Professional Documents
Culture Documents
net/publication/319672399
CITATION READS
1 8,462
2 authors:
All content following this page was uploaded by Mahmoud S.A. Shaheen on 26 October 2018.
ABSTRACT
The European racking federation design guidelines FEM10.2.07 (2012) contains provisions for the
analysis and design of drive-in and drive-through racks. It specifies different methods for carrying
out the design, and the design verifications differ depending on the method chosen. However, there
appears to be no published studies that compare the analysis and design methods specified in
FEM10.2.07 and give recommendations for using a particular method over the other. In addition,
the design guidelines are unclear about the definition of the effective length to use in the design of
uprights. Depending on the way the effective length provisions are interpreted it leads to
significantly different solutions. This paper studies the different design methods specified in
FEM10.2.07 and exemplifies the differences, advantages and disadvantages of each method, and
puts forward recommendations for design depending on drive-in rack characteristics.
1 INTRODUCTION
Worldwide, steel storage racks are used extensively in industry for storing goods on pallets. They
are heavily loaded and resist significant forces; yet are designed to be as light as possible. Among
the many types, drive-in racks are storage rack configurations that allow forklift trucks to drive
directly into storage lanes of stacked rows as shown in Figure 1.
Goods
Pallet
Forklift
Rails truck
Upright
frame
Drive-in racks can typically be 3 to 7 pallets deep. To allow forklift passage, drive-in racks in the
down-aisle direction can be only braced at the back (spine bracing) and at the top (plan bracing). As
a result, uprights have substantial unbraced lengths which make them prone to buckling in the
down-aisle direction. Whereas, in the cross-aisle direction, each two uprights are interconnected by
diagonal bracings members to form upright frames, thus producing higher shear stiffness in the
cross-aisle direction as compared to the down-aisle direction. In addition, owing to the friction
between the pallets and the supporting rail beams, the pallets are believed to act partially as an
elastic structural member. Yet, their interaction with the rack is not guaranteed. Therefore, the
European racking federation design guidelines FEM10.2.07 [1] does not permit taking into account
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
the stabilizing effects provided by the pallets. Adding further complexity, the nonlinear behaviour
of the critical connections such as: a) upright to base plate assembly, and b) upright to portal beam
has to be accounted for in the design of racks.
According to FEM10.2.07, Method 1, the most involved method, is a nonlinear 3D analysis which
takes into account second order effects both at the global frame (P- ∆) and local member (P- δ)
levels. Also, it requires the global geometric (frame) imperfection and local member (bow)
imperfections to be included in the global analysis model. Method 2, which is similar to Method 1,
requires a 3D second order analysis with global frame imperfections to be included in the model.
However, Method 2 does not take local member imperfections into account in the analysis but
rather uses design buckling curves to account for bow imperfections implicitly. Method 3 is a 3D
first order analysis method which is considered to be the simplest and least accurate among the
specified analysis methods, and hence has not been included in this study. In the literature, there
appears to be no published studies that compare the analysis and design methods specified in
FEM10.2.07 or give recommendations for using a particular method over the other. In addition, the
design guidelines are unclear about the definition of the effective length to use in the design of
uprights as far as Method 2 is concerned. This paper presents a case study through which the
different design methods are examined to clarify the advantages and disadvantages of each method,
and also to present recommendations for the most accurate definition of the upright effective
(unbraced) length to use in the design of drive-in racks.
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
section properties used in the finite element models are shown in Figure 4 and listed in Table 1
respectively.
Fig. 2. Illustration of the drive-in rack FE model Fig. 3: Plan dimensions for the drive-in rack
(a) (b)
Fig. 5. Pallet load arrangement for (a) fully loaded case, and (b) pattern loaded case.
1 1
𝛷 = √(2 + 𝑛 ) × (2𝛷𝑠 + 𝛷𝑙 ) ≥ 1/500 (1)
𝑙
where Φs = maximum specified installation out of plumb divided by the height, Φl = looseness of
the top tie (portal) beam end connector, and nl = number of interconnected lanes per drive-in block.
However, FEM10.2.07 permits Φl to be set to zero if the connection looseness is taken into account
in the global analysis model. Gilbert and Rasmussen [7] studied the influence of the portal beam to
upright bolted moment connection on the behaviour of drive-in racks. Through four different
connection models, the most accurate model was to consider the multi-linear moment rotation curve
(as shown in Fig. 19 in [7]) in the global analysis model, which therefore has been adopted in the
FE models used in this research. The maximum specified installation out of plumb angle Φs is taken
as 1/500 as per Annex C of FEM10.2.07. Consequently, the global out-of-plumb sway angle Φ is
calculated as 0.00346 radians. The global sway imperfection then has been applied using an
equivalent closed system of horizontal loads acting on the uprights at the rail beam levels.
To determine the rack buckled mode shape, an elastic buckling analysis was carried out using the
finite element program ABAQUS with the ultimate design loads shown in Figure 5. In the buckling
analysis (Figure 6a), the structure was fully laterally braced at the top of the rack, and the rotational
restraints at the top and bottom connections of the uprights were modelled as per the rotational
stiffness presented in [7, 10]. Also, Figure 6a shows that the upright on grid B4 is the most critical
upright with the critical cross section as highlighted. Figure 6b shows the pre-load and post-load
deformed geometries: the grey lines represent the scaled buckled shape imposed onto the structure
before applying the load, whereas the green lines represent the deformed geometry after the load
was applied. It is obvious that this method is far more sophisticated than Method 2 as it requires
greater effort in calculating many different parameters. For more details about the method for
calculating the amplitude of local member imperfections, the reader can refer to Clause 5.3.2.1(11)
of EN1993.1.1 and Shaheen [11].
(a) (b)
Fig. 6. Illustration showing the pre-load and post-load deformed geometries of the DIR: (a) Buckling load
analysis (b) Pre-load and post-load deformed shapes for the gravity load analysis.
(a) (b)
Fig. 7. Fully loaded case (a) axial load diagram, and (b) bending moment diagram for the critical upright at grid B4
(a) (b)
Fig. 8. Pattern loaded case (a) axial load diagram, and (b) bending moment diagram for the critical upright at grid B4
5 DESIGN VERIFICATION
According to FEM10.20.07 (2012), the design verification calculations mainly depend on the
method of analysis chosen. If the global analysis model is created according to Method 1 then
FEM10.2.07 requires the design verifications to be only based on the nominal cross-section
capacity. Whereas, if the analysis model is created according to Method 2, FEM10.2.07 requires the
design verification to be based on the member capacity. According to Method 2, the member
capacity checks require the determination of the effective length of the upright that shall be used to
calculate the critical buckling stress in either a flexural, torsional, or flexural-torsional buckling
mode. The determination of the effective buckling length is discussed in the following section.
𝐸𝐼
𝐿𝑒𝑓𝑓 = 𝜋√𝑃 . (2)
𝑐𝑟
The elastic buckling load can be determined using different approaches. One approach is the
analytical evaluation by which the differential equation of the deformed shape is defined and solved
in terms of the buckling load and the amplitude of deformation. Another approach for determining
the buckling load is by numerical calculation which is sometimes called linear eigenvalue analysis.
Nowadays, the numerical calculation of the elastic critical buckling load of a structure is a standard
feature of most commercial softwares. The second method for determining the effective length is
based on direct evaluation using the curvature of the column. Based on the latter method, the
effective length is calculated as the distance between the contra-flexure points as shown in Figures
9(b) and 9(c).
Fig. 9. (a) Upright 2D model for the determination of the effective length as per FEM10.2.07, (b) buckling
shape of the upright, (c) determination of the effective length between points of contra-flexure.
𝑁∗ 𝑀∗
+ , (3)
Φ𝑁𝑠 Φ𝑀𝑠
𝑁∗ 𝑀∗
+ Φ𝑀 . (4)
Φ𝑁𝑏 𝑏
In Eqs (3) and (4), M* is the 2nd order elastic moment with and without the additional moment due
to local member imperfection, as per Method 1 and Method 2 respectively, Ns is the axial capacity
of the cross-section, Ms is the moment resistance of the cross-section, Nb is the member axial
capacity, Mb is the member moment resistance, and Φ is the strength reduction factor. As shown in
Figure 10(b), the stress ratio in the advanced analysis was capped at 100% utilization ratio since the
failure of the rack was triggered by material yielding. In contrast, the stresses at failure according to
Method 1 exceeded 100% because the FEM10.2.07 guidelines do not consider material nonlinearity
in the global analysis model but instead they compensate for the effect of these by using an inflated
local member imperfection amplitude which is about four times the admissible local bow
imperfection specified in AS4100 for vertical columns. However, as shown in Figure 10(b),
Method 1 reaches 100% utilization at a load factor of 1.95 which is very close to the ultimate load
factor obtained using the advanced analysis method. According to Method 2, two curves were
plotted using two different effective lengths (i.e. 3420mm and 1975mm). When using Leff =
3420mm, the critical cross-section reaches its ultimate strength at a load factor of 1.3. Whereas,
when using Method 2 with Leff = 1975mm the critical cross-section reaches its ultimate strength at a
load factor of 1.7. This implies that using Method 2 with the effective length of the upright
determined by the curvature method in conjunction with using the design interaction equations in
the standards underestimates the load carrying capacity of the upright by 50% ((1.95-1.3)/1.3) when
compared to Method 1 results. Whereas, determining the effective length based on the maximum
axial load in the upright underestimates the upright’s capacity only by 15% ((1.95-1.7)/1.7) when
compared to the results of Method 1. It is therefore concluded that the latter approach is more
appropriate if Method 2 is used for the design of drive-in racks.
Figures 11a and 11b show the utilization ratios for the drive-in rack when the failure of the structure
is governed by global buckling of members in either the flexural, torsional, or flexural-torsional
buckling modes. For the two cases studied, viz. full load (Figure 11a) and pattern load (Figure 11b),
the stress utilization ratios based on Method 1 are consistently lower than the values obtained by
Method 2. For instance, when the full design load is applied (i.e. at 100% of the load), the stress
ratios, according to Method 1 and Equation 3, for the fully loaded and the pattern loaded models are
approximately 59% and 66% respectively, whereas the stress ratios, according to Method 2 and
Equation 4, are approximately 66% and 74% respectively. This means that by using Method 1 the
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
rack structure can resist an additional 12% of the applied loads in the fully and pattern loaded racks
before reaching the same utilization ratios achieved by Method 2.
Another case study was performed in which the uprights were governed by the local distortional
capacity of the upright cross-section. It was found that Method 2 could give very different results
when compared with the case when the uprights were governed by the global flexural behaviour of
the uprights as presented in the previous sections. More details about the distortion failure of the
upright cross-section can be found in [11].
(a) (b)
Fig. 10. (a) Load displacement curve based on the GMNIA analysis method, (b) Load factor-utilization
ratio curve for upright B4 at the critical section (RL+0.00) for the fully loaded case model.
(a) Critical section at the base (b) Critical section at RL+1.078m from the base
Fig. 11: Utilization ratio curve for upright B4 at the critical section for (a) fully loaded case, and (b) pattern loaded case
6 SUMMARY
The design of drive-in racks as per FEM10.2.07 was carried out using Method 1 and Method 2.
Different load cases were examined including fully racks and pattern loaded racks. It was found that
carrying out the design according to Method 1 was sophisticated as it required more calculations for
determining the design actions in the members, especially the uprights. Whereas, Method 2 seemed
to be computationally more efficient when compared to the analysis procedures required by
Method 1. According to Method 1, the design verification is based on the cross section capacity as
compared to Method 2 which is based on the member capacity. Different approaches were used to
define the effective length for determining the member capacity of the uprights according to
Method 2. When the effective length was defined as the distance between the points of contra-
flexure of the deformed shape of the uprights it showed that the capacity of the uprights were
underestimated by 50% when compared to GMNIA analysis results. Whereas, when the effective
length of the upright was back calculated based on the maximum force in the upright segments, it
showed that the capacity of the uprights were underestimated by 15%. Thus, this proposes that the
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
latter approach is more appropriate to adopt if the designer prefers to use Method 2 due to its
computational efficiency.
When using Method 2, If the stress utilization ratio at the ultimate design loads is less than or equal
to the ultimate capacity (i.e. utilization ratio ≤ 100%) then the design can be verified to be adequate
and there is no need for further refinements. However, if under full ultimate loads the utilization
ratio using Method 2 exceeds the design capacity by a small amount about (i.e. 10% ~ 12%), then
the designer has the option either to increase the capacity of the uprights, or to re-analyze the rack
using Method 1 which could demonstrate that the design of the uprights is adequate, thus leading to
a more economic design.
REFERENCES
[1] FEM10.2.07, The Design of "Drive-in and Drive-through racking" Pallet Racking, European
Racking Federation- ERF/FEM R&S. 2012, European Racking Federation, Birmingham.
[2] RMI, Specification for the design, testing, and utilization of industrial steel storage racks-2008
edition, Rack Manufacturers Institute (RMI), Charlotte, USA. 2008.
[3] EN15512, Steel static storage systems - Adjustable pallet racking systems - Principles for structural
design, European Committee for Standardization (CEN), Brussels, Belgium. 2009.
[4] AS4084, Steel Storage Racking, Standards Australia, Sydney, Australia. 2012.
[5] Gilbert, B.P. and K.J.R. Rasmussen, Finite Element modelling of steel drive-in rack structures.
Research Report R901. 2009, School of Civil Engineering, The University of Sydney, Australia.
[6] Gilbert, B.P. and K.J.R. Rasmussen, Drive-In Steel Storage Racks I: Stiffness Tests and 3D Load-
Transfer Mechanisms. Journal of Structural Engineering, 2012. 138(2): p. 135-147.
[7] Gilbert, B.P. and K.J.R. Rasmussen, Bolted moment connections in drive-in and drive-through steel
storage racks. Journal of Constructional Steel Research, 2010. 66(6): p. 755-766.
[8] EN1993.1.1, Eurocode 3 Design of steel structures. General rules and rules for buildings. 2005.
[9] AS4100, Steel Structures, Standards Australia, Sydney, Australia. 2012.
[10] Gilbert, B.P. and K.J.R. Rasmussen, Determination of the base plate stiffness and strength of steel
storage racks. Journal of Constructional Steel Research, 2011. 67(6): p. 1031-1041.
[11] Shaheen, M.S.A., Comparison between the analysis and design methods for design of steel drive-in
rack structures, Masters of Engineering thesis. Univeristy of Sydney. Australia. 2013.
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)