y l
Mzl 9 zl
(21)
[SM]
Fx2 dx 2
Fy 2 dy2
Fz 2 dz 2

Mx2 9x 2
My 2 8y 2
Mz 2 ez 2
4. Trans f ormation of the 12 x 12 Elbow Stiffness Matrix

Let the 12 x 12 stiffness matrix computed in eq. (12) be denoted as ~M(x. Y. zB.
to calI attention to the fact that it is written with reap~ct to the x, y, z axes
at ends A and B as shown in Figure 2. (Note that the directions x. y, z at end A
are not coincidentwith the directions x, y, z at end B.) It ia desired to for-
mulate the e1bow stiffness matrix in terme of the Xm. Ym, Zm member axes , shown
in Figure 2, that are commonly used to deacribe the stiffness matrix for straight
members directed trom A to B.

If ~M(Xm' Ym' zmD denotes the 12 x 12 elbow stiffness matrix referred to the
Xm, Ym' zm axes,then it can be shown that

(22)

where,

[
~~(a) o
R(a)
0
0
(23)
CR] - o R( -a)
o 0
and ,

-c0"2a
a
si~ (24)

o
 -9-
F 1/2

8 being the elbow bend angle and 0 in eq. (23) standing for the 3 x 3 null matrix.
The 12 x 12 matrix [R] i8 computed on lines 1135 through 1156 of the subroutine in
Table II and the matrix product in eq. (22) is computed on lines 1157 through 1168.
It should be noted that in the subroutine the array SM(I,J) is used to represent
first [SM(x, y, zTI and then ~M(xm, Ym, zm)]. to reduce core storage. SM(I,J) as
given on line 1168 represents the final values of the 12 x 12 elbow element stiff-
nese matrix ueed to replace that of a straight member e1ement when desired.

5. Computer Subroutine

Table II presents the FORTRAN IV listing of the subroutine used to generate the
12 x 12 elbow element stiffness matrix, Q;M(xm, Ym, zmD • The elbow element
geometry is defined by entering the coordinates of the end points A and B along
with the tangent intersection point P, in an overall global coordinate system X,
Y, Z as shawn in Figure 2. Once entered, the elbow bend angle, 8, and radius, R,
is computed and the local Xm. Ym, Zm axes are uniquely determined in relation to
the elbow as defined in Figure 2. These computations are performed on lines 127
through 162. Other portions of the subroutine have been explained in the previous
sections.

The fo11owing list defines the subroutine variables required as input data from
which aIl other variables are calculated:

l Elbow Element Number

JJ(I),JK(I) A End Joint Number, B End Joint Number
IX(I) Torsional Constant (Polar Moment of
Inertia)
IY(I) Bending Moment of Inertia
AX(I) Cross-Sectional Area
E Modulus of Elasticity
G Shear Modu1us
AC(I) Inside Pipe Diameter of Elbow
AE(I) Outside Pipe Diameter of Elbow
AR(I) Pipe Wall Thickness of Elbow
X(JJI),Y(JJI),Z(JJI) Global X, Y, Z Corrdinates of End A
X(JKI) ,Y(JKI) ,Z(JKI) Global X, Y, Z Coordinat es of End B
XP,YP,ZP Global X, Y, Z Coordinates of Point P
 -10-
F 1/2

REFERENCES

ASME Boiler and Pressure Vessel Code, Section III, Division l,

Subsection NC-3600, New York, 1974

WEAVER, W., "Computer Programs for Structural Analyses", Van

Nostrand, Princeton, NJ, 1967

MORRIS, D.L., "Curved Besrn Stiffness Coefficients", Journal of

the Structursl Division, ASCE, May 1968, Pgs. 1165-1174

NORRIS, C.H., WILBUR, J.B., UTKU, S., "Elementary Structural

Analysis", Third Edition, McGraw-Hill, New York, 1976 Pgs. 380
and 489

[5) SEELY, F.B., SMITH, J.O., "Advanced Mechanics of Materia1s",

Second Edition, John Wiley, New York, 1952, Pgs. 167-173

BROCK, J .E., "Expansion and Flexibility" in Piping Handbook,

Fifth Edition edited by R.C. King, McGraw-Hi11, 1967
 -11-
F 1/2

F1exibi1ity Matrix Coefficients For

Equations (9) and (10)

Al _ (R3 +~) (~_ sin29) +~ (~+ sin29 )

El AE 2 4 GAs 2 4

(cos9 - 1)

R2
A3 - El (1 - cos9)

A4 _ R) (9 _ 2sin9)J R3 +~)(~+ Sin29) +~(~_ Bin29)

El "'\ El AE 2 4 GAs 2 4

A5 - ~~ (sin9 - 9)

R
A6 = El 9

ANI _ R3 ( ~ _ sin29)
El 2 4
+GJ
R
3
( 39 _ 2sin9
2
+sin29)
4
+~
GAs

R2 R2) Bin 29 R2
AN2 - ( GJ - El - 2 - + GJ (cos9 - 1)

2
AN3 _ R
2
(~_ sin29) + R ( ~ + sin29 _ Bin9)
El 2 4 GJ 2 4

AN4 _ ~ ( !!. + sin29) +~ ( ~ _ Sin29)

El 2 4 GJ 2 4

AN5 -(~
GJ
-~)
El
 -12-
f 1/2

Table Il

Fortran IV Listing of the Subroutine Used

ta Generate the 12 x 12 Pipe Elbow Stiffness Matrix
REAIl(9,1100) I,JJ(I),JK(I), AX(I), IX(I),IY(I), UlPR0127
1 AC(I),AE(I),AR(I) HlPROl28
IF (JJ(I) .LI. JK(I» GO TO 130 IMPROl29
JJI - JK(I) IMPROl30
JK(I) = JJ(l) IMPROl31
JJ(I) = JJI IMPR0132
130 JKI JK(I) IMPROl34
JJI JJ(I) IMPROl35
XCL X(JKI)-X(JJI) IMPR0136
YCL Y(JKI)-Y(JJI) INPR0137
ZCL Z(J~I)-Z(JJI) IMPR0138
L(l) = SQRI(XCL*XCL+YCL*YCL+ZCL*ZCL) IMPROl39
CX - XCL/L(l) IMPROl40
CY = YCL/L(I) IMPROl41
CZ = ZCL/L(l) IMPROl42
REAU (Y,1070) I,Xl',YP,ZP IMI'ROl53
YP5 - YP-Y(JJI) IMPROl57
XPS = Xl'-X(JJI) IMPROl58
zrs = Zr-Z(JJI) IMPROl59
AP - SQRT(XPS*XPS+YPS*YPS+ZP5*ZP5) IMPR0160
THETA(I) = 2.*ACOS«XCL*XPS+YCL*YPS+ZCL*ZPS)/(L(I)*AP» IMPR0161
RAIlIUS(I) = AP/TAN(TI1CTA(I)/2.) HlPROl62
2 lOf F (I ) = 1 • 6 5 *( **
(A E ( 1 )+ AC (I ) ) / 2 • ) 2 • / (4 • * A R ( 1 ) * R AllI US (I ) ) Hl PRO 2 2 7
S F ( 1 ) =.9/ ( ( 16 • *P,AIJ 1 US ( 1 ) *AR ( 1 ) / (AC (1) +A E (I ) )** 2. ) **.66666666667 l MPRO 2 28
IF(FF(I).LT.I.)FF(l)=I. IMPR0229
lF(Sf(I).LI.I.)SF(I.)=I. II-IPU0231
C CO?lJ'UIJ: ELllOli 5TIFFHE5S HAlnx. INPR0291
300 CALL STlFF2(SII,RAllIUS(I),TfIETA(I),AX(I),FF(I),IY(I),IX(I),E,G) 1MPR0292
S IIIlIUl U T 1 ~ EST 1 F F 2 ( S 11 , Il , 0 , A , F F , YI, X J , E , C ) 1 M PRO 980
IMPLICIT ~EAL*8(A-I1,O-Z) 1MPR0981
C INCLlllJES SlIEAR DEfORHATI011 ANI> FLI:XlijlLITY FACTORS. 111PR0983
lJUICNSlotJ RE(l2,12), S~I(12,12), StlX(12,12) nlpR0984
lJ HI ENS ION 1\ (3 , ) ), C ( 3 , 3 ), U ( 3 , 3), V ( 3 , 3 ), C 1 (3 , 3), C 2 ( 3 , ) ), C 3 ( 3 , 3 Hl PRO 985
1) 1~fP[(0986
I>Hlt:tJSIIlN XH!(3,3), XI12(3,3), XI13(3,3) 1MPR0987
TII~TA = U 1MPR0989
XI = Y1/rf I~PR0990
AS = A*.504 IMPR0992
AS!J = AS IHpR0993
REl = K/(r,*X1) 1HPR0994
R 2 E 1 = p, E 1 *1: IMPR0995
R3U = R2U*n l~PR0996
RAt: - Il/(A*!:) 1NPR0997
S1M2u4 = Sl~(U+U)/4 HIPRO 9 y 8
02 - 0/2 111PR0999
RCAS rr/(G*AS) HIJ'R 1 000
CUf,O = COS(u) IIIPRI001
S U:O = S lN (0) lMPRI002
RGJ = Il/ (C*XJ) UlPRI003
P.2G.! = Il*IlGJ HIPRI004
IUCJ = It*1l2GJ l!IPIt 1 005
AI (R3El+llAE) * (U2-SIN20 I,)+!tCAS* (U2+SIN204) IHPRI006
A2 z (1',3E1+RAI~-llC;AS)*SIII()*SINU/2.+R3El*(C(JSO-1.) 1MPRI007
Al R2EI*(I-CUSO) lMPRI008
A4 n3EI*(U-2*SIN(O»+(RlEI+KAE)*(U2+SIH204)+UCAS*(02-SIN204) IHPRIOOY
AS R2[1*(SI~(O)-O) IHPRI010
Ab RE1*0 IHI'RIOII
 -13-
F 1/2

Table i l (COJ1t ' ù)

ANI - R3EI*(UZ-SINZ04)+R3GJ*(OZ*3.-2*SINO+SINZ04)+R*O/(G*ASN) IMPRIOIZ

AN2 (R2GJ-R2EI)*SINO*SINO/2.+R2GJ*(COSO-l.) IMPRI013
AN3 - R2EI*(02-S1N204)+R2CJ*(OZ+SIN204-S1NO) IMPRIOl4
AN4 - KEI*(02+S1NZ04)+RGJ*(U2-S1N204) IHPRI015
AIlS - (RGJ-REl)*SINO*SINU/2. I11PRIOl6
AN6 - REI*(OZ-SIN204)+RGJ*(OZ+SIH204) IHPII017
o • AI*(A4*A6-A5*A5)+AZ*(A3*A5-A2*A6)+A3*(A2*A5-A3*44) IHPRI018
B(1.I) - (A4*A6-A5*A5)/1J 1~IPRI019
6(1,2) (A3*A5-A2*A6)/1J IMPRI020
8(1,3)· (A2*A5-A3*A4)/JJ I~H'RI021
8(2,2) - (AI*A6-A3*A3)/lJ IHPP.IOZ2
n (2,3) - (AZ*A3-AI*A5)/D HIPRIOZ3
!I"(3, 3) - (AI*A4-AZ*AZ)/1) IMPRIIJZ4
5(2,1) - B(I,Z) IMPRIOZ5
K(3,1) - 8(1,3) 1MPRIOZ6
»(3,2) - B(2,3) IMPRIOZ7
ON • ANI*(AN4*AN6-AN5*AN5)+ANZ*(AN3*AN5-AN2*AN6)+AN3*(ANZ*AN5-AN3*IMPRI028
1 AN4) 1MPII029
C(I,I)· (AN4*AN6-AN5*AN5)/ON HIPRI030
cU,Z) - (AN3*AN5-ANZ*AN6)/lJtl IHPRI031
CC 1,3) - (ANZ*AN5-AN3*AN4)/IlN INI'HI03Z
C(Z,2) D (ANl*AN6-AN3*AN3)/Otl IMPRI033
CCZ,3) = (ANZ*AN3-ANI*AN5)/lJN HlPRI034
CC3,3) = (ANI*AN4-AIIZ*AN2)/lJN 1)IPRI035
CC2,1) C(I,2) IMPRI036
C(3,1) - C(I,3) IHPRI037
C(l,2) - C(Z,3) IMPR1033
U (1,1) - -COSO IIIPRI039
U(Z,I) - -SINO IMPRI040
V(3,1 ) = -SINO*R I:!PRIOl,1
V(I,Z) - SINO IHPRIOl,2
U(Z,Z) -COSO IMI'RI043
U(3,2) R*(I-CuSO) IHPRI044
VCI,3) - a IMPII045
V(Z,3) - 0 II1Pl\1046
V(3,3) - -1 IMPRl047
V(I,I) - -1 I111' RI048
V(Z,I) • K*SINO IHI'RI049
V(3,1 ) - R*(I-COSO) INPRI050
V(I,Z) - a I11PRI051
V(Z,2) - -COSO 1MPRI05Z
V(3,Z) -SINO IMPRI053
V( 1 , 3) - a HI P!{ 1054
V(Z,3) • SINO IMPRI055
V(3,3) - -COSO IMPRI056
CALL CGGlIli1I (GI,GZ,G3,B,U) 1~11'!l1057
CALL GGGHlill (XIlI,X1l2,XlI3,C,V) HIl'RI05B
DO la 1=1;12 IHPRI059
DO 10 J-I,12 IMPRI060
SM(I,J) - 0.0 1MPRI061
ID RE(I,J) - O. 1~PRI062
SM ( 1 , 1 ) G1 ( l , 1 ) H! l' P. 1 063
StI(I,2) = GI(I,Z) Hll'RI064
SH(Z,I) = Gl(I,Z ) PIPRI065
S!I(1,6) = GI(I,3 ) 1~!PP.I066
SN(6,1) = C1( I ,3) 1HPRI067
S fi '( 1 , 7) - C Z ( 1 , 1 ) l}! l' 1\1 06 8
SI1(7,1) = CZ(I,I) I?11'RI069
SI\(I,8) CZ(I.Z) 1:II'RI070
Sil ( 8 , 1 ) - G 2 ( l , 2 ) PlI' RIO 71
 -14-
F 1/2

T~ble II (Cont 'd)

5M(I,12) = C2(I,3) 1'11'1: 1 07 2

511(12,1) = G2(l,3) l':I'RI073
511(2,2) = GI(2,2) 1~il'l:1D74
SU(2,6) GI<2,3) 1'-1 l' R 1 CI 75
511(6,2) 2 GI(2,3) l'Il' Hl (17[,
S1I(2,7) = G2(2,1) 1!!1' !', 1 077
SH(7,2) G2(2,1) HIPR1078
5"(2,8) = C2(2,2) l'fPP-l079
SU(8,2) 2 C2(2,2) l'!I'P.I fJHO
511(2,12) = G2(2,3) l'II'Rl0S1
SH(12,2) = C2(2,3) l'Il' ~ 1 () 1.: 2
SH(3,3) XIlI(I,I) 1'11'Rl0113
SH(3,4) • XIlI(I,2) UII'Kl084
SH(4,3) • X!II(I,2) HII':~ 1085
SH(3,5) - Xlll(l,3) HII'Rl086
5lf(5,3) Xtll(l,3) 1'11'~10n7
SIID,9) XH2(l,l) 1'IPI>.10HPo
SH(9,3) XIl2(I,I) UIPRlOH9
511(3,10) XH2(J,l) Hll'l·:l090
SH(IO,3) = X1l2( !, 2) IèlPR1091
St1(3,ll) = XI12(l,3) HIP\( 1 O~ 2
511(11,3) = XH2(!,3) H'I'~, 1 09 3
SM(4,4) Xlll(2,2) 1'1 PP.l 094
S:1(4,5) = XIll(2,3) l'H'!! 1 09 5
5l1(5,4) Xlll(2.3) HI P~ 1 09 6
SH(4,9) XII2(2,1) PIPR 1 09 7
SIl(9,4) = XIl2(2,1) FIPI~JOYH
5M(4,10) XIl2(2,2) IHPRI 09 ')
SI1(10,4) XH2(2,2) ' 1Nl'RI100
SI1(4,11) XI\2(t,3) IHPRIUlI
SI1(1I,4) XII2(l,3)
2 l'IPRll02
SI1(5,5) - XHl(3,3) H!PP.lI03
SI1(5,9) = XIl2(3,1) HIP\( 1104
SI1(9,5) • XII2(3,1) !lI l'Il 1105
511(5,10) - XIl2(3,2) l'IPR J 106
5}1 (10 , 5) = X Il 2 ( 3, 2 ) 11-11'1\1107
SI1(5,11) = XI\2(3,3) IN PI! 1108
SH(II,5) = XIl2(3,3) HIPRll0~
511(6,6) = GI(3,3) 1'"1I'Rl110
SI1(6,7) C2(3,1) 1 '1 PI: 1 III
SI1(7,6) - G2(3,1) HII'I',1112
SI1(6,8) - G2(3,2) HII' R 1113
511(8,6) G2(J,2) 1'1J'U 114
SH(6,12) = e 2 ( 3, 3) IHPR1115
SI1(12,6) • C2 (3, 3 ) 1:11'1\ 1116
SM(7,7) • C3(I,I) DIPR1117
511(7,8) = G3(I,2) DIPltl11H
511(8,7) = G3(1,2) 1 HP!: 1119
511(7,12) = C3(I,3) INl'Rl120
S!l(12,7) = C3(1,3) 1 Il P I~ 1 121
SH(B,B) = C3(2,2) HIPR1122
511(8,12) = C3(2,3) HIPRI123
SII(12,8) = C3(2,3) PIPIt! 124
S)1(9,9) = XH3(l,l) IHpl{ 112 <;
511(9,10) = X1l3(I,2) l:IPR 1126
5H(10,9) - X1l3(I,2) 111PR1127
SIl (9 , 11) - XII3 ( j ,3) I1of1'Rll111
SI1(1I,9) = XII3(! 3) 1 II Ptt 1129
511(10,10) = XI\3(~,2) IHI'R1130
5H(10,ll) = XIl3(2,3) 1:11'1' 1131
 -15-
F 1/2

Tabla II (Cont'd)

S1I(II,10) = X1I3(3,2) II1PRII32

Sil ( 1 l, 1 1) = XII3 ( 3 , J ) Il!PRl133
S LI ( 1 2 , 12) = G3 ( 3,3 ) 1:1PRl134
SU - SI~(TlIETA/2.) IMPRl135
CU = CUS(THf.TA/2.) IHPRl136
!tE(I,I) = SU EIPRl137
RE(2,2) = SU HIPRll38
RE(4,4) ~ su IHJ?R1139
IU~(5,5J SU PIl'Rl140
RE(7,7) -su HIPRl141
RE(8,S) • -su IMPI\1142
RE( 10, 10) = -so IHPR1l43
RE(II,I1) = -su IIIPR1144
lt E (5,4) CO HlPR1145
RE ( :l , 1) = CO HIPRI146
R1:(B,7)=CU IMPRl147
RE(II,IO) = CU HIPRl148
R1:(I,2) = -cu 1MPRIl49
RI::(4,5) - -CU 1MPRll50
Re(7,8) = -cu IMPRl151
RI:( 10,ll) = -cu 1MPRI152
RE(3,3) z 1.0 HIPRl153
RE(6,6) - 1.0 1MPRl154
RE(9,9) = 1.0 U!PR1155
RE(12,12) = 1.0 IlIPRl156
DU3IlL-I,12 1MI'Rl157
DU 30 J-I,ll 1MPRl158
SUII = O. 1llPR1159
D020L~-I,12 IMPRl160
20 SUN = SUM+St1(L,K)*Hf,(K,J) HIPR1161
30 SHX(L,J) = SUM IMPRl162
DO 50 1-1,12 UIPRl163
DO 50 J-l,12 HlPRl164
SULI = O. IMPRl165
DO 40 L-I,12 HIPRl166
40 SUM = SUM+ltE(L,I)*SLlX(L,J) IMPRl167
50 SlI(I,J) - SUt! 1HPR1168
RETURN 1MPRl169
END UIPRl170
SUHROUTINE CCGlIllll(GI,G2,C3,B,U) HIPR1171
IMPLICIT REAL*8(A-H,O-Z) IIIPR 11 7 2
DUIF.NSION Gl(3,3), C2(3,3), G3(3,3), B(3,3), U(3,3) U1PRl173
DU 10 1=1,3 HIPR 117 4
DU 10 J-l,) IMPRl175
ID f,l(I,J) - Jl(I,J) IMPR1176
DU 30 1-1,3 IMPRI177
DO 30 Jel,) 1MPRIl7B
SUM = O. HlPR Il 7 9
DO 20 [(-1,3 U!PRII80
20 SUM - SUII+CI(I,K)*U(J,K) IIIPR11B1
30 G2(I,J) - SUI! IMPRl182
DO 50 1-1,3 IMPR11B3
DU 50 J-1,) IMPR1184
Sur-! = o. IMPRl185
UO 40 K-l,3 IMPRl186
40 SUII = SUM+U(I,K)*G2(K,J) IMPR1187
50 G3(I,J) - suri 1MPR11BB
RETURN H!PRl189
ENU IMPRl190
 9~
~ d
.....--z.normal d
PIPE ELBOW dx Y
CROSS-SECTION
/ex \Sy
z.z
X
iOd;ol y"on 9\;01

-- - ~ F.
zl PIPE ELBOW
X Mz l
Fx1
..--
M

'"
~n
Mx1

Fz2
R Mz 2 ~

, '" / ' F.2

\MY2

Figure 1 - Pi
-~
pe Elbow El ement Definin , oments ~ Di splacements,
and
Rotations g Forces M
-
N
 -17-
F 1/2

x
Ym

y P (tangentia 1 intersection point)

PI PE ELBOW
CENTER LlNE

c
Figure 2 - Pipe Elbow Element Defining the Relationship Between the
Various Coordinate Axes