Advanced Steel Construction Vol. 7, No. 1, pp.

27-47 (2011) 27

DESIGN BY TESTING OF INDUSTRIAL RACKS

N. Baldassino and R. Zandonini *

Department of Mechanical and Structural Engineering, University of Trento,
38123 Povo, Trento, Italy
*(Corresponding author: E-mail: [email protected])

ABSTRACT: Industrial racks are one of the most common structures for the storage of palletised goods. The
behaviour of these structures, which are built-up from thin-walled cold-formed steel profiles, is quite complex. The
sensitivity of the uprights to buckling, the presence of the perforations on the uprights, the non linearity of the
connections, the frame sensitivity to the second-order effects and the influence of the imperfections are the main
sources of complexity. The large variability in terms of geometry of the profiles, of the joints and of the perforations,
and the complexity of the phenomena which affects the member behaviour do not yet allow performing a pure
numerical design, but call for tests aimed at the characterisation of the structural components. Traditionally, the
design of the racks is carried out by a procedure combining experiments and numerical analysis. This approach
follows the so-called 'design by testing'. This paper intends to provide an overview of the experimental part of the
approach. The specifications in the European and the North American standards are reviewed for the main structural
components, i.e., the uprights, the upright frames and the joints. Problems related to the clarity, accuracy and
completeness of the specifications are pointed out. The main results of some studies carried out by the authors
provide a key to better understanding of the importance of testing.

Keywords: Industrial racks, design by testing, experimental analysis, numerical analysis, member buckling,
semi-rigid joints

1. INTRODUCTION

In modern steel construction thin walled cold-formed profiles are becoming increasingly popular.
The particular manufacturing processes [1] adopted for these profiles, which consist of rolling,
pressing or bending brake of thin coils, lead to a great flexibility of the section shapes and to a
significant weight reduction [2]. In recent years, the improvements of manufacturing techniques,
the adoption of corrosion protection, the application of techniques of shape optimization and the
increased structural performance have led to a wider use of cold-formed profiles. Roof and wall
systems, trusses, wall framing, steel decking for composite structures, and industrial rack structures
are some typical applications [2]. The design of each category of light gauge structures requires that
some peculiar problems are mastered.

This paper focuses on the design of rack structures and, in particular, of pallet rack systems. These
structures are one of the most common industrial storage systems for palletised goods. These
systems are laid down along the aisles used to move the pallets. They are built up as a series of
braced upright frames in the cross-aisle direction, connected in the transverse direction (the
down-aisle direction) by beam elements (Figure 1). The efficiency of storage depends on the
capability of the rack system to adapt to the type and geometry of the pallets. This goal is achieved
mainly via a flexible geometry in terms of distance between upright frames and free height
available between beam levels.
28 Design by Testing of Industrial Racks

Upright
(column)

Down-aisle
direction
Base plate
Cross-aisle direction Stringer (beam)

Figure 1. A Typical Steel Storage Pallet Rack System

The uprights are open thin-walled members. Their section shape is a 'C like' section stiffened by
lips and, in some cases, by rear flanges and additional lips. The uprights usually contain holes
and/or perforations at regular spacing to allow for the beam and the bracing connections. Beams,
which usually have boxed cross sections, are provided by end-plates to allow for a mechanical
connection to the uprights. The connection to the pavement is built-up by means of base plates,
which can be either bolted or welded to the upright, and anchored to the floor. The bracings in the
cross-aisle direction are usually made of open or closed cross section profiles eccentrically bolted
to the uprights. In the down-aisle direction, racks are often unbraced frames: the need of efficiently
storing and handling the pallets prevents the location of bracing systems in this plane. Therefore,
the frame lateral stability in the down-aisle direction is generally provided by the degree of
continuity offered by the beam-to-column and by the base-plate joints.

From the structural point of view racks can be considered as framed steel structures, characterised
by a great sensitivity to second-order effects. For the purpose of global analysis, second-order
analysis of the whole system (3-D analysis) or of the sub-frames in the planes parallel and
perpendicular to the aisles (2-D analysis) may be performed. In the latter case, a semi-continuous
frame model has to be considered in the down-aisle direction. The design of these structures is quite
complex. The system members are prone to different forms of buckling, as all the thin walled
members, while the perforation of the uprights adds further complexity to the instability
phenomena. Moreover, both beam-to-column and base-plate joints are partial strength joints and
exhibit a highly nonlinear response. Therefore, the analysis is fairly sophisticated and difficult to be
mastered by numerical methods only. This is reflected in the design philosophy adopted by the
main Standards for the design of pallet racks [3,4,5]. All Standards adopt a 'design by testing'
approach which combines the results of experiments on components and sub-frames with the
theoretical criteria developed and codified for traditional cold-formed members. Suitable tests are
specified to define the main behavioural parameters of the key components or the sub-assemblages.

This paper intends to provide a concise overview of the experimental approach adopted by the
European and the North American Standards for industrial racks [3,5]. The overview focuses on the
key rack structural elements, i.e., the uprights, the upright frames and the beam-to-column and base
joints. Problems related to the clarity, accuracy and completeness of the specifications are pointed
out. Grounds for improving the Standards are identified. Furthermore, the main results of some
studies carried out by the authors provide a better understanding of the importance of testing.
 N. Baldassino and R. Zandonini 29

2. UPRIGHTS

The columns of the pallet racks are mostly cold-formed open thin-walled steel sections. Their shape
depends on the type of connection with the bracing members (cross-aisle direction) and with the
beams (down-aisle direction). Complexity is also due to the constant process of innovation
associated with competitive design. Moreover, in order to facilitate flexibility in the location of the
connections, the upright is perforated along its length. Shape and distribution of perforations add
complexity to the problem of defining the response of the upright itself. The behaviour of the
uprights is in fact significantly affected by different forms of buckling. In addition to local and
global buckling phenomena, distortional buckling should be considered. Interaction between the
different buckling modes may also occur.

Since the 60's, several studies were carried out aimed at investigating the behaviour of upright
sections. The research work focused on different issues, such as the possible buckling modes [6,7,8]
and their interaction [9,10], the influence of the residual stresses [11,12], and the influence of the
perforation [13]. These studies combined experimental and numerical analyses.

The complexity of the phenomena characterising the upright response and the wide range of cross
sectional shapes make it difficult to identify general solutions to the problem. The main Standards
for the design of thin walled structures provide simplified equations which allow tackling the most
common design cases. The effective width approach is adopted to account for local buckling; more
complex, but conservative, equations are also proposed for checking distortional buckling, while
separate formulas are given for verifying the overall instability.

For the more complex upright sections, the European and the North American Standards prescribe
rational analysis or an experimental approach. The finite element method [14], the finite strip
method [15], the generalised beam theory [16,17,18] or the direct strength method [10] can be
adopted as effective and rational tools of analysis.

The influence of the perforations is taken into account by the Standards only by modifying the
effective area. Weakening of the resistant section, localised alteration of the material properties and
stress concentrations due to the discontinuity are the principal consequences of perforations. The
variety of the geometry and shape of both the perforations, the upright cross sections and the
interval between perforations make each case unique and do not enable identification of general
equations. Various numerical and experimental studies about the effects of perforations on the
failure mode and collapse load were recently carried out [19-22]. Simplified relationships were also
established, which do have the strong limitation that they are related to particular perforation
geometries and sections. Therefore, the most appropriate solution to analyse the behaviour of the
perforated sections seems to be the experimental approach, already specified by the European and
the North American Standards [3,5] for perforated members. In particular, the European Standard
[3] prescribes how to investigate the upright behaviour under compression and under bending, with
the aim to assess the effective properties of the section. In the following, the case of an upright in
compression is used for illustrative purposes.

The European Standard [3] specifies different compression tests aimed at investigating the
sensitivity of the upright to local and distortional buckling. A minimum number of three tests is
specified, in order to enable statistical evaluation of the parameters under consideration. The
stub-column test, i.e., a compression test on a short specimen, is adopted for singling out the effect
of local buckling. The length of the specimen should to be selected so that, on the one hand, local
buckling is the sole buckling mode present and, on the other hand, there exists a uniformly
compressed central zone. A further requirement of including at least 5 pitches of perforations is
30 Design by Testing of Industrial Racks

given. The ends of the specimen are welded to steel plates which allow the connection to the test rig
(Figure 2). The end fixtures should provide hinged restraints.

A key point in compression tests is the selection of the point of application of the load. Since the
section properties change along the upright due to the presence of the perforations, it is not possible
to identify a centroidal axis of the member. In other words, concentrically loaded specimens do not
exist. The problem is further complicated when local buckling takes place, inducing additional
changes of the effective cross section.

Figure 2. Stub-column Test Set-up

Baldassino and Hancock [23] studied the influence of the load eccentricities on the upright
response under compression. They analysed a particular section profile and compared the
experimental results of specimens with and without perforations. The non negligible influence of
the load eccentricity on both the collapse load and failure mode was unveiled. Such an influence is
strongly dependent on the length of the specimen. No general rule can be given for the load
eccentricity. Therefore, the European Standard [3] recommends applying the load in the location
between the centroids of the gross and net areas that provides the 'maximum failure load'. In order
to satisfy this condition, a series of preliminary tests is required.

The final purpose of the stub-column tests is the assessment of the upright effective area (Aeff),
which is determined as:

Rk
A eff  (1)
fy

where Rk is the characteristic value of the collapse load and fy is the yield stress of the steel.

The effective area Aeff takes into account, in a simplified way, the effect of local buckling, the
perforations and the cold-work process.

The North American Standard [5] prescribes a similar approach. The main difference is the end
restraint conditions, which are fixed instead of hinged. The tests allow the definition of a reduction
factor Q, obtained as the ratio between the collapse load and the maximum resistance of the
minimum net section (fy*Anet,min). In design, this Q-factor enables the determination of the effective
properties of the cross-section.
 N. Baldassino and R. Zandonini 31

The particular shape of the upright, with additional lips or rear flanges and additional lips, certainly
improves the performance of the section against overall buckling, but increases its sensitivity to
distortional buckling. Occurrence of distortional buckling generally triggers rapid failure: the
associated in-plane deformation can induce remarkable membrane stresses which can lead to the
yielding with subsequent failure of the stiffeners.

In order to investigate this feature, both the European and the North American Standards [3,5]
specify the use of numerical or experimental approaches. Furthermore, the European Standard [3]
provides a procedure that combines theoretical and experimental analysis and appears the most
suitable for perforated members.

The European Standard [3] specifies two experimental approaches involving a quite different
amounts of work. In both cases the tests are aimed at investigating the sensitivity of the upright to
distortional buckling in the down-aisle direction. The test should be performed considering, as
much as possible, the actual restraint conditions provided by the bracing system in the cross-aisle
direction and by the beams in the down-aisle direction. In racks, the peculiar type of
beam-to-column joints and the regularity of the perforation on the upright allow modifying the
beam levels during the structural life of the storehouse. The uncertainty in the beam positions
suggests disregarding, in the tests, the restraint offered by the beams. Both proposed experimental
approaches adopt this simplification.

The 'simpler' approach consists of a compression test of an upright specimen of length equal to that
of the single bracing panel closest to one meter. The provision 'closest to one meter' is of unclear
origin and seems not reasonable in the case of bracing panel lengths different from one meter.
Moreover, it seems to contradict a prescription of the same Standard which states that all bracing
lengths should be experimentally or numerically investigated. The specimen set-up and the test
procedure are the same as those prescribed for the stub-column test. Preliminary trial tests must be
carried out to select the point of application of the axial force. When distortional buckling governs,
its effect is accounted for by a suitable re-definition of the effective cross sectional area.

The second approach for the distortional check consists of the determination of the upright buckling
curve. The compression tests are performed so that a representative range of upright frame lengths
is covered. Two alternative test set-ups can be adopted, as shown in Figure 3. The failure load data
associated to the tested bracing lengths are evaluated following the prescription of the European
Standard [3]. The final goal of the test procedure is to plot a curve providing the variation of a
stress reduction factor (  ni ) with the non-dimensional slenderness (  ni ).

Concerning global buckling, no specific test procedure is prescribed by the Standards. The buckling
theory of thin walled members is adopted to define the curves for flexural or flexural-torsional
buckling. It is implicit that perforations do have a limited influence on the overall upright response.

This brief summary clearly points out the complexity of the procedure adopted by the Standards.
With the exception of the simpler upright geometries (simple section shapes and/or no perforations),
a significant effort is required on both the experimental and the numerical sides. Tests certainly
allow investigating the more complex problems but a strong limitation is the difficulty in
reproducing the actual restraint conditions. The number of tests is further increased by the need for
a statistical assessment. If the same upright section is used in different panel lengths or with
different wall thickness, the experimental effort increases proportionally. Although the numerical
approach makes it possible to reduce the number of tests, it must be viewed as a complementary
tool: due to the presence of the perforations, quite complex models are required which, in any case,
must be calibrated against experimental results.
32 Design by Testing of Industrial Racks

F support for the F

weight of the
frame

F F

Figure 3. Test Set-ups for the Evaluation of the Buckling Curve

3. UPRIGHT FRAMES

The stability of rack systems in the cross-aisle direction is provided by the bracing system which
connects the uprights. In this plane, the rack is hence comparable to a latticed column. The
particular features of these systems, which are characterised by uprights with open cross section, by
bracing members eccentrically connected to the uprights and, in the case of bolted connection, by a
non-negligible lack of fits, make them particularly sensitive to second-order effects. This aspect
stresses the need of an accurate evaluation of their deformability and of the selection of the most
appropriate method of analysis.

In lattice structures the evaluation of the elastic critical load Vcr requires taking into account both the
flexural deformability and the shear deformability. For these systems, Vcr is given by:

1
Vcr  (2)
1 1
*

Vcr S D

where V*cr is the elastic critical load disregarding the shear deformability of the bracing system and
SD is the shear stiffness per unit length of the bracing system, both of which can be determined
theoretically. Both the European and the North American Standards provide specific equations for
the most common types of bracing systems [3,5]. These equations are based on the assumption of
elastic behaviour of 'ideal' upright frame systems [6]. Factors such as the axial shortening of the
uprights, the eccentricities of the bracing, the connections lack of fit and the structural imperfections
are neglected. To analyse the influence of such factors, the European Standard [3] recommends the
experimental approach as an alternative to the theoretically evaluation. With this aim a testing
procedure is specified. A shear force (F) is applied to an upright frame segment composed by at least
two complete bracing panels (Figure 4). The purpose of the test is to determine the 'initial' frame
elastic shear stiffness. A value of F equal to 2*n kN is indicated, where n is the number of bracing
panels of the specimen. The displacement of the loaded upright is measured and the
load-displacement curve (F- curve) enables the evaluation of the shear stiffness SD as:
 N. Baldassino and R. Zandonini 33

k ti d 2
SD  (3)
h

where kti is the slope of the best-fit straight line of the experimental curve (F- curve), d is the
distance between the centroidal axes of the uprights (Figure 5) and h is the distance between the
restraints on the uprights (Figure 5).

Figure 4. Tests Set-up for Shear Test on Upright Frames Designed

at the University of Trento

The shear stiffness SD may be employed to determine a reduced bracing area or to characterise an
equivalent spring for the bracing connections.

The authors carried out an experimental campaign which comprised approximately 100 tests on 23
different typologies of upright frames. The specimens, consisting of 3 complete panels, involved
different upright dimensions, bracings systems and panel geometries. The ratio between the width
and the length of the panel (aspect ratio) ranged between 0,33 and 0,85. In all specimens, the
connections between the uprights and the bracing system were bolted. The load was increased up to
collapse, in order to obtain information on the overall behaviour of the frame in the full range of its
response. This information might be important for modelling the rack system.

Particular care was taken with the design of the test rig. One upright (upright B in Figure 4), is hinged
at its ends, while the other (upright A in Figure 4), rests on frictionless supports, so as to guarantee the
planarity of the frame. The tests were carried out for the two configurations shown in figure 5. Four
tests were carried out for each type of upright frame: three with configuration 1 and one with
configuration 2.

Test configuration 1 Test configuration 2

F F

d
F F
h

Figure 5. Test's Configuration

The experiments also allowed for a check of the applicability of the European testing procedure.
Moreover, they enabled understanding the effectiveness of the bracing system in both configurations
and the influence of the connections between the upright and the bracing members.
34 Design by Testing of Industrial Racks

The results showed that:

- the test configuration affects in a negligible way the load-displacement curve in the load range
addressed by the European Standard (0-6 kN) [3]. As shown in Figure 6, more pronounced
differences were observed as the load increased due to the different load resistance mechanisms
associated with each configuration. Furthermore, a more favourable mechanism results in a
substantial increase of the collapse load (Figure 6 a);

- the connection system between the upright and the bracing members strongly affects the frame
performance. Figure 7 compares the test results related to two upright frames differing only in
the connection system between the uprights and the bracing members. The noticeable
differences in the load-displacement responses can be explained by observing that the
connection systems exhibit different local effects. Focusing the attention on a nodal point of the
lattice structure, it can be noted that the bracing members are eccentrically connected to the
uprights, which results in local torsion. As the distance between the bracing members increases,
the local torsion applied to the upright is more pronounced. The global response of the uprights,
which have a low torsional stiffness, is greatly influenced by these local effects occurring even
for very small loads. Torsion associated with eccentricities of the upright-bracing connections
reduces substantially both the stiffness and the strength and, therefore, should be absolutely
minimised;

- the torsional deformation of the uprights was observed near collapse in all the specimens, also in
the case of bracing systems with small eccentricities (e value in Figure 7). Such a phenomenon is
enhanced by the buckling of the compressed diagonal;

- the collapse was caused by the instability of the bracing members (Figure 8a) and involved
significant deformations of the connections (Figure 8b). In a few cases the connections failed
(Figure 8c).

a) b)
40 Load (kN) 50 Load (kN)
40
30
30
20
20
10
10
Displacement (mm) Displacement (mm)
0 0
0 10 20 30 40 50 0 10 20 30 40

Figure 6. Typical Experimental Load-displacement Curves

 N. Baldassino and R. Zandonini 35

20 Load (kN)

15

10 F Mt e
F
5
Displacement (mm)
0
0 5 10 15 20 25 30

Figure 7. Influence of the Connection System between the Bracings and the Uprights

a b c

Figure 8. Typical Collapse Modes

The evaluation of the test results in accordance with the European Standard [3] made it possible to
point out a few critical aspects of the recommended procedure.

The load-displacement curves (F- curves) are non-linear from the very beginning. This leads to
difficulties in applying the procedure of the Standard [3], which requires the determination of a best
fit straight line.

The theoretical values of the shear stiffness SD are substantially higher than the experimental ones
(up to 60 times). The 'uncertainty' on the evaluation of the experimental shear stiffness and the
assumptions underlying the theoretical SD are the two main causes explaining such an enormous
difference.

Similar conclusions were reached in the study carried out by Godley and Beale [24]. The authors
analysed the response of upright frames characterised by different upright sizes, numbers of panels
and bracing patterns. The study comprised 80 tests on upright frames and numerous FE numerical
simulations. The work pointed out that parameters such as the connection eccentricity, the bolt
bending or the axial shortening of the uprights must be included in the theoretical analysis to obtain
satisfactory agreement between theory and experimental results. If such contributions are considered
the scatter between the theoretical and experimental values of SD is reduced. Further improvements
could be achieved if effects such as the local distortion of the uprights at the connection with the
bracings and the bolt looseness are taken into account.

The above considerations clearly indicate the need for testing. The theoretical equations for SD
should not be used directly. They rather enable the determination of an equivalent value of the
bracing area from experimental results. A numerical approach to the problem, whenever possible, is
quite complex due to the substantial influence of the local effects on the frame response. The
36 Design by Testing of Industrial Racks

reliability of the analysis requires the ability to reproduce the actual frame behaviour and, therefore,
leads to complex models, which must, in any case, be validated against the experimental results.

4. THE JOINTS

In racks, as in traditional framed structures, joints play a fundamental role in the frame performance.
This is clear in the down-aisle direction, since the degree of continuity of the beam-to-column and
the base-plate connections is fundamental in providing stability to the system. Therefore, an
adequate approximation of the joint behaviour is of vital importance for the analysis and design of
rack structures.

Studies carried out by different authors [25,26,27] enabled an overall understanding of the quite
complex response exhibited by the joint types usually employed in pallets racks. The presence of
lack of fit and the local deformation of both the connected members and the connector lead to
moment-rotation relationships that are highly nonlinear even for very small loads. Beside this
semi-rigid nature, the joints are 'partial strength joints'. Moreover, the variety of joint components
and connection systems is fairly large and in permanent 'evolution', due to the need of innovation
typical of a competitive market. As a consequence, the well-known models to estimate the joint
performance in traditional steel framed systems [28] cannot be straightforwardly extended to rack
joints. Furthermore, the complexity of the various phenomena involved in the joint response (e.g.,
large plastic deformation, highly localized contact forces, inelastic local buckling...) calls for
experimental analysis.

Tests to determine the behaviour of beam-to-column joints appear in all codes for pallet racks
[3,4,5]. The European Standard [3] also specifies a test to characterise the response of base-plate
connections.

The following two sections focus on joint performance and on its influence on the frame response.
From the perspective of design by testing, the code specifications to perform tests and evaluate
results are first addressed. The discussion is focused on the main issues of the experimental analysis
and the possible 'limits' of the code recommendations. Frame studies carried out by the authors are
presented in order to illustrate the sensitivity of the frame response to the joint behaviour and/or
modelling.

4.1 Beam-to-column Joints

In pallets racks, the connection between the upright and the beam is an easy-to-handle mechanical
connection. As already mentioned, the connected profiles and the connection itself may exhibit a
wide variety of shapes, dimensions and wall thickness of members, and of shape and dimensions of
connecting elements. Nevertheless, some common features can be identified, which are
summarised next.

Beams are usually built by welding two thin-walled channels to form a closed section. End
connectors are welded at the beam ends, positioned either symmetrically or asymmetrically with
respect to the beam axis. The connectors include tangs that accommodate into the holes located on
the front wall of the upright, thus creating the connection (Figure 9). This simple connection system
allows the rack levels to be adjusted to the storage needs. The geometries of the tangs,
beam-end-connector and holes on the upright are designed to facilitate the engaging and to reduce
as much as possible the lack of fit and the joint looseness. Moreover, connections are usually
provided with locking devices to prevent accidental unlocking or knocking-out of the beam.
 N. Baldassino and R. Zandonini 37

Tang

End connector

Beam
Upright

Figure 9. Typical Beam-to-column Connection

When subject to a bending moment, the connectors rotate outwards and tend to tighten up. The
moment transferred from the beam to the upright is associated with distortion of the end connectors
and local deformation of the upright wall. According to the traditional criterion, these joints would
be classified as semi-rigid [28] as pointed out by various authors [25,26,27]. Baldassino and
Bernuzzi [26] analysed the results of more than 200 tests on 52 types of beam-to-column
connections under bending. Their work showed that the considered connections were characterized
by a limited degree of flexural continuity. However, the results of an extensive numerical investigation
clearly indicated that such flexible joints have a significant influence on the frame performance.
Therefore, it was concluded that when the joint response falls in the flexible domain, the
semi-continuous frame model has to be adopted in order to accurately assess the frame response.

This central role in providing the rack lateral stability calls for specific analyses aimed at
investigating the joint performance under bending. With this aim, the European and the North
American Standards [3,5] provide recommendations for designing, performing and evaluating 'ad
hoc' tests to assess the stiffness and moment capacity of the joint. Although the purpose of the tests
is the same, the test set-up, procedure and method of evaluating the results are quite different.

The European Standard [3] specifies two types of tests: the bending test and the looseness test.
While the former is aimed at assessing the stiffness and the moment resistance, the latter enables
the determination of the initial looseness of the joint, an important parameter also for determining
the rack frame geometrical imperfection. Both tests adopt the same specimen geometry: a stub of
the upright section, connected to a counter frame, and a short length of beam, connected to the
central part of the upright (Figure 10). The load is applied at a distance of 400 mm from the face of
the upright. During the tests, the rotation of the end connector is measured (Figure 10). The bending
and looseness tests differ in the loading history. In the former, a first cycle of loading and unloading
allows for the lack of fit to take place before the load is increased up to the collapse of the specimen.
The latter involves a single load cycle up to a moment approximately equal to 10% of the joint
ultimate moment.

Figure 10. Typical Beam-to-column Connection

38 Design by Testing of Industrial Racks

Typical experimental responses from a bending and a looseness test are shown in figures 11a and
11b, respectively. The European Standard [3] provides detailed methods for evaluating the test data
and for defining a design moment-rotation relationship. The amplitude of the initial looseness is
obtained by means of a quite simple procedure. The intersection points between the rotation-axis
and the lines characterising the 'linear range' of the moment-rotation relation (i.e., points A and B in
figure 12) enable the determination of a segment (segment AB in figure 12), whose length is
assumed to be twice the value of the initial looseness (l).

The moment-rotation law is obtained by means of a time consuming procedure, which also
accounts for the deviations of the actual thicknesses and steel yield strength of the joint components,
with respect to their nominal values.

Two design models are possible: while one approximates the experimental results by a multi-linear
relationship, the other one adopts an elastic-perfectly plastic law. In both cases, it is necessary to
determine the design moment resistance (MRd), computed in accordance with the limit states
approach [3]. The experimental moment-rotation curves, as well as the related models, are assumed
to be bound by the MRd value.

The multi-linear law is selected by approximating the experimental moment-rotation relationship,

while strictly remaining below it (see the dashed line in Figure 13). No practical rules are given.

In the case of a bi-linear moment rotation relationship, the stiffness of the elastic range (kn,i) is
selected as the slope of the secant line passing through the origin and defining approximately equal
areas between it and the experimental curve (i.e., areas A1 and A2 in figure 14). To reduce the
deviation between the experimental and numerical results, the stiffness should be selected by
considering as additional criterion the condition kn,i  1,15 ·MRd/i (Figure 14). The stiffness to be
adopted in design (km) is the average value of at least three tests.

a) b)
1,2 Moment 0,6 Moment
(kNm) (kNm)
1,0 0,4
0,8 0,2
0,6 Rotation (mrad)
0,0
0,4 -8 -6 -4 -2 0 2 4 6
-0,2
0,2
Rotation (mrad) -0,4
0,0
0 10 20 30 40 50 60 70 80 -0,6

Figure 11. Typical Test Results of a Beam-to-column Connection:

a) Bending Tests and b) Looseness Test
 N. Baldassino and R. Zandonini 39

A
B 
l

Figure 12. Evaluation of the Initial Looseness

M M

MRd MRd
Average curve
Test results
Multi-linear
relationship
Average curve

 

Figure 13. Evaluation of a Multi-linear Moment-rotation Relatonship for

Beam-to-column Connection

M M
i/1,15

MRd MRd

A2 A1=A2 5%
tg  = km
A1 tg i = kn,i
i 
i  
Figure 14. Evaluation of the Initial Stiffeness

The initial looseness and linearised moment-rotation relationship define the beam-to-column joint
behaviour in bending. In the frame analysis, the initial joint looseness (l) is incorporated as an
additional sway frame imperfection. A more accurate approach is usually precluded by numerical
difficulties. The moment-rotation response is then simulated by means of rotational springs. The
European Standard [3] assumes a M- relationship symmetric for both positive and negative
bending moments. It should be mentioned that this assumption is not consistent with the
experimental results (Figure 15) [26], nevertheless it is be acceptable for design purposes.
Numerical studies carried out on typical rack configurations confirmed the negligible influence of
the joint symmetry assumption on the frame performance [26].
40 Design by Testing of Industrial Racks

1,6
Moment
1,2 (kNm)

0,8
0,4
Rotation (mrad)
0,0
-60 -40 -20-0,4 0 20 40 60 80
-0,8
-1,2
-1,6
Figure 15. Beam-to-column Joint Response under Positive and Negative Bending Moment

The North American Standard [5] specifies two different tests on beam-to-column joints: the
cantilever test and the portal test. The cantilever test, which is similar to the European bending test,
is aimed at assessing both the joint stiffness and moment capacity. However, it is mentioned that, in
this test, the joint deformability (or stiffness) might be non negligibly affected by the shear force.
Therefore, the portal test must be considered in order to provide a more accurate evaluation of the
joint stiffness (Figure 16a). The test is carried out on a substructure composed by four uprights,
hinged at the base, connected by two beams. The specimen is first subject to the vertical service
load, followed by the corresponding design horizontal load, applied at the beam level. The frame
sways and the beam-to-column connections rotate in opposite directions, so that the rotation caused
by the vertical load increases in one connection and decreases in the other one. The analysis of the
portal test results leads to an appraisal of the average value of the joint stiffness, accounting for the
joint behaviour asymmetry when the frame sways. However, Harris and Hancock [29] observed
that the portal test enables the evaluation of the joint behaviour only for relatively small joint
rotations, which does not fully reflect the actual joint response in a sway frame. In particular, they
point out that the portal frame results greatly depend on the level of joint bending moment under
the vertical loads, due to the non-linear joint behaviour (Figure 16b).

a) b)
M
Q Q
H

Initial rotation

Figure 16. Test Set-up for Portal Test

The remarkable influence of the joint response on the frame performance justifies the great
attention paid by the Standards to its characterisation. The authors carried out a numerical study
aimed at investigating such an influence [30]. Two rack configurations, involving different numbers
of bays and levels, were selected within a typical practical range. The attention was focussed on the
rack behaviour in the down-aisle direction. For both frame configurations, the same profiles
(uprights and beams) and connections (base-plate and beam-to-column) were considered. All the
joints were experimentally characterised in accordance with the European Standard [3]. The
 N. Baldassino and R. Zandonini 41

analyses were performed with the PEP program [31], which enables the incorporation of material
and geometrical non-linearity, semi-continuity and the influence of the axial load on the response of
the base joints. A simplified bi-linear relationship was assumed for the end connectors. The
influence of the beam-to-column joints on the frame performance was studied by selecting 3 joint
stiffness values: 2k, k, 0,5k, where k is the experimental joint stiffness. In addition, initial frame
out-plumb angles of 1/1000, 1/200 and 1/100 were considered to simulate a wide range of joint
looseness amplitudes. Three models were adopted for the base joint: hinge, fixed base and
semi-rigid. For the last one a simplified bi-linear curve was assumed. A total of 54 cases were
analysed.

The results of this study showed that:

- an increase of the beam-to-column joint stiffness may not lead to an increase of the load capacity.
This is mainly due to the change of the collapse mode, which may be associated with the
reaching of the maximum rotation capacity of the joint;
- influence of the initial out-of-plumb: the increase of the beam-to-column looseness leads to a
decrease of the load capacity, for realistic out-of-plumb angles. However, the results do not show
any clear relation between the out-of-plumb and the ultimate load;
- the single parameter affecting the frame performance the most appears to be the base restraint.

The above results point out the importance of the joint rotation capacity. No attention is given in the
Standard [3] to this parameter and such a gap needs to be filled. Indeed, the accuracy required in
the looseness tests seems to be insufficient, and a more accurate assessment of the end connector
stiffness is required. However, the results also seem to confirm that a more refined moment-rotation
law (i.e., the multi-linear model) does not lead to a real increase of the accuracy of the frame
analysis.

As pointed out in this section, the experimental procedures appear fairly demanding on both the test
performance and the evaluation of the test results. Moreover, even a small variation of the joint
configuration calls for a new series of tests. The numerical analysis based on the FE method [14]
seems to be a viable alternative, which would also allow for a deeper understanding of the
importance of the various factors involved. In recent years, some attempts were made along this
direction. The results obtained showed a general good agreement between the experimental and
numerical analyses, but also made it clear that the reliability of the numerical models strongly
depends on their calibration against experimental results. In addition, a labour-intensive and
time-consuming data input (including the FE mesh) is always required in order to adequately
capture the local phenomena, that affect the joint performance and trigger the frame collapse [32].
A convenient alternative is the so-called 'component method' [28], which builds up the joint
response on the basis of the behaviour of its components. Recently, Kozłowski and Ślęczka [33]
presented a preliminary research study of the application of the component method to rack joints.
This work involves five joint typologies, identifies the key joint components and proposes
equations to evaluate their stiffness and flexural resistance. Although the results are encouraging, an
in-depth study aimed at further validating the model for a wider range of joint typologies remains a
necessary pre-requisite for any possible use in design practice.

4.1.1 Base-plate connection

Base joints of rack frames are peculiar due to the configuration and components adopted. Therefore,
their response is different from that of base joints in traditional steel frames. If reference is made to
the Italian practice, base joints are typically built up from 'ad hoc' designed base-plate elements
bolted to the uprights and connected to the floor by means of fixing systems (chemical plugs,
42 Design by Testing of Industrial Racks

bolts…). Lack of fit between the base-plate element and the upright, local plastic deformations of
the upright at the connection to the base-plate element, local buckling of the upright, the
interactions between the aforementioned phenomena and the base plate deformation are the main
factors influencing the stiffness, resistance and ductility of the joint.

In order to account for the base joints response in the analysis, different approaches are prescribed
by the European and the North American Standards [3,5]. The European Standard [3] provides two
possible approaches: either to neglect any rotational restraint (i.e., model the joint as a hinge) or to
perform 'ad hoc' tests. In the latter case, to account for the influence of the axial load on the joint
response, it is recommended that tests are performed for at least two different axial load levels,
selected in the range up to the maximum design load for the upright.

The North American Standard [5] adopts a simplified approach: the base joint is modelled by a
linear moment-rotation (M-) relationship. The joint stiffness is determined by considering only the
elastic deformation of the concrete floor and does not depend on the axial load. The rotational
stiffness of the joint depends only on the dimensions of the upright. Such a difference between the
North American and the European Standards [5,3] can be explained by the different base joint
typologies employed.
In recent years, the authors carried out an experimental study of base-plate joints of Italian rack
frames [34]. The campaign comprised 200 tests on 25 different types of commercial connections,
performed in accordance with the European Standard [3]. The specimens consisted of two
stub-columns symmetrically connected to a concrete cube by the fixings adopted in the actual
structure (Figure 17).
Axial load on uprights (F1)

Hydraulic jack Concrete cube

Uprights

Loading cells

Transversal load (F2)

Counter frame

Figure 17. Test set-up for Tests on Base-plate Connections

The concrete cube simulating the floor surface is free to move horizontally in the direction of the
applied lateral force and restrained from rotating about the vertical axis. Two hydraulic jacks apply
the loads to the specimen: while the first one imposes the specified axial load F1 on the upright, the
second one applies a transverse load F2 to the centre of the cube. During the test, the rotation of the
column bases, with respect to the concrete cube, and the horizontal displacement of the cube are
measured.

The test results showed a non-negligible sensitivity to some features of the testing rig. Indications
on how to improve the test apparatus were obtained and summarised next:
 N. Baldassino and R. Zandonini 43

- the joint response is highly sensitive to small variations of the axial load F1 occurring during the
test, stemming from the modified system configuration due to force F2. Therefore, it is important
to have a reliable load control system to keep the axial load F1 constant;
- a vertical displacement of the concrete cube can occur, due to the unavoidable geometrical
imperfections of the base connections and/or eccentricities of forces F1 and F2 with respect to the
centre of this element. This effect requires an efficient vertical restraint.

Furthermore, the concrete cube tends to rotate near the collapse, thus inducing a parasitic bending
contribution triggering the premature collapse of the base joint. The magnitude of this perturbation
cannot be determined.

Typical moment-rotation relationships are presented in figure 18a. The rotation is evaluated as the
average value of the relative rotations between the cube and the base joint, measured on the right
and left hand sides of the concrete cube. The moment applied to the column base is calculated by
considering both the first-order contribution of the transversal load F2 and the second-order effect
of the axial load F1, due to the lateral displacement of the concrete cube.

a) b)
10 Moment 10 Moment
(kNm) (kNm)
2 1 2 1
8 8 X X
3
1 N=0,20 Nu
6 1 N=0,20 Nu 6 X3
2 N=0,40 Nu
4 2 N=0,40 Nu 4 3 N=0,60 Nu
3 N=0,60 Nu Nu = upright squash load
2 2
Nu = upright squash load Rotation (mrad) Rotation (mrad)
0 0
0 10 20 30 0 10 20 30
Figure 18. Typical Moment Rotation Relationship for Base-plate Connections:
a) Experimental Results and b) Simplified Relationships

Due to the large variability of the typologies tested and the range of axial load applied to the
specimens, it is not possible a direct comparison between the tests results. However, a few general
comments on the experimental results can be made:
- the influence of the axial load on both the stiffness and the ultimate moment is substantial. This
confirms the need to perform tests for an adequate number of axial load values within the range
of use of the upright;
- collapses were caused by plasticity and instability of the upright near to the base-plate.
Remarkable deformations of the fixing system to the concrete never occurred;
- the interaction between the base-plate element and the upright could significantly affect the
whole joint response. In particular, the local restraint offered by this element to the upright
deformation influences both the collapse mode and the joint stiffness [34].

For design purposes, test results should be 'transformed' into simplified moment-rotation
relationships accounting for the influence of the axial load on the base joint performance. The
European Standard [3] specifies a procedure to evaluate the tests results: for each axial load level,
the experimental curves are handled following the procedure prescribed for beam-to-column joints,
and an elastic-perfectly plastic moment-rotation relationship is obtained. Typical results of such
analyses are presented in figure 18b.
44 Design by Testing of Industrial Racks

A crucial question for design purposes concerns the way in which such simplified relationships
should be considered in numerical simulations. In principle, a domain M-N- should be defined,
with associated rules enabling the determination of the elastic stiffness. In order to simplify the
joint model, and in the absence of adequate knowledge, the European code [3] specifies a linear
interpolation between the ultimate moment and the joint stiffness for different axial load levels. No
rules are given for determining the rotation capacity.

As mentioned before, the North American Standard [5] adopts a very simple theoretical model for
the base joint, which relates the elastic stiffness to the joint geometry and the concrete elastic
modulus. Recently Sarawit and Peköz [35] carried out an extensive numerical study of base joints
for different upright sections, relative positions upright-base plate, and base plate thickness. The
study made it possible to improve the equation proposed in the Standard.

From the design point of view, the adoption of a base joint model, even in the simple form specified
by the European Standard [3], leads to a more reliable evaluation of the frame performance, but
also it significantly increases the computational effort and, requires codes able to incorporate
M-N-k- relationships to model the joint. Such a complexity can only be justified if the base joint
strongly affects the frame performance.

In order to investigate this issue, the authors recently carried out a numerical study of rack systems
with very flexible beam-to-column joints [34,36]. The influence of the base restraint, as well as of
the way the base joint response is incorporated into the down-aisle frame model, were investigated
using the program PEP micro [31]. Each frame configuration combined different base connections
and beam-to-column joints. Each joint type was experimentally characterised in accordance with
the European Standard [3]. The program linearly interpolates the base joint M- curves associated
with different axial load level and the analyses incorporated values of the out-of-plumb angle 
ranging from 0,001 to 0,01 rad. The beam-to-column joints were modelled as semi-rigid and a wide
range of stiffness was considered (starting from the experimental value). Four restraint conditions
at the base were investigated: hinge, fixed and two semi-continuous joints. The last ones differ in
the location of the rotational springs: one model concentrates all the joint deformation at the base,
while the other separates the contributions of the end-plate and the upright, locating them in the
appropriate positions.

The results of the study showed that:

- the frame response is strongly influenced by the base restraint condition, regardless of the
stiffness of the beam-to-column joint. If the load multiplier is considered, the assumption of a
fixed base, leads to an increase of the load multiplier by up to 209%, with respect to the hinged
base;
- the sway imperfection has a non-negligible influence on the frame response: this parameter
becomes less important when the end connector flexibility increases;
- the assumption of an hinged base, prescribed by the European Standard [3] in the absence of
tests results, is fairly penalising: the load multiplier for the actual base restraint is, on average,
103% larger than the hinged base one;
- the influence of the variation of the beam-to-column joint stiffness increases with the base joint
stiffness. As a consequence, it is not worth improving the beam-to-column joint performance
when the base joints are very flexible;
- the frame response determined by considering the semi-rigid base joint behaviour appears not to
be sensitive to the way in which the joint flexibility is incorporated into the model.

The above results clearly indicate the need of an experimental characterisation of the base joint. A
key parameter for the complete characterisation of the base joint is its maximum rotation capacity.
 N. Baldassino and R. Zandonini 45

Like for beam-to-column joints, no specific requirements are given by the Standard [3] for base
joints.

The FE numerical analysis is a useful tool to reduce the cost of experimental studies. Researches
carried out in the recent past [36,37], show the potential of this technique in predicting both the
joint overall response and collapse mode, even if the complex behaviour of the base joints leads to
vey complex numerical models. The need to reproduce the upright boundary conditions and the
local phenomena, due to the interaction between the base plate, upright and floor, requires
numerical simulations, involving very refined meshes and elements able to model contacts, which
should be accurately calibrated against the experimental results. This means that the role of tests
remains very important, since they cannot be fully replaced by FE numerical simulations.

5. CONCLUDING REMARKS

This paper provides a general overview of the 'design by testing' of steel storage pallet racks. The
experimental approach provided by the main standards, namely the European and the North
American Standards was presented and discussed. Attention was focused on the key rack
components: the upright under compression, the upright frame under shear, the beam-to-column
joints in bending and the base joints under compression and bending.

The following issues deserve to be specially mentioned:

- uprights: the experimental procedure proposed by the European Standard appears to be fairly
complex. A critical aspect for member testing is the selection of the point of load application.
The variable geometry of the cross section along the member length does not allow a definition
of a 'centroidal axis of the profile'. Clear guidance on this issue is missing;
- upright frames: the theoretical evaluation of the shear stiffness substantially overestimates the
experimental value. Key factors, such as eccentricities of bracing members, bending deformation
of the bolts or lack of fits, are not taken into account by the theoretical method. Either
experimental values or values yielded by 'realistic' numerical analysis need to be used in design;
- beam-to-column joints: since this play a fundamental role in the stability of rack frames in the
down-aisle direction, an adequate characterisation is required. Nevertheless, the results of frame
analyses seem to indicate that a refined moment-rotation law does not lead to a real increase in
accuracy. Bi-linear joint models are satisfactory. Although the approach proposed by the
European code is fairly demanding, it pays no attention to the joint rotation capacity, even if
extensive numerical studies carried out by the authors showed the importance of this parameter;
- base-plate connections: the experimental characterisation is the most appropriate way to capture
the behavioural complexity of these joints, as specified by the European Standard. Code
recommendations should be improved to include a more complete characterisation of the joint
response, namely the influence of the axial load and the maximum rotation.

Design by testing is a demanding procedure due to both the number and complexity of tests
required. At present, numerical tools do not seem to be a fully viable alternative: tests are always
required to validate numerical models. However, a reduction of the experimental effort may be
foreseen as a consequence of the continuous and fast advances in both software and hardware
capabilities.
46 Design by Testing of Industrial Racks

Industrial rack design would greatly benefit from a deeper knowledge of several issues that need
further investigation. For instance, the influence of the perforations on the upright response, the
definition of 'realistic' numerical models for upright frames, the application of the 'component
method' to rack beam-to-column and base joints, the criteria to estimate the base joint response
accounting for bending, axial force, stiffness and rotation capacity.

REFERENCES

[1] Ghersi, A., Landolfo, R. and Mazzolani, F.M., "Design of Metallic Cold-formed
Thin-walled Members", Spon Press, London and New York, 2002, pp. 164.
[2] Hancock, G.J., "Design of Cold-formed Steel Structures", Australian Institute of Steel
Construction, 2nd Edition, 1994, pp. 240.
[3] CEN, "EN 15512, Steel Static Storage Systems - Adjustable Pallet Racking Systems -
Principles for Structural Design", CEN European Committee for Standardization, 2009, pp.
137.
[4] AS, "AS 4084 - Steel Storage Racking", AS Standards Australia, 1993, pp. 41.
[5] RMI, "MH 16.1- Specification for the Design, Testing and Utilization of Industrial Steel
Storage Racks", RMI - Rack Manufactures Institute, 2008, pp. 59.
[6] Timoshenko, S.P. and Gere J.M., "Theory of Elastic Stability", 2nd Edition, McGraw-Hill
Book Company, New York, 1961, pp. 541.
[7] Hancock, G.J., "Ligth Gauge Construction", Progress in Structural Engineering and
Materials, 1997, Vol. 1, No. 1, pp. 25-30.
[8] Trahair, N.S., "Flexural-torsional Buckling", E & FN Spon, London, UK, 1993, pp. 420.
[9] Gioncu, V., "General Theory of Coupled Instabilities", Thin-Walled Structures, 1994, Vol.
19, No. 2-4, pp. 81-128.
[10] Pekoz, T., "Development of a Unified approach to the Design of Cold-formed Steel
Members", American Iron and Steel Institute, Research Report, CF87, No. 1, 1987.
[11] Weng, C.C. and Peköz, T., "Residual Stresses in Cold-formed Steel Members", Journal of
Structural Engineering, ASCE, 1990, Vol. 116, No. 6, pp. 1611-1625.
[12] Moen, C.D., Takeru Igusa and Schafer, B.W., "Prediction of Residual Stresses and Strains
in Cold-formed Steel Members", Thin-Walled Structures, 2008, Vol. 46, No. 11, pp.
1274-1289.
[13] LaBoube, R.A., Yu, W.W., Langan, J.E. and Shan, M.Y., "Cold-formed Steel Webs with
Openings: Summary Report", Thin-Walled Structures, 1997, Vol. 27, No. 1, pp. 79-84.
[14] Zienkiewicz, O.C., "The Finite Element Method", 3rd Edition, McGraw-Hill Company,
London, UK, 1977, pp. 787.
[15] Cheung, Y.K., "Finite Strip Method in Structural Analysis", Pergamon Press, Oxford, New
York, 1976, pp. 232.
[16] Schardt, R., "The Generalised Beam Theory, Instability and Plastic Collapse of Steel
Structures", Proceedings of the M.R. Horne Conference, University of Manchester,
Granada, London, 1983, pp. 469-475.
[17] Davies, J.M. and Leach, P., "First-order Generalised Beam Theory", Journal of
Constructional Steel Research, 1994, Vol. 31, No. 2-3, pp. 187-220.
[18] Davies, J.M., Leach, P., and Heinz, D., "Second-order Generalised Beam Theory", Journal
of Constructional Steel Research, 1994, Vol. 31, No. 2-3, pp. 221-241.
[19] El-Sawy, K.M. and Nazmy, A.S., "Effect of the Aspect Ratio on the Elastic Buckling of
Uniaxially Loaded Plates with Eccentric Holes", Thin-Walled Structures, 2001, Vol. 39,
No. 12, pp. 983-998.
[20] Moen, C.D. and Schafer, B.W., "Experiments on Cold-formed Steel Columns with Holes",
Thin-Walled Structures, 2008, Vol. 46, No. 10, pp. 1164-1182.
 N. Baldassino and R. Zandonini 47

[21] Moen, C.D. and Schafer, B.W., "Elastic Buckling of Thin Plates with Holes in
Compression or Bending", Thin-Walled Structures, 2009, Vol. 47, No. 12, pp. 1597-1607.
[22] Eccher, G., Rasmussen, K. and Zandonini R., "Elastic Buckling Analysis of Perforated
Thin-walled Structures by the Isoparametric Spline Finite Strip Method", Thin-Walled
Structures, 2008, Vol. 46, No. 2, pp. 165-191.
[23] Baldassino, N. and Hancock, G.J., "Distortional Buckling of Cold-formed Steel Storage
Rack Sections including Perforations", Proceedings of the Fourth International Conference
on Steel and Aluminium Structures (ICSAS '99), Elsevier, Espoo, Finland, 20-23 June
1999, pp. 131-138.
[24] Sajja, S.R., Beale, R.G. and Godley, M.H.R., "Shear Stiffness of Pallet Rack Upright
Frames", Journal of Constructional Steel Research, 2008, Vol. 64, No. 7-8, pp. 867-874.
[25] Markazi, F.D., Beale, R.G. and Godley, M.H.R., "Experimental Analysis of Semi-rigid
Boltless Connectors", Thin-Walled Structures, 1997, Vol. 28, No. 1, pp. 57-87.
[26] Baldassino, N. and Bernuzzi, C., "Analysis and Behaviour of Steel Storage Pallet Racks",
Thin-Walled Structures, 2000, Vol. 37, No. 4, pp. 277-304.
[27] Baldassino, N. and Zandonini, R., "Performance of Base-plate Connections of Steel
Storage Pallet Racks", Proceedings of the International Colloquium dedicated to the 70th
Anniversary of Professor Victor Gioncu, May 7-8 2004, Timisoara, Romania, pp. 11-20.
[28] CEN, "Eurocode 3 – Design of Steel Structures – Part 1-8: Design of Joints", CEN
European Committee for Standardization, 2005, pp. 133.
[29] Harris, E. and Hancock, G., "Stability Testing of Sub-assemblages of High Rise Steel
Storage Racks", Proceedings of the International Conference on Advances in Structures
(ASSCCA '03), Sidney, Australia, 22-25 June 2003, Vol. 1, pp. 251-256.
[30] Giovannini, M., "On the Response of Industrial Racks" (in Italian), Master Thesis,
University of Trento, 2008, pp. 137.
[31] Galea Y. and Bureau A., "PEP Micro – Manuel d'Utilisation", Centre Tecnique Industriel
de la Construction Métallique (CTiCM), France, 1998.
[32] Bajoria, K.M. and Talikoti, R.S., "Determination of Flexibility of Beam-to-column
Connectors used in Thin Walled Cold-formed Steel Pallet Racking Systems ", Thin-Walled
Structures, 2006, Vol. 44, No. 3, pp. 372-380.
[33] Kozłowski, A. and Ślęczka, L., "Preliminary Component Method Model of Storage Rack
Joint", Proceedings of Connections in Steel Structures V, Amsterdam, June 3-4 2004, pp.
253-262.
[34] Baldassino, N. and Zandonini, R., "Performance of Base-plate Connections of Steel
Storage Pallet Racks", Proceedings of Fifth International Conference on Coupled
Instabilities in Metal Structures (CIMS2008), Gregory J. Hancock Symposium, Sydney,
Australia, 23-25 June, 2008, pp. 119-130.
[35] Sarawit, A.T. and Peköz, T., "Cold-formed Steel Frame and Beam-column Design", Report
of a Research Project AISI-RMI, Report 03-03, 2003, pp. 330.
[36] Scandola, A., "Analysis of Industrial Racks" (in Italian), Master Thesis, University of
Trento, 2007, pp. 129.
[37] del Coz Díaz, J.J., García Nieto, P.J., Betegón Biempica, C. and Fernández Rougeot, G.,
"Non-linear Analysis of Unbolted Base Plates by FEM and Experimental Validation",
Thin-Walled Structures, 2006, Vol. 44, No. 5, pp. 529-541.