Documente Academic
Documente Profesional
Documente Cultură
Cornell University
HOLLISTER HALL, ITHACA NY 14853-3501
October 6, 2000
TO:
Prof. T. Pekz
FROM:
SUBJECT:
INTRODUCTION
Design of cold-formed steel pallet rack frames has traditionally been empirical due to
their complex behavior and lack of information with regard to the influence of various
parameters on the overall system strength. The behavior and ultimate strength of a typical
pallet rack frame are characterized by many parameters such as flexibility of beam to
column joints, column base fixity, perforations in their column members, local buckling
of member components, geometric and material imperfections and complex buckling
behavior of column members.
The recommendations of the current American pallet rack frame design specification
(RMI 1997) are not very different from its earlier versions which were formulated based
on limited experimental studies carried out at Cornell University (Pekz 1975). Since its
earliest edition, many developments have taken place in the manufacturing and use of the
rack frames. However, very little efforts are made to improve their design procedure
reflecting the changes in the technology. A review of the current specification (RMI
1997) reveals that its design recommendations are empirical and conservative with regard
to the estimation of frame strength under static loading condition. With the help of new
analytical methods available today such as FEM, it is possible to study the behavior of
rack frames more easily and with greater reliability. This would result in more accurate
and economical designs of the rack frames.
As an attempt towards achieving this objective, the RMI and the AISI have jointly
sponsored a project on Cold-Formed Steel Frame and Beam-Column Design with Prof.
Pekz of Cornell University as principal investigator. This report presents the results of
the research carried out during October 1998 September 2000 on the above project.
This section presents the details and results of the present work in two parts (i) Analysis
and (ii) Design.
3.1
Analysis
Experimental methods and Finite element technique are the two procedures available to
study the complex behavior of pallet rack frames. In the present study it is proposed that
the experimental methods be used to understand the behavior of components of rack
frames (e.g. beam-to-column joint, column base etc.) and finite element method to study
the overall system behavior. The details of the work are presented below.
3.1.1
Behavior of beam-to-column joints and column bases of different types of pallet rack
frames are studied in the present work as described below.
3.1.1.1
The connection between the shelf beam and column members of pallet rack frames is
generally flexible. The influence of joint flexibility on the overall frame behavior and
strength is significant. However, due to a very large variety of connection types used in
practice no attempt has been made so far to standardize these joint types nor to develop a
general expression for the joint stiffness which could be used for all types of connections.
The current RMI specification accounts for the effect of joint flexibility on column
strength by modifying the pallet beam stiffness and in turn modifying the column end
restraint offered by the beam member. The joint stiffness required to do so is to be
determined experimentally by individual manufacturer. The specification suggests using
the secant stiffness corresponding to 0.85 times the ultimate moment capacity of the joint
as determined from physical tests. While the specification procedure is simple to use,
there may be cases where the above assumption/simplification does not hold true. Hence,
it is always rational and safe to use the correct joint stiffness value in the frame analysis
by adopting joint M relationship valid through the entire load history.
In the present study such an expression for joint stiffness which can be used in the frame
analysis is developed. For this purpose, the experimental data available in terms of joint
moment-rotation history of a variety of joints as provided by different manufacturers is
used. The following six types of joints are considered in the study.
Type of Frame/Joint
Column
ALDR
Type B
Type C
Type D
U 3.0
U 3.5
U 5.0
3.75
4.09
5.22
3.00
3.50
5.00
In each of the Type D joints, again two different shapes are considered. They only differ
in the configuration. Their geometric dimensions are same as given above.
By trail and error, a tri-linear M equation has been established to fit the experimental
data of the above joint types. The corresponding joint stiffness is given by the following
equations.
K = 88.5 xDb
= 34. 5xDb
= 0.0
for
for
for
0. 0 M < 0.75M u
0. 75M u M < 0.99M u
0.99 M u M
(3.1)
The joint stiffness as given by Eq 3.1 is expressed in terms of the beam depth and
ultimate moment capacity of the joint as it was understood from previous experimental
and numerical studies that the joint stiffness is primarily due to distortion of a small
portion of the column member confined to the region of magnitude equal to beam depth
as shown in Fig 3.1.1.1.1 (This figure is taken from study on FE models presented in
section 3.1.2.1).
Fig 3.1.1.1.1
Eq 3.1 is compared with the experimental results in Fig 3.2.
Beam-column joint stiffness
Proposed
Type B
25
20
20
16
Moment (kip-in)
Moment (kip-in)
ALDR
15
10
Proposed
12
0
0
0.05
0.1
Joint Rotation (rad)
(a)
0.15
0.2
0.05
0.1
0.15
0.2
(b)
4
U 3.0 (Type 1)
24
Proposed
10
MOment (kips-in)
Moment (kip-in)
20
16
12
8
4
8
6
4
2
0
0
0.05
0.1
0.15
0.2
0.02
(c)
0.04
0.06
Rotation (rad)
0.08
0.1
(d)
U 3.5 (Type 2)
Proposed
U 5.0 (Type 1)
12
14
10
12
MOment (kips-in)
MOment (kips-in)
U 3.0 (Type 2)
12
8
6
4
2
U 5.0 (Type 2)
Proposed
10
8
6
4
2
0.02
0.04
0.06
0.08
0.1
0.02
0.04
0.06
0.08
0.1
Rotation (rad)
Rotation (rad)
(e)
(f)
Fig 3.1.1.1.2
It should be noted that the proposed expression for joint stiffness is not a general equation
applicable for all types of joints. This can be used only after validating with more
experimental data of wider range of joint types. It may also be noted that the above
equation requires the ultimate moment capacity known a priory. The joint stiffness as
given by the above expression also needs to be validated against the effect of moment to
shear ratio on the joint.
3.1.1.2
The column base stiffness of pallet rack frames is characterized by the base plate
dimensions, number, dimensions and layout of bolts, ratio of moment to axial load at the
column base and foundation characteristics. The degree to which each parameter effects
the base stiffness depends on the way the column is connected to the foundation. The
total base rotation can be divided into the following three components.
1. Base rotation due to deformation of foundation : Until no tension is developed at the
column base, the entire base rotation is due to deformation of foundation. The effect
of deformation of bolts is negligible in this case as they dont resist any external load.
An exception to this case is separation of base plate (due to lifting of plate ends
along the periphery of plate) from the foundation under very high axial load
especially when thick base plate is used. Such cases need to be analyzed carefully.
2. Base rotation due to bending of base plate : When the lateral load on the column (or
eccentricity of axial load) is large enough to cause tension at the column base, the
base plate separates from the foundation in the tension region. This happens so
irrespective of the number and arrangement of bolts. In such cases, the joint rotation
will primarily be due to the bending of plates. The effect of deformation of bolts on
base rotation could be neglected for all practical purposes under the assumption that
they are overdesigned. In such case, perfect contact between the base plate and
foundation at the locations of the bolts may be assumed for analysis purpose.
However, if the bolts fail in tension no contact between the base plate and foundation
in the tension zone can be assumed.
3. Base rotation due to deformation of bolts : The effect of extension of bolts on the
base stiffness is considerable when some portion of the column base experiences
tension. However, it is always accompanied by bending of base plate and as the
moment to axial load ratio increases, this effect becomes negligible. In order to study
the effect of bolt extension on base stiffness, the problem can be divided into the
following two cases.
When the bolt tension is smaller than the bolt capacity : In this case both the bolt
deformation and bending of base plate contribute to the base rotation. The lay out
of bolts also influences the base flexibility.
When the bolt tension exceeds the bolt capacity : This is a special case of category
2 as discussed above
In the present work, the base stiffness characteristics due to bending of base plate
(category 2) is studied. The following assumptions are made in the study.
For this purpose, a parametric study involving a large number of nonlinear finite element
analyses of a 3x3x0.091 lipped channel cantilever column has been carried out. The
length of the column is taken as 30``. All the influencing parameters are considered in the
study. They include : Axial and lateral loads on the column, base plate configuration and
number of bolts. Four types of base plates (8x5, 8x8, 6x6 and 7x7) with 3
thickness values (0.25, 0.375 and 0.5) in each category are considered in the study.
Each column was analyzed for 4 different lateral load to axial load ratios, ranging from
1.5% to pure moment case. Fig 3.1.1.2.1 shows the schematic diagram of the model used
in the parametric study.
P
C.G
3.0``x3.0``x0.7`` Column
Spring (Bolt)
Base Plate
Contact Elements
(To model gap)
Foundation block
(Fixed at bottom)
Fig 3.1.1.2.1
7
For the purpose of analysis, ABAQUS, a general purpose FE commercial software has
been used. The features of the FE model used in the study are described below.
Four noded shell element (S4R5) : Used to model the cantilever column and base and top
plates.
3D solid elements : To model the foundation. The foundation was assumed made of steel
and 1.0`` deep. The base of the foundation was assumed to be fixed in the study. (A later
study of one column case (wherein the base was modeled as 6`` deep concrete block with
fixed base) showed that the influence of the depth and material of the foundation is
insignificant. However this needs to be verified by further study).
Nonlinear spring elements : To model the anchor bolts. Two springs are used to model
each bolt one to model shear and one to model the stiffness in the axial direction.
However, the stiffnesses of the springs in both the directions are taken to be very high in
the present study assuming that the connections are over designed as discussed before.
Contact elements : To model the gap between the base plate and the foundation.
The moment-rotation relationships for various base plate configurations as obtained from
the finite element analysis are plotted in Figs 3.1.1.2.2-5. Each of Fig. 3.1.1.2.2(a)-5 (a)
contains 12 graphs showing the effects of axial load and plate thickness on column base
flexibility for a given plan area of base plate. Graphs corresponding to very high axial
load (Plat =1.6% of Pax) case are shown in larger scale separately in Fig 3.1.1.2.2(b)-5 (b).
The base moment used to develop these graphs includes the P- nonlinear effect. Also
shown in the above plots is the column base stiffness value as recommended by the
current American pallet rack design specification (RMI 1997).
The following are a few observations made based on the study.
The moment-rotation relationship of a typical pallet rack frame base is generally nonlinear
Higher the axial load on the column stiffer is the column base.
While the axial load increased the base stiffness by about 20% when the base plate is thin
(0.25``, thinnest of plates studied), its effect is found to be insignificant in the case of thicker
plates (Figs 3.1.1.2.2(a)-5 (a)).
when the axial load on the column is accompanied by only a small amount of lateral load
(1.6% of axial load), the column base stiffness is found to be very close to that of RMI
specification value (Figs 3.1.1.2.2(b)-5 (b )).
For lateral load to axial load ratios other than 0.016, the initial stiffness of column base is
found to vary from 0.3-0.7 times the RMI specification value, depending on the base plate
configuration and amount of axial load on the column (Figs 3.1.1.2.2(a)-5 (a)).
The effects of base plate configuration (plan dimensions) and number of bolts on the base
stiffness are found to be negligible. However in the case of thinner plates, smaller plate
configurations seem to help increase the base stiffness slightly.
Further study aimed to verify the above observations over wider range of parameters and
to quantify the base flexibility caused by the deformation of foundation and extension of
bolts is underway. The future study will consist of both experimental and finite element
studies.
(P_lat=0.33P_ax)
(P_lat=0.16P_ax)
50
Moment (kips-in)
40
30
20
10
0
0
0.01
0.02
0.03
0.04
0.05
Fig 3.1.1.2.2a
K_RMI
(P_ax=60P_lat)
10
Moment (kips-in)
0
0
0.001
0.002
0.003
0.004
Fig 3.1.1.2.2b
(P_ax=0.0)
(P_lat=0.016P_ax)
K_RMI
(P_lat=0.33P_ax)
(P_ax=6P_lat)
50
30
20
10
0
0
0.01
0.02
0.03
0.04
0.05
Fig 3.1.1.2.3a
K_RMI
(P_ax=60P_lat)
10
Moment (kips-in)
Moment (kips-in)
40
0
0
0.001
0.002
0.003
0.004
Fig 3.1.1.2.3b
10
(P_ax=0.0)
(P_lat=0.016P_ax)
K_RMI
(P_lat=0.33P_ax)
(P_lat=0.16P_ax)
50
30
20
10
0
0
0.01
0.02
0.03
0.04
0.05
Fig 3.1.1.2.4a
K_RMI
(P_ax=60P_lat)
10
8
Moment (kips-in)
Moment (kips-in)
40
0
0
0.001
0.002
0.003
0.004
Fig 3.1.1.2.4b
11
(P_ax=0.0)
(P_lat=0.016P_ax)
K_RMI
(P_lat=0.33P_ax)
(P_lat=0.16P_ax)
50
30
20
10
0
0
0.01
0.02
0.03
0.04
0.05
Fig 3.1.1.2.5a
K_RMI
(P_ax=60P_lat)
10
8
Moment (kips-in)
Moment (kips-in)
40
0
0
0.001
0.002
0.003
0.004
0.005
0.006
Fig 3.1.1.2.5b
12
3.1.2
Frame Behavior
The objective is to establish guide lines to prepare FE models of CFS pallet rack frames
to study their behavior considering various influencing parameters. The following two
models are proposed in the present study. ABAQUS a commercial finite element
software is used to validate the proposed models.
3.1.2.1
In this model the entire frame is modeled using thin shell elements. All the important
parameters that effect the frame behavior are automatically accounted for in the model
resulting in an accurate numerical model very similar to physical model of the frame. The
features and validation of this model as developed in the present study are discussed
below in detail.
Modeling features
Column and Beam members : General thin shell elements with 6 dof are used to model
the columns, beams and base plates. Holes in the columns are modeled by defining the
thickness of the corresponding elements very small (~ 0.0000001 inch).
Beam-to-column joint : This is the most critical part of the modeling. The 3D nature of
the connection behavior makes the exact joint modeling very complex, computationally
expensive and practically prohibitive. Moreover, in the case of cold-formed steel
members, very small or negligible out of plane stiffness of the plate element of the
column (like flange or web) to which the beam is connected is the primary source of the
joint flexibility. Hence, modeling of tabs/keys and other accessories could be ignored
without any loss of accuracy. The following two joint models are experimented in the
present study.
Joint model-1: In this model the end of the beam is connected directly to the flange or
web of the column and the connection is treated as continuous. The end plates and
tabs/keys are ignored completely. This is also termed as continuous model. This
model is simple to use and will be of great help when modeling large scale frames.
Joint model-2 : In this model the beam ends are connected to end connecting plates
which in turn are connected to columns at a few points. The gap between the column
and endplate is modeled using contact elements. This model is also called
discontinuous model. Tabs are not included in this model.
Column base : It is suggested to use contact elements to model the discontinuity between
the base plate and the foundation. The foundation bloc and the anchor bolts can be
modeled by using solid and beam elements respectively.
13
Validation
The following pallet rack components and sub-assemblies are chosen for the validation
study. The results of the analytical study obtained based on complete nonlinear
(geometric + material nonlinearities) FE analysis are compared with their experimental
results (Pekz 1975).
List of analyses carried out :
(i)
(ii)
(iii)
(iv)
(v)
Stub column
1 case
Cantilever beam test
2 cases
(a)
Direct connection between the beam end and the column flange
(b)
Using Contact element between the beam end and the column
flange
ALDR type frame under gravity load
3 cases
(a)
Direct connection between the beam end and the column flange
(b)
Using Contact element between the beam end and the column
flange
(c)
Using Contact element between the beam end and the column
flange with 1 inch in 10 ft out of plumb of columns
ALDR type frame under (gravity load + Lateral load)
2 cases
(a)
Direct connection between the beam end and the column flange
(b)
Using Contact element between the beam end and the column
flange
AHDR type frame under (gravity load + Lateral load)
1 case
Fig 3.1.2.1.1 shows the details of the column and beam members of ALDR and AHDR
frames. The corner radius and the effect of cold forming are neglected in the present
study.
Column post-ALDR frame
3.00"
0.687"
0.72"
1.57"
0.531"
2. 91"
0.534"
0.091"
0.71"
0.71"
0. 091"
14
0. 070"
0. 133"
1. 50"
3. 75"
2. 0"
0.063"
1. 675"
2. 75"
Column perforations
Fig 3.1.2.1.1
Discussion of results
Stub column
Analysis on the stub column of an ALDR (Fig. 3.1.2.1.1) column was carried out under
pure compression. The objective is to test the models performance against estimating the
section strength of columns accounting for the effect of local buckling and holes.
Residual stresses and imperfections are not included in the study.
The ultimate load of the stub column as obtained by the FE analysis is found to be 25.7
kips as against to its experimental value of 22.8 kips with a difference of +10%.
The deformed shape and von-Mises stress (equivalent uni-axial stress for a combined
state of stress; determined from the maximum distortion energy theory of failure,
according to which the failure by yielding occurs when, at any point, the distortion
energy for unit volume in a state of combined stress becomes equal to that associated
with yielding in simple tension) of the stub column at failure are shown below in Fig
3.1.2.1.2 and Fig 3.1.2.1.3 respectively.
16
3
2.5
2
1.5
1
0.5
0
0
Fig 3.1.2.1.4
*
0.01
0.02
0.03
0.04
0.05
17
The difference in the load estimates of the two models is found to be very small (3.5%).
Hence, the difference in the experimental and analytical strengths is attributed to
ignorance of the geometric imperfections and the residual in the analytical study. It may
also be noted from the plot that on contrary to the suspicion, the behavior is softer when
the holes are modeled with very thin elements. This may be due to greater loss of section
stiffness caused by out of plane bending of the negligibly thin elements in the first model.
Hence, it can be concluded that the holes could be modeled successfully with negligibly
thin elements with out appreciable error to predict the behavior and ultimate load.
However it may be advisable to do a similar study in more complicated situations as well
before proceeding to the actual analysis. This is further explained in the following
sections.
Cantilever beam test
The objective is to study the performances of different shell element models in predicting
the beam-to-column joint behavior of pallet rack frames. The details of the model used
(joint between ALDR column and beam Fig 3.1.2.1.1) for this purpose are shown Fig
3.1.2.1.5.
30. 0"
24. 0"
Fig 3.1.2.1.5
The following are a few observations made with regard to the models performance in
predicting the joint behavior.
18
Fig 3.1.2.1.6
A very good correlation is found between the Analytical and Experimental results
with respect to the deflection at the tip of the cantilever beam as shown in Fig
3.1.2.1.6.
The difference between the strength estimates of the two joint models is negligible.
Figs 3.1.2.1.7 and 3.1.2.1.8 show the deformation configurations of the joint as
predicted by the two models in subultimate and near failure.
19
(Joint Model-1)
(Joint Model-2)
Fig 3.1.2.1.7
20
Fig 3.1.2.1.8
As shown in Figs 3.1.2.1.7 and 3.1.2.1.8, irrespective of the joint model adopted, the
two flanges of the column distorted to different amounts reflecting the semi-rigid
nature of the joint. This is due to small out of plane stiffness of the column flanges.
The failure was initiated by the yielding of the material near the beam-to-column joint
as expected. The distribution of von-Mises shear stress at failure is shown in Fig
3.1.2.1.9.
The model is also able to predict the failure load with sufficient accuracy as compared
to the experimental load. The ratios between the analytical load and the experimental
load are found to be 1.09 (corresponding to vertical deflection=4.5 inch, which is very
large) and 0.98 for the Joint model-1 and Joint model-2 respectively. The failure
mode as predicted by the finite element analysis (yielding of the beam-column joint)
is also found to be in agreement with the experimental observation.
From the results discussed above it is clear that the proposed shell element model is able
to predict the behavior and strength of pallet rack frame components and subassemblages accurately. This model could be used as an alternative to physical tests in
future studies.
21
Fig 3.1.2.1.9
Fig 3.1.2.1.10
22
62. 0"
62. 0"
56. 0"
93. 0"
93. 0"
column joints. As shown in fig., the frame was restrained with respect to the out of plane
deflection at each story level.
23
Fig 3.1.2.1.12a
Fig 3.1.2.1.12b
24
The ultimate loads and failure modes of all the frames as obtained from nonlinear finite
element analysis are summarized in Table 3.1.2.1.1 along with their experimental
observations.
Table 3.1.2.1.1
Frame
Loading
ALDR
Gravity
ALDR
Gravity +
Horizontal
AHDR
Gravity +
Horizontal
Joint Model
Model-1
Model-2
**
Model-2
Model-1
Model-2
Model-1
Pult ,ana
P
ult , exp t
1.58
1.47
1.10
*
1.12
*
1.09
*
1.13
Failure Mode
Analytical
Exptl
Yd of joints
Yd of joints
Yd of joints
Yd+LB
of lowest
story beam
Permanent
sway
Permane
nt sway
Permanent
sway
Permane
nt sway
LB Local Buckling
First, the frame was analyzed using Joint model-1. As shown in Table 3.1.2.1.1, the
ratio between the analytical and experimental failure load was found to be very high
(1.58). Interestingly, the failure modes as predicted by the FEM analysis and that
observed in the test are also noted to be different.
The frame was re-analyzed using the Joint model-2 (though the earlier experience with
the cantilever beam problem (section2.3) showed only 10% difference in the ultimate
loads of the two joint models). The ratio between the analytical and experimental failure
loads with joint model-2 is found to be 1.47, with an improvement of about 10% as
compared to the joint model-1. While it is consistent with the previous observation made
at component level (cantilever problem), the reasons for a large difference in
experimental and analytical values remained undetermined at this time
While no convincing reasons could be seen for poor performance of the model, it was
decided to re-analyze the frame considering the out of plumb of the column uprights,
which was ignored, in the previous analyses. The initial configuration of the frame was
remodeled with columns 1 inch out of plumb for 10 ft of height as recognized in the RMI
specification. The Joint model-2 was used to model the connection. The ratio between the
analytical and experimental failure loads was found to be 1.10. The local imperfections
and the residual stresses are attributed for the conservative analytical estimates (by 10%).
It is learned from the above 3 analyses that the frame is very sensitive to the
imperfections, which must be included in the analytical study.
The vertical deflections at mid section of the bottom story beam as estimated by the
FE analysis is compared with the experimental observations in Fig 3.1.2.1.13.
25
Fig 3.1.2.1.13
The von-Mises stress contour (at failure) in the bottom story beam and in the interior
column joint is shown below in Fig 3.1.2.1.14. It may be noted that while the stresses
in the beam were below the yield stress (49 kips), the stress in the joint exceeded the
material stress indicating the joint failure which is in contrary to the experimental
observation. This is different from the failure mode observed in physical test (Table
2.4.2.1).
26
Fig 3.1.2.1.14a
Fig 3.1.2.1.14b
The deformed shape of the frame bottom interior joint at failure are shown in Fig 3.1.2.1.15.
27
Fig 3.1.2.1.15a
Fig 3.1.2.1.15b
28
The frame was analyzed for gravity and lateral load combination. The loading was
applied in the same sequence as it was applied in physical test. The connection was
modeled using both the joint models and the results are compared with experimental
values in Table 3.1.2.1.1.
As reported in Table 3.1.2.1.1, the difference between the failure loads of the two
joint models is negligibly small, both are found to be close to the experimental value.
The conservative estimates of FE analysis could be attributed to the ignorance of
imperfections.
As shown in Fig 3.1.2.1.16-17, the performance of the model in predicting the loaddeflection behavior is very satisfactory.
The failure of the frame was found to be due to large lateral deformations associated
with yielding of interior joint of the lowest storey. This agrees with the experimental
failure mode.
The deformed shapes of the frame at failure as observed in the analytical and
experimental studies are shown in Fig 3.1.2.1.18-20.
Fig 3.1.2.1.16 Load Vs Vertical deflection at the midsection of bottom story beam
29
Fig 3.1.2.1.17
Fig 3.1.2.1.18
30
Fig 3.1.2.1.19
31
Fig 3.1.2.1.20
32
The frame was analyzed for gravity and lateral load combination as in the previous
example. However, in view of negligible difference in the performance of the two
joint models as noted in the previous examples, joint model-1, which is simple to use,
was adopted in this analysis.
As noted in Table 3.1.2.1.1, the comparison between the analytical and experimental
estimates of frame strengths is very good.
The lateral displacement at the top story beam level and the vertical deflections at
the mid section of bottom story beam are plotted against the load in Fig 3.1.2.1.21. As
shown in the figure, the model is able to predict the frame behavior through the entire
load history.
Fig 3.1.2.1.21a
33
Fig 3.1.2.1.21b
Conclusion
The foregoing results demonstrate clearly the accuracy of the proposed shell element
model to predict the behavior and strength of pallet rack frames accounting for various
parameters. However, it was also observed that modeling of full frame by shell elements
is tedious, time expensive and requires experienced analysts. These drawbacks limit the
use of shell element model to a reference/benchmark numerical model. This model best
serves in validation studies (as an alternative to experiments) to evaluate the performance
of other simpler numerical or analytical techniques.
3.1.2.2
While the shell element model discussed above is versatile, rational and accurate, it is not
suitable for routine design works and in situations where large number of frames are
required to be analyzed. Hence, there is a need of a simpler analytical model. Such a
model is explored below.
The objective of the study presented in this section is to develop a simple finite element
model capable of estimating the behavior and ultimate load capacity of pallet rack frames
accurately. This helps to generate large amount of data of Pu and Mu of pallet rack
frames, more efficiently, required to quantify the approximations in the current RMI
specification and suggest modifications if required.
34
Modeling Features
The influencing parameters and how they are incorporated in the present FE model
developed in ABAQUS are described in the following table.
Table 3.1.2.2.1
Parameter
Column members
Pallet beams
Joint Flexibility
Warping d.o.f.
(at column base)
Local Buckling
Imperfections
1. Fixed
2. Free
35
Validation
The two pallet rack frames (ALDR and AHDR) used before in the validation study of
shell element model are analyzed using the beam element model with the above features
and the results are summarized in Table 3.1.2.2.2. In the case of ALDR frame the frame
capacities as evaluated by the current RMI design rules are also compared. The results
presented for AHDR frame correspond to warping fixed case. (It was observed later that
in the case of frames made of AHDR type columns, the end condition wrto warping dof
does not have significant effect on the frame strength).
Table 3.1.2.2.2a
Gravity
Free
Fixed
Shell
element
model
0.75
1.08
1.10
Gravity +
0.80
Lateral
*
with respect to lateral load
Table 3.1.2.2.2b
**
0.91*
1.09*
**
0.98
1.06**
with respect to gravity load
RMI
Specification
0.43
0.52
0.42
0.50
Load Combination
(Pu,beam /Pu,shell )
1.10
1.05
The effect of warping constraint on the frame strength depends on the type of column
members. The difference between the strengths of warping free and warping fixed
cases may vary from 4% (AHDR type columns) to 25% (ALDR columns).
The limited parametric study shows that the warping fixed case simulates the
experimental condition better (Table 3.1.2.2.2a).
The beam element model with the proposed features (Table 3.1.2.2.1) is capable of
estimating the frame strengths accurately as compared to experimental and rational
shell element based FE results.
The current RMI specification underestimates the strengths of pallet rack frames by
about 50% (more results on RMI procedure are presented in section 3.2.1). This
establishes the need to review and improve the current design procedure (RMI 1997).
The foregoing discussion qualifies the proposed use of beam element model in future
studies on development of design procedures for rack frames. The following section
discusses the details of the work done in this direction.
36
3.2
Design
As noted before, the current American pallet rack design procedure (RMI 1997) is very
conservative. This is attributed to the following inconsistencies in the procedure. The
st
specification suggests linear (1 order elastic) frame analysis to obtain the member
forces. It uses effective length procedure to include the frame effect. The effective length
factors are calculated either from rational buckling analysis or by using the alignment
charts. It also suggests a set of fixed effective length factors in lieu of the above
procedures. The effect of beam-to-column joint flexibility is accounted for indirectly by
reducing the beam stiffness in the calculation of column effective factor. The
specification recommends the linear beam-column interaction equation of the current
cold-formed steel design specification (AISI 1996) to check the member safety. To
account for moment magnification due to the presence of axial load on the column it uses
the simple amplification factor derived from linear elastic buckling analysis. The effects
of local buckling and holes in the columns are considered using effective section and
Q-factor approaches respectively. While the specification procedure has been made
conservative in the absence of knowledge of the effect of various parameters on the
overall frame behavior, it is not known to what magnitude each of these parameters
effects the accuracy of the design procedure.
To quantify the conservatism in the current procedure caused by the use of approximate
effective length factors, linear beam-column interaction equation and linear moment
amplification factor, a large parametric study involving nonlinear finite element analysis
of five types of frames has been carried out. The details of the analyses are given below.
Frame Type
AHDR
Type B
Type D
U 3.0
U 3.5
U 5.0
Column
Beam
3.75
4.09
3 x 3 x 0.091
3 x 3 x 0.091
3 x 3 x 0.091
3.00
3.50
5.00
37
The frame capacities as estimated by the RMI specification are compared with the finite
element results in Tables 3.2.1.1-6.
Table 3.2.1.1a
Frame
AHDR
2Bx3S
0.75
0.75
0.72
Remarks
U 5.0
0.70
4Bx3S
0.66
0.66
0.70
0.68
0.66
6Bx3S
0.72
0.72
0.74
0.72
0.67
4Bx4S
0.81
0.80
0.82
0.75
0.67
4Bx6S
0.62
0.56
0.74
0.74
0.72
6Bx6S
0.63
0.56
0.69
0.86
0.85
Table 3.2.1.1b
Frame
AHDR
2Bx3S
0.55
Loading
Gravity only (all but
one span case)
Notional load = 0.0
Base fixity
Kb = KRMI
Effective length factors
Kx = 1.7
Kphi = 0.8
Ky = 1.0
0.63
0.62
Remarks
U 5.0
0.63
4Bx3S
0.49
0.57
0.59
0.59
0.59
6Bx3S
0.53
0.63
0.62
0.62
0.60
4Bx4S
0.59
0.69
0.69
0.64
0.60
4Bx6S
0.46
0.49
0.62
0.64
0.65
6Bx6S
0.46
0.49
0.58
0.74
0.76
Loading
Gravity only (all but
one span case)
Notional load = 0.0
Base fixity
Kb = KRMI
Effective length factors
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
38
Table 3.2.1.2a
Frame
AHDR
2Bx3S
0.76
0.93
0.84
Remarks
U 5.0
0.69
Loading
Gravity+ 1.5% Lateral
4Bx3S
0.80
0.72
0.96
0.85
0.72
6Bx3S
0.88
0.84
0.91
0.80
Kb = KRMI
4Bx4S
0.84
0.76
0.96
0.88
0.75
4Bx6S
0.90
0.75
1.00
0.95
0.82
6Bx6S
0.88
0.86
0.95
0.90
0.88
Kx = 1.7
Kphi = 0.8
Ky = 1.0
Base fixity
Table 3.2.1.2b
Frame
AHDR
2Bx3S
0.57
0.80
0.74
Remarks
U 5.0
0.63
Loading
Gravity+ 1.5% Lateral
4Bx3S
0.60
0.62
0.82
0.74
0.66
6Bx3S
0.68
0.75
0.80
0.73
Kb = KRMI
4Bx4S
0.63
0.67
0.82
0.77
0.69
4Bx6S
0.68
0.66
0.86
0.83
0.74
6Bx6S
0.68
0.77
0.82
0.79
0.81
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
Base fixity
39
Table 3.2.1.3a
Frame
AHDR
R = (Ucal / UFEM)
Product D
B-type
U 3.0
U 3.5
Remarks
U 5.0
Loading
2Bx3S
0.86
0.80
0.78
0.84
0.85
4Bx3S
0.84
0.70
0.87
0.72
0.87
6Bx3S
1.17
0.80
0.83
0.93
0.93
Kb = KRMI
4Bx4S
0.92
0.69
0.92
0.90
0.97
4Bx6S
0.87
0.87
0.92
1.04
0.95
6Bx6S
0.89
0.90
0.90
Kx = 1.7
Kphi = 0.8
Ky = 1.0
Table 3.2.1.3b
Frame
AHDR
2Bx3S
0.72
R = (Ucal / UFEM)
Product D
B-type
U 3.0
U 3.5
0.73
0.72
0.78
Remarks
U 5.0
0.81
Loading
Gravity+ 15% Lateral
4Bx3S
0.71
0.65
0.80
0.67
0.83
6Bx3S
1.03
0.75
0.78
0.88
0.89
Kb = KRMI
4Bx4S
0.78
0.64
0.84
0.83
0.92
4Bx6S
0.75
0.81
0.85
0.97
0.90
6Bx6S
0.82
0.84
0.85
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
Base fixity
40
Table 3.2.1.4a
Frame
AHDR
2Bx3S
0.64
U 5.0
0.76
0.71
0.76
Remarks
4Bx3S
0.71
0.73
0.70
0.65
6Bx3S
0.78
0.76
0.73
0.67
4Bx4S
0.85
0.66
0.78
0.71
Loading
Gravity only
(All spans)
Notional load = 0.0
Base fixity
Kb = KRMI/2
4Bx6S
0.92
1.04
0.94
0.71
6Bx6S
0.95
1.04
0.93
0.93
Table 3.2.1.4b
Frame
AHDR
2Bx3S
0.50
Kx = 1.7
Kphi = 0.8
Ky = 1.0
U 5.0
0.67
0.64
0.66
Remarks
4Bx3S
0.52
0.61
0.61
0.59
6Bx3S
0.57
0.64
0.63
0.60
4Bx4S
0.62
0.55
0.67
0.64
4Bx6S
0.68
0.88
0.80
0.64
6Bx6S
0.70
0.88
0.80
0.84
Loading
Gravity only
(All spans)
Notional load = 0.0
Base fixity
Kb = KRMI/2
Effective length factors
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
41
Table 3.2.1.5a
Frame
AHDR
2Bx3S
0.79
U 5.0
0.79
0.69
0.83
Remarks
Loading
Gravity+ 1.5% Lateral
4Bx3S
0.79
0.79
0.87
0.71
6Bx3S
0.82
0.84
0.89
0.76
4Bx4S
0.85
0.84
0.88
0.84
Kb = KRMI/2
4Bx6S
0.94
0.97
0.98
0.91
6Bx6S
0.92
0.98
0.99
0.96
Kx = 1.7
Kphi = 0.8
Ky = 1.0
Table 3.2.1.5b
Frame
AHDR
2Bx3S
0.59
U 5.0
0.67
0.63
0.72
Remarks
Loading
Gravity+ 1.5% Lateral
4Bx3S
0.59
0.67
0.76
0.65
6Bx3S
0.63
0.73
0.78
0.68
4Bx4S
0.65
0.72
0.77
0.76
Kb = KRMI/2
4Bx6S
0.72
0.83
0.85
0.82
6Bx6S
0.70
0.84
0.87
0.86
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
42
Table 3.2.1.6a
Frame
AHDR
2Bx3S
0.90
U 5.0
0.91
0.74
0.93
Remarks
Loading
Gravity+ 15% Lateral
4Bx3S
0.81
1.08
0.90
0.85
6Bx3S
1.12
0.92
1.11
1.64
4Bx4S
0.92
0.98
1.33
1.47
Kb = KRMI/2
4Bx6S
0.95
0.97
1.01
1.92
6Bx6S
1.10
1.01
2.00
Kx = 1.7
Kphi = 0.8
Ky = 1.0
Table 3.2.1.6b
Frame
AHDR
2Bx3S
0.76
U 5.0
0.82
0.69
0.86
Remarks
Loading
Gravity+ 15% Lateral
4Bx3S
0.70
0.99
0.83
0.81
6Bx3S
0.93
0.85
1.02
1.56
4Bx4S
0.79
0.89
1.24
1.40
Kb = KRMI/2
4Bx6S
0.81
0.89
0.94
1.83
1.89
Kx = Align. charts
Kphi = 0.8
Ky = 1.0
6Bx6S
0.95
1.30
0.94
43
In order to quantify the net effect of various approximations in the RMI procedure (i.e.
other parameters such as local buckling, holes in the column etc. as well), the results of
ALDR frame presented in previous section (Table 3.1.2.2.2) could be used. As noted
before, the net effect of all the approximations is to underestimate the frame strength by
40%-50%. However, as this observation is based on single frame analysis, a more
detailed study in this direction covering wider range of products is required to quantify
the actual effect of the approximations.
From the foregoing discussion it may be said that the RMI specification rules are not
adequate to model various parameters accurately. Though the numbers shown above as to
the unconsrvatism of the RMI are based on the preliminary study and need verification,
they do give approximate idea and it could be said that the net effect of approximations of
RMI will not be less than 30% with regard to strength estimates. This suggests the need
of a better design procedure which is developed in the following section.
44
3.2.2
The objective is to develop a more rational and accurate design procedure for CFS pallet
rack frames by making use of the existing design equations. The beam-column
interaction equations and moment amplification factors of the current AISI and AISC
specifications are used for this purpose. A total of five possible methods devised by
different combinations of these equations are experimented to arrive at the most
acceptable procedure. The details of the procedures are presented in table 3.2.2.1. It may
be noted that the effect of local buckling is not included in the study.
Table 3.2.2.1
Method
Method 1
P M
+
= 1. 0
Pn M n
(RMI-Proc.I)
Method 2
M = Mlin = Elin *P
= 1 / (1P/P ex)
(RMI- Proc.II
with modified Ks)
P
M
+
= 1. 0
2Pn M n
for
Method 3
(AISC proc. I)
Remarks
P
0 .2
Pn
P 8 M
+
= 1.0
Pn 9 M n
for
M = Mlin = Elin *P
= 1 / ( 1(P/H)(/L) )
P
> 0 .2
Pn
Method 4
(AISC proc.II)
Mu
0.99M u
0.75 M u
Ki
Ks
ROTATION
46
Frame
Type
Confg
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
2x3
0.68
0.12
0.89
0.11
0.88
0.46
0.85
1.42
0.84
1.16
4x3
0.71
0.13
0.94
0.12
0.90
0.77
0.86
2.36
0.85
2.14
6x3
0.78
0.14
1.03
0.13
0.99
1.28
0.90
3.07
0.90
3.07
4x4
0.85
0.15
1.11
0.14
0.98
1.32
0.87
2.42
0.86
2.19
4x6
0.92
0.16
1.21
0.15
0.91
1.32
0.75
1.85
0.75
1.69
6x6
0.95
0.17
1.25
0.16
0.90
1.73
0.74
2.42
0.74
2.17
2x3
0.80
0.15
1.06
0.13
1.05
0.43
1.04
0.78
1.04
0.78
4x3
0.72
0.13
0.96
0.13
0.96
0.73
0.94
6.26
0.93
5.6
6x3
0.76
0.14
1.02
0.13
1.00
0.99
0.93
4.14
0.93
3.55
4x4
0.66
0.12
0.87
0.11
0.83
1.69
0.72
6.7
0.72
5.62
4x6
1.05
0.19
1.40
0.18
1.03
3.66
0.79
5.77
0.79
5.37
6x6
1.04
0.19
1.40
0.12
0.97
2.09
0.74
5.19
0.74
2.91
( BxS )
A
H
D
R
U
3.0
2x3
4x3
0.70
0.07
0.93
0.06
0.93
0.34
0.93
1.12
0.92
0.86
6x3
0.73
0.13
0.97
0.12
0.97
1.04
0.93
3.85
0.93
3.71
3.5
4x4
0.78
0.14
1.03
0.13
0.96
0.89
0.89
1.50
0.88
1.41
4x6
0.94
0.17
1.26
0.16
6x6
0.94
0.17
1.25
0.16
2x3
0.72
0.13
0.95
0.12
0.95
0.25
0.95
0.39
0.95
0.39
4x3
0.66
0.12
0.87
0.11
0.87
0.29
0.87
0.62
0.86
0.56
6x3
0.67
0.12
0.89
0.11
0.89
0.33
0.89
0.91
0.88
0.81
5.0
4x4
0.71
0.13
0.95
0.12
0.90
3.42
0.91
2.03
0.91
2.03
4x6
0.71
0.13
0.95
0.12
0.89
0.83
0.71
3.13
0.71
2.91
6x6
0.93
0.16
1.25
0.16
0.95
2.33
0.79
3.35
0.79
2.89
47
Frame
Type
Confg
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
2x3
0.73
0.32
0.95
0.32
0.84
0.60
0.78
0.71
0.77
0.68
4x3
0.83
0.35
1.10
0.35
0.93
0.72
0.87
0.79
0.86
0.76
6x3
0.81
0.34
1.06
0.34
0.88
0.71
0.83
0.79
0.81
0.72
4x4
0.85
0.34
1.12
0.34
0.89
0.73
0.82
0.85
0.81
0.80
4x6
0.96
0.37
1.26
0.37
0.91
0.92
0.80
1.11
0.79
1.01
6x6
0.95
0.32
1.25
0.32
0.87
0.84
0.76
0.99
0.74
0.86
2x3
0.72
0.27
0.96
0.27
0.86
0.49
0.82
0.54
0.81
0.52
4x3
0.81
0.32
1.08
0.32
0.94
0.62
0.88
0.70
0.86
0.67
6x3
0.81
0.30
1.08
0.30
0.92
0.59
0.86
0.68
0.83
0.62
4x4
0.87
0.33
1.15
0.33
0.94
0.70
0.86
0.82
0.84
0.77
4x6
0.91
0.34
1.21
0.35
0.85
0.87
0.71
1.10
0.70
0.99
6x6
0.91
0.34
1.21
0.34
0.83
0.92
0.68
1.09
0.66
1.00
2x3
0.90
0.25
1.21
0.25
1.09
0.45
1.06
0.50
1.03
0.47
4x3
0.93
0.27
1.24
0.27
1.09
0.54
1.05
0.60
1.02
0.55
6x3
0.94
0.27
1.26
0.28
1.07
0.54
1.02
0.61
1.00
0.56
3.5
4x4
0.92
0.31
1.23
0.32
1.02
0.67
0.96
0.78
0.94
0.71
4x6
0.98
0.36
1.30
0.36
0.95
0.92
0.84
1.06
0.83
0.99
6x6
1.01
0.37
1.34
0.38
0.93
0.96
0.83
1.16
0.81
1.05
2x3
0.70
0.30
0.94
0.30
0.88
0.47
0.85
0.56
0.83
0.52
4x3
0.74
0.30
0.99
0.30
0.90
0.51
0.86
0.61
0.84
0.57
6x3
0.72
0.33
0.95
0.33
0.87
0.57
0.82
0.68
0.79
0.63
5.0
4x4
0.75
0.30
1.00
0.30
0.89
0.54
0.83
0.68
0.80
0.62
4x6
0.82
0.32
1.09
0.32
0.92
0.69
0.72
0.92
0.76
0.87
6x6
0.94
0.32
1.25
0.32
0.91
0.86
0.80
0.99
0.79
0.90
( BxS )
A
H
D
R
U
3.0
48
Table 3.2.2.4
Frame
Type
Confg
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
2x3
0.78
0.49
1.03
0.50
0.89
0.79
0.84
0.88
0.82
0.82
4x3
0.79
0.46
1.03
0.47
0.87
0.77
0.83
0.83
0.79
0.75
6x3
0.83
0.48
1.09
0.50
0.90
0.82
0.84
0.86
0.82
0.80
4x4
0.85
0.47
1.11
0.49
0.88
0.82
0.82
0.90
0.80
0.84
4x6
0.95
0.49
1.22
0.51
0.89
0.93
0.80
1.03
0.78
0.94
6x6
0.92
0.46
1.20
0.47
0.85
0.87
0.76
0.97
0.74
0.88
2x3
0.78
0.43
1.04
0.44
0.92
0.67
0.88
0.73
0.86
0.69
4x3
0.79
0.40
1.04
0.41
0.89
0.65
0.84
0.72
0.81
0.65
6x3
0.85
0.48
1.11
0.50
0.94
0.80
0.88
0.86
0.86
0.80
4x4
0.84
0.40
1.10
0.41
0.89
0.69
0.83
0.86
0.81
0.71
4x6
0.96
0.48
1.26
0.50
0.91
0.92
0.78
1.07
0.75
0.93
6x6
0.98
0.49
1.29
0.51
0.91
0.97
0.76
1.07
0.74
0.97
2x3
0.83
0.36
1.11
0.38
0.99
0.57
0.95
0.59
0.92
0.56
4x3
0.88
0.44
1.15
0.45
0.98
0.72
0.95
0.75
0.92
0.71
6x3
0.90
0.38
1.17
0.40
0.99
0.64
0.96
0.68
0.92
0.62
3.5
4x4
0.89
0.38
1.17
0.40
0.96
0.67
0.91
0.72
0.89
0.67
4x6
0.98
0.50
1.29
0.52
0.95
0.96
0.85
1.04
0.83
0.95
6x6
0.98
0.48
1.29
0.50
0.92
0.92
0.83
1.02
0.81
0.93
2x3
0.70
0.45
0.92
0.47
0.86
0.65
0.82
0.73
0.79
0.68
4x3
0.72
0.40
0.95
0.41
0.86
0.60
0.81
0.67
0.79
0.62
6x3
0.75
0.41
1.00
0.43
0.87
0.63
0.84
0.72
0.81
0.66
5.0
4x4
0.83
0.44
1.11
0.46
0.97
0.70
0.90
0.80
0.88
0.75
4x6
0.91
0.42
1.20
0.44
0.99
0.73
0.86
0.89
0.84
0.81
6x6
0.95
0.47
1.25
0.49
0.92
0.91
0.82
1.02
0.80
0.93
( BxS )
A
H
D
R
U
3.0
49
Table 3.2.2.5
Frame
Type
Confg
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
Pr
Mr
2x3
0.90
0.85
1.03
0.89
0.99
1.01
0.96
1.03
0.89
0.93
4x3
0.81
0.86
0.89
0.89
0.87
0.99
0.86
1.00
0.80
0.92
6x3
1.13
0.81
1.30
0.84
1.22
0.99
1.17
0.99
1.10
0.90
4x4
0.93
0.83
1.04
0.88
0.98
1.00
0.94
0.99
0.88
0.91
4x6
0.95
0.86
1.07
0.89
0.96
1.02
0.92
1.02
0.86
0.94
6x6
1.11
0.82
1.25
0.86
1.12
0.98
1.07
0.93
1.00
0.91
2x3
0.91
0.83
1.05
0.88
1.02
1.00
1.00
1.00
0.92
0.91
4x3
1.09
0.79
1.24
0.83
1.21
0.96
1.18
0.97
1.09
0.87
6x3
0.92
0.85
1.05
0.90
1.01
1.02
0.99
1.03
0.91
0.94
4x4
0.99
0.88
1.12
0.92
1.05
1.04
1.03
1.07
0.95
0.96
4x6
0.98
0.88
1.10
0.92
0.99
1.05
0.92
1.05
0.87
0.97
6x6
1.42
0.84
1.59
0.88
1.42
0.99
1.32
1.00
1.25
0.93
2x3
0.94
0.85
1.08
0.89
1.05
1.02
1.04
1.03
0.96
0.93
4x3
0.90
0.82
1.03
0.86
0.99
0.98
0.98
0.99
0.90
0.89
6x3
1.12
0.79
1.28
0.83
1.23
0.95
1.20
0.95
1.11
0.86
3.5
4x4
1.34
0.76
1.55
0.81
1.45
0.92
1.43
0.94
1.32
0.85
4x6
1.02
0.86
1.14
0.89
1.04
1.04
0.99
1.03
0.92
0.95
6x6
1.01
0.86
1.14
0.89
1.02
1.06
0.97
1.03
0.91
0.94
2x3
0.74
0.83
0.86
0.88
0.85
0.98
0.83
1.00
0.77
0.91
4x3
0.86
0.81
0.99
0.86
0.97
0.97
0.95
0.99
0.88
0.89
6x3
1.65
0.73
1.90
0.77
1.86
0.87
1.81
0.89
1.67
0.80
5.0
4x4
1.49
0.73
1.71
0.78
1.67
0.88
1.58
0.88
1.48
0.81
4x6
1.93
0.76
2.23
0.80
2.09
0.91
1.96
0.93
1.81
0.84
6x6
2.01
0.78
2.28
0.82
2.06
0.94
1.98
0.96
1.83
0.86
( BxS )
A
H
D
R
U
3.0
50
Table 3.2.2.6
The ratios reported are with respect to the axial load capacity : (Pr = Pcal / Pfem )
Loading
Gravity
Gravity+
0.75% lat.
Gravity+
1.5% lat.
Gravity+
15% lat.
Table 3.2.2.7
Statistics
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Min
Max
Avg
Min
Max
Avg
Min
Max
Avg
Min
Max
Avg
0.66
1.05
0.80
0.72
1.01
0.85
0.70
0.98
0.86
0.74
2.01
1.13
0.87
1.40
1.06
0.94
1.34
1.14
0.92
1.29
1.13
0.86
2.28
1.29
0.87
1.05
0.93
0.83
1.09
0.92
0.85
0.99
0.92
0.87
2.09
1.21
0.71
1.04
0.86
0.68
1.06
0.85
0.76
0.96
0.85
0.86
1.98
1.17
0.71
1.04
0.85
0.66
1.03
0.83
0.74
0.92
0.82
0.80
1.83
1.09
The ratios reported are with respect to the moment capacity : (Mr = Mcal / Mfem )
Loading
Gravity
Gravity+
0.75% lat.
Gravity+
1.5% lat.
Gravity+
15% lat.
Statistics
Method - 1
Method -2
Method - 3
Method - 4
Method - 5
Min
Max
Avg
Min
Max
Avg
Min
Max
Avg
Min
Max
Avg
0.12
0.19
0.14
0.25
0.37
0.32
0.36
0.50
0.44
0.73
0.88
0.82
0.06
0.18
0.13
0.25
0.37
0.32
0.38
0.51
0.46
0.77
0.92
0.86
0.25
3.66
1.25
0.49
0.96
0.69
0.57
0.96
0.77
0.87
1.06
0.98
0.39
6.26
2.82
0.54
1.16
0.81
0.59
1.07
0.85
0.88
1.07
0.99
0.39
5.62
2.47
0.47
1.05
0.74
0.56
0.97
0.78
0.80
0.97
0.90
51
While the results presented in the above tables are based on preliminary investigation, it
appears at this stage that the AISC procedure with Kx = 1.0; Kphi =0.8 in conjunction with
either initial joint stiffness (method 3) or with secant joint stiffness (method 4) may be
used to predict the pallet rack frame capacities with sufficient accuracy.
4
The following objectives have been set for the future study.
The effect of beam end shear and the ratio of the shear to moment on the beam-tocolumn joint stiffness by modeling the joints more accurately. A more refined finite
element modeling including the endplates, tabs and actual hole shape is required for this
purpose. Such a study will shed light on several important topics as
the correlation of the results from a cantilever test and a portal test
behavior under reversal of loading on the joint (such as the case when the
beams are subjected to reverse curvature).
The behavior of pallet rack frames with semi-rigid beam-column joints and flexible
column bases was studied by experimental and numerical (FEM) investigations. A
general M relationship was established to model the beam-to-column joint stiffness
of pallet rack frames. The column base flexibility of rack frames caused by base plate
bending was studied in detail and quantified in terms of the current specification value.
Guidelines were framed to carry out nonlinear FE analysis of rack frames accounting for
various influencing parameters.
52
A critical review of the current American rack frame design specification (RMI 1997)
was carried out and found to be conservative by about 30% with regard to strength
estimates. The sources of conservatism in the specification were identified and an attempt
was made to improve the same. A few suggestions were made to deal with the buckling
and nonlinear behavior of frames in a more rational way. These include modified
effective length factors, beam-column interaction equation and moment amplification
factor. The use of Kx = 1.0; Kphi =0.8 as effective length factors in conjunction with the
AISC (or AISI) beam-column design procedure and AISC B2 -fator was found to give the
most acceptable frame strengths. Plans for the future work were discussed.
53
References
Bibliography
1. ASCE (1997) Effective length and Notional load approaches for Assessing Frame
Stability : Implications for American Steel Design, Special publication of ASCE
(Task Committee on Effective length factor).
2. Chen, W. F. (1985) Connection Flexibility and Steel Frames, Proceedings of a
session sponsored by structural Division of the ASCE in conjunction with the ASCE
Convention in Detroit, Michigan.
3. Chen, W. F. and Atsuta, T. (1975) Theory of Beam-Columns Volume 1: In-plane
Behavior and Design, Advanced Book Program, Mc Graw Hill Book Company.
4. Chen, W. F. and Atsuta, T. (1976) Theory of Beam-Columns Volume 2: Space
Behavior and Design, Advanced Book Program, Mc Graw Hill Book Company.
5. Chen, W. F. and Lui, E. M. (1991) Stability Design of Steel Frames, CRC Press.
6. Christopher, J. E., and Bjorhovde, R. (1999) Semi-rigid Frame Design Methods for
Practicing Engineers, Engineering Journal AISC, Vol.36, No. 1, 12-28.
7. Galambos, T. V. (1960) Influence of Partial Base Fixity on Frame Stability, Journal
of Structural Engineering ASCE, Vol. 86, No. ST3, May 1960.
8. Lau, S. M., Kirby, P. A. and Davison, J. B. (1999) Semi-rigid Design of Partially
Restrained Columns in Non-sway Steel Frames, Journal of Constructional Steel
Research, Vol.50, 305-328.
9. Narayanan, R. (1985) Steel Framed Structures : Stability and Strength, Elsevier
Applied Science Publishers.
54
10. Ronald D. Zieman, William McGuire and Gregory G. Dierlein (1992) Inelastic Limit
States Design. Part I: Planar Frame Studies, Journal of Structural Engineering
ASCE, Vol. 118, No.9, 2532-2549.
11. Ronald D. Zieman, William McGuire and Gregory G. Dierlein (1992) Inelastic Limit
States Design. Part II: Three-Dimensional Frame Study, Journal of Structural
Engineering ASCE, Vol. 118, No.9, 2550-2568.
12. Salmon, C. G., Schenker, L. and Jhonston, B. G. (1955) Moment-Rotation
Characteristics of Column Anchorages, Proceedings of ASCE, Vol. 81, No. ST3,
April 1955.
13. White, D. W., and Clarke, M. J. (1997) Design of Beam-columns. I: Philosophies
and Procedures, Journal of Structural Engineering ASCE, Vol. 123, No.12, 15561564.
14. White, D. W., and Clarke, M. J. (1997) Design of Beam-columns. II: Comparison of
Standards, Journal of Structural Engineering ASCE, Vol. 123, No.12, 1565-1575.
55
APPENDIX
Effect of Nonlinear base stiffness on frame strength
As shown in the report (sec 3.1.1.2), the column base stiffness of a typical rack frame is
nonlinear in nature. Hence, the corresponding moment-rotation relationship through
complete load history is required for proper consideration of its effect on the behavior
and failure strength of a frame. However, before developing such an expression, it is
deemed to quantify the effect of nonlinear base stiffness on the frame strength as
compared to the capacity of the frame with linear base stiffness. This study enables to
assess the actual need of a nonlinear expression for base stiffness.
For this purpose, a large number of pallet rack frames of different types are analyzed by
FEM for two base stiffness characteristics: Linear and Nonlinear. The details of the study
are discussed below.
Parameters
Types of Frames
Frame configuration
Loading
4 types
6 types
4 types
This resulted in FE analyses of 192 rack frames (96 in each of the two base stiffness cases).
FE modeling features : Used beam element model with the following features.
Shelf Beams
Columns
Column perforations
Beam-column joints
Column base fixity
Linear column base
for
for
for
The nonlinear base stiffness as expressed by these expressions is compared with the
actual moment-rotation relationship (as obtained by FEA) in the following figure.
Base Fixity RMI vs FEM (7x7x0.375)
K_RMI
Approx. (nonlinear)
FEM
Approx. (linear)
50
Moment (kips-in)
40
30
20
10
0
0
0.01
0.02
0.03
0.04
57
APPENDIX
58
= bd2 /1440) and Kx = 1.5, Kphi = 1.5 for hinged base. The results were found to be
accurate enough for all practical purposes.
It was also found, based on parametric study (the base stiffness as the variable) on AHDR
frames of two configurations (2Bx3S and 4Bx4S), that the effective length factors Kx =
1.0, Kphi = 1.0 could be used to predict the frame strength accurately when the base
stiffness is in the range of fixed to 1/10th of the RMI value. As the base stiffness
approaches hinged case the effective length factors are found to be increased to Kx = 1.5,
Kphi = 1.5 in order to estimate the frame strengths accurately.
For column base fixities between hinged and 1/10 th of RMI value, a relationship
between the column base stiffness and the effective length factors needs to be established.
However, it was deemed to quantify the practical range of column base stiffnesses before
taking up further work in this direction.
Objectives
To evaluate the performance of the Notional load approach (considering it as a required
or an acceptable alternative procedure) in the case of frames with practical column base
fixities and other load combinations that have not been covered so far.
59
INTERIM REPORT
By
Kotha S. Rao, Ph.D.
Professor Teoman Pekz
(Proncipal investigator)
October 6, 2000
60
INTERIM REPORT
By
Kotha S. Rao, Ph.D.
Professor Teoman Pekz
(Proncipal investigator)
October 6, 2000
61
CONTENTS
1
INTRODUCTION
3.1
Analysis
3.2.1
3.2.2
3.2.2.2
3.2
Stub column
16
18
Design
3.2.1
3.2.2
2
3
6
13
13
13
14
23
34
35
36
37
37
45
52
52
REFERENCES / BIBLIOGRAPHY
54
APPENDIX
55
62