Flexural Behavior of SSR Beam Connections - Zhao2014

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of Constructional Steel Research 99 (2014) 161–175

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Flexural behavior of steel storage rack beam-to-upright connections


Xianzhong Zhao a,b,⁎, Tuo Wang a, Yiyi Chen a,b, K.S. Sivakumaran c
a
College of Civil Engineering, Tongji University, Shanghai 200092, China
b
State Key Laboratory for Disaster Reduction in Civil Engineering, Tongji University, Shanghai 200092, China
c
Department of Civil Engineering, McMaster University, Hamilton, Ontario L8S 4L7, Canada

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the flexural behavior of connections associated with the cold-formed steel storage pallet
Received 18 October 2013 racks. This experimental study considered seventeen groups of beam-to-upright connections with different
Accepted 2 April 2014 constructional details. Various influencing parameters, such as upright profile and thickness, pallet beam profile
Available online 4 May 2014
and the number of tabs in the beam-end-connector were investigated. Each group considered three identical
specimens subjected to monotonic, static, hogging loading in a single cantilever test setup. The displacements
Keywords:
Storage pallet rack
and strains, including rosette stains, were measured at strategic locations. The results show that the deformation
Beam-to-upright connection modes of the connections were similar prior to failure, while the failure modes depend on the relative thickness
Beam end connector between upright and the beam-end-connector. Based on moment–rotation curves, the initial stiffness, the
Moment–rotation moment capacity, and the associated connection rotations were established. These values have been compared
Experimental analysis with corresponding international design specifications. Ensuing discussions shed light on the impact of parameters,
such as thickness of upright profile, depth of pallet beam profile, construction details of beam-end-connector on the
stiffness and moment capacity of such connections.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction the tabs used in the tongue and slot design. The studs are press fitted
into pre-punched holes of the end-connector bracket. Class (D) — dual
Steel storage rack structures are widely used in industry for storing integrated tab design. The tabs are both formed and punched out of
and retrieving goods often placed on pallets. The structural members of the bracket so that they remain joined to the connector bracket at two
a typical pallet rack consist of cold-formed steel uprights, which are points. The characteristics of above classes of designs may be combined
often perforated along the web and/or the flanges, and cold-formed to generate a sophisticated end-connector design.
steel welded beams often in the shape of rectangular or non-rectangular The storage racks under consideration are essentially un-braced
boxes. The beams are connected to uprights by end-connectors, which frames in the down-aisle direction so that the lateral stiffness and the
are pre-welded onto the ends of beams. The end-connector tabs are strength of the racks in the down-aisle direction are largely provided
inserted into the upright's perforations and the beam is pushed down to
engage and connect the beam to the upright. Furthermore, a locking pin
(safety device) is often used to prevent the end-connector from
pulling out of the upright in the presence of an uplift force. A typical
beam-to-upright connection and its elements are shown in Fig. 1.
The commercially available beam-end-connectors come in different
shapes and sizes, and whose design has been classified by Markazi et al.
[1] as follows: Class (A) — tongue and slot design. In this design the
cantilevered tab which is formed and punched out of the end connector
bracket (tongue) is slipped into the upright slots to form the connection.
Class (B) — blanking design. In this class of design, a blanking operation
produces the end connector tabs which, depending upon the design of
the upright, interlock either parallel or perpendicular to the web of
the upright. Class (C) — stud-incorporated design. The studs replace

⁎ Corresponding author at: College of Civil Engineering, Tongji University, 1239 Siping
Road, Shanghai 200092, China. Tel.: +86 21 65985089; fax: +86 21 65984976.
E-mail address: x.zhao@tongji.edu.cn (X. Zhao). Fig. 1. A typical beam-to-upright connection.

http://dx.doi.org/10.1016/j.jcsr.2014.04.007
0143-974X/© 2014 Elsevier Ltd. All rights reserved.
162 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

Notations

d the actual horizontal distance between the loading


point and the centroid of the upright
db the depth of the beam
dc the depth of the beam-end-connector
e eccentricity
fy the yielding point of the steel used
fu the tensile strength of the steel used
h the distance between the beam top and bottom flanges
K0 the initial stiffness of the connection
K0a the average value of the initial stiffness of the
connection
KEN the stiffness obtained per EN 15512 Fig. 3. Portal frame test set-up.

KENa the average value of the stiffness obtained per EN 15512


KRMI the stiffness obtained per RMI
are neither easy nor effective to determine the rotational stiffness and
KRMIa the average value of the stiffness obtained per RMI
the moment capacity of any connection. Due to intricate features and
M the bending moment applied to the beam-to-upright
the complex shapes of the uprights, the beams and the beam-end-
connection
connectors, it is almost impossible to obtain accurate flexural character-
Mc the bending moment capacity of the connection
istics of storage rack beam-to-upright connections utilizing analytical
Mca the average value of bending moment capacity of the
methods [2,3]. Consequently, the international storage rack design
connection
specifications, such as Australian Standard AS 4084 [4], the Rack
P the load applied to the beam
Manufacturers Institute (RMI) specification [5] and the European
tt the thickness of the tab
Standard EN 15512 [6], recommend carrying out experiments in order
tu the thickness of upright
to obtain realistic flexural properties of an upright–beam connection.
ϕ the rotation of the beam-to-upright connection
Two alternative experimental setups, i.e. cantilever test and
ϕb the rotation of the beam
portal frame test, are described in these specifications [4–6], and
ϕb′ the rotation of beam axis including the component of
the researcher or the designer has the liberty to choose one of the
beam flexural deflection
test setups. The cantilever test, as shown in Fig. 2(a), is included in
ϕc the rotation of the upright
all three specifications although some minor differences exist in
ϕm the rotation at the maximum moment
the specimen dimensions and testing procedures. Many researchers
ϕma the average value of the rotation at the maximum
have used the cantilever test method to investigate the rotational
moment
behavior of beam-to-upright connections of storage racks under
monotonic or cyclic loadings [1,7–10]. In the cantilever test, the
connection can be loaded to failure, and the results are well suited
by the rotational stiffness and the flexural strength of the beam-to- for the analysis and design of storage rack members and their con-
upright connections. As such, appropriate frame analysis and design nections. Bajoria and Talikoti [11] had proposed a double cantilever
models for such down-aisle storage rack require the moment–rotation test setup, which is as shown in Fig. 2(b). In the double cantilever
characteristics of the beam-to-upright connections. In general, the tra- test, the end-connectors will experience three types of forces name-
ditional analytical techniques, including the finite element method, ly, moment, shear and an axial pull, which somewhat represents the

Fig. 2. Cantilever test set-up.


X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 163

a) The Uprights

b) The Beams

c) The Beam-End-Connectors
Fig. 4. The structural components used in this study.

actual field conditions. The shear-to-moment ratio in an actual rack side of the portal frame, which may have different characteristics, this
frame may be better reproduced by the double cantilever test. test will not establish the exact connection behavior [7]. The portal
However, the connections on either ends of the setup in a double frame test results are usually suitable for sway analysis of gravity loaded
cantilever test may not be symmetrically loaded or deformed, partic- rack structures [12]. Harris [13] compared the results based on these
ularly in the non-linear range, which may lead to obtaining two test techniques and noted that the cantilever test based connection
erroneous moment–rotation characteristics for such connections. stiffness values are typically half of that based on the portal frame test.
The portal frame test, as shown in Fig. 3, is included in the Australian Obviously, the difference between the cantilever test and the portal
Standard AS 4084 [4] and in the RMI specification [5], but not in the frame test lies on the corresponding test setups. The reasons for these
European Standard EN 15512 [6]. As the portal frame test method can differences were explained by Harris [13], and were further explained
only determine the average stiffness of the connections and does not by Gilbert and Rasmussen [14] as follows: When the portal beam is
distinguish between the behavior of the two connections on either gravity loaded with its service pallet load, both the left and right
164 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

Table 1
Mechanical properties of the structural components.

Flat portion Corner

fy (MPa) fu (MPa) fy (MPa) fu (MPa)

Uprights C5 289 552 422 524


C5* 374 561 476 546
C4 395 549 457 552
C4* 363 512 429 513
C3 336 530 422 520
Beams 268 483 403 434
Beam-end-connectors 269 456 476 493

Notes: The uprights C5* and C4* are 2.5 mm and 2.0 mm thick, respectively. All other
uprights are 1.8 mm thick.

connectors are subjected to closing up moments (where the beam


ends which were perpendicular to the upright begin to rotate towards
the upright). If the portal frame is subsequently subjected to a horizontal
loads, one connector will experience increasing moments, whereas other
connecter will experience decreasing moments, resulting in one end
experiencing increasing closing up while the other connector starts
Fig. 6. Arrangement of displacement transducers.
opening (unloading) (see Fig. 3). It was observed [13,14] that the
unloading stiffness is typically significantly higher than the loading
stiffness, resulting in two different stiffness values for the right and left maximum moment of connections. This investigation provides
connectors. From the above discussion it appears that the cantilever additional information such as the parameters controlling the failure
test is more appropriate for obtaining the moment–rotational behavior mode and the importance of the tabs in the response of the beam-to-
of beam-to-upright connections of steel storage pallet racks. upright connection to the rack connection database. This series of tests
This paper reports a recent experimental evaluation of the flexural was performed in order to establish a base series for further research
behavior of locally manufactured beam-to-upright connections when work on storage rack structures.
subjected to increasing static hogging loading. The main purpose of
this investigation is to establish the moment–rotation relationship of 2. The experimental investigation
specific connection details, which will be the base of design formula
for predicting the performance of boltless connections. The connection 2.1. The structural components and the test specimens
tests presented herein used the cantilever test method, which has
been utilized by many researchers; however, special boundary Parameters governing the behavior of beam-to-upright connection
conditions and refined measuring methods were incorporated in this include the profile and thicknesses of the upright, beam and the
study. The connections were subjected to relatively large deformations beam-end-connector. The geometry and the number of tabs, and the
in order to acquire the whole range of moment–rotation relationships, number of contact planes between the beam-end-connector and
which will be of great use for seismic analysis and for progressive upright also influence the connection behavior [1,3,8–10]. Investigation
collapse analysis of rack structures. This paper reports results on 51 of the influence of all of the parameters at the same time is an arduous
individual tests, comprising of 17 groups of three identical tests each, task. Prabha et al. [10] established that the most influential parameters
and the results include the deformation modes, the failure mechanisms, governing the flexural behavior of beam-to-upright connections are the
the connection stiffness, the moment capacity, and the rotations at the thickness of upright (tu), the depth of beam (db) and the depth of the

Fig. 5. The test set-up.


X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 165

beam-end-connector (dc). This investigation considered three different This study utilized the single cantilever test method, and thus, each
upright profiles identified in Fig. 4 as C3, C4 and C5, and three kinds of specimen consisted of an 800 mm long beam piece connected to a
beam profiles tagged as B105, B120 and B145 as shown in Fig. 4. The 760 mm high upright using a beam-end-connector, which was welded
uprights and beams of specimens were fabricated to shape from cold- symmetrically to the beam ends. Different combinations of uprights,
formed steel sheets. The three beams, having a depth of 105, 120, and beams and beam-end-connectors were considered which resulted in
145 mm, respectively, were of thickness 1.5 mm, and all of the uprights, 17 groups of tests. Three identical tests were performed for each
except upright groups C4* and C5*, were of thickness tu = 1.8 mm. The group of specimen, which resulted in a total of fifty-one individual
upright group C4* was of thickness tu = 2.0 mm, and the upright group tests. For convenience of identification and communication each
C5* was of thickness tu = 2.5 mm. The study considered three kinds of specimen was given a specimen ID of the form [Cx-By-zT], where [Cx]
beam-end-connectors shown as 3T, 4T and 5T in Fig. 4, and they identifies the upright which was one of C3, C4, C4*, C5, or C5*. Similarly,
contained 3 tabs, 4 tabs, and 5 tabs, respectively. Note that the number [By] indicates the beam type and [zT] indicates the beam-end-
of tabs of the beam-end-connector directly reflects the depth of the connector, as identified in Fig. 4.
connector (dc). The beam-end-connectors were made of 3.0 mm hot-
rolled steel sheet, and the depths of 3T, 4T and 5T connectors were, 2.2. The test setup and the test procedure
150, 200, and 250 mm, respectively. The material properties of uprights,
beams and beam-end-connectors were established based on coupon The connection test was conducted within a multifunctional reaction
tests. The flat portions and the corners of these structural components frame specially designed for this research program. The schematic
were considered. Three coupons each were tested for each category, diagram of the test set-up and a photographic image of it are shown in
and the average yield strength and the tensile strength of these Fig. 5. The 760 mm high upright had plates welded at its ends and was
elements based on three identical coupon tests are listed in Table 1. bolt fasten to cantilever support beams, thereby creating a near fixed

a) Arrangement of strain gauges.

b) Arrangement of strain rosettes.


Fig. 7. Arrangement of strain gauges and strain rosettes.
166 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

boundary conditions. The 800 mm long beam piece was connected to the 2.3.2. Strain measurements
upright at its mid-height. The free end of the beam was laterally braced Each test contained sixteen strain gauges and ten strain rosettes.
to restrain the out-of-plane movement of the beam. Two pieces of stain- Strain gauges were attached to the beams, whereas the strain rosettes
less steel plates were fastened to the beam free ends. The lateral braces, were attached to the upright and to the end-connector. Four sets of
which had four sleek ball tips, were then snug tight to these stainless strain gauges were used along the length of the beam, which were locat-
steel plates, so that the beam free end can freely slide up or down on ed on the beam cross-sections at 10, 60, 110, and 160 mm from the face
the braces. This specially designed lateral bracing arrangement is consid- of the upright. Fig. 7(a) shows the strain gauge locations, which are at
ered to have eliminated the frictional forces between the beam free end the top and bottom flanges of the beam. In general, the strain gauges
and the braces. Single 20 kN electric actuator applied the load to the were fixed at the center of the sub-element of the flanges, except for
beam. The upper end of the actuator was supported on a frictionless slid- beams B120 on which the strain gauges were fixed at 15 mm from the
er mechanism, thus the actuator was free to move horizontally and webs. As shown in Fig. 7(b), eight strain rosettes were attached onto
maintain the applied load in a vertical position at all times during the the web of the upright within the connection zone and two strain
test, when the beam undergoes vertical deflections. Two LVDTs, shown rosettes were attached at the junction of end-connector and beam.
as A and B in Fig. 5 monitored the horizontal movements of the actuator.
The bottom end of the actuator was connected to the test beam using a
hinge mechanism, thereby only a vertical force is transferred to the 3. The experimental observations and the results
beam (no bending moment is created at the loading point). The initial
distance between the loading point and the flange of the upright is 3.1. Deformation patterns and failure modes
400 mm. The load was applied slowly in order to have a quasi-static load-
ing condition. The actuator was controlled through displacement by a Fig. 8 shows the deformations of a typical connection [C5-B120-4T]
servo-valve controlled motor using a speed of 0.5 mm/min before during test. Overall, the connections with similar configuration and
reaching the maximum load and a speed of 1–2 mm/min beyond peak profiles exhibited similar deformation patterns. Due to the hogging
loads. The values of the applied loads were obtained through a calibrated bending moment effects, the upper part of end-connector separated
load cell located at the lower end of the actuator. The test continued until from the upright while the lower part of the connector continued to
the load dropped to 75% of the peak load or the deformation of the be in contact with the upright flange. The end-connector plate attached
specimen was so large that it was unsuitable to carry load. to the beam simply rotated as a rigid body, perhaps because of the
stiffening effect provided by the beam. With increasing loads the
connection elements experienced significant deformations. The tabs
2.3. Instrumentations which are on the side leg of the connector, however, interacted with up-
right perforation walls and experienced distortions. Beam-end-
2.3.1. Displacement measurements connector and the flange of upright experienced flexural deformations
Fig. 6 shows the displacement measurement locations, and accord- as well. Recall that, except for uprights C5* and C4*, all other uprights
ingly eight displacement transducers were used to measure the deflec- were of wall thickness of 1.8 mm. The failure mode of these 1.8 mm
tions of the upright and the beam. The D1 & D2 and D3 & D4 uprights can be described as tearing of the upright wall near the perfo-
transducers measured the in-plane horizontal displacements of the top rations. The tearing often occurred at the topmost tab, which was sub-
and the bottom flanges of the beam, respectively, at the connection. jected to substantial flexural tension. Often, the experimental peak
The D5 & D6 transducers measured the corresponding in-plane horizon- load corresponded to the initiation of the tearing of the upright walls.
tal displacements of the upright. The distance between these pairs of Beyond this stage, the crack continued to develop leading to large defor-
transducers depended on the size of the beam under consideration. mations and reduction in load carrying capacity of the connections. This
The transducers D7 and D8 measured the vertical displacement of the mode of failure, which is called herein as failure mode — I, can be cate-
beam. As shown in Fig. 6, the transducer D7 was located at 160 mm gorized as due to failure of the thinner upright wall which is weaker
from the face of the upright, whereas transducer D8 was located than that of the connector tab. Fig. 9 shows the photographs of the con-
100 mm along the beam from D7. nector and the upright experiencing this failure mode. The failure of

a) Front view b) Side view c) Back view


Fig. 8. Deformation of a typical connection [C5-B120-4T] during test.
X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 167

connections having C5* upright, which is of thickness 2.5 mm, was due ineffective accompanied with redistribution of the forces towards the
to cracking of the top level tab. Once again, the experimental peak load center of the end-connector. If the specimen is continued to be loaded
approximately corresponded to the initiation of the crack on the con- the second connector tab or the corresponding upright wall would
nector tab. The crack continued to propagate leading to large deforma- crack. One other failure mode was observed during these tests, which is
tions and load reductions. Eventually, the tab fell off from the beam associated with the end connector 5T, which had five tabs. As shown in
end connector. This mode of failure, which is called herein as failure Fig. 4, this connector is 250 mm long, and thus, a substantial portion of
mode — II is, thus, due to weaker tab-stronger upright design. Fig. 9 it was extending beyond the height of the beam (like an extended end-
also shows the photographs of the connector and the upright experienc- plate connection). In these connections the connector itself distorted
ing this failure mode. and ruptured. Fig. 10 shows such failure of the end-connector.
Recall that the three specimens of series C4* upright were of thickness It is apparent that the most influencing parameter that determines
2.0 mm, which is in between C5* upright thickness (2.5 mm thick) and the failure mode of the connection type under consideration is the
thickness of all other uprights (1.8 mm thick), thus considered herein as ratio tu/tt, where tu is the upright wall thickness and tt is thickness of
moderate thickness. One specimen of C4* group failed with connector the tab of the beam-end-connector. In other words, the part with
tab crack, whereas the other two specimens failed due to tearing of the lower strength will fail first which leads to failure of the connection
upright walls. It is apparent that the crucial parameter that dictates the and the rack structure. Some other experimental investigation [16] on
failure mode is the thickness of upright relative to the thickness of tabs, a different kind of rack connections found that the failure of the connec-
which determines the relative strength between the upright wall and tion was controlled entirely by the tabs (failure mode — II). Whether the
the connection tab. In all of these failure modes, once the topmost tab failure mode I or the mode II is the preferred failure mode is a subject of
or the upright perforation wall is cracked, the connection became argument. Aguirre [16] favors the failure mode II in which tabs act as

a) Front view b) Tab distortion c) Upright wall torn


Failure mode - I: tearing of the upright wall

a) Front view b) Tab crack c) Upright wall intact


Failure mode - II: tab crack

Fig. 9. Failure modes of upright–beam connection.


168 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

between beam axis and upright axis at their intersection, given as


ϕ = ϕb − ϕc, where ϕb is the beam rotation and ϕc is the rotation
of the upright. Fig. 11 clarifies these rotations. Based on a refined
measurement method as shown in Fig. 6, the rotation of beam axis
at the connection ϕb was established based on deflection measurements
D1, D2, D3 and D4, as ϕb = 0.5[(D1 − D3) / h + (D2 − D4) / h], and
the rotation of upright axis at the connection ϕc was established from
measurements D5 and D6 as ϕc = (D5 − D6) / h. Here, h is the distance
between the beam top and bottom flanges.
It should be pointed that in some research studies, only the rotation
of beam axis [1] or the rotation beam-end-connector [15] was
established, which fail to account for the rotation component of the up-
right, and the flexibility of the beam and beam-end-connector. Abdel-
Jaber et al. [7] compared and analyzed the two alternative methods of
measuring the rotation of the beam and derived appropriate correction
formulae. The formulae were based on the assumption that the beam
and the beam-end-connector must be in the elastic state. In our study,
the rotation of beam axis ϕb′, which includes the component of beam
flexural deflection, was also obtained through ϕb′ = (D7 − D8) / 100,
where 100 is the distance between the displacement transducers D7
and D8. It was found that ϕb is almost the same as the modified ϕb′ at
the beginning of the loading stage, where ϕb′ was modified by excluding
the flexural rotation according to the correction formulae given be
Abdel-Jaber et al. [7]. This proves that rotation formulation presented
above is the proper way to obtain the true rotation of the beam-to-
Fig. 10. Beam-end-connector 5T failure mode (distortion and rupture).
upright connection.
Fig. 12 shows an idealized characteristic moment–rotation curve for
fuses thereby preventing an upright failure. It is argued that a tab failure
results in a beam failure, which can be easily replaced as compared to an typical beam-to-upright connections of steel storage racks. The connection
is expected to display a linear behavior in the beginning, represented by
upright failure, which may lead to overall rack structure collapse.
line OA. However, if there was a slack in the connection, the behavior
may be OA'A, representing the initial flexibility and stiffening later. Beyond
3.2. Moment–rotation curves the linear part OA, the moment–rotation relations typically exhibit
non-linear characteristic before the peak moment, though in Fig. 12
The moment M at the intersection of upright and beam axes can be it is shown as a straight line AB. Here, point B is the ultimate moment
obtained as M = P ∗ d, where, P is the applied load, and d is the actual of the connection. The reasons for this non-linear behavior may be
horizontal distance. Note that distance d changes during the test, due due to factors such as; (i) relative slippage between the upright
to the deformation of the connection. The moment at each load point and the beam end connector, (ii) yielding of the tabs, or some points
was calculated based on the load and the corresponding distance d, on the end-connector or the upright perforation walls due to localized
which was adjusted based on LVDT-B readings corresponding to that stress concentration, and (iii) geometrical non-linearity. Often point B
load. Recall that the initial distance to the load d was 400 mm plus corresponds to a failure of an element, which may correspond to crack-
half of the width of the upright web, and the maximum displacements ing of the upright perforation wall or the connection top tab. Beyond
recorded in LVDT-A & B were between 15 and 22 mm. Incorporation point B, the load-carrying capacity decreases moderately until initiation
of this adjustment increases the moment up to 5%. The rotation ϕ of of a second crack or some other failure, which is reflected by point C.
the beam-to-upright connection can be defined as the relative rotation Depending on the failure mode, the test may proceed to point D or
point E, when the connection is completely ruptured.
The moment–rotation (M–ϕ) curves of all fifty-one connection
specimens are plotted in Figs. 13–15. In these plots the horizontal axis
gives the rotations, as calculated per above discussion, in radians, and
the vertical axis gives the moments in kN·m. Figs. 13, 14, and 15 focus
on upright C5, C4, and C3, respectively. Salient observation points,

Fig. 11. Schematic diagram for the relative rotation of connection. Fig. 12. Characteristic moment–rotation curves for connections.
X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 169

a) C5-B120-4T b) C5*-B120-4T

c) C5-B105-4T d) C5-B145-4T

e) C5-B120-3T f) C5-B120-5T
Fig. 13. Moment–rotation curves for connections with a C5 upright.

such as, bottom beam-end-connector contact with upright, distortion of admitted that there may be a stiffness-changing point between OA for
tabs, crack of upright perforation at the top tab, crack of top tab, crack of some of the specimen of C3 or C4 series. Except for C5* specimens,
upright perforation at the second top tab, crack of the second top tab, which were with 2.5 mm thick upright, all other specimens exhibited
and rupture of beam-end-connector are also shown in these graphs. good deformational capability beyond peak moments, reflecting the
As evident from these moment–rotation graphs, in general, the ideal- gradual yielding of the elements. The C5* exhibited (Fig. 13(b)), some-
ized deformational trend shown in Fig. 12 was observed in all fifty- what of a sudden failure associated with the tab failure.
one tests. It was observed that, for most part, the initial portions of From these moment–rotation curves, beam-to-upright connection
these graphs are linear, somewhat similar to line OA. The reason may characteristics such as, the initial stiffness K0, the moment capacity Mc
be that the tabs and upright perforations were well designed, fabricated (peak moment) and the rotation ϕm at the maximum moment can
and constructed that there was no slack in-between. However, it is be determined. These values for the fifty-one specimens under
170 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

a) C4-B105-4T b) C4*-B105-4T

c) C4-B120-4T d) C4-B145-4T

e) C4-B105-3T f) C4-B105-5T
Fig. 14. Moment–rotation curves for connections with a C4 upright.

consideration are listed in Table 2. The first column of this table iden- connection is obtained as the slope of a straight-line through the origin
tifies the specimen. The column 2 shows the initial stiffness K0 values which isolates equal areas between it and the experimental curve. The
for each test specimen, which is the gradient of the initial straight line corresponding values are listed in column 6 of Table 2. For some
portion of the experimental curves. According to the RMI specification specimen in some group, for example C5-B120-4T-2, the stiffness is
[5] (which is the same as Australian Standard AS 4084 [4]), the stiffness, substantially different from those of the other two similar connections.
KRMI to be used in a linear analysis must be taken as the gradient of a line In the authors' view, the factors leading to this difference, in order of im-
passing through the original point and a point on the moment– portance, include different degrees of connectivity between the tabs
rotation curve at 85% of the maximum moment. The individual test and the perforations, variations in the actual geometrical dimensions,
values for KRMI have been shown in column 4 of Table 2. According to and perhaps the nominal scatter of steel mechanical properties. Howev-
the European Standard EN 15512 [6], the stiffness, KEN for such er, in general, for three nominally identical specimens in a group, their
X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 171

a) C3-B105-4T b) C3-B120-4T

c) C3-B145-4T d) C3-B105-3T

e) C3-B105-5T
Fig. 15. Moment–rotation curves for connections with a C3 type upright.

moment–rotation curves are similar before the moment reaches half of KEN is used for linear analysis of rack structures. KEN is found to be
the peak moment, which means that the initial stiffness is consistent more appropriate to predict the sway behavior of rack structures [17].
with the connection construction. Considering the three test specimens Usually K0 is not employed in the linear structural analysis model for
of each test group, the corresponding average values can be established, the design of rack structures, but it can be used as an important parameter
and have been given in columns 3, 5, and 7, respectively, of Table 2. The to describe the whole moment–rotation relationship, which is in reality a
average values have been identified in Table 2 with a suffix ‘a’. The ini- non-linear relation.
tial stiffness K0 established in this study is always much larger than the The peak moments reached by these connections are tabulated
stiffness KRMI or KEN obtained according to the RMI or EN 15512 speci- in Table 2 — column 8. The corresponding rotations are given in
fications, respectively. This is because the K0 calculated herein is the Table 2 — column 10. As evident from this table, consistent peak
tangent stiffness, while the other two are secant stiffnesses. KRMI or moments were reached in specimens within a group. The lowest
172
Table 2
The test results.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

Specimen ID. Initial stiffness Average initial stiffness Stiffness per RMI Average of KRMI Stiffness per EN15512 Average of KEN Moment capacity Average moment capacity Rotation at Mc Average of ϕm

K0 K0a KRMI KRMIa KEN KENa Mc Mca ϕm ϕma


(kN·m/rad) (kN·m/rad) (kN·m/rad) (kN·m/rad) (kN·m/rad) (kN·m/rad) (kN·m) (kN·m) (rad) (rad)

C5-B120-4T-1 175.3 183.4 83.1 95.5 102.1 110.8 2.40 2.21 0.044 0.040
C5-B120-4T-2 202.8 106.7 127.9 2.03 0.037
C5-B120-4T-3 172.0 96.8 102.3 2.21 0.039
C5*-B120-4T-1 247.3 238.1 116.4 117.4 144.8 144.9 2.70 2.88 0.038 0.037
C5*-B120-4T-2 230.0 99.8 130.5 2.95 0.040
C5*-B120-4T-3 236.9 136.0 159.3 2.97 0.035
C5-B105-4T-1 148.1 167.5 83.1 88.9 82.5 91.4 2.11 2.13 0.045 0.048
C5-B105-4T-2 196.8 95.1 103.9 1.95 0.037
C5-B105-4T-3 157.5 88.4 87.8 2.34 0.063

X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175


C5-B145-4T-1 192.6 180.0 87.2 87.7 98.6 110.5 2.21 2.39 0.037 0.042
C5-B145-4T-2 169.1 78.3 108.9 2.73 0.049
C5-B145-4T-3 176.9 97.7 124.1 2.23 0.039
C5-B120-3T-1 91.4 98.0 41.4 42.1 49.7 49.8 1.65 1.43 0.064 0.054
C5-B120-3T-2 99.4 43.7 50.9 1.27 0.043
C5-B120-3T-3 103.3 41.1 48.8 1.37 0.056
C5-B120-5T-1 185.5 198.5 127.1 130.4 138.9 143.3 2.71 2.83 0.039 0.042
C5-B120-5T-2 214.2 131.9 145.9 2.88 0.041
C5-B120-5T-3 195.6 132.1 145.1 2.90 0.048
C4-B105-4T-1 170.2 120.9 57.8 55.6 75.2 67.3 1.98 2.00 0.054 0.064
C4-B105-4T-2 94.0 58.8 68.7 2.07 0.066
C4-B105-4T-3 98.6 50.3 58.0 1.95 0.072
C4*-B105-4T-1 139.9 134.9 67.3 63.9 70.9 73.9 2.14 2.17 0.089 0.074
C4*-B105-4T-2 108.3 55.6 66.9 2.18 0.074
C4*-B105-4T-3 156.5 68.7 83.8 2.20 0.059
C4-B120-4T-1 181.6 143.7 105.0 84.5 104.4 94.6 2.40 2.42 0.078 0.058
C4-B120-4T-2 118.8 75.0 93.7 2.39 0.054
C4-B120-4T-3 130.6 73.5 85.7 2.47 0.043
C4-B145-4T-1 198.5 193.5 79.8 72.4 94.2 92.4 2.52 2.43 0.056 0.049
C4-B145-4T-2 238.6 80.5 114.1 2.59 0.043
C4-B145-4T-3 143.3 57.0 68.8 2.18 0.049
C4-B105-3 T-1 95.7 88.2 38.7 29.6 46.3 35.0 1.49 1.47 0.073 0.085
C4-B105-3T-2 84.2 28.9 33.2 1.56 0.097
C4-B105-3T-3 84.6 21.2 25.4 1.37 0.087
C4-B105-5T-1 162.6 173.0 50.7 59.0 68.7 71.7 2.19 2.22 0.072 0.094
C4-B105-5T-2 165.8 83.3 92.4 2.28 0.085
C4-B105-5T-3 190.7 43.1 53.9 2.17 0.126
C3-B105-4T-1 129.1 130.5 55.7 54,5 66.6 67.6 1.90 1.97 0.051 0.055
C3-B105-4T-2 126.6 47.4 59.5 2.01 0.060
C3-B105-4T-3 135.8 60.3 76.7 2.00 0.053
C3-B120-4T-1 112.3 182.4 66.4 80.8 83.5 105.0 2.21 2.18 0.053 0.043
C3-B120-4T-2 226.8 103.9 136.3 2.18 0.033
C3-B120-4T-3 208.2 72.1 95.3 2.15 0.043
C3-B145-4T-1 209.5 240.0 69.3 95.0 82.8 113.6 2.19 2.26 0.041 0.034
C3-B145-4T-2 296.6 123.0 141.5 2.33 0.030
C3-B145-4T-3 213.9 92.7 116.4 2.26 0.030
C3-B105-3T-1 79.7 89.3 24.9 26.2 30.2 31.5 1.35 1.33 0.066 0.064
C3-B105-3T-2 90.6 23.3 27.7 1.27 0.069
C3-B105-3T-3 97.9 30.3 36.6 1.37 0.056
C3-B105-5T-1 178.4 169.8 68.6 65.7 84.4 77.7 2.34 2.27 0.107 0.103
C3-B105-5T-2 175.1 62.9 76.9 2.18 0.107
C3-B105-5T-3 155.8 65.6 71.7 2.30 0.096

Notes: The uprights C5* and C4* are 2.5 mm and 2.0 mm thick, respectively. All other uprights are 1.8 mm thick.
X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 173

a) C5-B120-5T-3

b) C3-B105-3T-3
Fig. 16. Typical load–strain curves.

moment of 1.27 kN·m was reached in specimen C5-B120-3T-2, beam-B145 [6.02 kN·m]}, which were calculated based on the yield
whereas the highest moment of 2.97 kN·m was reached in connection of outmost fiber of the beam. The lowest resistance among the beams
C5*-B120-4T-3. The average moment resistances are given in is 3.7 kN·m which is associated with beam B105. Obviously, the exper-
Table 2 — column 9. The average rotations at peak loads are given in imental connection moment resistances are less than those of the
column 11. The theoretical moment capacities of the beams were beams. The pallet rack is a structure with weak connections and strong
established {beam-B105 [3.70 kN·m], beam-B120 [4.35 kN·m], and beams, which is different from the traditional building structures where
174 X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175

Fig. 17. A schematic diagram for additional moment from eccentricity.

capacity design dictates stronger connection — strong column–weak more, respectively. Similar trends were observed for connections with
beam design for enhanced ductility. a C4 or C3 type upright.
In spite of the minor variations in the test results between the spec-
imens of a group, the influence of different parameters on the initial 3.3. Strains
stiffness and moment capacity of the upright–beam connections can
be made using the average results given in Table 2. Not all strains are shown herein, and load–strain curves for the
beams of two specimens are shown in Fig. 16. The strains in the beam
3.2.1. Effect of upright thickness on the connection behavior section near the end-connector are irregular compared to the ones in
Table 2 compares the results of the first two groups of connections. other sections away from the end-connector. The reason may be that
These connections have the same beam depth of 120 mm (B120) and the welding between the beam and the end-connector, and the stress
four tab connectors (4T), however, the upright thickness was increased concentration in accord with Saint-Venant's principle, lead to uneven
from 1.8 mm (C5) to 2.5 mm (C5*). This resulted in an increase of distribution of forces near the beam end. For the specimen C5-B120-
average initial stiffness and the moment capacity of about 29% and 5T, the strain S2 in section I shows that it has yielded, perhaps because
30%, respectively. Similarly, comparing the C4 upright of thickness of stress concentration. For the specimen C3-B105-3T, strains in section
1.8 mm with that of the C4* upright having a thickness of 2.0 mm, con- I are smaller than the strains in other sections, which is opposite to
nected to beam B105 through connector 4T, shows an increase in initial engineering intuition. This also can be explained by the principle of
stiffness and the moment capacity of about 11% and 8%, respectively. Saint-Venant. The ratio of beam depth to end-connector depth for spec-
Naturally, the connections made of uprights with thicker sections imen C3-B105-3T is larger, comparing with specimen C5-B120-5T,
possess higher stiffness and higher moment capacity. which means that both the top and bottom tabs are closer to the top
flange and bottom flange of the beam, respectively. In reality, the forces
in tabs are not evenly transferred to the beam, but, transferred primarily
3.2.2. Effect of beam height on the connection behavior to the portions of the flange and the outside web closer to the tab (beam
For connections with a C5 type upright, the increase in beam depth web near upright web as shown in Fig. 6.). Essentially, there is an
from 105 to 120 mm has resulted in about 14% and 4% increase in the
average initial stiffness and moment capacity, respectively. While
further increase to a beam depth of 145 mm has increased the average
moment capacity, but the average initial stiffness has remained the
same compared with 120 mm. For connections with a C4 type upright
of thickness 1.8 mm and connector with number of tabs 4 (4T), the
increase in beam height from 105 mm to 120 mm has resulted in
about 19% and 21% increase in the average initial stiffness and moment
capacity, respectively. Further increase to 145 mm has increased the
average initial stiffness by about 35% and virtually no change in the
moment capacity. A similar trend was observed for connections with a
C3 type upright.

3.2.3. Effect of the number of tabs on the connection behavior


In all of the groups of specimens considered, the number of tabs is
the most influencing factor on the behavior of connections. Considering
the connections with a C5 type upright of thickness 1.8 mm and beam
height 120 mm, the increase in the number of tabs from 3 to 4 shows
an increase in the average initial stiffness of about 95% and an increase
in the moment capacity of about 55%, which are significant indeed.
When the number of tabs was increased to 5 the resulting average initial
stiffness and moment capacity were observed to be about 4% and 28% Fig. 18. Principal stresses in the connection zone.
X. Zhao et al. / Journal of Constructional Steel Research 99 (2014) 161–175 175

eccentric load transfer. This eccentricity results in much more uneven beam and the number of tabs. The influence of the depth of
distribution of stresses in the beam section I. For larger depth ratio beam on the moment capacity of connection is limited comparing
sections, where the tab is closer to the flange, or in sections far away with that of the thickness of upright and the number of tabs.
from upright, the impact of the eccentricity of the force in beam section (3) In an actual design, the upright and the tabs (or beam-end-
is much smaller. connector) must be implemented interactively. Tabs play an
The strain measurements marked as odd numbers are noted to be important role in the performance of such structures, thus their
much smaller than those marked as even numbers (see Fig. 7). The rea- design controls the behavior and the strength of the whole rack
son is that the force acting on the beam end-connector from the upright structure. Based on the analysis of results presented herein, four
is not in the vertical plane passing through the centroidal axis of the tab connectors are found to exhibit optimal performance.
beam profile as shown in 1-1 sectional view in Fig. 6. The forces acting
on tabs lead to additional moment due to eccentricity e, resulting in
uneven distribution of forces on the beam section together with the mo- Acknowledgments
ment from applied load. The mechanical explanation is shown in Fig. 17.
Not too surprisingly, in all cases, the strains at rosette points T6, T7 The test specimens used in this experimental program were provided
on the connector and T9, T10 on the upright near the top and bottom by WAP Logistics Equipment (Shanghai) Co., Ltd.
tabs (see Fig. 7) evolved much more rapidly than at the strain rosettes
located at other points. Utilizing the strain rosette measurements, typi- References
cal distribution pattern of principal stresses within the connection zone
[1] Markazi FD, Beale RG, Godley MHR. Experimental analysis of semi-rigid boltless
between the region of top and bottom flanges of beam is shown in connectors. Thin-Walled Struct 1997;28(1):57–87.
Fig. 18. In this figure, the arrows show the magnitude and the direction [2] Baldassino N, Bernuzzi C. Analysis and behavior of steel storage pallet racks. Thin-
of the principal stresses at the measurement points. The principal Walled Struct 2000;37(4):277–304.
[3] Baldassino N, Zandonini R. Design by testing of industrial racks. Adv Steel Constr
stresses at T9 and T10 points are much larger than that of other points 2011;7(1):27–47.
because of the interaction between the tabs and the upright walls. [4] AS 4084. Steel storage racking. Sydney (Australia): Standards Australia; 1993.
Thus the steel of upright wall near these two points is expected to [5] RMI. Specification for the design, testing and utilization of industrial steel storage
racks. Charlotte (USA): Rack Manufacturers Institute; 2008.
yield first, while the material at other points remains elastic for most
[6] EN 15512. Steel static storage systems – adjustable pallet racking systems – principles
of the test. The stress flow in the diagonal direction from points T1 to for structural design. Brussels (Belgium): European Committee for Standardization
T4 then to T10 is compressive, while the stress flow in the diagonal (CEN); 2009.
[7] Abdel-Jaber M, Beale RG, Godley MHR. A theoretical and experimental investigation
direction from points T3 to T4 then to T9 is tensile.
of pallet rack structures under sway. J Constr Steel Res 2006;62(1–2):68–80.
[8] Bernuzzi C, Castiglioni CA. Experimental analysis on the cyclic behaviour of
4. Conclusions beam-to-column joints in steel storage pallet racks. Thin-Walled Struct
2001;39(10):841–59.
[9] Krawinkler H, Cofie NG, Astiz MA, Kircher CA. Experimental study on the seismic
This paper concerns with the experimental evaluation of the flexural behavior of industrial storage racks. Report No. 41. California: The John A. Blume
behavior of beam-to-upright connections typically used in cold-formed Earthquake Engineering Center, Department of Civil and Environmental Engineering,
steel storage pallet racks. The investigation considered 17 groups of Stanford University; 1979.
[10] Prabha P, Marimuthu V, Saravanan M, Jayachandran SA. Evaluation of connection
three identical specimens, subjected to hogging static loading in a single flexibility in cold formed steel racks. J Constr Steel Res 2010;66(7):863–72.
cantilever test setup. Based on the load, displacement, strain measure- [11] Bajoria KM, Talikoti RS. Determination of flexibility of beam-to-column connectors
ments and experimental observations, the following conclusions may used in thin walled cold-formed steel pallet racking systems. Thin-Walled Struct
2006;44(3):372–80.
be made: [12] Sarawit AT, Pekoz T. Design of industrial storage racks. In: LaBoule RA, Yu WW,
editors. 16th International Specialty Conference on Cold-Formed Steel Structure
(1) Typical failure modes for beam-to-upright connections are Orlando (FL, USA); 2002. p. 369–84.
tearing of the upright wall and cracking of tab. The thickness of [13] Harris E. Sway behaviour of high rise steel storage racks. [Ph.D. thesis] University of
upright relative to the thickness of tabs determines the failure Sydney; 2006.
[14] Gilbert BP, Rasmussen KJR. Bolted moment connections in drive-in and drive-through
mode. Typical moment–rotation relationship is nonlinear,
steel storage racks. J Constr Steel Res 2010;66(6):755–66.
reaching the peak moment at the first crack. The tearing of [15] Markazi FD, Beale RG, Godley MHR. Numerical modelling of semi-rigid boltless
upright failure mode exhibited better post-peak behavior connector. Comput Struct 2001;79(26–28):2391–402.
(ductility) compared to the cracking tab failure mode. [16] Aguirre C. Seismic behavior of rack structures. J Constr Steel Res 2005;61(5):607–24.
[17] Abdel-Jaber M, Beale RG, Godley MHR. Numerical study on semi-rigid racking
(2) The initial stiffness and the moment capacity of connections are frames under sway. Comput Struct 2005;83(28–30):2463–75.
mainly determined by the thickness of upright, the depth of

You might also like