You are on page 1of 11

STRENGHT REDUCTION FACTORS FOR STORAGE RACKS

Carlos Aguirre A.
Federico Santa Maria University, Valparaíso, Chile
carlos.aguirre@usm.cl

ABSTRACT

Present Seismic Codes include Strength Reduction Factors and Displacement


Amplification Factors. In this paper, they were obtained from non linear analysis of
some typical rack structures subjected to selected earthquake records. The moment
rotation relationship for a typical connection was obtained in a previous experimental
research work performed at Federico Santa Maria University Laboratory.

The results show that drift displacements became much larger than the limits
prescribed in codes, it follows the design is normally controlled by displacements.
The Strength Reduction Factor obtained from the analyses are smaller than the
factor prescribed in the Chilean and in UBC Earthquake Code, which means that the
design under those codes factors, could become unsafe in some cases.

BACKGROUND

Storage racks are extremely flexible structures, especially under the action of lateral
loads. The behaviour in seismic zones is strongly dependent on the properties of the
connections; they determine the behaviour of the structure and the performance of the
structural system in the event of a destructive earthquake occurrence.

Rack structures in seismic zones are requested to comply with the local building codes,
so they must be engineered to meet the code requirements of the building structures.
Even though rack structures are quite different to buildings they use to be placed inside
a building, so it is necessary to control the lateral deflections in order to avoid the
hammering with the surrounding structure, and eventually the collapse of both
structures. The storage rack structure studied in this paper has been used successfully
in Chile for several years and they have survived the March 3, 1985 Chilean
earthquake. The beam is connected to the column by using hooks that are fabricated
with the beam; these hooks are inserted into columns slots, so they can be easily
disconnected from the column. Details and connecting elements are shown in Fig. 1, the
thickness of the elements is usually 2 mm. In Figure 1a, there is a detail of the beam to
column connection and in figure 1b is shown the curve of the joint subjected to a cyclic
load, obtained in a previous research work at Santa Maria Laboratory (Irisarri, 1998) and
a characteristic curve to be used in non linear analysis of the rack structures. The details
of these findings were presented in a previous paper during the Connections-V
Workshop held in Amsterdam on 2004.

Most of the seismic codes define the earthquake loads in terms of reduced spectra. The

AISC-ECCS Rack Structures – Aguirre C. March 2004 Page 1 of 11


reduction factor depends on the structural ductility and over strength properties of the
structure. Figure 2 shows the different kinds of behaviour, the actual one is the real
response of the structure (it is also shown an equivalent idealized bilinear model) and
the typical linear analysis performed at engineering offices.

2500

2000

1500
R2
1000

Moment [N-m]
R1
500 R 1

0
-0,03 -0,02 -0,01 R1 0 0,01 0,02 0,03 0,04 0,05 0,06 0,07
R1 -500

-1000

R2
-1500

-2000
Rotation [radians]

(a) Connection (b) Moment – Rotation relationship

Figure 1 – Test conducted at Santa Maria University

Q Even though this approach


simplifies the analysis, it does not
allow to estimate the actual
Linear behavior of the structure. For these
Qmax Behavior reason, the codes used to define
two factors, the Strength Reduction
Factor (R Factor) and the
Actual Behavior Displacements Amplification Factor
(Cd Factor, is the name given at
Qy NEHRP, 2001).
Idealized
Behavior
Qe PRELIMINARY ANALYSES

It was selected a typical storage


rack and two heights: three and six
Δe Δ y Δmax Δu Δ storage levels, that means 4.5 m
and 9 m respectively. The height
Figure 2 - Behavior of the Structure
limit establish for the Rack
Manufacturers Institute (2005), is 10 m for Racks constructed on public places. There
were conducted some preliminary elastic analyises, assuming standard support
conditions, that means the base of the columns were modelled as simple supports,
simple supports plus beams 20 cm above the ground and fixed supports. The
standard weight of the pallets is 1000 Kgf, so it is possible to have 2000 Kgf in each

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 2 of 11


level. A summary of the periods obtained from those analyses is presented on Table
1. It can be seen that storage racks are extremely flexible structures, with periods
about 1 to 3 seconds and larger in higher racks, which means there is a need to have
a special care to avoid hammering with the surrounding structure.

Table 1 – Period of Storage Racks


Structure Period ( seconds)

Three levels frame 0.94 – 1.18 – 1.50

Six levels Frame 2.07 – 2.34 – 2.74

PROCEDURE

There were conducted some elastic and non linear analyses on each structure under
the same seismic load. In a first step the earthquake demand was determined,
dynamic non linear analyses of the structures were conducted, under some selected
ground motion records ocurred during the last 25 years, the characteristics of the
records are shown on Table 2.

Table 2 - Characteristics of the records


Richter Mercalli Duration
Earthquake Year Component Amax
Magnitude Intensity (s)
Viña del Mar 1985 S20W 7.8 VI – VII 116 0.35 g
Llo-Lleo 1985 N10E 7.8 VIII 116 0.67 g
México 1984 EW 8.1 VIII – IX 62 0.17 g
Northridge 1994 Sylmar EN 6.8 VIII – IX 60 0.84 g
Kobe 1995 JMA N-S 6.9 VIII – IX 48 0.82 g

(a) Simple support (b) Beam above the supports (c) Fixed supports

Figure – 3 Frame geometry and column base support

The geometry and support conditions for the 3 levels rack is shown on figure 3. The

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 3 of 11


same conditions were applied for the six levels rack (9 meters heigth), so 6 structures
were analyzed. The section properties are shown on Table 3.
In a second step, the capacity of each structure was determined, two approaches
were applied: push over analyses first and non linear dynamic analyses under scaled
records after.

Table 3 – Section Properties


GEOMETRIC PROPERTIES
Dimension Column Beam
B [mm] 100.0 21.5
H [mm] 100.0 100.0
e [mm] 3.0 2.0
A [cm2] 10.0 8.03
Ix [cm4] 138.04 122.53
Iy [cm4] 129.81 28.51
3
W x [cm ] 27.61 24.51
3
W y [cm ] 22.23 10.93
Weight [Kg/m] 7.84 6.28

EARTHQUAKE DEMAND

Lateral Elastic Displacements - The lateral displacements envelopes when the


behaviour is elastic are shown in Figure 4.

900

800

700

600
Altura (cm)

500

400

300

200

100

0
0 50 100 150 200
Desplazamiento
Displacement [cm] (cm)
Viña del Mar Llo-lleo Northridge Kobe México

Figure 4 – 6 Levels rack displacements envelopes – Fixed Support at Base

The maximum displacement for the six levels rack was obtained with the Mexico
earthquake record and it is 18% of the height of the rack structure; this figure is much
larger than the limit prescribed at the Chilean NCh 2369 (INN, 2002) and the RMI

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 4 of 11


(RMI, 2005) Standards. For the less flexible three level racks, the maximum
displacement was obtained with Northridge or Llolleo earthquake records. All the
drift values obtained from analysis with different ground motion record exceed the
maximum drift limit (0.015H) stipulated by the Chilean Code.

Figure 5, shows the displacement response at the top of the rack for the simple
supported frames under the Chilean Earthquake record obtained in Viña del Mar city.
The same figure shows the maximum allowable drift according to the Chilean NCh
2369 Code. In both cases the maximum drift is exceeded, but due to the frequencies
content the three levels rack is more sensitive to the Viña del Mar earthquake record.

6.0%
Deformación sísmicaDrift
Maximum Seismic máxima
Desplazamiento (%H)

4.0% NCh NCh2369 Of2002


2369 Of.2002
2.0%

0.0%

-2.0%

-4.0%

-6.0%
0 10 20 30 40 50 60
Tiempo (s)
Time (sec)
Viña viga
Apoyo del Mar – Simple
3 pisos - ViñaSupport
del Mar3 levels – Simple
Apoyo viga 6Support 6 levels
pisos - Viña del Mar

Figure 5 – Displacements at the top level

The maximum drift, for every structure subjected to different earthquakes are
presented on Figure 6.

25%
Sismo Kobe
Kobe
Kobe
Sismo Llolleo
Llolleo
Llolleo
20% Mexico
Sismo Mexico
Drift (%H) (% H)

México
Nortridge
Sismo Northdrige
Viña del Mar
Northridge
Desplazamientos

15% Sismo Viña del Mar

10%

5%

T = 0.94 (s) T = 1.18 (s) T = 1.50 (s) T = 2.07 (s) T = 2.34 (s) T = 2.74 (s)
0%
E3P

Viga

SA 3

E6P

Viga

SA 6
BB6

SS6
FB6
SS3
BB3
FB3

Figure 6 – Maximum drift at the top level

The structures are presented from small to large periods, fixed base for 3 levels

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 5 of 11


(FB3) has the smallest period and simple support for 6 levels (SS6) has the largest
period.

Bending Moments and Rotations on the Elastic Connection – The envelopes of


moment in connections shows that the strength capacity of the connection (20000
Kg-cm) is exceeded with all the records. The rotation capacity of the connection (0.03
radians) is also exceeded with all the ground motions, except with the Mexico
earthquake.

Non Linear Dynamic Analysis. In order to consider the non linear behaviour, it was
necessary to include the non linear properties of the connection. The moment –
rotation relationship, obtained from cyclic tests by Irisarri (1998) is shown on Figure
7. A tri-lineal model can be used. There is an initial gap due to the fitting of the hooks
into the slots. The change in the curve slope is because of the yielding of the first
hook. Two equivalent models were explored, one of them is by adding a multi linear
plastic element between the column and the beam, another possibility is to include a
hinge in the beam.
The program does
not accept the
same value in
Moment (Kgf-cm)

different points of
the curve, so it
was necessary to
provide a small Column
slope near the
origin (10 kg-cm Beam
moment when the
rotation is 0.002
radians), to the Rotation (rad)
initial gap.
Figure 97– Moment rotation relationship

Non Linear Displacements.


900
Envelopes for the lateral
800
displacements are shown on
700
Fig. 8. The maximum
600
displacement for the six levels
Altura (cm)

500
rack with beam at the bottom
Height (cm)

400
occurred with Mexico record.
300
This earthquake is the worst
200
condition for the 9 levels rack.
100
The non linear drift is about 3%,
0
which is also larger than the 0 5 10 15 20 25 30 35
codes limits (INN, 2002; RMI, Desplazamiento
Displacement (cm)(cm)
1997). Kobe, Northridge and Viña del Mar Llo-lleo Northridge Kobe México
Mexico produce the collapse of
Figure 108 – Displacement Envelope – 9 meters-
the racks. Envelopes of rack. Beam at the base
displacements and base shears
are shown on Fig. 9.

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 6 of 11


0.45
Sismo Kobe
Sismo Llolleo
0.40
Sismo México
0.35 Sismo Northdrige
Sismo Viña del Mar

basal(Q/P)
(V/P)
0.30

0.25
Shear
0.20
Corte

0.15
Base

0.10

0.05
T = 0.94 (s) T = 1.18 (s) T = 1.50 (s) T = 2.07 (s) T = 2.34 (s) T = 2.74 (s)
0.00

BB6
FB6

SS6
BB3
FB3

SS3
(a) Lateral Displacements
6%
Sismo Kobe
Sismo Llolleo
5%
Sismo Mexico
(% H) (% H)

Sismo Northdrige
4%
Desplazamientos

Sismo Viña del Mar

3%
Drift

2%

1%

T = 0.94 (s) T = 1.18 (s) T = 1.50 (s) T = 2.07 (s) T = 2.34 (s) T = 2.74 (s)
0%
BB6
FB6

SS6
BB3
FB3

SS3

(b) Base Shear


Figure 9 – Displacements and Base Shear Envelopes
CAPACITY OF THE STRUCTURES
6000
Pushover Analyses.
Triangular and uniform 5000
lateral load pattern were
(kgf)

applied to the structures 4000


(Kg-f)
Basal

above mentioned. SAP 3000


2000 and connection
Shear
BaseCorte

Irisarri’s bilinear model 2000


(1998) were used. In
1000
this model, the yielding
starts at θy=0.012 0
radians and the ultimate 0 2 4 6 8 10 12 14 16
Desplazamiento (cm)
rotation is θu=0.03. The Displacement (cm)
plastic moment is 2000 Simple Support
Simplemente Apoyado Simple Support
Simplemente and
Apoyado conBase
Viga aBeam Fixed
20 cm de la base. Empotrado

Kg-cm.
Figure 10 – Pushover Curves

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 7 of 11


Typical shear displacements curves for the six level racks, considering the 3 support
conditions at the base are shown on Fig. 10. A summary of the largest displacements
and base shear are given on Table 4.

odel
Table 4 – Maximum Shear and Displacements

SUPPORT CONDITION Δmax (cm) %H Qy (kgf)


Simple Support (4.5 m) 15.20 3.3 % 2150
Beam at the Base (4.5 m) 15.46 3.4 % 2850
Fixed (4.5 m) 12.6 2.8 % 4180
Simple Support (9.0 m) 22.5 2.50 % 1910
Beam at the Base (9.0 m) 25.9 2.88 % 2560
Fixed (9.0 m) 26.6 2.95 % 3200

Scaled Records – A second approach to determine the capacity was by scaling the
amplitudes of the ground motion records, as far as the collapse or the structure or the
0.03 radians rotation 40
(cm)

limit is reached in some


Desplazamiento (cm)

30
connection. In the case 20
Displacement

of Kobe, Northridge and 10


Mexico was not 0
necessary to increase
-10
the amplitude of the
-20
ground motion record.
-30
There were performed 0 10 20 30 40 50 60
30 analyses (6 Tiempo (s)
Time (sec)
structures, 5 records),
Sismo
Llolleode Llo-lleo
1.5 (SismoSismo
1.5 de Llo-lleo
Llolleo)
11 of them did not
reach the ultimate Figure
Fig 14
11––Time
ure 13 Timehistory
history response
response forfor Llolleo
Llolleo record
condition and 10 of record
on onofthe
the top thetop of therack
9 meters 9 meters rack
them belong to Chilean
records. The only exception was the 3 levels rack with fixed base under the Mexico
earthquake. Fig.11 shows one of the cases processed with the Chilean Earthquake.

DESIGN FACTORS

In order to obtain the design R and Cd factors, it is necessary the following


definitions, associated to Fig. 2:

Δu Rμ =
Qmax Qy
μ= Ω=
Δy Qy Qe

Ductility Ratio - Strenght Factor due to Ductility - Over strength Factor

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 8 of 11


1. – Displacement Amplification Factor. – It was obtained from the pushover
analyses, the results and the Displacement Amplification Factor [Cd], according to
equation (4) (Uang, 1991) are shown in Table 5.

Cd = μ ⋅ Ω (4)

Table 5 – Displacement Amplification Factor


Height (m) Over Strenght Factor (Ω) Ductility Ratio (μ) Cd
3 Levels 1.85 1.90 3.49
6 Levels 1.60 1.82 2.91

NEHRP Recommended Provisions (BSSC, 2001) suggest Cd=3.5 for Rack


Structures, which seems to be appropriate for the 3 levels rack, for taller racks is less
which is a conservative approach.

2. – Strenght Reduction Factor. – It was determined by two approaches: Pushover


Analyses and non Linear Analyses with Scaled Records. It can be shown the
following equation can be used to determine the Strength Reduction Factor [R]

R = Rμ ⋅ Ω (5)

2.1. – Pushover Analyses Results. The average R factor for the 3 levels rack was
2.32, for the 6 levels rack the average was 2.83; the codes normally required a value
around 4. The Over Strength is smaller for larger racks but the ductility ratio is larger
for large racks.

2.2. – Scaled Non Linear Analyses Results. In most of the cases the ultimate
strength condition of the structure was obtained with the original record. In those
cases where there was no failure, the record was scaled until the structure reached
the ultimate
condition.
A summary of
results is
presented on
Fig. 12 for
R Factor

both types of
analyses. It
can be seen
that Pushover
R Factors are
smaller than R Period (sec)
factors Non Linear Dynamic Analyses Pushover Analyses
obtained with
non linear Figure 12 – Strength Reduction Factors
analysis.

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 9 of 11


There are several reasons that explain the differences: first under monotonic loading
the plastic moment reaches by the connection is 20000 Kg-cm larger than the plastic
moment obtained under cyclic loading, which is 17683 Kg-cm, because of the cyclic
deterioration of the plastic moment.

CONCLUDING REMARKS

1. The average displacement produced for Chilean Earthquakes is 4% of the


rack height. México (1985), Northridge (1994) and Kobe (1995) earthquakes
produce an average displacement 10% of the total rack height. Those figures
are larger than the limits of the codes, so the design is controlled by the lateral
deflections.
2. Chilean Earthquakes do not take the rack to the strength limit condition that
means the connections never reach the maximum rotation (0.03 radians).
3. No matter the R factor value, the lateral displacements obtained from time
history exceed the limit values. As a consequence displacements control the
seismic design. P – Δ effects should be included in the analysis.
4. The Average Base Shear reaches 40% of the weight of the structure.
5. The Strength Reduction Factor obtained with Chilean earthquakes is about 2,
which is half of the maximum R factor specified in the Chilean code. More
destructive earthquakes give larger R factors. The factor is sensitive to
earthquake characteristics and to the structure properties.
6. The Chilean Codes don’t specify an R value for rack storage structures. The
American “NEHRP Recommended Provisions for Seismic Regulations for New
Buildings and Other Structures, 2001” suggest Cd= 3.5. The analyses show
that Cd becomes 3.49 for 3 levels racks and 2.91 for six levels racks.
7. It seems appropriate to include a beam above the support because it reduces
the lateral displacements and improves the dissipation mechanism, because
of the larger redundancy.

REFERENCES

ƒ Building Seismic Safety Council (2001) – “NEHRP Recommended Provisions for


Seismic Regulations for New Buildings and Other Structures”, USA
ƒ Instituto Nacional de Normalización INN (2002) – “Diseño Sísmico de Estructuras e
Instalaciones Industriales” Norma Chilena NCh 2369-Of.2002
ƒ Irisarri Harding, Alejandro (1998) – “Conexiones Semirígidas en Estructura de Acero
y su influencia en el Análisis”. Memoria Para Optar al Titulo de Ingeniero Civil,
Departamento de Obras Civiles, Universidad Técnica Federico Santa María,
Valparaíso, Chile. .
ƒ Rack Manufactures Institute RMI (1997) – “Specification for the Design, Testing and
Utilization of Industrial Steel Storage Racks”, USA
ƒ Uang C. M (1991). – “Establishing R (or Rw) and Cd Factors For Building Seismic
Provisions”. Journal of Structural Engineering, ASCE, 117(1), 19-28.

Key words: Steel Design, Rack Structures, Non Linear Behaviour, Semi-Rigid

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 10 of 11


Connections.

AISC-ECCS Rack Structures – Aguirre C. March 2008 Page 11 of 11

You might also like