Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Contents lists available at ScienceDirect

Mechanical Systems and Signal Processing

journal homepage: www.elsevier.com/locate/ymssp

Behavior of Industrial Steel Rack Connections

S.N.R. Shah a, N.H. Ramli Sulong a,n, R. Khan b, M.Z. Jumaat a, M. Shariati a
a
Department of Civil Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
Mechanical Engineering Department, College of Engineering, Al-Imam Muhammad Ibn Saud Islamic University, 11432 Riyadh, Kingdom
of Saudi Arabia

a r t i c l e i n f o abstract

Article history: Beam-to-column connections (BCCs) used in steel pallet racks (SPRs) play a significant role
Received 15 June 2015 to maintain the stability of rack structures in the down-aisle direction. The variety in the
Received in revised form geometry of commercially available beam end connectors hampers the development of a
18 August 2015
generalized analytic design approach for SPR BCCs. The experimental prediction of flex-
Accepted 23 August 2015
Available online 26 September 2015
ibility in SPR BCCs is prohibitively expensive and difficult for all types of commercially
available beam end connectors. A suitable solution to derive a particular uniform M–θ
Keywords: relationship for each connection type in terms of geometric parameters may be achieved
Cold-formed steel racks through finite element (FE) modeling. This study first presents a comprehensive
Beam-to-column connections
description of the experimental investigations that were performed and used as the
Double-cantilever test
calibration bases for the numerical study that constituted its main contribution. A three
Finite element modeling
Parametric analysis dimensioned (3D) non-linear finite element (FE) model was developed and calibrated
Moment-rotation curve against the experimental results. The FE model took into account material nonlinearities,
geometrical properties and large displacements. Comparisons between numerical and
experimental data for observed failure modes and M-θ relationship showed close
agreement. The validated FE model was further extended to perform parametric analysis
to identify the effects of various parameters which may affect the overall performance of
the connection.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Steel pallet racks (SPRs) have exhibited considerable growth in the last few years, even if they comprise a structural
system that is relatively unique compared with traditional steel structures. SPRs are made of cold-formed steel. Pallet rack
designs, which can be easily expanded, come in numerous configurations to handle any storage need. However, SPR
structure comprises two main directions, namely, cross-aisle and down-aisle directions. Bracing systems are generally
placed only in the cross-aisle direction. Given the requirement to organize racking systems such that products are efficiently
stored and sufficiently accessible, the presence of bracing systems is generally hampered in down-aisle direction, and hence,
lateral stability is provided by the degree of continuity offered by beam-to-column connections (BCCs) and base plate
connections. Thus, SPR BCCs are significantly responsible for the overall stability of the rack structure.
The SPR BCCs are considered as semi-rigid boltless connections and their behavior is governed by the restraining effect of
the connection device which is called ‘beam end connector’. The important part of the beam end connector is the ‘tab’. These

n
Corresponding author at: Department of Civil Engineering, University of Malaya, Malaysia.
E-mail addresses: [email protected] (S.N.R. Shah), hafi[email protected] (N.H. Ramli Sulong).

http://dx.doi.org/10.1016/j.ymssp.2015.08.026
0888-3270/& 2015 Elsevier Ltd. All rights reserved.
726 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

tabs are inserted into column perforations to establish the BCC. SPR BCCs usually experience an initial looseness of the
connection due to the absence of bolts and welds. This feature of SPR BCCs need special design consideration. The prediction
of the behavior for all SPR BCCs through a generalized analytic approach is very difficult because of the wide variety of the
types and sizes of commercially available beam end connectors. Thus, the most recent design codes, such as that of the RMI
[1], EN 15512 [2], and AS 4084 [3], suggest individual experimental testing and define testing protocols with a marginal
difference to predict the moment–rotation (M–θ) behavior of any SPR BCC. However, these experimental investigations are
expensive and difficult to repeat. Therefore, a suitable solution is to derive a particular uniform M–θ relationship for each
connection type in terms of geometric parameters through finite element (FE) modeling. Several studies have been pre-
sented on numerical analysis of steel connections in last few years [4–7]. A considerable number of studies performed in last
few years are available on the experimental testing of SPR BCCs [8–14], however, a very few studies numerically discuss the
behavior of SPR BCCs solely [10,15,16].
This paper investigates the structural behavior of SPR BCC. Eight different sets of specimens were tested. Each set was
composed of four trials of each specimen taking the total number of tests to thirty-two. Double cantilever test set-up was
used and the M-θ behavior and major failure modes were investigated. Then, a 3D nonlinear elasto-plastic FE model was
developed according to the geometry of tested specimens using the commercial FE software ABAQUS [17]. The results of the
developed FE models were compared with the experimental results, and overall close agreement was achieved. Then, a
parametric study was conducted to compute the strength and stiffness of the connection by varying the dimensions of those
influential parameters which may affect the strength and stiffness of SPR BCCs. This study demonstrates that how the
variations in geometric characteristics of components can modify connection behavior.

2. Experimental Investigations

2.1. Material Properties

Cold formed steel sections were used for columns and beams. The beam end connectors were manufactured of hot rolled
steel. The material properties of the column, beam and beam end connector were obtained through the tensile coupon test
and are given in Table 1.

2.2. Specimen Details

A total of thirty-two tests were carried out. The specimens were distinguished by two different column thicknesses, four
different beam depths and the number of tabs in the beam end connector being either four or five. The cross-section of the
column is illustrated in Fig. 1(a). Box-beams with four different depth values, namely B1, B2, B3 and B4 were used for
experimental testing. Beams B1 and B2 had a four tab beam end connector, while for B3 and B4, the connector had five tabs.
The cross section of the box-beam is represented in Fig. 1(b). The dimensions of the columns and beam sections are given in
Table 2. The geometry of the beam end connectors was distinguished by the number of tabs in the beam end connector.
Connector ‘A’ had four tabs and connector ‘B’ had five tabs. The depths of connectors ‘A’ and ‘B’ were 200 mm and 250 mm,
respectively. The thickness of the beam end connectors was 4 mm. The cross-section of the beam end connector is shown in
Fig. 1(c). All the dimensions of the specimens are the measured values.
For a clarified representation of the specimens under investigation, each specimen was given a specific specimen ID and
a letter, which is listed in Table 3. For example, in the specimen ID ‘2.0UT-92BD-4T’, 2.0UT represents the column thickness
as being 2.0 mm, 92 BD represents the depth of beam as being 92 mm and 4T represent the number of tabs in the beam end
connector, which is four.

2.3. Testing Arrangement

In this study, the double-cantilever test method was adopted to predict the M-θ behavior of the connection. The column
was first aligned and vertically levelled exactly below the loading apparatus. Two beams were then connected to the left and
right sides of the center of the column. The lateral movement of beams was restricted by restraining them by means of two
rectangular hollow sections welded to the angle sections and bolted to the strong floor. The unconnected ends of the beams

Table 1
Material properties of specimens.

Member Young’s Modulus (E) Poisson’s ratio (ν) Yield strength (fy) Ultimate strength (fu) Plastic strain at UTS
(GPa) (MPa) (MPa)

Column 210 0.3 459 575 0.2272

Beam 353 497 0.3275
Beam-end-Connector 263 365 0.4182
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 727

Fig. 1. Cross-section of the connection components. (a) Column, (b) Beam, (c) Beam end connector.

Table 2
Dimension details of columns and beams.

Column Beam

A B ID Width ‘b’(mm) Depth ‘h’(mm) Thickness ‘t’(mm)

Thickness ‘ t’ (mm) 2.0 2.6 B1 40 92 1.5

Flange width ‘bt’ (mm) 67.6 68.3 B2 40 110 1.5
Web ‘w’ (mm) 112.2 113.1 B3 50 125 1.5
Height ‘h’ (mm) 500 500 B4 50 150 1.5

Table 3
Details of specimens’ ID.

Letter assigned to specimens Specimen details

A 2.0UT-92BD-4T
B 2.0UT-110BD-4T
C 2.0UT-125BD-5T
D 2.0UT-150BD-5T
E 2.6UT-92BD-4T
F 2.6UT-110BD-4T
G 2.6UT-125BD-5T
H 2.6UT-150BD-5T

Beam end connector

50 mm

L/4 L/4 L/4 L/4

Fig. 2. Schematic diagram of test set-up.

728 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

were supported by roller supports on the left and right sides. The effective distance between the roller supports was 2 m.
The tabs of the beam end connectors were reversely hooked in the column perforations. A locking pin was used to avoid any
change in the position of the column or the connector due to accidental uplift. The load was applied in a displacement
control based method using a 50 kN hydraulic actuator controlled by the computer at a rate of 3 mm/min until the con-
nection failure. The load was applied to the top of the column which causes compression in the top of the beam end
connector and tension at the bottom. Two digital inclinometers were placed on the top flanges of the beams on either side at
a distance of 50 mm from the face of the column to directly record the rotation of beams in degrees. Deflection mea-
surements using linear variable differential transducers (LVDTs) having a measurement range from 50 mm–200 mm were
also installed. Two LVDTs were installed at a distance of L/4 from the center of roller support, on the beams on either side to
observe the deflection in the beams. One LVDT was placed at the bottom of the column to observe central deflection. For the
tests involving beams B1 and B2, three strain gauges were pasted. One strain gauge (S1) was pasted in the column web near
the top surface of the beam end connector to estimate the tensile strain. The other two strain gauges (S3 and S4) were
pasted near the bottom slots of both the left and right beam end connectors in the tension region. For beams B3 and B4, four
strain gauges were installed. Three strain gauges were pasted in the same position as for B1 and B2, whereas an additional
strain gauge (S2) was installed in the lower portion of column web near the bottom surface of the beam end connector.
Readings from the strain gauges and LVDTs were recorded onto the computer system through data logger. The schematic
diagram of test set-up is presented in Fig. 2.

3. Experimental Results

3.1. Moment-Rotation (M-θ) behavior and stiffness

The moment was calculated by the following equation:

P L w
Moment (M ) = × −
2 2 2 (1)

where L is the length between the supports, w/2 is the half width of the column; as the bending moment is to be calculated
in the beam end connector, the half width of the column is subtracted from L. The stiffness was calculated using equal area
method. In this method, the rotational stiffness is taken as the slope of the line passing through the origin, which isolates
equal areas between it and the experimental curve, below the design moment. The average test results are given in Table 4.

3.2. Failure Modes

During experimental investigations, collectively, among all the experiments, three failure modes were observed:
(i) tearing of the column material, (ii) yielding of the beam end connector, and (iii) fracture or yielding of the tabs. In the
specimens A and B, the pre-dominating failure mode was the failure of the tabs. The top tabs on both sides in compression
zone were completely ruptured. The bottom two tabs were deformed and disengaged from the column slots after tearing
the column flange. At failure, the beam end connector experienced a noticeable deformation. The failure phenomenon of
specimens E and F showed significant distortion of all tabs and the beam end connector, however, the noticeable difference
was the increase in failure load as compared to the specimens A and B. In the tests conducted on specimens C and D, the
complete rupture of top tabs was not observed. The tabs in the tension zone of the beam end connector came out by tearing
the column flange. The tests on specimens G and H exhibited that the last tabs in the tension side initially disengaged and
finally came out of the column slots, completely. No beam failure was noticed in any test. A typical failure of connection is
presented in section 5.1 as a comparison with the failure achieved through FE analysis.

Table 4
Average Test Results.

Specimen Failure Load (kN) Ultimate Moment Capacity (kNm) Rotation (Radians) Stiffness (kNm/rad)

A 4.89 2.31 0.10 32.3

B 5.19 2.45 0.10 36.2
C 6.27 2.96 0.094 54.29
D 8.05 3.80 0.088 79.21
E 5.17 2.44 0.10 35.9
F 5.44 2.57 0.10 42.8
G 7.54 3.56 0.091 69.54
H 10.04 4.74 0.082 97.86
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 729

Fig. 3. Geometry of tabs.

4. FE Analysis

FE simulations were performed to evaluate the structural behavior of SPR BCCs. A 3D nonlinear FE model of the column,
beam, and beam end connector assembly was developed using the commercial FE software ABAQUS [17] for both four- and
five-tab connectors.

4.1. Connection Modeling

Four- and five-tab beam end connector assemblies were modeled. Element C3D8R (i.e., continuum 3D with eight nodes
and reduced integration) was used for all the three parts of an assembly. A bilinear elasto-plastic material model was
adopted to estimate the deformation behavior of the column, beam, and beam end connector assemblies. The material
properties of all the three components listed in Table 1 were used for FE modeling (FEM). Plastic strain (Table 1) was
calculated through the Ramberg–Osgood method [18]. The tabs of the connector were modeled using SOLID element with a
thickness of 4 mm (Fig. 3) to replicate the original experimental conditions. Considering tabs as a structural solid element
complicated the FE model. However, this approach replicated an actual system, and thus, the deformation behavior of tabs
could be predicted.

4.2. Surface Interaction

Contact nonlinearity was incorporated into the FE model by defining the interactions among column, beam end con-
nectors, and tabs. The surface-to-surface interactions (front and side) between the column and the beam end connector
were defined through tangential frictionless behavior, as shown in Fig. 4(a). Similarly, the surface-to-surface interactions
between the column and the tabs were defined in two ways: (i) normal hard contact, as shown in Figs. 4(b) and 4(c), and (ii)
tangential frictionless contact, as shown in Fig. 4(d). The former was defined to avoid the relative movement between the
surfaces of the column and the tab, whereas the latter was defined to restrain the sideways movement (normal to long-
itudinal axes) of the column. Future research may incorporate precise surface-to-surface tangential contact through the
Coulomb friction model. The weld between the beam end connector and the beam was modeled using the BEAM element
which simulates the welded connection. In future works, more advanced analysis can be performed through coupling
elements that replicates the exact type of welded connections in structures. The complete geometry of the modeled con-
nections is presented in Fig. 5.

4.3. Loading and boundary conditions

In experimental setup, the roller and lateral supports were used to restrain the vertical and lateral movements of beams,
respectively. Further, in double-cantilever test setup, the column moves only up and down as a rigid body. Therefore, similar
boundary conditions were applied to the end of beams and the column. During test setup, a displacement controlled loading
730 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Fig. 4. Surface to surface interaction among components.

was applied. Similar loading procedure was adopted for FE analysis. Displacement was applied in 10,000 steps and con-
trolled by providing the logical limits of time (step size) from 0 to 1. This made the every sub-step equals to 1/10,000. The
step size was further sub-divided and this undertaking resulted in a displacement rate of 70  10 4 mm/sub-step size. This
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 731

Fig. 5. Complete geometry of the modeled connection (a) Connection with four tab beam end connector, (b) Connection with five tab beam end connector.

Fig. 6. Division of the connection components into a four-sided region (a) Division of the column, (b) Division of the beam end connector.

small value was selected to avoid strain rate effects on deformation behavior. A displacement controlled loading was applied
on the top surface of the column in the negative y axis.

4.4. Model Discretization

All components of the SPR assembly were discretized using the mapped discretization scheme to enhance computational
accuracy. Mapped discretization was achieved by efficiently dividing individual components into a four-sided region, as
shown in Figs. 6(a) and (b).
In the experimental study, excessive deformation was observed at the joining regions of the tab and the beam end
connector. High-density mesh was utilized in these regions to capture the deformation phenomenon precisely. The meshed
components are shown in Figs. 7(a)–(c).
732 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Fig. 7. Discretized components of the SPR assembly: (a) column, (b) front view of the beam end connector, (c) back view of the beam end connector to
show tabs.

Convergence issues in the simulations because of contact, geometric and material nonlinearities were successfully
resolved by reducing time step size and increasing mesh density in critical regions. Fig. 8 shows the meshing of the entire
connection.

5. FE Results

5.1. Failure Modes

The FE model exhibited close agreement with the experimental results. Similar to the experimental investigations,
deformation of almost all tabs was observed in the specimens with relatively shallow beams and thin columns, however, the
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 733

Fig. 8. Meshing of connection.

Fig. 9. Comparison of the connection failure. (a) Experimental failure, (b) FE analysis failure.

tabs in the tension zone experienced higher stress and attempted to tear down the column slots. Moreover, the deformation
of the beam end connector observed in the experimental and FE results were similar. The FE model also captured the
distortion of the column wall and the phenomenon of tabs attempting to exit column slots. However, the distortion of
column slots captured in the FE model was less intense than that in the experimental investigation. A comparison of failures
in both investigations, namely, experimental and FE analysis, is presented in Figs. 9(a) and (b).

5.1.1. Failure of the beam end connector

Similar to the experimental investigations, the FE analysis indicated that the beam end connector experienced a
noticeable deformation in the tension zone. The gradually increased loading increased the gap between the column and the
tension zone of the beam end connector at connection failure. Fig. 10(a) shows the failure of the beam end connector ‘A’
during experimental investigations. The von Mises stress distribution and the failure of the beam end connector A in the FE
model is shown in Fig. 10(b) to compare them with the experimental results.
734 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Fig. 10. Comparison of failure of the beam end connector ‘A’ (a) Experimental failure, (b) FE analysis failure.

Fig. 11. Failure of top tabs.

5.1.2. Failure of Tabs

The FE model was unable to predict the complete rupture of top tabs for specimens A and B, however a noticeable
deformation of top tabs was observed as shown in Fig. 11. The distortion levels at the bottom tabs were almost similar to
those observed in experimental testing. The only difference between the two types of specimens (A and B) was the failure
load. A high beam depth (110 mm) caused the specimen to sustain a marginally large amount of load. The FE models
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 735

Fig. 12. Comparison of the maximum deformation of the column at the connection failure (a) Experimental failure, (b) FE failure.

prepared for rest of the specimens predicted the same level of deformation of tabs as observed in the experimental
investigations.

5.1.3. Failure of column

The third type of failure observed in the experimental investigations was the tearing of column flange and cut in the
holes. According to von Mises stress distribution, the column was under high stress in the tension zone of the connection.
Consequently, the portion of the column near the tension zone of the beam end connector experienced higher stress than
that in the compression zone. The tabs were reversely inserted into the column perforations. Thus, a large stress con-
centration was observed at the portion of the column where the tab came in contact with the column slot in the tension
zone, as opposed to other parts of the column web.
During experimental investigations, for almost all specimens, the tabs in tension zone slit the column slots and clearly
came out by tearing the column. The FE analysis predicted the tearing of column slots, however the intensity of tearing was
less than that observed in the experimental testing. Maximum column deformation at the connection failure occurred in the
tension zone. A comparison of column failures during experimental and FE investigations and stress distribution observed in
FE analysis is provided in Figs. 12(a) and (b).

5.2. Moment-Rotation (M-θ) behavior and stiffness

The FE model predicted the experimental behavior to a large extent. The stiffness of the specimens matched well with
that in the experimental test result. However, the ultimate moment capacity of the connection obtained through the FE
model for specimen A was slightly higher than that obtained through the experiments. This is because the rupture of top
tabs was not captured by the FE model. Further, a slight difference between the column slot failures in the two types of
investigations increased the moment capacity of the connection predicted by the FE model.
A comparison of the M–θ graphs plotted for the experimental and FEM investigations is provided in Fig. 13(a–d). Four
specimens with varying column thickness values, beam depths, and numbers of tabs in the beam end connector were
compared to illustrate the agreement between the experimental and FE analysis results. The major difference between the
two types of curves is the increased moment capacity in ultimate stage of most of the FE curves which is because the less
intensity of column slots failure was predicted by FE model. The FE curve of specimen A showed a different behavior from its
initial stages as compared to the experimental curve because the FE model could not captured the complete rupture of tabs
736 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Fig. 13. Comparison of the M–θ graphs of the experimental and FE investigations (a) specimen A, (b) specimen D, (c) specimen E, (d) specimen H.

Table 5
Comparison of the moment capacity and stiffness of the tested connection in the experimental and FE investigations.

Specimen ID Ultimate Moment Capacity (kNm) Stiffness (kNm/rad)

Experimental FEA Experimental FEA

A 2.31 2.49 32.3 37.93

D 3.80 3.93 79.21 82.89
E 2.44 2.63 35.9 44.02
H 4.74 5.02 97.86 101.4

at connection failure. Table 5 shows a comparison between the moment capacity and stiffness obtained from the experi-
mental investigation and those obtained from the FE investigation.

6. Parametric analysis

The FE model that was developed to validate the experimental testing was further extended for parametric analysis. The
analyzed parameters are as follows: variation in column thickness, beam depth and number of tabs in the beam end
connector, variation in the thickness of the beam end connector, variation in the welding position of beams to the beam end
connector, and variation in the spacing between the tabs by keeping the depth of the connector similar to that during the
experimental test. Different sizes of specimens were chosen for parametric analysis to observe the connection response.
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 737

Fig. 14. Effect of variation in parameters on the M–θ behavior (FE analysis).

Fig. 15. Failure of the beam end connectors (a) Specimen I, (b) Specimen J.

6.1. Effect of varying column thickness, beam depth and number of tabs in the beam end connector

Fig. 14 and Table 5 shows the effect of variations in column thickness, beam depth, and number of tabs in the connector,
as observed through FEM. The FEM results indicated that increasing the sizes of the members resulted in an increment in
the strength and stiffness of the connection.
According to the FEM results, by keeping the depth of beam and number of tabs constant, a comparison of specimens A
and E showed that by increasing the column thickness from 2.0 mm to 2.6 mm, the moment capacity and stiffness of the
connection are increased by 6% and 14%, respectively. By keeping the column thickness constant, a comparison of specimens
A and D showed that a combined effect of increasing beam depth (from 92 mm to 150 mm) and the number of tabs (from
four to five) showed an increase of 37% and 54% in the moment capacity and stiffness, respectively, of the connection.

6.2. Effect of variation in the thickness of the beam end connector

The thickness of the beam end connector was 4 mm for all the specimens tested in the experiment. The FE analysis was
performed by varying the thickness of the beam end connector by 750%, that is, 2 mm- and 6 mm-thick beam end con-
nectors were used. The geometries of the remaining connection components were the same as those for the experimental
testing of specimen H. The letter IDs ‘I’ and ‘J’ were assigned to the specimens with 2 mm- and 6 mm- thick beam end
connectors, respectively.
Reducing beam end connector thickness from 4 mm to 2 mm (specimen I) significantly affected the failure mode of the
beam end connector. The FE analysis showed that the failure of the beam end connector was the major failure mode along
with an increased level of deformation of the last two tabs on either side in the tension zone of the beam end connector. The
fourth tab from the top was more deformed as compared to the deformation observed during experimental testing of
specimen H. Fig. 15(a) shows the failure mode of the 2 mm-thick beam end connector.
738 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

Fig. 16. M–θ graphs of the variation in the thickness of the beam end connector.

Fig. 17. Failure modes of the specimens with various welding positions of beam to the beam end connector (a) Specimen K, (b) Specimen L, (c) Specimen M.

Fig. 15(b) shows the response of the specimen ‘J’ after the FE analysis. Increasing the thickness of the beam end connector
from 4 mm to 6 mm improved the connection performance and significantly minimized the deformation of the beam end
connector and tabs while keeping all other dimensions the same as those of the experimentally tested specimen H. The
stress was higher in the top two tabs in the compression zone and bottom tabs in the tension zone. The failure mode was the
deformation of bottom tabs. However, the deformation was less than that of the specimen H tested experimentally.
Fig. 16 shows a comparison of the M–θ graphs for the specimens H, I and J. The influence of the dimensional changes of
the beam end connector highly influenced the stiffness of the connection. Taking 4 mm thickness of the beam end connector
as a base, a 50% decrease in the thickness of the beam end connector decreased the moment capacity and stiffness of the
connection by 18% and 42%, respectively. An increase in the thickness of the beam end connector by 50% increased the
moment capacity and stiffness of the connection by 14% and 49%, respectively.

6.3. Effect of welding position of beam to the beam end connector

Simulating the experimental investigations involving specimen E, where the position of tabs was reversed and the
distance of top flange of beam was 40 mm from the top of the beam end connector, parametric analysis was performed on
specimen E in order to assess the effect of welding position of beam to the beam end connector.
For parametric analysis, several models were prepared and the top flange of beam was welded at different distances from
the top of the beam end connector. Keeping the rest of the dimensions of the connection similar to specimen E, the welding
distance of top flange of beam from the top of the beam end connector was changed from 40 mm to 10 mm (specimen K),
30 mm (specimen L), and 70 mm (specimen M). In specimens K and L, the beam was up-welded whereas in specimen M, the
beam was down-welded. According to Markazi et al. [8], a beam is considered up-welded when the horizontal axis of
symmetry of the beam is off set upwards from the beam end connector's mid height. A beam is considered down-welded
 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740 739

Table 6
Strength and stiffness of the specimens with various welding positions of beam to the beam end connector.

Specimen Ultimate Moment capacity (kNm) Stiffness (kNm/rad)

K 1.73 23.28
L 2.21 29.8
M 2.81 55.04

Fig. 18. M-θ behavior of specimen N.

when the horizontal axis of symmetry of the beam is off set from the beam end connector's mid-height, either below or just
above it [8].
Welding the beam 10 mm below the top surface of the beam end connector led to a premature failure of the connection,
as shown in Fig. 17(a). The tabs in compression zone were highly distorted at relatively lower value of failure load. The
bottom two tabs were less distorted than the tabs in the compression zone. The stress distribution in the top tabs was highly
non-linear and considerably critical at the weld between the beam and the beam end connector near the second tab from
the top. The distortion in the beam end connector was significant in the compression zone. In addition, a highly non-
uniform stress distribution near the second tab from top to the bottom flange of the beam was observed which influenced
the behavior of weld between the beam and the beam end connector. The governing failure mode in this type of specimen
was a pre-mature failure of the tabs and the beam end connector in the compression zone. FE analysis of Specimen L
showed a slight variation in failure mode compared with that of the specimens E and K. The failure of connection occurred
due to significant distortion of second tabs from the top. The deformation of the beam end connector in the tension zone
was also noticeably higher than that observed for the specimens E and K, as shown in Fig. 17(b). Specimen M showed a
relatively even distribution of load between the tabs in tension and compression zones, hence the performance of the tabs
was better. Further, the beam end connector was remained in contact with the face of the column at the connection failure
and deformation was minimized as compared to the reference specimen E. The major failure mode was the deformation of
tabs. The failure of specimen M is shown in in Fig. 17(c). The ultimate moment capacity and the stiffness of the specimens
with different welding position of beam to the beam end connector obtained through the M-θ curves are tabulated
in Table 6.

6.4. Spacing between the tabs in the beam connector and column perforations

Spacing between the tabs in the beam end connector was 50 mm for all the specimens used in the experimental
investigation. The FE analysis was performed by reducing the spacing between the tabs in the beam end connector by 20%,
that is, 40 mm, and by keeping the connector depth the same as that of the specimen E used in the experimental test. The
distance between the column perforations was also adjusted according to the selected spacing between the tabs. The effect
of decreased spacing only was analyzed. An increase in spacing between tabs by 20% may result in increased connector
depth. The geometries of the remaining connection components were kept the same as those for the experimental testing of
specimen E and the specimen with 40 mm tab spacing is named as ‘specimen N’.
Reducing the spacing between tabs from 50 mm to 40 mm (specimen N) improved the distortion intensity of the top two
tabs compared with the FEM results achieved for specimen E. However, a significant increase in failure intensity of the beam
end connector and the tabs in tension zone was observed. Both edges of the beam end connector (inner and exposed) were
significantly distorted, which caused a drop in failure moment and influenced the stiffness of the connection. Fig. 18 shows
the M-θ behavior of specimen ‘N’. Taking specimen E as a reference, a 20% decrease in the tab spacing reduced the ultimate
moment capacity and stiffness of the connection by 8% and 32%, respectively. Based on the observations of this study, a
spacing of 50 mm between tabs is suitable for a beam end connector depth of 200 mm.
740 S.N.R. Shah et al. / Mechanical Systems and Signal Processing 70-71 (2016) 725–740

7. Conclusion

The behavior of SPR BCCs is naturally complex. The application of this type of boltless connection is limited. The design
problems of these connections in terms of use in engineering practice are related to the structural analysis of this connection
type and the reliability of the analysis. This study presents an experimental and numerical evaluation of SPR BCC behavior
under static loading. A 3D nonlinear FE analysis of such complicated connections is performed, and the results are discussed.
Each analysis is compared with the test result. After validating the FE models, the study is further extended by performing a
parametric analysis. The study revealed that several factors that govern the strength and stiffness of SPR BCCs exist. The
following conclusions can be drawn from this study.
The governing failure in the experimental analysis was tab failure. The connections sustained further load even after the
distortion of the beam end connector had started. Connection failure actually occurred when the tabs were disengaged from
the column slots that cut the column flange and caused a drop in load. Column failure was the secondary failure mode. The
intensity of this failure was minimized in connections with deep beam sections and a large number of tabs in the beam end
connector. Connection efficiency was enhanced by increasing column thickness, beam depth, and the number of tabs in the
connector. The parametric analysis showed that an increase in beam end connector thickness enhanced the performance of
the connection. Regarding the position of weld of beam to the beam end connector, the findings revealed that the up-
welding of beam causes a non-uniform stress distribution in the tabs which compels the connection towards a pre-mature
failure. Thus, up-welding of the beam is detrimental and reduces the performance of the connection. Down-welding of
beam resulted in a better performance of connection in terms of strength and stiffness. According to findings of this study,
keeping the spacing between the tabs of the beam end connector approximately one-fourth of the total depth of the beam
end connector may result in satisfactory performance of the connection.
The increasing size of parameters increased the stiffness of connection at a higher rate as compared to the strength. This
indicates that the elastic properties of SPR BCCs rely on geometrical parameters.

Acknowledgment

The study presented herein was made possible by the University of Malaya Higher Impact Research (HIR) Grant: UM.C/
HIR/MOHE/ENG/57 and PPP grant (project no. PG055/2013B). The authors would like to acknowledge the support. The study
presented herein was made possible by the University of Malaya Higher Impact Research (HIR) Grant: UM.C/HIR/MOHE/
ENG/57 and PPP grant (project no. PG055/2013B). The authors would like to acknowledge the support.

References

[1] Rack Manufacturers Institute (RMI). Specification for the design, testing and utilization of industrial steel storage rack. Material Handling Industry,
MH16.1 2012. Charlotte, NC.
[2] EN 15512. Steel static storage systems–Adjustable pallet racking systems–Principles for structural design. European Committee for Standardization
2009. Brussels, Belgium.
[3] Standards Australia. Steel storage racking. AS4084, 2012. Sydney, Australia.
[4] A. Saritas, A. Koseoglu, Distributed inelasticity planar frame element with localized semi-rigid connections for nonlinear analysis of steel structures,
International Journal of Mechanical Sciences 96-96 (2015) 216–231.
[5] S. Bograd, P. Reuss, A. Schmidt, L. Gaul, M. Mayer, Modeling the dynamics of mechanical joints, Mechanical Systems and Signal Processing 25 (8) (2011)
2801–2826.
[6] M. Eriten, M. Kurt, G. Luo, M.D. McFarland, L.A. Bergman, A.F. Vakakis, Nonlinear system identification of frictional effects in a beam with a bolted joint
connection, Mechanical Systems and Signal Processing 39 (1-2) (2013) 245–264.
[7] L.R.O. de Lima, S.A.L. de Andrade, S. da, P.C.G. Vellasco, L.S. da Silva, Experimental and mechanical model for predicting the behaviour of minor axis
beam-to-column semi-rigid joints, International Journal of Mechanical Sciences 44 (6) (2002) 1047–1065.
[8] F.D. Markazi, R.G. Beale, M.H.R. Godley, Experimental analysis of semi‐rigid boltless connectors, Thin‐ walled structures 28 (1) (1997) 57–87.
[9] C. Bernuzzi, C.A. Castiglioni, Experimental analysis on the cyclic behaviour of beam-to-column joints in steel storage pallet racks, Thin-Walled
Structures 39 (10) (2001) 841–859.
[10] K.M. Bajoria, R.S. Talikoti, Determination of flexibility of beam to column connectors used in thin walled cold formed steel pallet racking systems, Thin
walled structures 44 (2006) 372–380.
[11] L. Sleczka, A. Kozłowski, Experimental and theoretical investigations of pallet racks connections, Advanced Steel Construction 3 (2) (2007) 607–627.
[12] A. Kozłowski, L. Sleczka. Preliminary component method model of storage rack joint. In: Proc., Connections in Steel Structures V 2004. Amsterdam.
[13] A. Filiatrault, P.S. Higgins, A. Wanitkorkul, J. Courtwright, Experimental Stiffness of Pallet-Type Steel Storage Rack Teardrop Connectors, Practice
Periodical on Structural Design and Construction 12 (4) (2007) 210–215.
[14] X. Zhao, T. Wang, Y. Chen, K.S. Sivakumaran, Flexural behavior of steel storage rack beam-to-upright connections, Journal of Constructional Steel
Research 99 (2014) 161–175.
[15] F.D. Markazi, R.G. Beale, Godley MHR. Numerical modelling of semi-rigid boltless connectors, Computers and Structures 79 (2001) 2391–2402.
[16] P. Prabha, V. Marimuthu, M. Saravanan, S.A. Jayachandran, Evaluation of connection flexibility in cold formed steel racks, Journal of Constructional
Steel Research 66 (7) (2010) 863–872.
[17] ABAQUS Software version 6.11 2012. Dassault Systmes.
[18] RambergW, OsgoodWR. Description of Stress-Strain-Curves by Three Parameters. Technical Report, Technical Note No. 902, NACA, 1943.