Sunteți pe pagina 1din 6

INELASTIC ANALYSIS AND DESIGN: FRAMES WITH MEMBERS IN

MINOR-AxIS BENDING

By Ronald Do Ziemian1 and Alan Ro Miller/ Associate Members, ASCE

ABSTRACT: To extend previous research on obtaining designs that rely on a second-order inelastic hinge
analysis to demonstrate that a structure is adequate in resisting the effects of factored loads, this paper presents
a study of a series of two-story planar steel frames where all columns are subjected to combinations of axial
force and minor-axis bending. Inelastic designs, which are prepared to satisfy both strength and serviceability
limit states requirements, are compared to designs obtained using a conventional second-order elastic analysis
in conjunction with the AISC LRFD Specification. The degree of accuracy and reliability of using a concentrated
plasticity analysis to model system behavior is also established. The study indicates that for a majority of the
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

frames investigated, a second-order inelastic hinge analysis can be used effectively to design steel structures. It
is also shown that in some cases a distributed plasticity analysis may be required.

INTRODUCTION Load and Resistance Factor Design (AISC LRFD) Specifica-


tion (Load 1994). The degree of accuracy and reliability of
Engineers continue to research and develop more compre- using a concentrated plasticity analysis to model system be-
hensive methods of inelastic analysis that take full advantage havior is established by comparing results with two different
of the power and interactive graphics available within modern levels of distributed plasticity analysis methods. Consistent
computing platforms (Porter and Powell 1979; Orbison 1982; with previous work, this study indicates that, in general, in-
AIMashary and Chen 1989; King et al. 1992; White 1988; elastic analysis can be used effectively to design steel struc-
Attalla et al. 1994; Clarke 1994). Where these methods of tures that are efficient, rational, and reliable. It is shown further
analysis provide realistic assessments of the ultimate load be- that in some cases, a distributed plasticity analysis is required
havior of steel structures, they inherently afford an ideal basis to accurately capture the limit state behavior of steel structures
for a limit states design process. This behavior includes a non- with members subjected to combinations of axial force and
linear load response, which can be attributed to second-order minor-axis bending.
effects associated with changes in the structure's geometry and
material yielding of structural components. Producing a design DESCRIPTION OF ANALYSIS PROGRAMS
that accounts for this type of limit state response necessitates
the use of either conventional linear elastic analysis, supple- The analysis and design routines primarily used in this study
mented by empirical or judgmental allowances for nonlinear- are contained in the computer program, CU-STAND (Hsieh et
ity, or a nonlinear analysis. al. 1989). The program contains incremental analysis routines
This paper extends research that focuses on obtaining de- that can calculate realistic limit state responses for two- and
signs that rely on a second-order inelastic analysis to demon- three-dimensional steel frames subject to static loads. Nonlin-
strate that a structural system and its components are adequate ear geometric behavior is included by use of element geomet-
in resisting the effects of factored loads (Ziemian et al. ric stiffness matrices combined with an updated Lagrangian
1992a,b). A majority of these studies focus on cases where all formulation. Material nonlinear behavior is incorporated by
members are subjected to major-axis bending. Other previous using a concentrated plasticity (inelastic hinge) model. This
studies indicate that the response of beam-columns subjected model employs a full plastification surface to account for the
to minor-axis bending in frames of little to no redundancy can material nonlinear interaction of axial force and bending. To
be very sensitive to the refinement of the inelastic analysis account for the destabilizing effects of thermal residual
procedure employed (White et al. 1991). It has also been stresses, a tangent modulus approach is used. This approach
shown that significant differences may exist between these reduces the elastic modulus of a member according to the
varying analytical results and current design requirements, in- Heavy or Light Gravity Loads
cluding column curves and interaction equations (White and
I I I I I I +1 2.50 kN
Chen 1993). Using a series of 16 two-story, two-bay planar
frames (see Fig. 1), with all columns oriented to resist com-
Heavy or Light Gravity Loads
binations of minor-axis bending and axial force, this paper
provides additional insight into this area, especially for frames j j I I I j j ±29.19 kN
,-
of a significantly higher degree of redundancy.
E
--I-- cry =250 MPa (36 ksl)
In this study, a second-order concentrated plasticity (inelas-
tic hinge) analysis is employed to obtain designs that satisfy
both strength and serviceability limit states requirements.
-o
%
I

I
=345 MPa (50 ksl)
'-.
These designs are compared with member proportions ob-
tained using a conventional second-order elastic analysis in ~o-----..j..o-------__-jl
r 0 • • •
Pinned or
Fixed Base
Symmetric: 10.36 m 10.36 m
conjunction with the American Institute of Steel Construction Unsymmetrlc: 6.10 m 14.63 m

'Asst. Prof. of Civ. Engrg., Bucknell Univ., Lewisburg, PA 17837. Gravity Load Condition
'Struct. Engr., Pennoni Associates, Philadelphia, PA 19130. Hea Light
Note. Associate Editor: W. Samuel Easterling. Discussion open until Dead Load:
July I, 1997. To extend the closing date one month, a written request Btm Level 74.66 kNlm 22.40 kNlm
must be filed with the ASCE Manager of Journals. The manuscript for Top Level 34.84 kNlm 11.20 kNlm
this paper was submitted for review and possible publication on February Live Load
Btm Level 39.81 kNlm 11.94 kNlm
2. 1996. This paper is part of the Jourrud of Structural Engineering, Top Level 18.58 kNlm 5.97 kNlm
Vol. 123, No.2, February, 1997. ©ASCE. ISSN 0733-9445/97/0002-
0151-0156/$4.00 + $.50 per page. Paper No. 12555. FIG. 1. Dimensions and Loads of Low-Rise Planar Frames

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 1997/151

J. Struct. Eng. 1997.123:151-156.


amount of axial force present. The program can also incor- M
porate the effects of initial imperfections such as member-
sweep and frame out-of-plumb. , Inelastic Hinge (CU-STAND)
To verify the accuracy of the results obtained by the
inelastic hinge analysis, two types of distributed plasticity
analyses are also performed. In the inelastic zone analysis, tr·"Clnelastic Zone (NIFA)
which is performed for all of the frames, a fiber element model
is used to include explicitly the spread of yielding throughout
. ~ __ . _ . . __ .2[ =- ~.~(jy
the cross section and length of a member, as well as to account
directly for the effects of thermal residual stresses. The in-
.
elastic zone analysis routines used in this study are contained
in the computer program, NIFA, developed at the University
of Sydney (Clarke 1994). For some of the frames a quasi-
: ~J_/,,- -H-
plastic hinge analysis (Attalla et al. 1994) is also performed.
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

In a quasi-plastic hinge analysis, element stiffness is also grad-


ually reduced to account for the effects of partial yielding but
without the complexity of a fiber element model. This analysis eP End Rotation, e
method also incorporates the effects of thermal residual
stresses in a more direct manner than does the inelastic hinge FIG. 2. Moment-Rotation Relationships for Minor-Axis Bend-
model. For the planar frames investigated in this study, a Ing
quasi-plastic hinge analysis provides the accuracy of an in- P
elastic zone model with the computational efficiency of an Pcr/Py
inelastic hinge model.
1.0+--~------\,---- I
With respect to the inelastic hinge analysis used to design
the frames in this study, two observations basic to this paper
0.8
%
are noted. First, for cases of pure minor-axis bending, the in-
W10x49
elastic hinge model does not fully capture the potentially sig- Or = 0.3oy
0.6 00 = Ll1000 sinusoidal
nificant level of gradual yielding that may actually exist for with 20 elements
members with a large shape factor, which equals ZlS, where
0.4
Z is the member's plastic section modulus and S is the elastic
section modulus (see Fig. 2). Second, for cases of axial com- 0.2
pression, the second-order inelastic hinge model does have an
inherent column curve that, due to the use of a tangent mod-
ulus approach, is not limited to being Euler elastic-perfectly 40 80 120 160 200
plastic (see Fig. 3). However, the tangent modulus model does Ury
not represent the additional distributed plasticity effects caused FIG. 3. Column Curves for Minor-Axis Bending
by the combination of bending and axial compression. For a
column slenderness between 60 and 100, this limitation can
result in the inelastic hinge model's curve being liberal in 1.0
comparison to the AISC LRFD and inelastic zone column
curves. r Nominal Surface
(CU-STAND)
It should be noted that the inelastic hinge column curve does
not approach unity, Pc,IPy = 1.0, as Llry approaches zero be- P/Py
cause of an attribute of the full plastification surface employed
in CU-STAND. As a compromise to providing a single con- 0.6
tinuous surface, which results in significant computational sav-
ings by avoiding problems often encountered with faceted sur-
faces (Orbison et al. 1982), the CU-STAND full plastification 0.4 Design Surface
surface undervalues the theoretical axial yield load by about (<I>b = 0.90, <l>c = 0.85)
7% (see Fig. 4). Noting this variation, Orbison (1982) shows
that this surface is accurate for light- to medium-weight Amer-
ican wide flange sections. In any regard, this distortion of the 0.2
full plastification surface and its effect on the inelastic hinge
column curve is significant only in the range of L/ry less than
50. Further, it is not of concern in this study because in all 0.0 +-~--.------r--.--------r--L-+­
cases the AISC LRFD plastic design requirement of limiting 0.0 0.2 0.4 0.6 0.8 1.0
axial force in the columns to 0.75<1>cAg(J"y is strictly adhered.
MylM py
DESCRIPTION OF FRAMES AND LOADING
FIG. 4. Comparison of Nominal and Design Full Plastlflcation
As shown in Fig. 1, a series of planar frames representative Surface.
of bents found in typical low-rise industrial buildings is stud-
ied. By varying combinations of geometric configuration (un- of-plane behavior. Beams are oriented with their webs in the
symmetrical and symmetrical), base fixity (pinned and fixed), plane of the frame (resulting in major-axis bending) and the
material yield strength [248.2 and 344.7 MPa (36 and 50 ksi)], columns are oriented with their webs perpendicular to the
and load intensity (heavy and light), sixteen different frames plane of the frame (resulting in minor-axis bending). All con-
are investigated. The frame geometries are similar to a set nections are assumed to be rigid. Only bare steel frames are
originally studied by Ufland and Birnstiel (1982). considered; no composite interaction with floor or roof deck
All members are assumed to be fully restrained against out- is modeled. The distributed gravity loads shown in Fig. 1 are
152 I JOURNAL OF STRUCTURAL ENGINEERING I FEBRUARY 1997

J. Struct. Eng. 1997.123:151-156.


lumped according to tributary length and applied as concen- _ _ _.....:1.:.::.0T_-;:i~~#6
trated loads at the beam quarter points.

DESIGN REQUIREMENTS
In designing the frames, both strength and serviceability re- ____ !i~;;I1 d t
Applied Load
0.6 Ratio, A
quirements are satisfied. According to the AISC LRFD Spec- #6 I~
ification, the underlying strength requirement observed is W36x210 I~

2: 'YiQi :5 <j>Rn (1) #3

where Qi = nominal load effect; 'YI =load factor; R n =nominal cry = 250 MPa

resistance; and <l> = resistance factor. The left-hand side of (1)


represents the factored load effects and the right-hand side
represents the design resistance terms. This basic strength re-
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

quirement must be satisfied at both a local level (e.g., connec- -2.0 -1.0 0.0 1.0 2.0
tions and members) and at a global level (e.g., subassemblages d t (cm)
and overall system). It should be noted that by performing a
second-order inelastic analysis, (1) is satisfied explicitly at FIG. 5. Load-Displacement Curve and Hinge Formation Se-
both levels. This is in contrast to contemporary elastic analysis quence (1.2 Dead + 1.6 Live)
and design approaches, where (1) is satisfied only implicitly # - #6-1.00
at the global level. It is generally assumed that if the member-
C4 0'~5 #4 - 0.99
by-member checks are met, then the overall system strength 1.0 #1 _ #3-0.99
requirement for the structure is satisfied. P/Py Design Full 0.7
In this study, the following AISC LRFD factored load com- / Plastlflc8tlon
#5-0.98
binations are investigated: 0.8 . -!.-. Surface C1

1.2 dead + 1.6 live 1.2 dead + 0.5 live + 1.3 wind (2)
\ .. ---....,.
0.6 \ Initia "'"'-,
To insure that the designs of these industrial buildings have (Yield of C5(btml'
adequate stiffness to function properly at service loads, three i'iect1on
0.4 \ C2(top)
requirements are imposed: (1) total lateral drift and interstory _ Applied Load
drift due to the unfactored wind load are limited to H1250, \ =
Ratio 0.90
where H is either the height of the structure or the story height 0.2
(ASCE 1988); (2) beam deflections under unfactored live
loads are limited to Ll360, where L is the beam span; and (3)
plastic hinges are prohibited from forming under service loads. 0.0 -F'~--,-~-r-'-.-'''=r=~---,->.--,­
0.0 0.2 0.4 0.8 1.0
DESIGN ALTERNATIVES
FIG. 6. Force Point Traces for Column Sections (1.2 Dead +
As a basis for comparison, two least-weight designs that 1.6 Live)
satisfied the previous limit states requirements are prepared for
each frame REPRESENTATIVE EXAMPLE
1. Integrated Inelastic: A design obtained using all of the Representative of the general findings, which will be in-
AISC LRFD Specification provisions for plastic analysis cluded in the next section, the results for one of the sixteen
with the exception of the beam-column interaction equa- frames investigated are provided. In this example, frame U-
tion. In place of this design check, a second-order in- P36H, an unsymmetrical frame with pinned base conditions,
elastic hinge analysis is used to demonstrate the ability a material yield strength of O'y = 242.2 MPa (36 ksi), and heavy
of the frame and its members to resist factored loads, gravity loads, is presented.
thereby satisfying the underlying strength requirement of For the factored gravity load combination given in (2), the
(1). In applying this integrated analysis and design ap- load-displacement response curve and hinge formation se-
proach, all member and frame destabilizing effects are quence shown in Fig. 5 are calculated for the Integrated In-
assumed to be represented on the load effects side of (1), elastic design alternative. Note that the member sizes obtained
and hence, the resistance terms reduce to expressions for using this design approach are also shown in Fig. 5. As in-
member cross section design strengths. Only compact dicated in the response curve, the first plastic hinge forms in
sections are used to assure adequate ductility during any the lower left beam at a load just above service gravity (11. =
substantial inelastic deformations. A member's cross sec- 0.73). After a significant level of nonlinear response, the limit
tion design strength is incorporated into the analysis by of resistance of the Integrated Inelastic design is reached at
reducing the nominal full plastification surface used in full application of the factored gravity load. As shown by the
the inelastic hinge model by the appropriate AISC LRFD unbounded lateral deflection, the frame fails by sidesway in-
resistance factors (see Fig. 4). A limit of resistance is stability. It should be noted that in this failure mode there are
then defined as whichever comes first: (1) a limit point not a sufficient number of plastic hinges to define a kinematic
in a key load-displacement response curve (e.g., applied mechanism in the classical sense. Thus, driven by the P-Ll
load versus horizontal displacement); or (2) the load at effect, both geometric and material nonlinear destabilizing in-
which the slope of a key load-displacement curve be- fluences contribute to the strength limit state of the frame.
comes less than 5% of its initial value (e.g., load versus The force point traces shown in Fig. 6 indicate how the
midspan deflection of beam). combination of axial force and minor-axis bending moment at
2. Elastic-LRFD: A design that satisfies all of the AISC selected column cross sections changes as the factored gravity
LRFD Specification provisions and is based on a second- load is proportionally applied. The highlighted point on each
order elastic analysis. curve corresponds to the force distribution at 90% of the fac-
JOURNAL OF STRUCTURAL ENGINEERING 1 FEBRUARY 1997/153

J. Struct. Eng. 1997.123:151-156.


tored gravity load. Force point traces that reach the design full TABLE 1. AISC LRFD Interaction Equation Check (1.2 Dead +
plastification surface signify the formation of plastic hinges at 16 Live)
these cross sections. Two additional curves are also shown in Muy
Fig. 6. The first curve defines where initial yielding of the ..!:JL
p
AISC LRFD
Member K 4>c • 4>b Mny equation H1-1
column sections occurs. As indicated in the figure, portions of (1) (2) (3) (4) (5)
most of the column force point traces lie outside of this curve,
indicating a significant level of partial yielding. Again it is Cl 2.0 0.47 0.19 0.64 (pass)
C2 2.0 1.04 0.29 1.30 (fail)
noted that this behavior is not captured by the inelastic hinge C3 2.0 1.07 0.85 1.83 (fail)
analysis and hence, the analysis may be overpredicting the C4 1.0 1.76 0.05 1.80 (fail)
strength of the design. C5 1.0 0.84 0.87 1.61 (fail)
The second additional curve is a plot of the AISC LRFD C6 1.0 0.43 0.93 1.26 (fail)
interaction equation (Eq. HI-I)
uy )
Inelastic Hinge
-P-
u
+ -8 ( -
M- Pu
s 1.0 for-- ~ 0.2 (3a)
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

<Pcp. 9 <PbM.y <Pcp. 1.0 0.96-v


--_..._YO.94
Inelastic
uy Quasi- Inelastic
-P-
u
+ ( -M-) s 1.0 for -P-
u
< 0.2 (3b) Hinge Zone
2<Pc P• <PbM.y <Pcp.
The curve shown is for a beam-column of zero unbraced
length (kL =0). Where it lies almost entirely inside the design Applied 0.6
Load
full plastification surface, the interaction equation would lie
Ratio, A
even further inside for unbraced lengths greater than zero (kL
> 0). Realizing that the resistance of a compact section mem- 0.4
ber in compression and bending is limited by its stability as a
member of the structural system and its cross section strength, 0.2
the specification currently requires all members to satisfy the
interaction equation. If compression members are modeled by
an adequate number of elements (in this study, four discrete
elements), the Integrated Inelastic design method includes both -2.0 0.0 2.0 4.0
member and system stability checks directly in the analysis,
and hence, these slenderness effects need not be included in ~ t (cm)
the limitation on cross section strength. Where the specifica- FIG. 7. Load-Displacement Curves from Distributed Plasticity
tion requires that the interaction equation be satisfied, regard- Analyses (1.2 Dead + 1.6 Live)
less of the analysis type employed, it is inherently preventing
member forces from reaching the design full plastification sur- ence in ultimate load at about 4-6%. To indicate the extent
face. More importantly, it is eliminating the possibility of de- of partial yielding in the frame and provide a potential reason
tecting inelastic force redistribution. The AISC LRFD checks for the distributed plasticity analyses predicting lower ultimate
do include some elastic member and system destabilizing ef- loads, the relative percentages of section area yielded at the
fects which, strictly speaking, are only applicable when elastic limit load are also shown in Fig. 7. The locations of highly
analysis is used. While the specification permits the use of yielded regions calculated by the inelastic zone analysis are
plastic analysis, it appears only viable for structures with mem- consistent with the hinge locations shown in Fig. 5. It should
bers resisting little axial compression (e.g., continuous beams be noted that the potentially beneficial effects of strain hard-
of compact cross section). ening are not incorporated in any of the analyses.
Although the inelastic hinge analysis indi~ates that the de-
sign can resist the factored load effects, it is shown in Fig. 6
SUMMARY OF ALL FRAMES STUDIED
that several members violate the AISC LRFD interaction equa-
tion. Checks made using (3) are summarized in Table 1. The Results of the 16 frames studied are summarized next. In
effective length factors of k = 2.0 for the pinned-based, bot- all cases, differences in the general character of the Integrated
tom-story columns and k = 1.0 for the upper-story columns Inelastic design and the corresponding Elastic-LRFD design
can be shown to be liberal when compared to effective length are modest. Specific member sizes obtained for both designs
factors computed from either an inelastic eigenvalue analysis of each of the 16 frames are reported by Miller (1995).
or the use of alignment charts. Using the force distribution
obtained from the inelastic hinge analysis, five of the six col- Load at First Plastic Hinge
umns significantly violate the interaction equation.
To reproportion the Integrated Inelastic design to satisfy (3) For all frames, the first plastic hinge forms in the Integrated
requires an additional 23% of steel weight. When subjecting Inelastic design shortly after application of the service gravity
this reproportioned frame, which is the Elastic-LRFD design, load, which is approximately 70% of the full factored gravity
to a second-order inelastic hinge analysis, the first plastic hinge load. In contrast, plastic hinges do not form in the Elastic-
forms at a load in excess of the factored gravity load (h. = LRFD designs until after application of the full-factored grav-
1.02) and the limit of resistance does not occur until 1.13 times ity load.
the factored gravity load.
To confirm the adequacy of the Integrated Inelastic design, Ultimate Load
two distributed plasticity methods of analysis, inelastic zone
and quasi-plastic hinge, are employed. Resistance factors are In all cases, members could be proportioned in the Inte-
incorporated into these analyses by reducing the yield strength grated Inelastic design so that the system reaches a limit of
of the material by a factor of 0.9. Fig. 7 provides response resistance just at application of the factored gravity load. In
curves computed by each of the analysis methods. The re- most cases, the Elastic-LRFD design does not reach its limit
sponse curves for the three analyses are similar with a differ- state of strength until a ratio of 1.2 or more of the factored
154/ JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 1997

J. Struct. Eng. 1997.123:151-156.


gravity load is applied. The factored wind load combination TABLE 2. Comparison with Inelastic Zone Analyses (1.2 Dead
does not control member sizes in either design approach. In + 1.6 Live)
all analyses, the AISC LRFD resistance factors and load fac- Applied Load Ratio at Limit of Resistance
tors are incorporated. Inelastic hinge Inelastic zone
Frame (CU-STAND) (NIFA)
Service Load Deflections (1 ) (2) (3)
In all cases, the designs obtained using the Integrated In- U-F36H 1.01 0.99
elastic approach experience greater lateral and beam deflec- U-F36L 0.99 0.98
tions under service loads. Although beam deflection limita- U-P36H 1.00 0.94
U-P36L 1.01 1.01
tions do not control the design of any members in either U-F50H 1.01 0.97
approach, the lateral deflection requirements do control mem- U-F50L 1.01 0.99
ber sizes in eight of the Integrated Inelastic designs and six of U-P50H 1.01 0.86
the Elastic-LRFD designs. U-P50L 1.01 0.99
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

Total Steel Weight Comparisons Inelastic


The frames designed using the Integrated Inelastic approach Hinge
weigh on average approximately 15% less than those obtained
using the Elastic-LRFD approach. For those frames subjected
to heavy gravity load combinations, the average difference in
weight is the greatest at approximately 20%.
Applied
Compliance with AISC LRFD Interaction Equation Load
Ratio, A
In all of the Integrated Inelastic designs, at least two col-
umns do not satisfy the strength interaction equation provided 0.4
in the AISC LRFD Specification. In all cases, beam member
sizes proportioned according to the Integrated Inelastic design
approach satisfy all of the AISC LRFD requirements. 0.2
Comparison with Inelastic Zone Analysis
Table 2 provides a comparison of the ultimate loads pre- -10 -8 -6 -4 -2 o 2
dicted by the inelastic hinge analysis and the inelastic zone
analysis for the unsymmetrical frames. Resistance factors are ~t (cm)
incorporated into both analyses. Results obtained for the sym- FIG. 8. Load-Displacement Curves for Frame U-P50h (1.2
metrical frames are discussed next. In most cases, the inelastic Dead + 1.6 Live)
hinge analysis does provide accurate and reliable results. How-
ever, in one case, frame U-P50H, the inelastic zone analysis TABLE 3. Effects of Initial Imperfections on Symmetric
indicates that a significant level of partial yielding of the col- Frames (1.2 Dead + 1.6 Live)
umns will result in an ultimate load value 13% lower than that
Applied Load Ratio at Limit of Resistance
predicted by the inelastic hinge model. As shown in Fig. 8,
the quasi-plastic hinge analysis provides results similar to Plastic hinge Plastic hinge Plastic zone
those of the inelastic zone analysis. It should be noted that Frame without .:1 0 with .:1 0 = H/500 with .:1 0 = H/500
when the frame is redesigned so that the inelastic zone analysis (1 ) (2) (3) (4)
indicates a limit of resistance at an applied load ratio of A = S-F36H 1.00 1.00 0.90
1.0, several of the columns continue to exceed the limits of S-F36L 1.01 1.01 0.99
the AISC LRFD interaction equation. S-P36H 1.02 1.00 0.91
S-P36L 1.02 1.02 1.04
S-F50H 1.03 1.03 0.95
Effects of Initial Imperfection S-F50L 1.02 1.02 1.00
S-P50H 0.99 0.94 0.85
When designing the frames according to either the Inte- S-P50L 1.03 1.03 1.00
grated Inelastic or the Elastic-LRFD methods, frame geome-
tries were not modified to account for the effects of initial
imperfections. Using out-of-plumb imperfections of H1500 suIts are also obtained from the quasi-plastic hinge analyses.
(Code 1992), which are considered in both horizontal direc- For the unsymmetrical frames, an initial out-of-plumb of HI
tions, additional inelastic hinge and inelastic zone analyses are 500 does not significantly change the ultimate loads and cor-
performed. As shown in Table 3, the inelastic hinge analyses responding failure modes.
indicate that initial imperfections will have a significant effect
on only one of the symmetric frames, S-P50H. For the inelas- CONCLUSIONS
tic zone analyses, four of the eight symmetric frames have a
computed limit of resistance significantly less than that pre- This study demonstrates that for a majority of the frames
dicted by the inelastic hinge method. It should be noted that investigated, a second-order inelastic hinge analysis, which
two factors are contributing to this reduction in calculated can provide an accurate assessment of overall system response,
strength; incorporation of the effects of partial yielding and can be used to efficiently design steel structures with members
the inclusion of initial imperfections. Unfortunately, reliable subjected to minor-axis bending. It has been shown that in
inelastic zone results could not be obtained for the symmetric some cases, especially in the cases of heavily loaded sym-
frames without initial imperfections, and hence it is not clear metric frames, a distributed plasticity analysis is required to
which, if either, of the two factors is controlling. Similar re- predict limit state behavior accurately. In such cases, the au-
JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 1997/155

J. Struct. Eng. 1997.123:151-156.


thors recommend a quasi-plastic hinge analysis be used. More Orbison, J. G. (1982). "Nonlinear static analysis of three-dimensional
broad sets of studies including different frame geometries and steel frames," PhD thesis, Cornell Univ., Ithaca, N.Y.
Orbison, J. G., McGuire, W., and Abel, J. F. (1982). "Yield surface ap-
variations in live-to-dead load ratios will need to be conducted plications in nonlinear steel frame analysis." Computational Methods
before any general recommendations can be made as to when in Appl. Mech. and Engrg., No. 33, 557-573.
a concentrated plasticity analysis is ensured to be sufficient. Porter, F. L., and Powell, G. H. (1979). "Static and dynamic analysis of
It appears that second-order inelastic analysis cannot be frame structures." Earthquake Engrg. Res. Ctr., Univ. of California,
used to full advantage if designs are required to satisfy all of Berkeley, Calif.
the current AISC LRFD member strength equations. In any White, D. W. (1988). "Analysis of monotonic and cyclic stability of steel
frame subassemblages," PhD thesis, Cornell Univ., Ithaca, N.Y.
regard, the effectiveness of an inelastic design method can be White, D. W., and Chen, W F. (1993). Plastic hinge based methods for
realized only when: (l) used in conjunction with performance- advanced analysis and design of steel frames-an assessment of the
type strength criteria such as the prevention of system and state-of-the-art. Struct. Stability Res. Council, Bethlehem, Pa.
member instability under factored loads; and (2) coupled with White, D. W., Liew, J. Y. R., and Chen, W. F. (1991). "Second-order
necessary serviceability requirements. inelastic analysis for frame design: a report to SSRC task group 29 on
It has also been shown that initial imperfections such as recent research and the perceived state-of-the-art." Rep. CE-STR-91-
12, Purdue Univ., West Lafayette, Ind.
Downloaded from ascelibrary.org by Ondokuz Mayis Universitesi on 11/10/14. Copyright ASCE. For personal use only; all rights reserved.

frame out-of-plumb may need to be considered, especially Ziemian, R. D., McGuire, W, and Deierlein, G. G. (1992a). "Inelastic
when designing symmetrical frames subjected to symmetrical limit states design: Part 1-planar frame studies." J. Struct. Engrg.,
loading conditions. ASCE, 118(9), 2532-2549.
Ziemian, R. D., McGuire, W, and Deierlein, G. G. (1992b). "Inelastic
limit states design: Part II-three-dimensional frame study." J. Struct.
ACKNOWLEDGMENTS Engrg., ASCE, 118(9), 2550-2568.
The authors wish to thank Prof. Murray Clarke of the University of
Sydney for his assistance with the inelastic zone analyses and Dr. Mourad APPENDIX II. NOTATION
Attalla, formerly of Cornell University, for his assistance with the quasi- The following symbols are used in this paper:
plastic hinge analyses.
Ag = gross area of section;
APPENDIX I. REFERENCES H = height of structure;
k = effective length factor;
AIMashary, F., and Chen, W F. (1989). "Simplified second-order inelas- L = total span length;
tic analysis for steel frame design." Rep. CE-STR-89-33, Purdue Univ., M.y = minor-axis nominal flexural strength;
West Lafayette, Ind. M uy = minor-axis required flexural strength;
ASCE Task Committee on Drift Control of Steel Building Structures-
Committee on Design of Steel Building Structures. (1988). "Wind drift
Mp =
plastic moment capacity;
design of steel-framed buildings: state-of-the-art report." J. Struct. My = minor-axis moment;
Engrg., ASCE, 114(9), 2085-2108. M;ld =
moment at initial yield;
Attalla, M. R., Deierlein, G. G., and McGuire, W (1994). "Spread of Pcr = member axial force at critical load;
plasticity: quasi-p1astic-hinge approach." J. Struct. Engrg., ASCE, p. = nominal axial strength;
120(8), 2451-2473. P u = required axial strength;
Clarke, M. J. (1994). "Plastic zone analysis of frames." Advanced anal- Py = axial yield load;
ysis of steel frames: theory, software and applications, W. F. Chen and Qi = nominal load effect;
S. Toma, eds., CRC Press, Boca Raton, Fla., 259-319. R. = nominal resistance;
Code of standard practice for steel buildings and bridges. (1992). Am.
Inst. Steel Constr., Chicago, Ill.
ry = minor-axis radius of gyration;
Hsieh, S.-H., Deierlein, G. G., McGuire, W, and Abel, J. F. (1989). S = elastic section modulus;
"Technical manual for CU-STAND." Struct. Engrg. Rep. No. 89-13, Z = plastic section modulus;
School of Civ. and Envir. Engrg., Cornell Univ., Ithaca, N.Y. "VI = load factor;
Iffland, J. S. B., and Birnstiel, C. (1982). "Stability design procedures At = top story lateral displacement;
for building frameworks." AISC Proj. No. 21.62, Am. Inst. Steel A, = initial lateral imperfection;
Constr., Chicago, Ill. 80 = initial member imperfection;
King, W S., White, D. W, and Chen, W. F. (1992). "Second-order in- e = member end rotation;
elastic analysis methods for steel-frame design." J. Struct. Engrg., e = rotation at which plastic hinge forms;
p
ASCE, 118(2),408-428. A = applied load ratio;
<T = material yield strength;
Load and resistance factor design specification for structural steel build-
y
ings. (1994). Am. Inst. Steel Constr., 2nd Ed., Chicago, Ill.
<T = residual stress;
r
Miller, A. R. (1995). "Advanced second-order inelastic analysis of steel
structures with columns experiencing minor-axis bending subject to <P = resistance factor;
strength limit state requirements," MS thesis, Bucknell Univ., Lewis- <l>b = resistance factor for flexure; and
burg, Pa. <Pc = resistance factor for compression.

156/ JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 1997

J. Struct. Eng. 1997.123:151-156.

S-ar putea să vă placă și