Professional Documents
Culture Documents
00
PergamonPESS plc
Abstract-The potential application of elastic-plastic analysis and design methods to tubular steel
H-frame structures is discussed. The ma~cmati~l theory is su~uently presented for a large deflection
beam element with elastic-plastic material characteristics. The numerically integrated beam element and
associated computer program are based on an incremental Newton-Raphson solution methodology. The
beam element was used to study the failure characteristics of a 500 kV tubular steel H-frame structure
and a substation take-off structure. Computed results were found to agree well with full-scale structure
test results.
design conditions [2,3]. In general, the size of tubular and differential surface area; and A is the current
transmission structures is controlled by large un- body force vector defined relative to the reference
balanced longitudinal loads. Following a failure, configuration. This equation is an integral statement
tubular steel structures behave quite differently from of equiiib~um, and states that the virtual work
lattice transmission structures. Self-supporting lattice performed by the internal stresses must equal the rate
structures generally behave in a brittle manner fol- of virtual work performed by the externally applied
lowing a failure and will not resist a signi~cant load loads. S~~Ii~lly, eqn {I) may be written as
in a post-failure state. Conversely, tubular steel trans-
mission structures fail in a ductile manner and can SkV,= SlPr, (2)
sustain a significant load following a failure. In the
light of the inherent ductility manifested by tubular
or alternatively,
electric transmission structures, it may be beneficial
to take advantage of this feature in the design
srit;- W,=O, (3)
process. This is particularly true for low probability
of occurrence loads in which the main objective is to
where
contain the failure within three or four structures.
A prerequisite for taking advantage of the ductile
failure characteristics of tubular electric transmission SW*= P&W d4 -I- 8% d4, (4)
structures is to be able to compute the pre- and j4 s B0
post-yield response of not only a single structure, but
also of several structures representing a segment of is the rate of internal virtual work, and
an electric transmission line. The required compu-
tational capabilities encompass static and dynamic
(5)
response with associated elastic and elastic-plastic
material behavior, small strains with large rotations,
and elastic/elastic-plastic stability phenomena. is the rate of external virtual work. If equilibrium
The purpose of this paper is to describe a nonlinear between the surface tractions and internal stresses is
beam element which has been developed and applied not satisfied, eqn (3) will not be equal to zero.
to the post-yield analysis of tubular steel electric Consider a Taylor’s series expansion of eqn (3)
transmission structures. The element is based on a about a state which is not in equilibrium, then
numerical integration of the governing equations and
is applicabIe to compact, cio~d-polygonal sections. “(s~,-&s~)=‘(s~,-~s~)
Comparisons between computed results and experi-
mental data obtained from two full-scale structure + -$ ‘(6 F& - 6 l&) dy -I- O(d&‘, (6)
tests conducted at the Transmission Line Mechanical
Research Facility (TLMRF) [4] are presented.
where the left superscripted number 1 or 2 denotes
GOVERNING EQUATIONS
the configuration to which the kinetic and kinematic
variables and integrals are referenced (note that con-
In this section the governing equations are figuration 1 is not in ~~~b~urn). If co~guration 2
presented and cast in a form compatible with an in eqn (6) is chosen to be a configuration which is in
incremental-Newton-type solution scheme. The equilibrium, then
Lagrangian statement of the rate of virtual work is [5]
‘(SlP,--SW,>=0 (7)
p,$ y d& + Y&i&dB,
j& j and eqn (6) becomes
_j @EdA,+ j &Wd&, 0)
L 8,
where p, is the material density in the reference The right side of eqn (8) represents the amount by
configuration; $ is the current acceleration vector at which configuration 1 is out of equilibrium and can
a material point; & is a virtual velocity vector at a be considered an unbalanced force. The left side of
material point; B, is the material volume in the eqn (8) represents a system of equations with un-
reference configuration; T is the 2nd Viola stress known displacement increments &. The displace-
tensor; S& is the virtual rate of the Green strain ment increments dp are estimates of the dispiace-
tensor &; A, is the surface of the reference ments required to bring the unbalanced forces into
~nfiguration; t, is the current surface traction vector equilibrium.
at a point on the surface, defmed relative to the Equation (8) can be generalized to yield a
reference configuration unit outward normal vector Newton-Raphson-type solution scheme in which it-
Elastioplastic analysis of tubular transmission structures 605
erations are performed to ensure ~~lib~urn; that is, is the effective load vector. In eqn (15), (F,,) is the
vector of externally applied node point loads, and
{Tf is a vector whose components are the 2nd Piola
stress tensor components. The [B], [G] and [D]
matrices are defined sym~lically as
where the superscriptj denotes the configuration used
to compute the derivatives. The superscript i denotes W1 = WI{@I (16)
the number of the equilibrium iteration. The im-
proved displacement after i ~~lib~~ iterations is (Sy:) = [Gl(SP] (17)
obtained as
(dT) = W]{dE), (18)
ELASTIC-PLASTICCONSTITUTMT
EQUATIONS: IDI MATRIX
+ [~lr[Sl~Gl d(%)
1 (14)
components of’, and g is the deviator of g which computed using the equation
represents the instantaneous center of the yield sur-
face; a, is the instantaneous radius of the yield surface {dE”“} = dA{S}, (32)
defined by where
?ss)=&Si. (24)
d’=$Hq,(r~)~+{SJT[C]{S}' (33)
/I is a parameter ranging from zero to one which
controls the amount of isotropic and kinematic hard- The following steps are used to integrate the consti-
ening used to describe the material, for p = 1.0 the tutive equations at each element integration point:
equations reduce to isotropic hardening, for /I = 0.0
the equations reduce to Ziegler’s modified kinematic (1) compute the new estimate for the radius of the
hardening, for 0.0 < B < 1.0 the equations represent yield surface using eqns (32) and (33) in combination
combined kinematic and isotropic hardening. with eqn (27);
The only non-zero components of the stress and (2) compute the new estimate for the stress com-
strain tensors, T and g, for a beam element are ponents using eqn (25);
T”, T’*, T” and E,,, E,*, E,,. Therefore, for the (3) compute the new estimate for the stress coordi-
special case of a beam element, eqns (19) through (2 1) nates defining the center of the yield surface using eqn
may be written in matrix form as (26);
(4) compute a new estimate a(q) for the radius of
C 1
PAN W[Cl the yield surface using the results from steps (2) and
WI (25)
w = [C]-~~~,(r1)2+{~}T[C]{~} (3) in conjunction with eqn (24);
(5) determine the ratio of a,(q) computed in step
(1) and a(q) computed in step (4); that is,
(26)
ratio = g ;
da,(ttI= HB h-3 (27)
1
YOO
[Cl= 0 G 0 (28) In step (l), dn is computed using a constant value
for H. The value at the beginning of the interval or
[ OOG at the point of transition from elastic to elastic-
plastic response during the interval is used. However,
Y = modulus of elasticity the value of H is updated during the integration of
G = shear modulus eqn (27). This technique provides consistent and close
approximations to the yield surface radius for curved
{dZ-}r= {dZ-“, dT12, dT’?} (29) stress-strain curves.
The stress and stress coordinates for the center of
{dEjr = {dE,,, 2dE,,, 2dE,,} (30) the yield surface computed in steps (2) and (3) are
tangent approximations. To ensure that the stresses
lie on the yield surface, they are scaled radially back
(31) to the yield surface according to steps (4) through (6).
This technique is essentially a tangent approximation
with radial correction (predictor-corrector) as dis-
The components of the plastic strain increments are cussed in [8].
TROID OF CROSS-SECTION
I- 1 2 2 3 xyj= ST + UI;,
u -UC-x u,,,-x u,,, (344 (38)
u2= uf- x’tl, Wb) where Sy is the Kronecker delta function, Substi-
tuting eqn (38) into eqn (37) yields the following
u’=u;+x%,. (34c) relationship for the virtual rate of Green’s strain:
The subscript c indicates that the subscripted variable
s~~==fIsv,,i+6~i,j+6~“,jU:j+6~=,juI”i}. (39)
is on the axis of the element which passes through the
--
centroid of the cross-section. These equations do not initial
small
account for warping of the cross-section in torsion, or displacement
deformation
for lateral shear deformation. These two phenomena contribution
contribution
are generally unimportant in tubular electric trans-
mission structures.
The deformation of the centroidal axis at any point For an updated Lagrangian analysis technique [5],
along the length of the beam is related to the node the initial displacement contribution is zero, and eqn
point displacement through the interpolation (39) reduces to
equations [9].
Pa)
For a beam element, the only non-zero virtual rate of
8, =fsn” +f6c’O Wb) Green’s strain components are 68,, , CL??,,
and Sl?,, .
Therefore
24;=f,ti2 +f2ziS +f3c6 +f4ii12 (35c)
W-4
[El= [ 00 0 0
0 X2h.I
-x3&., 0
0 0
0 0
0 0
0 0
0 a..,
-x3&1 0
0 0
0. I
WI
608 ROBERT
LEMASTER
er al.
The initial stress contribution to the tangential The integrals contained in eqns (14) and (15) are
stiffness matrix comes from the term evaluated numerically using a Newton-Cotes algor-
ithm, which can be written as
d(6~~)=f(6V”Jidu”lj+6VmI,du”l,). (46)
where Wi, Wj, W, are the weights applied to values
Using eqn (46), eqn (45) may be written as of the function F evaluated at the sample points
x’, x2, x3. This integration method requires the evalu-
ation of the [B], [D], [G] and [S] matrices at each
sample point in the element. Typical sample point
locations are shown in Fig. 2. The actual number of
= ~(6V”IiT”du”lj+6V”ljT”dumIi)d(~B,) sample points and their locations depends on the
I number of sides making up the polygonal cross-
section.
= {6V}T [GIT[sl[Gld(kB,){d~}, (47)
500 kV H-FRAME STRUCIWRE
f 5. I -Xtr,,,,
0 fl,,
0 0
0 -f,. I
[G] = 0 0
0 0
0 0
0 0
0 0 0 0 0 0 0 0
T” 0 0 T12 0 0 T13 0 0
0 T” 0 0 T’2 0 0 T13 0
0 0 T” 0 0 T’* 0 0 T13
T’* 0 0 0 0 0 0 0 0
0 T12 0 0 0 0 0 0 0 (51)
0 0 T12 0 0 0 0 0 0
T'-' 0 0 0 0 0 0 0 0
0 T13 0 0 0 0 0 0 0
0 0 T13 0 0 0 0 0 0
Elastic-plastic analysis of tubular transmission structures 609
0 MATERIAL NONLINEARITIES
* ELASTIC/ELASTIC-PLASTIC BUCKLING
\
NUMERICAL INTEGRA~ON POINTS
while the structure was exposed to a high transverse loads were held constant at the 100% load level. The
wind and unbalanced outside conductor load combi- structure remained intact and did not experience a
nation. The faiiure shown in Fig. 5 occurred in an catastropic failure during the loading sequence.
x-brace member loaded in combined bending and Eighty-eight electric resistance strain gages were
compression. The stress-strain curve of the material placed at strategic locations on the structure. Each
obtained from a sample taken from the failed x-brace gage was arranged to measure axial and flexure
member is shown in Fig. 6. Note that although the strains and utilized a separate Wheatstone bridge
0.2% off-set yield stress value for the material is circuit. Strain versus percent load plots were obtained
approximately 411 MPa (60 ksi), plastic deformation for each measurement location.
actually begins at approximately 240 MPa (35 ksi). The response of the structure was computed using
The element did compute an elastic-plastic in- the geometric and material nonlinear beam element.
stability in the x-brace at the same load level and at Figure 8 shows the geometry and location of the
the location where the x-brace failed during the test. numerical integration points associated with the
However, the program also indicated that the main dement. During the analysis, the loads were applied
compression shaft-which was loaded in combined to the structure in 10% increments, with an average
bending, compression, and torsion-would buckle at of four equilibrium iterations performed following
a 70% load level. This phenomena was not observed each increment. Execution time on a VAX 11/780
during the test, and is currently under investigation. computer was approximately one CPU hour.
An evaluation of the elastic-plastic analysis results
indicated that the most significant yielding of the
SU~TA~ON TAKEOFF STRUCTURE material occurred at the mid-span of the cross-beam.
This was confirmed by the test results which indicate
The test structure was a 115-kv substation take-off the development of a plastic hinge at the mid-span of
structure as shown in Fig. 7. The member shapes were the cross-beam.
hollow rectangular sections assembled from ASTM In Figs 9 through 14, six pairs of analysis and test
A36 plates welded together. The structure was ex- results are presented for the mid-span of the cross-
posed to a complex combination of transverse, longi- beam. The first plot in each pair shows the computed
tudinal, and vertical loads. The 100% level associated stress versus strain curve at the integration point
with these loads was dictated by current design closest to a strain gage installation. This plot shows
practice and was sufficient to cause the extreme fibers the degree of yielding occurring at that point in
in the main cross member to just reach the material the beam. In general, the yielding is very moderate
yield stress. The loads were increased simultaneously and does not exceed the 0.2% offset strain for the
from the 0 to 100% levels with hold points at 50, 75, material.
90 and 95%. From the 100% level, all loads except The second plot in each pair shows both the
the ground wire attachment points were increased computed and measured strains as a function of
simultaneously from the 100 to 200% load levels in percent load. These figures show that the computed
10% increments. The ground wire attachment point elastic-plastic strain growth followed the test
C.&S.
*8,$---D
610 ROBERT LEMASTER
et al.
data fairly well-the two exceptions are shown in indicated that the structure, as modeled, was unstable
Figs 12(b) and 14(b). in the 190% load range. However, the test structure
Figure 1l(b) demonstrates a significant amount of was capable of carrying the 200% level loads.
strain growth at each of the test hold points. This The results indicated that the structure had devel-
strain growth is attributed to creep in the test struc- oped a plastic hinge at the mid-span of the beam.
ture. The slope of the test strain versus percent load After the plastic hinge was fully developed, the beam
curve between the hold points agrees with the com- began to carry the loads in tension instead of ben-
puted slope, which indicates that the elastic-plastic ding. The transition between the bending and tension
response between the hold points was being load carrying mechanisms is the point at which the
computed accurately. No creep mechanism was structural software fails to converge. An adaptive
accounted for in the analysis. numerical algorithm which determines the load appli-
The structural analysis terminated while at- cation increment based on solution characteristics
tempting to achieve an equilibrium configuration at would have facilitated this type of computation and
a load level of 190%. The software termination is currently being implemented.
Elastic-plastic analysis of tubular transmission structures 611
r-r-i--
5oo
----I--
0 I I ! I
-.I
1.22 m
jr.0”)
it-4
1.22 n
(4‘~S=
LONG~TUO~NAL
I/ TRANSVLRSE
Fig. 8. Geometry and numerical integration points associated with nonlinear beam element.
614 ROBERTLEMASTERer al.
-50
-100
1 I
-150
-200
-250
E
-0.175 -0.150 -0.125 -0.100
STRAIN
-0.050
(10-2)
Xl o-2
0.00
- TEST
+ ANALYSIS
-0.05
-0.16
-0.20
PERCENT LOAD
Fig. 9(b). Comparison of computed and measured strain for location BI.
Elastic-plastic analysis of tubular transmission structures 615
-250
(10-2)
STRAIN
x10-2
0.00
-0.05
-0.10
2
a
z
* -0.15
-0.20
-0.25
0 100
PERCENT LOAD
Fig. 10(b). Comparison of measured and computed strain for location B2.
616 ROBERTLEMASTERet al.
(103
STRAIN
-0.05
0 50 100 150
1
200
PERCENT LOAD
Fig. II(b). Comparison of measured and computed strain for location B4.
Elastic-plastic analysis of tubular transmission structures 617
260
200
f 160
scn
s
E
u) 100
50
STRAIN (103
- TEST
+ ANALVSIS
I I I I I 1 I I I I I I I , I I
0 50 100 150
PERCENT LOAD
Fig. 12(b). Comparison of measured and computed strain for location BS.
618 ROBERTLEMAWERet al.
200
K 150
3
v)
E
a 100
50
0 I I I I I I , I I 1 I 1 I 1 1 I .
(lo-q
STRAIN
__ TEST
+ ANALYSIS
-0.05
PERCENT LOAD
Fig. 13(b). Comparison of measured and computed strain for location B6.
Elastic-plastic analysis of tubular transmission structures 619
STRAIN (103
x10-2
0.00
- TEST
+ ANALYSIS
-0.05
E
-0.10
l--
0 50 100 150 200
PERCENT LOAD
Fig. 14(b). Comparison of measured and computed strain for location B8.
620 ROBERTLEMASTER
er al.
Acknowledgements-The work reported in this paper was 4. TLMRF the Transmission Line Mechanical Research
supported by the Electric Power Research Institute under Facility. Electric Power Research Institute, Palo Alto,
contract RP2016-03. The data from the 500-kv H-frame CA (1983).
structure were obtained during a co-sponsored research test 5. K. I. Bathe, Finite Element Procedures in Engineering
between Valmont Industries, Inc., Valley, Nebraska, and Analysis. Prentice-Hall, Englewood Cliffs, NJ (1982).
EPRI. The data from the sub-station take-off structure were 6. M. Tanaka, Elastoplastic constitutive laws based on
obtained during a co-sponsored research test between the combined isotropic and kinematic hardening. Osaka
Western Area Power Authority, Denver, Colorado, and (Japan) University Faculty of Engineering Technology
EPRI. In addition to the financial support provided by these Reports, No. 25, 101-115 (1975).
organizations, the efforts of the TLMRF operations staff 7. R. A. LeMaster, Finite deformation-finite element
played an important role in the accomplishment of this formulations for elastic-visccplastic materials. Thesis
work. presented to the University of Tennessee, at Knoxville,
TN, in partial fulfillment of the requirements for the
degree of Doctor of Philosophy (1983).
REFERENCES 8. R. D. Kriea and D. B. Kriea. Accuracies of
numerical sohrtion methods for thi elastic-perfectly
Guidelines for transmission line structural loading, plastic model. J. Pressure Vessel Technol. 99, 510-515
p. 117. Prepared by the Committee on Electrical Trans- (1977).
mission Structures of the Committee on Analysis and 9. R. S. Barsoum and R. H. Gallagher, Finite element
Design of Structures of the Structural Division of the analysis of torsional and torsional-flexural stability
American Society of Civil Engineers. American Society problems. Int. J. Numer. Meth. Engng 2, 335-352
of Civil Engineers, New York (1984). (1970).
Design of steel transmission pole structures, p. 10. 10. M. A. Crisfield, A fast incremental/iterative procedure
Prepared by Task Committee on Steel Transmission that handles snap-through. Comput. Strucr. 13, 55-62
Poles of the Committee on Analysis and Design of (1981).
Structures of the Structural Division of the American 11. J. Podovan and S. Tovichakchaikul, Self-adaptive
Society of Civil Engineers. American Society of Civil predictor-corrector algorithms for static nonlinear
Engineers, New York (1978). structural analysis. Compnr. Struct. 15, 365 (1982).
Tapered Tubular Steel Structures-ANSI/NEMA 12. E. Riks, An incremental approach to the solution of
ITI-1983, p. 6. American National Standards Institute. snapping and buckling problems. Inr. J. Solids Struct.
Inc., New York (1984) IS, 529-551 (1979).