Professional Documents
Culture Documents
net/publication/273370130
CITATIONS READS
15 1,728
2 authors:
Some of the authors of this publication are also working on these related projects:
Progressive collapse mechanism in timber buildings (focus on timber connections) View project
All content following this page was uploaded by Benoit P Gilbert on 02 May 2018.
Abstract: The paper summarizes the main new design provisions included in the recently revised Australian standard for steel storage racks.
The standard features multitiered analysis provisions ranging from basic linear-elastic analysis-based provisions to highly advanced integrated
design-analysis [geometric and material nonlinear analysis with imperfections (GMNIA)] provisions that allow the analysis and design to be
completed in one step. The GMNIA provisions distinguish between beam element–based and shell element–based analysis according to cross
Downloaded from ascelibrary.org by Griffith University on 06/25/13. Copyright ASCE. For personal use only; all rights reserved.
section slenderness and provide rules for the imperfections to use for the two types of analysis, including imperfections in the local and
distortional buckling modes for the shell element–based analysis. The selection of the system-based reliability (resistance) factor (fs ) is
discussed. The standard is seen as the most advanced design code of its type currently available for frame-type steel structures. The paper also
provides an in-depth discussion about the use of linear and nonlinear elastic analysis methods for the design of steel storage racks and how
torsion may be considered in determining design capacities while not in the structural analysis. DOI: 10.1061/(ASCE)ST.1943-541X.0000665.
© 2013 American Society of Civil Engineers.
CE Database subject headings: Standards and codes; Australia; Steel structures; Structural analysis; Storage facilities.
Author keywords: Steel storage racks; Revised Australian standard AS4084; Advanced analysis.
contains provisions for the shear stiffness test that are based on those The last, often referred to as “advanced analysis,” incorporates
of EN15512, as well as provisions for a new test arrangement the dominant nonlinear effects and is capable of accurately pre-
based on Australian research (Gilbert 2009; Gilbert and Rasmussen dicting the behavior and strength of structural steel frames. The
2011). revised standard considers two types of geometric and material
nonlinear analysis with imperfections (GMNIA): one (GMNIAc)
assuming the cross section is compact and based on beam elements
Analysis Types
and one (GMNIAs) assuming the cross section is noncompact (or
There is little consistency in the literature in the definitions of the slender) and based on full discretisation of members into finite strips
various types of analysis. However, a logical and coherent naming or finite elements.
convention was introduced in Part 6 of Eurocode3 (CEN 2007), In the cases of LA, LBA, GNA, and GMNIAc, the structural
which has also been adopted in the revised standard for steel storage analysis may be assumed to be based on beam elements. The
racks. The revised standard features the following analysis types: structural model may be two dimensional (2D) or 3D and may
• Linear (elastic) analysis (LA); consider torsion, although most analyses at present do not consider
• Linear buckling analysis (LBA); torsion, particularly warping torsion. A design may also require
• Geometric nonlinear analysis (GNA); and several types of analysis to be carried out, e.g., LA and LBA. Thus,
• Geometric and material nonlinear analysis with imperfections there are a large number of potential combinations of structural
(GMNA). analyses. However, from a practical viewpoint, it may be assumed
forcesb Analysis may, or may not, consider torsional (twist) rotations and torsional
internal actions.
Gross cross section properties may be used, ignoring perforations.
GMNIAc and Inelastic Yesc Out-of-plumba or Equilibrium is obtained in the deformed frame configuration.
GMNIAs equivalent horizontal Plasticity modeling should be based on flow theory.
forcesb Modeling of the nonlinear stress–strain relationship should be based on
recognized models for hot-rolled or cold-formed steel, as appropriate.
Residual stresses should be modeled directly or indirectly (e.g., through the
stress–strain curve).
Type c analysis:
Cross-section is compact in accordance with AS 4100 and nonperforated
except at connection points.
Analysis may be based on prismatic beam-elements.
Analysis should consider torsional (twist) rotations and torsional internal
actions, including warping torsion, unless fully laterally restrained as per
Clauses 5.3 and 5.4 of AS4100.
Type s analysis:
Cross-section is slender according to AS4100 or cold-formed.
Analysis is based on discretisation of pallet beams and uprights into shell
finite elements or finite strips modeling perforations.
Analysis incorporates local and distortional geometric imperfections.d
a
Out-of-plumb should be modeled in accordance with Section 3.3.
b
Equivalent horizontal forces should be modeled in accordance with Section 3.3.
c
Member geometric imperfections should be modeled in accordance with Section 3.3.
d
Local and distortional geometric imperfections should be modeled in accordance with Section 3.3.
that when several types of analysis are employed in a given design, horizontal forces irrespective of the type of structural analysis in
they will be based on the same structural model. Typically, if GNA is the revised standard.
employed in conjunction with LBA, it may be assumed that if the Out-of-plumb is derived from the lack of verticality arising
GNA is 2D, then so is the LBA, and that if the GNA does not during erection (ws ) and the looseness of the pallet beam to the
consider torsion, then neither does the LBA. upright connector (wl ). Testing provisions are included in the revised
The requirements for each method of analysis with regard to standard for determining the latter.
geometric and material modeling are detailed in Table 1. The frame imperfections (wi ) for unbraced frames specified in the
previous Australian standard for steel storage racks (Standards
Geometric Imperfections, Residual Stresses, and Australia 1993), the American RMI specification (RMI 2008), and
Notional Horizontal Forces the European steel storage rack specification (CEN 2009) are sum-
marized in Table 2. The previous Australian standard (Standards
Frame Imperfection
Australia 1993) specified three tolerance grades and different frame
The down-aisle and cross-aisle stability strengths are reduced by
imperfections for each grade, as shown in Table 3. The frame im-
out-of-plumb, shown as wi in Fig. 2(a) for down-aisle imperfections,
and henceforth referred to as “frame imperfections.” Frame im- perfection in the previous Australian standard was a function of the
perfections may be incorporated into the geometry of the structural number of bays. The Australian and European specifications add
model; however, it is usually more convenient to construct the upright-to-pallet-beam connector looseness (wl ) to the frame im-
model assuming vertical uprights and to incorporate the effect of perfection, whereas the American specification does not. Fig. 3
the frame imperfections by means of notional horizontal forces shows the frame imperfection (wi ) of the previous Australian, current
adjusted to provide the same first-order base moments as those European, and current American specifications as a function of the
produced by the combinations of out-of-plumb and vertical pallet number of bays (n) for two values of connector looseness
loads, as shown in Fig. 2(b). Out-of-plumb is modeled as notional (wl 5 0:0025 and wl 5 0).
Fig. 2. Frame imperfection (out-of-plumb) and notional horizontal forces (unbraced frames)
Table 2. Frame Imperfection for Unbraced Frames Specified in Current significantly greater than that specified by the American specification,
Australian (AS4084), European (EN15512), and American (RMI) Spec- as shown in Fig. 3(a).
ifications (Down-Aisle Direction Only) It can be seen from Fig. 3 that the number of bays (n) significantly
Specification wi changes only the frame imperfection (wi ) for n , 5, a situation of
little practical significance. Consequently, the revised standard
AS4084 wi 5 1=2 c0 ð1 1 1=nÞ 1 wl
adopts the expression for the frame imperfection given in the Eu-
c0 5 0:01 (Tolerance grade I), 0.007 (Tolerance
ropean specification
grade II), and 0.005 (Tolerance grade III)
No connector looseness test: wl 5 0:01 wi ¼ ws þ wl ð1Þ
Range (wlmin 5 0; wlmax 5 0:01):
0:0025 # wi # 0:02
where wl 5 connector looseness to be obtained from tests, or taken as
EN15512 wi 5 ws 1 wl
wl 5 0:01 if no test is performed; and ws 5 out-of-plumb erection
ws 5 1=350 5 0:003
tolerance as shown in Table 3. The revised standard imposes
Range (wlmin 5 0; wlmax 5 0:001):
a specific requirement of ws $ 1=250 for GNA, which is greater than
0:003 # wi # 0:0013
the requirement of ws $ 1=350 of the European specification. This
RMI wi 5 1=240 5 0:0042
increased requirement for GNA is incorporated to ensure that suf-
ficient bending moment develops in uprights with dominating axial
forces. The requirement is a substitution for the requirement in
Table 3. Tolerance Grade in Current Australian (AS4084) Specification Clause 8.4.2.2 of AS4100 (Standards Australia 1998), which stip-
Maximum ulates a separate check of the compression member capacity, in-
allowed cluding the determination of the effective length of the upright. Such
Tolerance Type of unit load out-of-plumb a requirement is not included in the revised standard.
grade handling equipment (down-aisle) The out-of-plumb for braced frames (both the down-aisle and the
cross-aisle direction) in the revised standard is the same as that
I Manually operated equipment guided by 1/500 specified in EN15512
operator (e.g., wide and narrow aisles) 0:5
wi ¼ 1 þ 1
II Manually operated equipment guided by 1/750
2ws ð2Þ
electrical or mechanical devices (e.g., very 2 nf
narrow aisle)
III Fully automatic operated equipment guided 1/1,000 where wi # 2ws and ws $ 1=500 and, in the down-aisle direction,
by electrical or mechanical devices (e.g., nf 5 number of upright frames in one row of bays and, in the cross-
crane installation) aisle direction, nf 5 number of upright frames connected together
(e.g., by top ties, run spacers, or intermediate floors) and acting
together.
It follows from Fig. 3(b) that the frame imperfections of the three
specifications are close for n $ 5 and no connector looseness. In Member Imperfection
this case, AS4084 (Standards Australia 1993) Tolerance Grade II The effects of member imperfection must be considered in the de-
most closely matches the European and American specifications. sign, either by direct or indirect modeling in the structural analysis
When substantial connector looseness is present, the frame imperfec- or by the use of strength curves in the design strength check. When
tions specified in the Australian and European specifications are using LA or GNA, member imperfections are not required to be
EI p ¼ 0:8EI ð3Þ
for all members and connections. This provision originates from the
AISC-360 specification (AISC 2010), which, in lieu of modeling
member imperfections, allows the flexural rigidity (EI) to be reduced
as per
8 p
Fig. 3. Frame imperfection (out-of-plumb) wi , down-aisle direction >
> 0:8EI for N # 0:5
only >
< Ny
EI p ¼ ð4Þ
>
> p p Np
>
: 0:8EI 4 N 1 2 N for . 0:5
incorporated into the structural model because their effect is Ny Ny Ny
accounted for by the use of column and beam strength curves in
determining the member design capacities (Nc , Mbx , and Mby ). where N p 5 design axial force and Ny 5 Afy 5 squash load. This
When GMNIA structural analysis is used for determining in- provision was also included in the North American specification
ternal actions, the geometric imperfections of uprights must be in- for cold-formed steel structures (NAS 2007). The square bracket in
cluded in the structural model. The revised standard suggests the Eq. (4) accounts for the second-order effects caused by yielding.
following methods for implementing member imperfections: Although this yielding-induced reduction of EI may be required for
1. Performing an LBA with restraints to prevent sway, scaling hot-rolled steel structural frames, cold-formed steel racks are unlikely
the buckling mode to an amplitude of d0 , and superimposing to be greatly influenced by such second-order effects, and hence, for
the scaled buckling mode onto the perfect straight-member cold-formed racks, the reduction of EI may be taken simply as
geometry; or EI p 5 0:8EI. Studies by Sarawit and Pekoz (2006) support this ap-
2. Off-setting nodes relative to the straight member geometry by proach, which has also been adopted in the revised standard.
d0 . In this case, recognizing that failure of the frame inevitably
occurs mainly by buckling of uprights in the lower two Local and Distortional Cross Section Imperfections
storeys, it is sufficient to introduce geometric imperfections In GMNIAs of slender sections, the strength may be affected by
into the uprights of these storeys only. It is also usually local and/or distortional buckling of the cross section. In this case,
sufficient to introduce two additional nodes per member as geometric imperfections must be incorporated to trigger these
shown in Fig. 4. modes of buckling. This is usually achieved most conveniently by
The magnitude of the member geometric imperfection shall be in superimposing geometric imperfections in the shapes of the local
accordance with the maximum allowable erection member out-of- and distortional buckling modes onto the perfect geometry. The
straightness, typically h=1,000 as per AS4100 (Standards Australia local and distortional buckling modes may be determined by con-
1998), where h is the length of the member, or the damage of uprights ducting an LBA with appropriate restraints to suppress frame and/or
tolerated during service. The damage levels tolerated in the re- member buckling and by varying the thickness of the cross section
vised standard are d0 5 h=200 and d0 5 h=333 for the down-aisle to switch between local and distortional buckling modes.
stress of the cross section as determined from a rational analysis. on GNA or LA with moment amplification, the member strength is
Although the Walker expression was originally derived for local determined by using the interaction equations of structural steel
imperfections, it also provides reasonably accurate design strength design standards, which, in turn, require the axial and flexural
predictions when applied to distortional buckling (Lecce and member strengths to be calculated according to the standard, thus
Rasmussen 2006), i.e., the magnitude of the distortional buckling accounting for the effect of yielding and combined actions. When
imperfection may be determined from based on GMNIA, second-order geometric effects and yielding
are accounted for, and the ultimate strength of the frame is de-
0:5
fy termined directly. In this case, there is no need to check the
Sod ¼ 0:3t ð6Þ member strength according to a structural design standard, except
fod
that a resistance factor (fs ) for the frame must be applied to the
where fod 5 elastic distortional buckling stress of the cross section ultimate load capacity, as discussed in the later section on
as determined from rational analysis. Figs. 5(a and b) show the GMNIAc.
definitions of the magnitudes of local and distortional geometric
imperfections, respectively. The revised standard specifies Eqs. (5)
Member Buckling Modes in Bending and Compression
and (6) for the magnitudes of the local and distortional geometric
imperfections, respectively. Singly symmetric open sections are particularly prone to flexural-
The revised standard requires residual stresses to be modeled di- torsional buckling when subjected to compression and bending.
rectly or indirectly (e.g., through the stress–strain curve) in GMNIA. Assuming the symmetry axis is the x-axis, in compression, the elastic
The standard does not contain specific provisions for the modeling of modes of buckling are the flexural buckling mode about the y-axis
residual stresses, whereas it is expected that the commentary of the (buckling stress foy ) and the flexural-torsional buckling mode about
standard will contain much more detailed guidance. It is generally the x- and z-axes (buckling stress foxz ).
accepted that through-thickness residual stresses lead to a gradual In bending, when bent about the y-axis, the critical mode is a
transition from the linear range to the yield plateau in tensile coupon flexural-torsional mode, irrespective of whether bending induces
tests and that they can be accounted for by modeling the gradual compression or tension at the free edges of the flanges. This mode
transition when specifying material properties in GMNIA. Membrane must be considered in design when Iy . Ix . When bent about the x-
residual stresses need to be modeled directly, particularly for the axis, the critical mode is also a flexural-torsional mode. This mode
design of storage racks built from hot-rolled or fabricated sections. must be considered in design when Ix . Iy .
Although cold-formed sections also feature membrane residual
stresses arising primarily from the plastic forming process, they are
relatively small in magnitude and their effect is likely to be offset by Linear Analysis and Geometric Nonlinear Analysis
the beneficial enhancement of the strength of the corner properties Flexure Only in Linear Analysis and Geometric Nonlinear
arising from the plastic working, which is usually not otherwise Analysis (Torsion Not Considered)
accounted for in the design of cold-formed rack structures. In the case of LA, the bending moments (Mxp and Myp ) must be
amplified to account for second-order effects. According to the re-
vised standard, this may be achieved by using an LBA to determine
the elastic buckling load (Ncr ) of the member, and then multiplying
the first-order bending moments by the amplification factor, which
in the revised standard is specified as
Ncr 1
or ð7Þ
Ncr 2 N p 1 2 1=lcr
where lcr 5 load factor obtained from the LBA when applying
factored design actions.
It may be assumed that when the analysis does not consider
torsion, the structural model is 2D, or, if a 3D model is constructed,
restraints are imposed to confine displacements to occur in a single
Fig. 5. Cross-sectional geometric imperfections plane at a time, not considering torsion. (Current specifications for
racks encourage this type of approach by allowing the down- and
standard and accounting for the effects of local buckling and Imperfections (Slender Cross Section)
perforations. If the upright is subjected to primary torsion GMNIAs assumes noncompact or slender cross sections, and hence,
action, the following interaction equation should be used: discretization of the cross section (e.g., into shell elements) is re-
quired. Frame, member, and cross-sectional (local and distortional)
Np Mp Myp Bp geometric imperfections must be included in the structural model, as
þ x þ þ #1 ð12Þ
fNs fMsx fMsy fBs described in the section Geometric Imperfections, Residual Stresses,
and Notional Horizontal Forces, as are perforations. The ultimate
where the bimoment (Bp ) has been added to account for the capacity of the frame is determined directly by advanced analysis, as
effect of warping torsion. The bimoment section capacity (Bs ) for GMNIAc Design Option 2, i.e., Eq. (14) must be satisfied.
is conservatively calculated as Connections may be modeled in one of two ways:
1. All components of the connections are discretised and mod-
Iw eled explicitly, e.g., beam end brackets may be modeled using
Bs ¼ fy ð13Þ
vmax shell elements, bolts may be modeled using beam elements,
etc. Where the connection relies on friction, contact elements
where Iw 5 warping rigidity; vmax 5 maximum value of may be required. The structural model should be verified
sectorial coordinate; and fy 5 yield stress. against connection tests conducted as per the revised standard.
2. The ultimate capacity of the frame is determined directly by
The model should be shown to replicate the experimentally
advanced analysis. In this case, the factored ultimate limit
obtained moment-rotation behavior and strength.
states loads are multiplied by a load increment factor (l),
2. The connections are modeled as nonlinear beam or connection
which is increased until the frame fails at l 5 lmax . Based on
nominal values, the frame can support the ultimate loads if elements, as in LA, GNA, and GMNIAc. In this case, it may be
lmax $ 1, and fails if lmax , 1. To account for random vari- necessary to insert solid elements into the members near the
abilities, the resistance factor of the system is taken as connections, allowing the connection elements to be inserted
fs 5 0:9, which is the same as the resistance factor used for between these. The connection elements should incorporate
members in AS4100 (Standards Australia 1998). Conse- the experimentally obtained moment–rotation relationships,
quently, the design is carried out by checking that according which should be scaled down by a capacity factor (f), and it
to the advanced analysis, the ultimate capacity of the frame is should be checked that the design value of the experimentally
reached for obtained connection strength is not exceeded. Particular at-
tention needs to be paid to the modeling of continuity/support
1
fs lmax $ 1 0 lmax $ ð14Þ conditions for the warping displacement at connection points.
fs The reduced shear stiffness of the upright frames derived from
the flexibility of the connections between the uprights and the di-
where fs 5 0:9. agonal bracing members should be considered in the structural
Choosing an appropriate system resistance factor requires a model or modeled directly as for GMNIAc.
comprehensive study of the system reliability of steel storage rack
structures, which is not available at present. For statically redundant
frames such as conventional steel storage racks, it is most likely
Design Provisions for (Doubly Symmetric) Hot-Rolled
conservative to use the member resistance factor for the resistance
Steel Uprights
factor of the system (Ellingwood 1994; Ellingwood 2000). A recent
study (Rasmussen and Gilbert 2011) on the down-aisle strength of
braced and unbraced steel storage racks has shown that on an average General
basis, the strengths obtained using GMNIAc and GMNIAs analyses
match the strengths obtained using LA and GNA analyses, although Heavily loaded racks may be constructed from hot-rolled (or fabri-
differences in strengths are observed when considering braced and cated) I-sections or tubular sections, the latter being cold formed or
unbraced frames separately. hot rolled. Collectively, these sections are referred to as “hot rolled.”
For both design check options, the semirigid characteristics of They share the common characteristic of being doubly symmetric
the pallet beam-to-upright and base plate connections should be and may be designed to AS4100 (Standards Australia 1998). Ac-
included in the structural model, and it should be checked that the cordingly, the flexural section capacity may reach the plastic moment.
jhu.edu/bschafer/cufsm/æ (Sep. 21, 2011). Ziemian, R. D., and McGuire, W. (2002). “Modified tangent modulus
Standards Australia. (1993). “Steel storage racking.” AS4084, Sydney, approach, a contribution to plastic hinge analysis.” J. Struct. Eng.,
Australia. 128(10), 1301–1307.
Standards Australia. (1998). “Steel structures.” AS4100, Sydney, Australia. Ziemian, R. D., McGuire, W., and Deierlein, G. G. (1992). “Inelastic limit
Standards Australia. (2005). “Cold-formed steel structures.” AS/NZS4600, states design. Part I: Planar frame studies.” J. Struc. Eng., 118(9), 2532–
Sydney, Australia. 2549.