Beruflich Dokumente
Kultur Dokumente
DOI 10.1007/s10518-005-0187-9
Springer 2005
1. Introduction
SteelPTFE sliding bearings have been widely used in the past to accommodate thermal movements and effects of pre-stressing, creep and shrinkage in bridges. More recently, they have been proposed as part of seismic
isolation systems, to support the weight of the structure, while relying upon
separate mechanisms, to provide the system with re-centring and additional
energy absorbing capability.
Several sliding isolation systems with restoring force have been proposed. Some of them have reached the stage of implementation, such as the
Resilient-Friction Base Isolation System (Mostaghel, 1984), based on the
elastic properties of rubber, the SMA Isolation System (Dolce et al., 2000),
based on the superelastic properties of shape memory alloys, and the Friction Pendulum System (Zayas et al., 1987), which exploits an articulated
slider moving on a spherical surface to provide restoring capability. The
most important advantage in using sliding isolation systems with (weak)
76
M. DOLCE ET AL.
77
order to fully investigate the effects of sliding velocity, contact pressure, air
temperature, number of cycles and state of lubrication on the mechanical
behavior of steelPTFE sliders. Based on the experimental outcomes, two
mathematical models of their frictional behavior, for conditions of interest
in seismic isolation, have been developed and calibrated. The rst model
is basically an extension of the model proposed in (Constantinou et al.,
1990), while the second model is derived from the one proposed in (Chang
et al., 1990). In this paper, the main results of the experimental tests are
described. Model predictions and experimental results are then compared.
2. Experimental Set up and Procedure
The experimental program on steelPTFE interfaces had two specic objectives: (i) investigating the variability of the sliding friction coefcient while
varying contact pressure, velocity, air temperature, displacement amplitude
and state of lubrication of steelPTFE interfaces, (ii) developing and calibrating a numerical model of the mechanical behavior of steelPTFE sliding bearings.
2.1. Materials
The materials used were as follows:
Unlled PTFE pads obtained from a 5.45 mm thick sheet with dimpled
recesses, whose dimensions and pattern are shown in Figure 1. The function of the dimple recesses is to retain the grease and to gradually introduce it between the sliding interfaces, during wear of PTFE.
Stainless steel plates (AISI 316/L) of 3 mm thickness, polished to mirror
nish, with less than 0.1 m surface roughness.
Silicone grease, of the type normally used in the bearing manufacturing.
78
M. DOLCE ET AL.
79
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
L/P
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
10/20/50
P c (MPa)
dd
fe
Wavef
(mm) (Hz)
vg
Cycles CDh
(mm/s) No.
(mm)
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
9.36/18.7/28.1
50
50
50
50
10
10
10
10
25
25
25
25
50
50
50
50
10
40
100
200
var.
var.
var.
var.
var.
var.
var.
var.
var.
var.
var.
var.
0.05
0.2
0.5
1
0.05
0.2
0.5
1
0.05
0.2
0.5
1
0.05
0.2
0.5
1
TR.
TR.
TR.
TR.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
SYN.
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
1000
2000
3000
4000
4200
4400
4600
4800
5300
5800
6300
6800
7800
8800
9800
10,800
80
M. DOLCE ET AL.
Triangular (saw-tooth) tests were carried out at constant displacement amplitude (50 mm), while increasing the sliding velocity from 10 to
200 mm/s. In the sinusoidal tests, the displacement amplitude was varied
from 10 to 50 mm, while increasing the frequency of loading from 0.05 to
1 Hz, thus producing peak sliding velocities ranging from 1 to 300 mm/s.
Five complete loading cycles were performed during each test. The cumulative distance during each series of test was approximately 10.8 m, as shown
in Table I. Thus, a total distance of about 32.4 m was covered by each set
of PTFE pads. The thickness of the PTFE pads was measured before testing and after their removal.
3. Experimental Results
3.1. Effect of type of test
Representative frictional force-displacement loops of the single interface are
shown in Figures 3 and 4. They refer to tests conducted on non-lubricated interfaces under the same contact pressure (28.1 MPa), air temperature (10 C) and displacement amplitude (50 mm), only differing for the
wave form and the loading frequency. Figure 3 compares the frictional
behavior exhibited by the PTFEsteel interfaces in (a) triangular and (b)
sinusoidal tests at the same peak velocity, equal to 10 mm/s (i.e. test No. 1
vs. test No. 13, according to Table I). Figure 4 refers to (a) triangular and
(b) sinusoidal tests conducted at the higher peak velocities: 200 mm/s and
300 mm/s, respectively (i.e. test No. 4 vs. test No. 16, according to Table I).
For the sinusoidal tests, the peak velocity is dened as the average velocity
in the displacement range corresponding to force levels greater than 95%
of the maximum frictional force.
Two phenomena are clearly visible in the triangular tests, one at the
start of sliding, the other at every motion reversal. The rst phenomenon
is generally taken into account through the denition of a breakaway friction coefcient, also known as static friction coefcient, to distinguish it
from the sliding (kinetic) friction coefcient. The second phenomenon, generally referred to as stick-slip, corresponds to a short duration increase of
the frictional force, followed by a rapid decrease. Both the observed experimental phenomena can be related to (i) a momentary sticking of the interfaces and to the (ii) acceleration impulse occurring at the start of the test
and at every motion reversal, especially in the triangular tests.
The examination of Figures 3 and 4 clearly highlights the dependence
of the friction coefcient from sliding velocity. In each triangular test, the
forcedisplacement loops are practically rectangular, in accordance with
friction Coulombs law, but the friction force increases while increasing the
frequency of loading in the different tests. In the sinusoidal tests, on the
81
(mm)
60
0
160
(sec)
(KN)
-60
s t ick slip
break-away friction
4
2
0
-60
-40
-20
20
40
-2
60
(mm)
-4
-6
(a)
stick slip
(KN)
(mm)
60
0
0
160
(sec)
-60
6
4
2
0
-60
-40
-20
0
-2
20
40
60
(mm)
-4
(b)
-6
contrary, the friction coefcient varies during the motion, reaching its maximum at the maximum velocity (i.e. at zero displacement). Furthermore the
friction coefcient tends to decrease during continuous loading cycles, this
effect being related to self-heating of the sliding interfaces. Indeed, the rate
of decrease of the friction coefcient is quite negligible in the tests al low
82
M. DOLCE ET AL.
(mm)
60
0
20
(sec)
break-away friction
(KN)
-60
stick slip
8
6
4
2
0
-60
-40
-20
-2
20
4
-4
40
60
(mm)
-6
3
-8
stick slip
(a)
(KN)
(mm)
60
0
0
20
(sec)
-60
8
6
4
2
0
-60
-40
-20
-2
-4
20
4
40
60
(mm)
-6
(b)
-8
83
Fr
N
(1)
84
M. DOLCE ET AL.
P= 9.36 MPa
(%)
(a) 25
P= 18.72 MPa
P= 28.1 MPa
exponential
20
logarithmic
logarithmic
15
exponential
10
logarithmic
exponential
5
(mm/sec)
0
0
(%)
(b) 25
50
100
150
P= 9.36 MPa
200
250
P= 18.72 MPa
300
350
P= 28.1 MPa
exponential
20
logarithmic
logarithmic
15
exponential
10
logarithmic
5
exponential
(mm/sec)
(c) 25
50
(%)
100
150
P= 9.36 MPa
200
250
P= 18.72 MPa
300
350
P= 28.1 MPa
exponential
20
logarithmic
15
logarithmic
exponential
10
logarithmic
5
exponential
(mm/sec)
0
0
50
100
150
200
250
300
350
Figure 5. Variation of the friction coefcient with sliding velocity, air temperature
and bearing pressure, for non-lubricated interfaces. Comparison between analytical
laws and experimental results. Air temperature equal to: (a) 10 C, (b) 20 C and
(c) 50 C.
85
(%)
-10C
+20C
+50C
12
10
8
6
4
2
0
9.36
18.72
28.08
(MPa)
(%)
(a)
300
-10C
+20C
+50C
250
200
150
100
50
0
9.36
(b)
18.72
28.08
(MPa)
Figure 6. (a) Absolute and (b) percent increment of the sliding friction coefcient in
seismic with respect to service conditions, as a function of contact pressure, for three
different air temperatures.
During its service lifetime, the sliding isolator acts as a usual sliding
bearing, subjected to service load and thermal movements at very low
velocities. The operating conditions of sliding bearing contained in various codes (AASHTO, 1999), (BS 5400, 1983), (CEN 1337, 2000) are
quite different. CEN considers lubricated steelPTFE interfaces only, while
BS and AASHTO allow the use of both lubricated and non-lubricated
interfaces. The maximum allowable contact pressure is assumed equal to
24 MPa by AASHTO (in absence of specic wear tests), 45 MPa by BS
and 60 MPa by CEN. For unlled PTFE sliding against stainless steel without lubrication, the service friction coefcient suggested by AASHTO (in
absence of tests), for temperatures 25 C, is equal to 15% at 10 MPa
and 10% at 20 MPa (or more). At 20 C, instead, the service limit state friction coefcient is assumed equal to 6% at 10 MPa and 3% at 20 MPa (or
86
M. DOLCE ET AL.
87
(%)
20
15
10
P2,v1
P3,v1
0
-10
(%)
(a)
20
20
P1,v2
15
P2,v2
50
(C)
P3,v2
10
(P1 =9.36MPa, v2 =316mm/s): = 1.96E-1 3.95E-4*T + 1E-6*T2
(P2 =18.7MPa, v2 =316mm/s): = 1.42E-1 3.95E-4*T + 1E-6*T2
(P3 =28.1MPa, v2 =316mm/s): = 1.18E-1 3.95E-4*T + 1E-6*T2
0
(b)
-10
20
50
(C)
Figure 7. Sliding friction coefcient at (a) very low and (b) very high velocities (i.e. 8
and 316 mm/s, respectively) as a function of air temperatures, for three different normal pressure values (i.e. 9.36, 18.72 and 28.1 MPa, respectively). Comparison between
experimental results and model predictions.
88
M. DOLCE ET AL.
experimental values drawn from the whole set of experimental data are 1.17
(0.09) and 0.89 (0.054), respectively.
3.5. Effect of lubrication
Figures 8 and 9 show the sliding friction coefcient for a number of signicant tests on both non-lubricated (Figure 8) and lubricated (Figure 9)
interfaces. Each diagram of gure 8 and 9 refers to a different air temperature: (a) 10 C, (b) 20 C and (c) 50 C, respectively, as well as to a
different set of PTFE pads. Each series of data, instead, refer to a different
normal load, resulting in 9.36, 18.7 and 28.1 Mpa contact pressure, respectively. In each diagram the experimental data have been gathered in four
groups, based on the peak sliding velocity. Each group, therefore, corresponds to tests repeated under similar test conditions, after a certain number of cycles. The displacement amplitude is generally different from one
test to another, within each group. The order of execution of the tests is
given by the number beside each point.
By comparing Figures 8 and 9, it turns out that lubrication reduces friction coefcient by about 5 times at 10 and 20 C, and almost 8 times
at 50 C. The sliding friction coefcient of lubricated steelPTFE interfaces
results always below 4%, which is the limit value for sliding isolation devices
prescribed by the new Italian seismic code (Ordinanza 3274, 2003). The scatter in each group is practically negligible for non-lubricated interfaces, while
it is not for lubricated steelPTFE interfaces. For these latter, the tendency is
a progressive reduction of the friction coefcient while increasing the number of cycles performed, due to the following reasons: (i) the introduction,
in the sliding interfaces, of the grease contained inside the dimpled recessed,
during the wear of the PTFE pads and (ii) the test sequence (see Table I),
where tests wiith increasing displacement amplitude were carried out. Actually, before the start of each test sequence, grease was spread over the PTFE
pads, to ll the dimpled recesses, but a certain amount of grease was also
spread over the stainless steel sheets. When increasing displacement amplitude, it is reasonable to believe that more grease, from the stainless steel
sheets, was introduced in the sliding interfaces.
The quite negligible scatter in the experimental results relevant to nonlubricated interfaces (see Figure 8) implies that the wear of PTFE (see
below) do not affect signicantly their frictional behavior. For the lubricated steelPTFE interfaces, friction is much lower and the wear of PTFE
is expected to be much less than for non-lubricated interfaces.
Figure 10 summarizes the results of all the tests on lubricated PTFE
steel interfaces. The sliding friction coefcient is reported as a function of
(peak) velocity and contact pressure, for three different temperature values:
(a) 10 C, (b) 20 C and (c) 50 C, respectively. In the gure, experimental
89
(%)
20
15
6
10
17
22
20KN
30KN
8
11
24
27
40
43
10
13
18
23
26
34
39
42
15
28
31
44
47
30
29
45
12
14
46
33
38
50
10
50
50
10
25
10
25
50
25
50
10
12
15
40
32
30
67
77
60
155
152 (mm/sec)
20
15
10
2
1
17
13
22
29
18
10
23
26
34
39
42
45
(mm)
30KN
20KN
10KN
(%)
(a)
11
14
12
28
24
27
30
40
43
46
44
15
31
47
33
38
50
10
50
50
10
25
10
25
50
25
50
10
12
15
40
32
30
68
79
61
160
157 (mm/sec)
20
10KN
(%)
(b)
20KN
15
2
10
13
17
22
29
33
38
50
10
18
7
23
(mm)
30KN
11
24
27
12
15
28
31
44
47
14
10
26
40
43
30
46
45
34
39
42
10
50
50
10
25
10
25
50
25
50
12
15
40
34
31
70
80
60
162
159 (mm/sec)
(c)
(mm)
data and model predictions are compared. As can be seen, the friction
coefcient of lubricated interfaces follows a trend similar to that described
for non-lubricated interfaces. It tends to increase when increasing sliding
90
M. DOLCE ET AL.
10KN
(%)
20KN
38
33
17
1
42
18 23
45
13
50
50
46
24
34
29
22
44
43
39
30KN
40
25
10
31
12
11
10
28
30
27
26
47
15
14
0
50
(a)
10
12
15
31
31
10KN
64
50
79
25
60
17
33
1
22
38
29
20KN
160
30KN
11
13
40
25
50
(mm)
156
(mm/sec)
(%)
10
10
7
18
2
45
10
23
39
24
26
42
15
14
28
27
40
12
30
44
31
43
46
34
47
50
10
50
50
10
25
10
25
50
25
50
(mm)
10
12
15
40
32
30
65
80
61
159
154
(mm/sec)
20KN
10KN
(%)
(b)
30KN
1
17
6
22
2
13
29
39
18
23
38
45
33
8
10
42
26
40
24
11
43 14
44
12
15
28
47
46
27
31
30
34
(c)
50
10
50
50
10
25
10
25
50
25
50
(mm)
10
15
40
31
31
66
80
62
161
158
(mm/sec)
velocity and to reduce when increasing air temperature and contact pressure. However a greater scatter can be noted, probably due to the testing
sequence and the difculty of getting uniform distribution of lubricating
grease. The behavior of the friction coefcient at 10 C is anomalous, as
it increases while increasing contact pressure. Nevertheless, some interesting
91
(a) 4
log@28.1MPa
log@9.36MPa
log@18.72MPa
P= 9.36 MPa
25
20
15
10
P= 18.72 MPa
5
0.000
5 0.000
1 00.000 1 5 0.000
2 00.000 2 5 0.000
3 00.000 3 5 0.000
(m
m
ec)
/s
P= 28.1 MPa
(mm/sec)
0
0
100
150
200
(%)
(b)
50
250
300
350
P= 9.36 MPa
25
20
15
10
P= 18.72 MPa
5
0.000
log@28.1MPa
5 0.000
1 00.000 1 5 0.000
2 00.000 2 5 0.000
3 00.000 3 5 0.000
(m
m
ec)
/s
P= 28.1 MPa
log@9.36MPa
log@18.72MPa
(mm/sec)
0
0
100
150
200
250
300
350
(%)
(c) 4
50
P= 9.36 MPa
25
20
15
10
P= 18.72 MPa
5
0.000
5 0.000
1 00.000 1 5 0.000
2 00.000 2 5 0.000
3 00.000 3 5 0.000
(m
m
ec)
/s
P= 28.1 MPa
log@28.1MPa
2
log@9.36MPa
log@18.72MPa
(mm/sec)
0
0
50
100
150
200
250
300
350
Figure 10. Variation of the friction coefcient with sliding velocity, air temperature
and bearing pressure, for lubricated interfaces. Comparison between analytical laws
and experimental results. Air temperature equal to: (a) 10 C, (b) 20 C and
(c) 50 C.
92
M. DOLCE ET AL.
93
(2)
(3)
(5)
94
M. DOLCE ET AL.
The coefcients i have been obtained from a multivariate nonlinear regression, using a statistical analysis package (SPSS, 1999).
The good t of the regression, for both analytical laws, is apparent
in Figure 5, which refers to non-lubricated interface. The proposed logarithmic model, however, is more accurate in capturing the experimental
behavior in the low velocity range than the Constantinous model, which
underestimates the friction coefcient in that range.
For lubricated PTFEsteel interfaces (see Figure 10), the accuracy of the
model predictions is decidedly worse, due to the great scatter in the experimental results.
Tables II and III show the values of the model parameters for the nine
combinations of bearing pressure and air temperature considered during
the tests on non-lubricated and lubricated interfaces, respectively. In the
tables there are also compared the experimental and analytical maximum
friction forces at high peak velocities (160 mm/s, precisely). As can be seen,
the error is at most equal to 5% for non-lubricated interfaces, while it
ranges between 1% and 43% for lubricated sliding interfaces. It should be
considered, however, the different order of magnitude of the frictional force
for non-lubricated and lubricated interfaces, whose average values are equal
to about 2.5 and 0.45 kN, respectively. Thus, a 5% error implies an absolute error of 0.125 kN, in the rst case, while a 43% error implies, in the
second case, an absolute error of 0.193 kN. In any case, the maximum frictional force provided by the model can be reliably used in seismic isolation
design.
Table II. Model parameters relevant to non-lubricated interfaces, for nine different combinations of bearing pressure and air temperature values.
Experimental test
P
(MPa)
T
( C)
Fra
9.36
9.36
9.36
18.7
18.7
18.7
28.1
28.1
28.1
10
20
50
10
20
50
10
20
50
Constantious model
max
(%)
(kN)
min
(%)
1.82
1.70
1.60
2.72
2.49
2.20
3.27
2.95
2.74
8.43
6.68
6.12
6.23
4.49
3.93
4.87
3.13
2.56
19.61
18.40
17.76
13.91
12.70
12.10
11,48
10.26
9.66
0.020
0.018
0.013
0.022
0.020
0.015
0.024
0.022
0.017
Logarithmic model
Fra
(kN)
Er.
(%)
Fra
(kN)
Er.b
(%)
1.92
1.78
1.64
2.74
2.47
2.27
3.40
3.02
2.76
5
5
3
0
0
3
4
2
1
0.024
0.026
0.026
0.015
0.018
0.018
0.014
0.015
0.016
0.058
0.035
0.031
0.055
0.030
0.016
0.039
0.021
0.006
1.80
1.68
1.59
2.64
2.39
2.20
3.27
2.90
2.62
1
1
1
3
4
0
0
2
4
Comparison between experimental and analytical maximum frictional force at high peak
velocities (160 mm/s precisely)
a
Peak friction force at 160 mm/s maximum sliding velocity.
b
Percent error between experimental and numerical peak friction force.
95
Table III. Model parameters relevant to lubricated steelPTFE interfaces, for nine different
combinations of bearing pressure and air temperature values.
Experimental test
Constantious model
Logarithmic model
P
(MPa)
T
( C)
Fra
(kN)
min
(%)
max
(%)
Fra
(kN)
Er.b
(%)
Fra
(kN)
Er.b
(%)
9.36
9.36
9.36
18.7
18.7
18.7
28.1
28.1
28.1
10
20
50
10
20
50
10
20
50
0.23
0.26
0.16
0.57
0.45
0.27
0.98
0.59
0.54
2.64
1.93
1.21
2.15
1.43
0.72
2.04
1.32
0.61
3.30
2.47
1.77
2.95
2.12
1.42
3.12
2.29
1.59
0.012
0.013
0.004
0.017
0.018
0.009
0.026
0.026
0.017
0.32
0.24
0.14
0.58
0.42
0.24
0.93
0.68
0.46
41
9
15
1
8
10
5
15
16
0.002
0.001
0.001
0.001
0.000
0.000
0.002
0.002
0.002
0.022
0.017
0.009
0.021
0.016
0.009
0.020
0.015
0.007
0.32
0.24
0.16
0.81
0.55
0.32
0.97
0.71
0.47
43
10
2
42
21
18
1
19
13
Comparison between experimental and analytical maximum frictional force at high peak
velocities (160 mm/s precisely)
a
Peak friction force at 160 mm/s maximum sliding velocity.
b
Percent error between experimental and numerical peak friction force.
The accuracy of the proposed model in capturing the actual frictional behaviour of steel-PTFE sliding bearings is conrmed by Figures 11 and 12, which compare the experimental and numerical force
displacement loops of non-lubricated and lubricated interfaces, respectively. The experimental force-displacement relationships shown in Figures 11 and 12 refer to the second cycle of the tests No. 13 and
16 of Table I, at low and very high peak sliding velocities (i.e. about
15 mm/s and about 316 mm/s, respectively). Both tests have been conducted with the same displacement amplitude (50 mm), contact pressure (18.72 MPa) and air temperature (20 C). In the construction of
the numerical relationships of Figures 11 and 12, reference was made
to the displacement-time histories, as drawn from the experimental output.
As far as non-lubricated interfaces are concerned, the accordance
between experimental observations and model predictions is almost perfect,
especially at low velocities. At high velocities, the numerical model is not
able to capture the decay of the friction coefcient due to self-heating, as
expected. Larger differences between experimental and numerical results are
observed for lubricated interfaces, due to the big scatter in the experimental outcomes and to the less sensitivity of the model to velocity variation
during the applied sinusoidal displacement.
As previously noted, the gap between minimum and maximum sliding
friction coefcient is signicant for non-lubricated interfaces, while being
negligible for lubricated steelPTFE interfaces. Figures 11 and 12 clearly
prove this. As a consequence, it can be said that the behavior of lubricated
96
M. DOLCE ET AL.
steelPTFE sliding bearings tends to be the same under seismic and service
conditions, while considerable differences, in terms of maximum force and
energy loss, are found for non-lubricated steelPTFE sliding bearings.
97
5. Conclusion
A comprehensive program of experimental tests on unlled PTFE pads
sliding against polished stainless steel has been carried out. The effects
on the sliding friction coefcient of (i) type of test, (ii) sliding velocity,
(iii) contact pressure, (iv) air temperature, (v) conditi on of interfaces (i.e.
lubricated or not) and (vi) number of cycles have been investigated. The
following most important experimental ndings have been obtained:
(1) The coefcient of friction increases rapidly with velocity, up to a certain velocity value, beyond which it remains almost constant. Such value
is around 150 mm/s, regardless air temperature and bearing pressure. The
maximum velocities occurring in steelPTFE sliding bearings under an
earthquake are surely greater than 150 mm/s. Therefore, the design value of
the frictional force in seismic applications can be assumed to be independent from frequency of loading and displacement amplitude.
(2) The sliding friction coefcient of steelPTFE interfaces reduces while
increasing pressure. The reduction rate, however, depends on both sliding velocity and air temperature. It increases while increasing velocity and
while decreasing air temperature. By referring to 20 C air temperature,
18.7 MPa contact pressure and sliding velocities 150 mm/s, maximum
variations in the frictional force of steelPTFE sliding bearings of the order
of 30% are expected for 50% variations of the contact pressure, regardless
the state of lubrication of the interfaces.
(3) The sliding friction coefcient decreases while increasing the air temperature. Its rate of reduction is greater when passing from low-to-medium
temperatures, than when passing from medium-to-high temperatures. Moreover, it depends on sliding velocity, while being practically independent
from contact pressure. At sliding velocities of interest for seismic applications, the reduction rate of the friction coefcient with temperature is of
the order of 0.150.3%/ C, for non-lubricated interfaces. As a consequence,
upper and lower bound analysis is needed, in order to evaluate the maximum forces on the structural elements and the maximum displacements in
the isolation system. To this end, reference can be made to the lambda-factors (max,t and min,t ) provided by AASHTO, to estimate the values of the
friction coefcient at extreme design temperatures. Based on the experimental results of the present study, the lambda-factors suggested by AASHTO
lead to very accurate predictions for non-lubricated interfaces (differences
less than 10%), while they result too conservative for lubricated interfaces
(overestimations up to 40%).
(4) The coefcient of friction tends to decrease during continuous
loading cycles at high velocities, due to self-heating of the sliding interfaces. The phenomenon is exhausted in a few cycles, due to the attainment of a new thermal equilibrium with the ambient. Based on the
98
M. DOLCE ET AL.
available experimental outcomes, the overall decay of friction with continuous loading cycles is of the order of 2530%, for non-lubricated
interfaces.
(5) Lubrication considerably reduces (by 58 times, depending on temperature) the frictional resistance of steel-PTFE sliding interfaces and, as
a consequence, the wear of PTFE. In addition, the use of silicone grease
strongly diminishes the gap between maximum and minimum friction
coefcients and, then, the differences in the structural response between
seismic and service conditions. On the other hand, the use of silicone
grease increases the sensitivity to temperature variations of the mechanical
behaviour of steelPTFE sliding bearings.
A mathematical model of the frictional behavior of steelPTFE sliding
interfaces has been presented, which takes into account the dependence of
the frictional force on sliding velocity, contact pressure and air temperature.
The proposed model describes the relationship between friction coefcient and sliding velocity through a logarithmic analytical law. The model
is characterised by two parameters, which are analytically expressed as a
function of contact pressure and air temperature, through a second-order
polynomial equation.
The accuracy of the proposed model in capturing the actual frictional
behaviour of steelPTFE sliding bearings has been veried by comparing
the experimental and numerical forcedisplacement loops of non-lubricated
and lubricated interfaces. It has been proved that the maximum frictional
force provided by the model can be reliably used in the design of seismically isolated structures.
Acknowledgements
The authors are indebted with Ing. Roberto Marnetto (TIS SpA), and
Mr Domenico Nigro (University of Bailicata) which have cooperated in the
setting up of the testing apparatus. This work has been partially funded by
MIUR, COFIN 2002.
References
AASHTO American Association of State Highway and Transportation Ofcials (1999)
Guide Specications for Seismic Isolation Design, 2nd Edition. American Association of
State Highway and Transportation Ofcials, Washington, DC.
CEN Comite Europeen de Normalisation. TC 167 Structural bearings (2000) EN 1337
Structural bearings, Part 2: Sliding elements, Brussel, Belgium.
BS British Standards Institution (1983) BS 5400: Steel, concrete and composite bridges,
London, UK.
99
Computers and Structures Inc. (2002) SAP2000 Analysis Reference Manual, Version 8.0,
Berkeley, CA.
Constantinou, M.C., Caccese, J. and Harris, H.G. (1987) Friction characteristics of PTFE
steel interfaces under dynamic conditions. Earthquake Engineering and Structural Dynamics. 15(6), 751759.
Constantinou, M., Mokha, A. and Reinhorn, A.M. (1990) PTFE bearings in base isolation:
modelling. Journal of Earthquake Engineering 116(2), 455472.
Chang J.C., Hwang J.S. and Lee G.C. (1990) Analytical model for sliding behaviour of
Teon-stainless steel interfaces. Journal of Engineering Mechanics 116, 27492763.
Deb, S.K. and Paul, D.K. (2000) Seismic response of buildings isolated by sliding-elastomer
bearings subjected to bi-directional motion. Proceedings of the 12th World Conference on
Earthquake Engineering, Auckland, New Zealand.
Dolce, M., Cardone, D. and Marnetto, R. (2000) Implementation and Testing of Passive
Control Devices Based on Shape Memory Alloys. Earthquake Engineering and Structural
Dynamics. (29), 945958.
Dolce, M. (2001) Remarkable design examples concerning recent applications of innovative
anti-seismic techniques to bridges and viaducts in Europe. Proceedings of the 7th International Seminar on Seismic Isolation, Passive Energy Dissipation and Active Control of
Structures, Assisi, Italy.
Fan, F.G., Ahmadi, G. and Tadjbakhsh, I.G. (1988) Base isolation of a multistory building under harmonic ground motion A comparison of performances of various systems.
Tech. Report NCEER-88-0010, National Center for Earthquake Engineering, State University of New York, Buffalo.
Hwang, J.S., Chang, K.C. and Lee, G.C. (1990) Quasi-static and dynamic characteristics of
PTFE-stainless interfaces. Journal of Structural Engineering 116(10), 27472762.
Mokha, A., Constantinou, M. and Reinhorn, A.M. (1990) PTFE bearings in base isolation:
testing. Journal of Earthquake Engineering 116(2), 438454.
Mokha, A., Constantinou, M., and Reinhorn, A.M. (1993) Verication of friction model of
PTFE bearings under triaxial load. Journal of Structural Division, ASCE, 119(1), 240261.
Mostaghel, N. and Tanbakuchi, J.T. (1983) Response of Sliding Structures to Earthquake
Support Motion. Earthquake Engineering and Structural Dynamics. 11, 729748.
Mostaghel, N. (1984) Resilient-Friction Base Isolator. Report No. UTEC 84/97, Department
of Civil Engineering, University of Utah, Salt Like City, USA.
Ordinanza del PCM No 3274/2003 (2003) Primi elementi in materia di criteri generali per la
classicazione sismica del territorio nazionale e di normative tecniche per le costruzioni in
zona sismica, Roma, Italy.
SPSS Inc. (1999), SPSS Advanced models Version 9.0, Chicago, Illinois.
Takahashi Y., Iemura H., Yanagawa S. and Hibi M. (2004) Shaking table test for frictional
isolated bridges and tribological numerical model of frictional isolator. Proceedings of the
13th World Conference on Earthquake Engineering, Vancouver, Canada.
Tsopelas, P., Constantinou, M.C., Okamoto, S., Fuji, S. and Ozaki, D. (1996) Experimental
study of bridge seismic sliding isolation system. Engineering Structures 18(4), 301310.
Tyler, R.G. (1977) Dynamic tests on PTFE sliding layers under earthquake conditions. Bulletin of the New Zealand National Society for Earthquake Engineering 10(3), 129138.
Wen, Y.K. (1976) Method for Random Vibration of Hysteretic Systems. Journal of the Engineering Mechanic Division, ASCE, 102 (EM2).
Zayas, V., Low, S. and Mahin, S. (1987) The FPS earthquake protection system: experimental report. Report No. UCB/EERC-87/01, Earthquake Engineering Research Center, University of California, Berkeley.