Content uploaded by Arsen Adzhemyan
Author content
All content in this area was uploaded by Arsen Adzhemyan on Jun 26, 2018
Content may be subject to copyright.
THE USE OF EXPERIMENTAL RESULTS
IN SEISMIC ISOLATION DESIGN
Arsen ADZHEMYAN1, Gianmario BENZONI2, Giuseppe LOMIENTO3
ABSTRACT
The extensive results collected from prototype tests of Seismic Isolators and Energy Dissipators are only
partially utilized to support structural models and design procedures, which often rely on a number of
assumptions and simplifications that may significantly affect the prediction of the structural performance. It must
be recognized that a major difficulty for engineers is the translation of results of lengthy and expensive
experimental programs into design tools. Tests are, in fact, often proposed for a "macroscopic" validation of
devices’ performance. The experimental approach is somehow still evolving from what was traditionally
required for elastomeric bearings, with the attempt to "adapt" requirements to devices of completely different
concepts. Acceptance criteria enforced by codes often show lack of consistency with design approaches and
goals. Examples will be proposed to support the need for ad-hoc designed testing protocols that can provide clear
inputs to improved analytical tools.
Keywords: Keyword1; Keyword2; Keywords should use Times New Roman 10 pt. font, Italic; separated by
semicolon; Maximum 5
1. INTRODUCTION
Device testing has played a major role in the development, design, and acceptance of seismic isolation
since the time of implementation of earliest systems. The first building code incorporating regulatory
requirements for seismic isolation is the 1991 Uniform Building Code (UBC) (ICBO 1991). Prototype
testing were specified in the code to assess the isolation devices’ performance in regards of the
expected maximum seismic demand. The original prototype testing were based upon elastomeric
isolators, the most common technology available at that time. Since then, a number of isolation
devices have been proposed, relying on different functional mechanisms, with the most common being
today elastomeric bearings and friction isolators. On a parallel effort, several facilities with extensive
testing capabilities have been instrumented, in order to test actual devices to be used in construction
under the excepted seismic rates of loading. The availability of extensive experimental evidence from
full scale testing on seismic isolators determined the evolution in the design of modern seismic
isolation systems.
The experimental variability of the mechanical performance of seismic isolation devices was early
recognized. This evidence was found so relevant that upper and lower bound values of the isolators’
properties were soon required to be used when analyzing such systems in an attempt of allowing the
use of simplified structural analysis based on nominal viscoelastic property, or simplified bilinear
plastic models, rather than complex nonlinear models. The 2nd Edition of the AASHTO Guide
Specification for Seismic Isolation Design (AASHTO 1999) introduced the use of property
modification factors (λ) to account for this variability in the structural analysis of seismic isolation
systems. Cyclic motion, loading rate, variability in production, temperature, aging, environmental
1MS Civil Engineering, California State Polytechnic University, Pomona, USA, arsenadzhemyan@yandex.ru
2Research Scientist, Structural Engineering, University of California, San Diego, USA, benzoni@ucsd.edu
3Contacting author: Asst. Professor, Civil Engineering, California State Polytechnic University, Pomona, USA,
glomiento@cpp.edu
exposure and contamination were all considered as significant source of performance variability.
According to the AASHTO, nominal properties in the design process of typical isolation systems
needed to be based on prior prototype test results provided by manufacturers, while lower and upper
bound values of the same properties are then obtained by multiplying nominal values by minimum
(λmin) and maximum (λmax ) isolator-dependent property modification factors. Structural simulations
are finally performed with lower and upper bound values to encompass the entire expected
performance variability with simplified numerical models.
When the International Building Code (IBC) (ICC 2000) replaced the UBC, seismic isolation
provisions were almost identical to those of the latest UBC edition (ICBO 1997). Starting from 2003,
however, the IBC refers to the ASCE-7 for seismic isolation design and testing requirements, and the
latest version of the ASCE 7 (ASCE 7-16) introduces the use of property modification factors to
account for the isolator performance variability in the design process. The European PrEN 15129
(CEN 2008) adopts a similar approach, based on lower and upper bound representative values of the
isolators’ performance, accounting for manufacturing, temperature, and aging variations.
Despite the recognized difference in terms of performance variability for isolators of different types,
testing procedures remained almost unchanged in the years to follow. Most of the current standards
addressing seismic isolation recommend the use of lower and upper bound model parameters to
account for all the sources of performance variability. However, the types of prototype tests didn’t
show the same level of evolution over time, and remain mostly unchanged from 1991 to nowadays.
Changes were instead proposed from each individual regulatory agency in terms of number of tests
and criteria used for acceptance of the isolation devices, which resulted in notable differences among
current standards. The lack of significant updates in terms of types of prototype tests caused current
testing protocols being difficult to relate to the recently observed sources of performance variability
for different types of isolators. For friction isolators, for instance, each single prototype test triggers
different sources of variability that simultaneously affect the mechanical properties of isolators
(Benzoni and Lomiento 2016, Lomiento and Benzoni 2017), in ways that make the interpretation of
experimental data debatable. At the same time, since all the simultaneous sources of variability cause
difficult to predict variations in the mechanical properties, adequacy criteria are mostly used to avoid
that those variations exceed predetermined limits during prototype tests, rather than requiring that the
observed variation is properly implemented in the structural analysis.
This study is aimed at analyzing and comparing testing protocols and acceptance criteria from their
original formulation to some of the current regulations in use in US and Europe. Friction isolators are
used as reference devices. Experimental data from recently tested double-concave sliding isolators are
used to support the comparison between the testing protocols. A recent friction coefficient model
(Lomiento et al. 2013) is used in order to simulate the isolators’ performance for different prototype
tests. The prototype testing protocols of the original 1991 UBC, and the current AASHTO, ASCE-7,
and PrEN 15129 are presented and discussed in terms of similarities and ability to capture the actual
performance of friction isolation devices. Acceptance criteria are finally discussed on the basis of the
expected experimental behavior of friction isolators, in reason of their consequences on the structural
analysis of such isolation systems.
2. EXPERIMENTAL EVIDENCE AND MODEL FOR FRICTION ISOLATORS
A set of experimental data on double pendulum isolator is used to provide evidence of the expected
variability of sliding isolators’ performance for different prototype testing protocols. Experimental
tests on full scale pendulum isolators show that the restoring forces developed at the curved sliding
interfaces are mainly affected by the applied vertical force, while the main changes in performance are
related to variations of the coefficient of friction (Lomiento et al. 2013a). The four main effects
affecting the coefficient of friction µ of typical steel-polymer interfaces, are:
1. “Breakaway effect”, (sudden increase of µ at each motion beginning/reversal);
2. “Load effect”, (reduction of µ for increasing vertical compression load);
3. “Cycling effect”, (reduction of µ due to temperature increase for cyclic sliding);
4. “Velocity effect”, (increase of µ with increasing sliding velocity).
2
The performance of friction isolator is considered rate dependent due to the increase of the coefficient
of friction with the sliding velocity, which requires presudo-dynamic testing. Moreover, friction
isolator are sensitive to the bi-directional sliding. Due to the different direction of friction and
restoring forces (Lomiento et al. 2012, 2013b), the slope of the force-displacement loop, the effective
stiffness, and the damping all depend on the bi-directional displacement pattern. The effect of the bi-
directional motion on the effective stiffness is shown in Fig. 1 for a sample cloverleaf test. Fig. 1b
shows the force-displacement loop a cloverleaf test with longitudinal displacement component
identical to the displacement component used for the mono-directional test of Fig 1a. The reduction in
effective stiffness caused by the lateral displacement component in the bi-directional cloverleaf test is
24% (from 4.09 kN/mm to 3.05 kN/mm), which qualifies the device as having a direction-dependent
behavior (>10% difference for UBC, >15% difference for ASCE 7 and AASHTO).
(a) (b
Fig. 1 – Force-displacement loops for (a) mono-directional (a) and bi-directional cloverleaf test.
A detailed experimental model for the variation of the frictional properties was demonstrated suitable
for capturing most of the effects of load, velocity, and repetitive motion (cycling). The experimental
data used for the model calibration derived from an extensive mono- and bi-directional test campaign
conducted on full-scale double-concave friction isolators under a wide range of vertical loads and
velocity (Adzhemyan 2017). The proposed friction model accounts for load, velocity, and cycling
effects through the independent functions
( )
Nf
N
,
( )
vf
v
, and
( )
Cf
C
, respectively:
Friction model
( ) ( ) ( ) ( )
vfCf
Nfv
CN
v
CN
⋅⋅=
,,
µ
(1)
Vertical load function
()
ref
NN
s
N
eNf
/
0
−
⋅=
µ
(2)
Vertical load function
( ) ( )
1 /
ref
vv
vevf −
⋅−+=
γγ
(3)
Vertical load function
( )
( )
β
ref
CC
C
eCf
−
=
(4)
Cycling variable
( )
∫
=
t
t
dtNv
a
tC
0
2
2
π
(5)
where
µ
s0 = zero-load static coefficient of friction, N>0 vertical compression load on the isolator, Nref
= load associated to a 63% friction reduction, a = in-plane radius of the slider, v = sliding velocity,
γ
=
fast / slow friction coefficient ratio, vref = experimental value of v related to a 63% increment of the
coefficient of friction,
β
= exponential rate of the friction degradation determined from experimental
data, Cref = value of the variable C associated with a 63% friction reduction for cycling effects, A = in-
plane radius of the concave surface. In this study, values of a specific tested bearing are used, as
follows: R = 5000 mm, a = 323.1 mm, A = 802.4 mm μs0 = 0.14, Nref = 8600 kN, γ = 1.9, vref = 23
3
mm/s, Cref = 6 kN/mm/s,
β
= 0.2. Exemplificative experimental and predicted force-displacement
loops are presented in Fig. 2 for two different levels of vertical load (i.e. pressure p).
(a) (b
Fig. 2 – Exemplificative experimental and predicted force-displacement loops.
3. PROTOTYPE TESTING
The evolution of the prototype testing from the 1991 UBC to the current AASHTO, ASCE 7, and
PrEN is presented hereafter. The number of tests was greatly reduced over the years, and differences
and similarities between the testing protocols are analyzed for application to sliding isolation devices.
Despite the differences between the testing protocols, 8 categories of tests are commonly used in the
different standards, and specifically:
C1. Service (for wind loads or max non-seismic displacement)
C2. Benchmark (for reference values before degradation tests are performed)
C3. Lateral range (for change in lateral performance at different levels of seismic displacement)
C4. Degradation (for degradation of lateral performance due to repeated cycling: ≥10 cycles)
C5. Integrity (for verification of the integrity, after degradation tests are performed)
C6. Vertical range (for change of lateral performance under max and min vertical seismic loads)
C7. Bidirectional (for verification of the directional performance)
C8. Verification (for property verification at the end of the testing protocol)
The tests from each regulatory standard are here attributed to these 8 test categories. Prototype tests
are performed separately on two full-size specimens of each type and size of isolator unit of the
isolation system. The following notation and notes apply to all the testing protocols.
Legend:
DL, LL are Dead and Live loads
E, EM is the Seismic load, at the design and maximum level earthquake
W is the Wind load
ψ is the seismic combination factor for live loads
V is the peak velocity
D, DM, DTM is the total design displacement (design-level, maximum-level, and maximum-total including torsion)
SMS, SM1 is the MCE spectral acceleration values at short periods, and 1 sec period
Sl is a numerical coefficient for site-soil profile as set forth in Table No. A-23-U of 1991 UBC for seismically-
isolated structures
BM is the spectrum reduction factor that accounts for the effective damping of the isolation system.
B is a numerical coefficient related to the effective damping of the isolation system as set forth in Table No. A-
23-W of 1991 UBC
fI is the inverse of the isolation period TI
Notes:
(1) Applicable only to vertical load carrying devices.
(2) Applicable only if the force-deflection properties of the isolator units are dependent on the rate of loading.
Properly documented dynamic tests of scaled specimens may be used to quantify the properties of rate-
dependent systems.
4
(3) Bi-lateral tests to be carried out only if the force-deflection properties of the isolator units are dependent on
bilateral load. Properly documented dynamic tests of scaled specimens may be used to quantify the properties
of direction dependent systems.
(4) The bi-directional test shall be performed with a “clover-leaf” pattern. If testing equipment is unable to perform
test B, the test can be completed after a rotation of 90 degree of the bearing in order to involve a displacement
path perpendicular to the one verified with previous tests.
3.1 UBC (1991)
The first regulatory standard incorporating prototype testing requirements for seismic isolators was the
1991 UBC. Prototype tests were used for establishing and validating the design properties of the
isolation system, and complement manufacturing quality control tests. Three types of tests were
specifically required, namely A, B, and C, as summarized in Table 1. The number of test per specimen
ranges from a minimum of 8 tests, in case of a non-vertical load carrying device with not rate-
dependent and not direction-dependent behavior, to a maximum of 48 tests, for vertical load carrying
device with rate dependent behavior. This large variation in number of required tests was originally
motivated by the lack of experimental evidence of the behavior of relatively new isolation devices at
the time of the publication of the standards. It shall be noted that some of the test requirements, such
as the bi-directional testing requirements for large-scale isolators, were unattainable at that time
because of lack of suitable equipment.
Table 1 – Sequence and cycles of prototype tests from UBC 1991.
Type Test Category Vert Load
Lateral
Load
# of Cycles Max Displacement
Rate of
loading
(2)
A C1. Service DL W 20 -
½f
I
fI
2fI
B C3. Lateral range
C7. Bidirectional
DL
1.2DL+1.0LL+|E|(1)
0.8DL-|E|(1)
- 3
0.25D
0.5D
0.75D
1.0D
(0.25, 1.0)D(3)
(0.50, 1.0)D(3)
(0.75, 1.0)D(3)
(1.0, 1.0)D
(3)
½fI
fI
2fI
C C4. Degradation DL - 15Sl/B ≥ 10 1.0D
½f
I
fI
2fI
Static C6. Vertical range
1.2DL+1.0LL+|E|
0.8DL-|E|
- - 1.0DM -
3.2 AASHTO (2014)
The AASHTO 2014 prototype tests are presented in Table 2. Tests 2, 3, 4 and 7 are replacement of the
1991 UBC tests A, B, C and static. Test 6 is a property verification test, similar to test 3 of the ASCE
7. The number of tests ranges from a minimum of 12 to a maximum of 13 tests, in case of direction-
dependent behavior.
3.2 ASCE 7 (2016)
Differently from all the other standards the ASCE 7-16 uses the Maximum Considered Earthquake
(MCE) rather than the design earthquake to set load and displacement demand for the tests. Tests 1, 2,
4 and 5 replace the UBC tests A, B, C, and static. A type 3 test is introduced to evaluate the integrity
of the isolator at the maximum total displacement. The sequence of tests associated to the test type 2
for four levels of displacement can be replaced by two dynamic tests, where the four levels of
displacement are applied dynamically and consecutively in descending and ascending order (one cycle
5
per level of displacement). The number of tests ranges from a minimum of 6 to a maximum of 21
tests, in case of load-carrying devices with direction-dependent behavior.
Table 2 – Sequence and cycles of prototype tests from AASHTO (2014).
Type Test Category Vert Load
Lateral
Load
# of
Cycles
Max Displacement
Rate of
loading
(2)
1 C1. Service DL - 3 Thermal -
2 C1. Service DL W 20 - ≤0.075Hz
3 C2. Benchmark
C3. Lateral range DL - 3
1.0D
0.25D
0.5D
0.67D
1.0D
1.25D
fI
4 C4. Degradation DL - 20 1.0D fI
5 C5. Integrity DL W 3 - ≤0.075Hz
6
C5. Integrity
C7. Bidirectional
DL - 3
1.0D
(0.71, 0.71)D
(3)
fI
7 C6. Vertical range
1.2DL+1.0LL+|E|
0.8DL-|E|
- 1 1.0D -
Table 3 – Sequence and cycles of prototype tests from ASCE 7-16.
Typ
e Cat. Vert Load Latera
l Load # of Cycles Max Displacement
Rate of
loading(2
)
1 C1. Service DL+0.5LL W 20 - fI
2 C3. Lateral range
C7. Bidirectional
DL+0.5LL
1.2DL+0.5LL+|EM|+0.2S
(1)
0.9DL-|EM|(1)
-
3
or
1
0.25D
M
0.5DM
0.67DM
1.0DM
(0.25, 1.0)DM (3)
(0.50, 1.0)DM (3)
(0.75, 1.0)DM (3)
(1.0, 1.0)DM (3)
or
[1.0+0.67+ 0.5+0.25] DM
[0.25+0.5+ 0.67+1.0] DM
fI
3
C5. Integrity
C7. Bidirectional
DL+0.5LL - 3
1.0D
M
(1.0, 1.0)DM
(3) fI
4 C4. Degradation DL+0.5LL -
30S
M1
/(S
MS
B
M
)≥1
0
0.75DM fI
5 C6. Vertical range
1.2DL+0.5LL+|E
M
|+0.2S
(1)
0.9DL-|EM|
(1)
- 1 1.0DTM -
3.4 PrEN 15129 (2008)
The PrEN 15129 specifies different prototype tests for different types of isolators. Only the tests for
sliding isolators are presented hereafter. The PrEN requires the minimum number of tests in
comparison with the other standards, i.e. a total of 9 tests. Service, dynamic and seismic tests are
equivalent to the UBC tests type A, B and C, with the exception of the seismic test being dynamic
instead of static. Benchmark, integrity and verification tests are additional tests aimed at verifying the
performance of the isolator before and after the main tests are performed.
6
Table 4 – Sequence and cycles of prototype tests from PrEN 15129 (2008) for sliding isolators.
Type Test Category Vert Load
Lateral
Load
# of
Cycles
Max Displacement
Rate of
loading
Service C1. Service DL+LL - 20 Max non-seismic V=5mm/s
Benchmark C2. Benchmark DL+LL - 3 1.0D V=50mm/s
Dynamic C3. Lateral range DL+LL - 3
0.25D
0.5D
1.0D
fI
Integrity C5. Integrity DL+LL - 3 1.0D f I
Seismic C6. Vertical range
1DL+ψLL+|E|
1DL+ψLL -|E|
- 3 1.0D f I
Bi-dir. C7. Bidirectional DL+LL 3 (1.0D, 1.0D)(4) f I
Verification C8. Verification DL+LL 3 1.0D f I
2.5 Comparison
The number of tests required for sliding isolators, which have a rate-dependent and direction-
dependent behavior, is summarized in Table 5 for all the standards. The larger variety of tests, in terms
of different categories, is required by the PrEN 15129 which requires however the minimum number
of tests (9), while the larger number of tests (48) is required by the 1991 UBC. Prototype tests
significantly diminished in number from the 1991 UBC to the ASCE 7-16 and the AASHTO, which
are currently similar, both in terms of required number of tests and of test categories. Service tests,
lateral range, vertical range and bidirectional tests are common to all the standards, even if they can
vary in number and level of force/displacement.
Table 5 – Comparison between standards in terms of tests for sliding isolators.
Test Category
1991 UBC
AASHTO
ASCE 7
PrEN
C1. Service
3
2
1
1
C2. Benchmark
-
1
-
1
C3. Lateral range 36 5 6-12 3
C4. Degradation 3 1 1 -
C5. Integrity 0 2 1 1
C6. Vertical range 2 2 2 1
C7. Bidirectional 4 1 3-5 1
C8. Verification - - - 1
Total
48
14
14-22
9
Despite the similarities in terms of types of test, the acceptance criteria are significantly different, as
shown in the next section.
4. ACCEPTANCE CRITERIA
The isolators’ adequacy for all the standards is based on checks of the following 10 items:
- Slope ∆F/∆D for the mono-directional force-displacement loop
- Period TT, based on tangent stiffness of force displacement loop
- Effective stiffness keff = (F+-F-)/(D+-D-) where F=Fmax for keff,max, and F=Fmin for keff,min
- Effective damping βeff = (Area loop)/( 2πkeff,maxD2)
- Max service force Fs,max
7
- Max design force Fmax
- Max service displacement Ds,max
- Restoring stiffness kres produced by the concavity of the sliding surfaces
- Energy dissipated per cycle EDC
- Coefficient of friction µ
A total of 23 possible checks are performed based on the above 10 items, depending on the standard,
as shown in Table 6. It should be noted that the US standards are gradually evolving over time,
moving from checks based on linear effective parameters, such as the effective stiffness and damping
values, to nonlinear parameters characterizing the isolators’ performance, such as the restoring
stiffness (referred to as post-yield stiffness), and the coefficient of friction, checked only by the PrEN.
The checks in terms of force-displacement slope, stability and deterioration are common to all the
standards. Most of the quantitative acceptability checks are performed on lateral range, degradation
and integrity tests, even if types of checks might significantly differ from one standard to another one.
Moreover the isolators are checked for stability, absence of visible deterioration and vibration.
Table 6 – Comparison between standards in terms of tests for sliding isolators.
Parameter
Condition
UBC
AASHTO
ASCE 7
PrEN
∆F/∆D
∆F/∆D > 0
All tests
-
All tests
All tests
∆F/∆D ≥ W/80 between 0.5D and 1.0D
-
All tests
-
-
TT
≤ 6sec
-
All tests
-
-
keff
≤10% difference from average
C3
C3
-
-
≤10% difference between 2 specimens C3 C3
-
-
≤15% difference from average of 2 specimens
-
- C3, C5
-
≤±10% (average) difference from design value
-
C3 -
-
≤±20% difference from initial value C4 C4 C4
-
βeff ≤20% decrease from initial value C4 - C4
-
≤30% decrease from initial value
-
C4
-
-
Fs,max ≤ design value
-
-
-
C1
Fmax ≤±15% difference from design value
-
-
-
C3, C5, C6
≤ design value
-
-
-
C7
Ds,max ≤ design value
-
C1, C5
-
-
kres
λtest,min ≤ kres /nominal ≤ λtest,max (±30%)
-
-
C3, C4, C5
-
≤±10% difference between successive cycles
-
-
-
C3, C5, C6
≤±5% diff. between upper/lower portion of cycle
-
-
-
C3, C5, C6
≤±15% difference from design value
-
-
-
C3, C5, C6
≤±15% difference between 2 specimens
-
-
-
C3, C5, C6
EDC
λspec,min -5% ≤ EDC/average ≤ λspec,max+5% (±20%)
-
-
C5
-
λtest,min ≤ EDC/nominal ≤ λtest,max (±30%)
-
-
C4
-
≥ 85% design value
-
-
-
C3, C5, C6
µ
within design limits
-
-
-
All tests
5. COMPARISON BETWEEN STANDARDS
The sequence of protocol tests from all the standards is simulated by using the experimentally
calibrated friction model. Results are presented in the following, showing the variation of some of the
performance parameters in regards to the acceptance criteria. The analysis focuses on the following
main parameters: (i) effective stiffness keff, (ii) effective damping βeff, (iii) restoring stiffness kres, and
(iv) EDC. The variability of the parameters is presented versus the cycling variable, which is
8
representative of the cumulative heat flux generated by the sliding motion. The experimental variation
of the lateral force with the cycling variable is presented in Fig. 3a for a typical 3-cycle prototype test.
The test is preceded by an entrance loop, and followed by an exit loop to resolve disturbances of test
results at the beginning and the end of the motion. The difference between lateral and friction force is
given by the restoring force, whose effect is shown as vertical arrows. The decrease in force
throughout the test is due to cycling effects. The force variability predicted by the model is shown in
Fig. 3b, which clearly resembles most of the experimental force variability. A notable difference is due
to the local breakaway friction force spikes, which are only visible in the experimental data of Fig. 3a,
and have limited effects on the above mentioned performance parameters. The cumulative cycling
variable for the entire testing protocols is reported in Fig 3c, which shows that the ASCE 7 testing
protocol is the one generating the most heat flux, as it involves larger displacements (maximum
instead of design levels), while the PrEN is the one generating less heat flux overall.
(a) (b) (c)
Fig. 3 – Variation of the normalized force F/N versus the cycling variable c for a typical 3-cycle test: (a)
experimental, (b) model prediction; (c) cumulative cycling variable for the testing protocols
For the comparison between the tests of Figs. 4-7, the variation of the four performance parameters is
normalized to the value at the beginning of each test, while black lines indicate the acceptance criteria.
5.1 Effective stiffness
The largest variation in effective stiffness is associated with the UBC testing protocol, while the least
variation is given by the PrEN. For UBC and AASHTO, service tests show reduction of stiffness up to
32% for service test, while degradation tests show up to 23% reductions. The reductions for ASCE and
PrEN are always less than 20%. The effective stiffness is considered important as an acceptance
parameter by the UBC and AASHTO. The analyzed isolator would satisfy all the effective stiffness
acceptance requirements, except for the one associated to the degradation test of the AASHTO
protocol, mostly due to the high cycling effects associated with the test.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
050 100 150 200 250 300 350
Acceptance Effective stiffness
Cumulative cycling variable, c,
MW/m2
Lateral range
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0150 300 450 600 750 900
Acceptance Effective Stiffness
Cumulative cycling variable, c,
MW/m2
Lateral range
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0200 400 600 800 1000
Acceptance Effective Stiffness
Cumulative cycling variable, c,
MW/m2
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
040 80 120 160 200
Acceptance Effective Stiffness
Cumulative cycling variable, c,
MW/m2
a b c d
Fig. 4 – Effective stiffness and acceptance criteria for: (a) UBC, (b) ASCE 7, (c) AASHTO, (d) PrEN.
9
5.1 Effective damping
Similarly to the effective stiffness, the largest variation in effective damping is given by the UBC (up
to -43%) and AASHTO testing protocol (up to -48%), while the smallest variation comes from the
PrEN (-39%). The damping ratio appears more affected by the degradation induced by the cycling
effect in comparison with the effective stiffness, which results into non-acceptable results for the
UBC, ASCE and AASHTO testing protocols. The highest reduction in damping ratio for the ASCE
and the PrEN comes from bi-directional tests, which are not considered for acceptance.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0150 300 450 600 750 900
Accepta nce Effe ctive Damping ,
Cumulative cycling variable, c,
MW/m2
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
050 100 150 200 250 300 350
Accepta nce Effe ctive Damping
Cumulative cycling variable, c,
MW/m2
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0200 400 600 800 1000
Accepta nce Effe ctive Damping
Cumulative cycling variable, c,
MW/m2
Degradati on
Bi-Dir
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
040 80 120 160 200
Accepta nce Effe ctive Damping
Cumulative cycling variable, c,
MW/m2
Bi-Dir
a b c d
Fig. 5 – Effective damping and acceptance criteria for: (a) UBC, (b) ASCE 7, (c) AASHTO, (d) PrEN.
5.1 Restoring stiffness
The cycling effect is again the main source of variation for the restoring stiffness during single tests, as
it reduces the slope of the force-displacement loop. The restoring stiffness increases from the 1st to the
following loops, due to the friction reduction caused by the cycling effect, which is more significant at
the beginning of the test. The restoring stiffness is only considered as an acceptance parameter by the
ASCE and PrEn, with the PrEN requirements being more restrictive and resulting in failure of passing
the test. The failing condition occurs in the low level (0.25D) displacement test, for which the
restoring stiffness suffers a >10% increase from the first to the second loop.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0150 300 450 600 750 900
Accepta nce Res toring Stiffne ss
Cumulative cycling variable, c,
Bi-Dir
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
050 100 150 200 250 300 350
Accepta nce Res toring Stiffne ss
Cumulative cycling variable, c,
Servi ce
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0200 400 600 800 1000
Accepta nce Res toring Stiffne ss
Cumulative cycling variable, c,
Degradati on
Lateral range
Integrity
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
040 80 120 160 200
Accepta nce Res toring Stiffne ss
Cumulative cycling variable, c,
Verti cal range
Lateral range
Integrity
a b c d
Fig. 6 – Restoring stiffness and acceptance criteria for: (a) UBC, (b) ASCE 7, (c) AASHTO, (d) PrEN.
5.1 EDC
Finally, the highest variability in performance parameters is associated with the EDC values. The
reduction of the EDC ranges from 52% and 60% throughout all the tests. The variability in EDC was
found to be below the acceptance limits of the analyzed standards, even if very close to the lower
boundaries for the PrEN requirements.
10
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0150 300 450 600 750 900
Acceptance EDC
Cumulative cycling variable, c,
MW/m2
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
050 100 150 200 250 300 350
Acceptance EDC
Cumulative cycling variable, c,
MW/m2
Degradati on
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0200 400 600 800 1000
Acceptance EDC
Cumulative cycling variable, c,
MW/m2
Degradati on
Integrity
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
040 80 120 160 200
Acceptance EDC
Cumulative cycling variable, c,
MW/m2
Verti cal range
Lateral range
Integrity
a b c d
Fig. 7 – Energy dissipated per cycle and acceptance criteria for: (a) UBC, (b) ASCE 7, (c) AASHTO, (d) PrEN.
6. DISCUSSION
From the analysis of the four testing protocols, the performance parameters most subjected to change
during tests are the effective damping ratio and the energy dissipated per cycle. Significant variations
of these parameters are evidenced in the degradation tests of the UBC and AASHTO standards, and in
the bi-directional tests required by the ASCE 7 and PrEN standards. The modification of the
degradation test in the ASCE, with respect to the original UBC, reduces the degradation of the
dissipation properties in order to match expected reduction from design seismic events. It is advisable
that the AASHTO will update testing protocols in a similar manner. Despite showing great reduction
in dissipation capability, the bi-directional tests are not considered in any acceptance criteria. When a
bi-directional analysis is performed, proper reduction of the dissipation capability shall be considered,
in order to account for the augmented cycling effect produced by bi-directional motion.
The effective stiffness of friction isolators appears less affected by all the sources of performance
variability. This is partially due to the relative high importance of the restoring force in comparison
with the friction force at the design displacement. However, significant variability is identified for
service tests, in which the displacement is well below the design displacement. This suggests that
appropriate property modification factors shall be defined for serviceability conditions, different in
values from what is used for design-level conditions.
The variation of the effective stiffness and damping are considered of secondary importance by the
recent ASCE 7 and PrEN. These standards implicitly suggest that nonlinear parameters such as
restoring stiffness, EDC, and coefficient of friction are more adequate representation of the
performance of sliding isolators. Among these, the restoring stiffness appeared significantly affected
by variations during service tests, for which multiple cycles of low-amplitude displacement are
performed. Increase in restoring stiffness up to 80% are evidenced for UBC service tests (Fig. 6). This
variation shall be also appropriately considered in the analysis, as it can affect the level of
displacement expected under service loads.
Overall, a large discrepancy in the acceptance criteria is evidenced among the analyzed standards.
Simulations show that the analyzed isolators would not be acceptable for any of the prototype testing
protocols, but for different criteria concerning the effective stiffness (AASHTO), the effective
damping (UBC, AASHTO, ASCE 7), and the restoring stiffness (PrEN). It is advisable that the testing
protocols will go through a standardization process, in order to avoid such discrepancies.
The degradation of the coefficient of friction due to cycling effects has been evidenced as the
predominant source of changes in the performance parameters. As already discussed in (Lomiento and
Benzoni 2017), a revision of the testing protocols is envisioned in order to determine an equivalence
between prototype tests and the expected seismic design conditions not only in terms of force,
velocity, and displacement demand, but also in terms of generated heat flux. Finally, the activation of
all the sources of friction variability in each test makes interpretation of the variability of the
performance parameters hardly predictable. Ad-hoc “property characterization” tests can be effectively
introduced in order to isolate each of the sources of variability. This will simplify the understanding of
the performance variability, and will allow the calibration of sophisticated predictive models, such as
the one used in this study.
11
7. CONCLUSION
The increased awareness of the changes in friction performance of seismic sliding isolators resulted in
modifications of the testing protocols over the years. The number of tests was greatly reduced, and the
acceptance criteria were updated to include nonlinear performance parameters along with linear
parameters such as effective stiffness and damping. This study compared testing protocols for friction
isolators from four regulatory standards: 1991 UBC, ASCE 7, AASHTO, and PrEN 15129. An
experimentally validated phenomenological model of the frictional performance of sliding isolators
was used to simulate testing protocols and investigate their mutual differences. The comparison
showed that there is large discrepancy between acceptance criteria for friction isolators in the different
standards, despite the prototype tests being similar in type amongst all the standards, and only slightly
different from the original prototype testing protocols of the 1991 UBC. The importance of the bi-
directional testing for such isolators has been recognized, and tests have been added in this regards,
even if they are not associated with any specific acceptance criteria. The energy dissipation capability
was found to be the performance that experienced the largest variations during individual tests. The
stiffness properties are instead mostly affected by variations in service test conditions rather than in
design test conditions. The simulations proved that the major source of performance variability is the
cycling effect, associate to the heat flow generated by the cyclic sliding. A revision of the testing
protocols is envisioned in order to make prototype testing representative of seismic design conditions
also in terms of heat flux, rather than just considering force, velocity and displacement demand.
8. REFERENCES
AASHTO (1999). Guide Specifications for Seismic Isolation Design. American Association of State
Highway and Transportation Officials (AASHTO), Washington, D.C.
AASHTO (2014). Guide Specifications for Seismic Isolation Design, 4th Edition. American Association of
State Highway and Transportation Officials (AASHTO), Washington, D.C.
Adzhemyan A (2017). Experimental Model for Double Concave Sliding Bearings. Master’s Thesis. Department
of Civil Engineering, California State Polytechnic University, Pomona, CA.
ASCE (2016). Minimum Design Loads for Buildings and Other Structures, American Society of Civil Engineers
(ASCE) ASCE/SEI 7-16, Reston, VA.
Benzoni G, Lomiento G (2016). Experimental assessment of property modification factors for friction isolators.
Proceedings of the 8th World Congress on Joints, Bearings and Seismic Systems for Concrete Structures.
Georgia, Atlanta.
CEN (2008). PrEN 15129. Anti-seismic devices. CEN/TC340.
ICBO (1991). Uniform Building Code, 1991 Edition. International Conference of Building Officials (ICBO),
Whittier, CA.
ICBO (1997). Uniform Building Code, 1997 Edition. International Conference of Building Officials (ICBO),
Whittier, CA.
ICC (2000). International Building Code, 2000 Edition. International Code Council (ICC), Falls Church, VA.
Lomiento G, Benzoni G (2017). Testing protocols and acceptance criteria for performance characterization of
pendulum isolators. Proceedings of the 16th World Conference on Earthquake Engineering. Santiago, Chile.
Lomiento G, Bonessio N, and Benzoni G (2012): Effects of Loading Characteristics on the Performance of
Sliding Isolation Devices. Proc. of the 15th World Conference on Earthquake Engineering. Lisbon, Portugal.
Lomiento, G, Bonessio N, and Benzoni G (2013a): Friction model for sliding bearings under seismic excitation,
Journal of Earthquake Engineering, 17 (8), 1162-91.
Lomiento G, Bonessio N., and Benzoni G (2013b): Concave sliding isolator’s performance under multi-
directional excitation, Ingegneria Sismica, 30 (3), 17-32.
12