11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

Experimental dynamic tests on the first structure in the world isolated with rubber bearings
M. Garevski
Institute of Earthquake Engineering and Engineering Seismology, University "St. Cyril and Methodius", Skopje, Republic of Macedonia

J. M. Kelly
EERC, University of California at Berkeley,California, U.S.A.

M. Bojadziev
IZIIS, University "St. Cyril and Methodius", Skopje, Republic of Macedonia

Keywords: rubber bearings, aging effect, dynamic full scale measurements, natural frequencies, damping, 3D mathematical model ABSTRACT: The results obtained by experimental dynamic ambent and forced vibration tests on a base-isolated structure are presented and discussed in this paper. Since the isolators of the structure were placed 28 years ago, one of the goals of these investigations was to discover whether the isolators have been affected by aging, i.e., whether there has been a change of the dynamic characteristics of the structure. Rocking motion due to low vertical stiffness of the rubber isolators was analyzed.The influence of the reinforced concrete plates covering the seismic gap on the stifness characteristics of the structure was also examined. Based on these tests, some of the necessery characteristics of the isolators to be used for formulation of the 3D mathematicla model of the school were obtained. 1 INTRODUCTION The primary school "Pestalozzi" in Skopje built in 1969 is the first building in the world for which natural rubber isolators were used for its protection against strong earthquakes (Kelly, 1981, Izumi, 1988). Since the isolators on which the school is placed are more than 28 years old, a project is currently being carried out jointly by IZIIS (Institute of Earthquake Engineering and Engineering Seismology, University "St. Cyril and Methodius, Skopje, Republic of Macedonia) and EERC (University of California at Berkeley) to evaluate the state of the isolators after passing of so many years. The behaviour of the school building under the effect of strong earthquakes shall be analyzed by using a 3D mathematical model that shall include all the present characteristics of the isolators. To accurately formulate this model, it is necessary to consider the effect of aging of the isolators, i.e., define their present stiffness characteristics and equivalent damping. For the purpose of defining the mentioned characteristics, experimental laboratory tests of two isolators taken from the foundation of the school building "Pestalozzi" are planned to be performed in the beginning of June 1998. To define the dynamic characteristics, in-situ, ambient and forced vibration tests were performed obtaining thus the fundamental and some of the higher frequencies and the corresponding mode shapes for the three orthogonal directions of the school-building.Two torsional modes of vibration and equivalent dampings for all the measured mode shapes were also defined. The results from the ambient and forced vibration tests of the school building "Pestalozzi" are given in the subsequent text. First of all, the school building is described and the characteristics of the bearings are given. Then, the technical data on the equipment used for measuring of the dynamic characteristics of the 1

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

building are presented and the measuring procedures are described. At the end of the paper, some of the results from the performed measurements are presented and discussed. 2 DESCRIPTION OF THE BUILDING The building which is subject of the investigations, i.e., the primary school "J.H. Pestalozzi" in Skopje was donated to Skopje by the Swiss Government after the catastrophic earthquake of July 26, 1963. The school building consists of several units of different number of storeys and outline. The base isolation system has been applied only for the main school building, while the other units are founded in the classical way. 2.1 Description of the structural system of the base isolated part The main school building consists of a ground floor and two storeys. The proportions of the stricture at plan are 13.0 m/61.5 m, while its height is 10.0 m. It represents a reinforced concrete box system. The bearing wall system is composed of shear walls with a thickness of 0.18 m. The floor slab is 0.20 m thick. The building has a strip foundation forming a beam grid which is sufficiently rigid to sustain all the effects from the upper structure without heavier deformations. Along the edges of the building, there is a seismic gap with a width of 0.3 m. The gab is covered with reinforced concrete plates. The structure is base isolated by its placement on special rubber bearings incorporated between

Figure 1. Rubber bearings placed between the strip foundation and the building.

the foundation structure and the first floor slab (Fig.1). There is a total of 54 bearings made of natural rubber, with a square shape, a side of 0.7 m and a height of 0.3 m. Each bearing transfers an axial force amounting to approximately 500 kN. The rubber bearings are produced by the Swiss firm HUBER-SUHNER from Zurich. The isolators do not contain reinforcing steel plates. They are fabricated by gluing together several rubber layers. The vertical stiffness of such bearings is therefore not much greater than the horizontal one. In addition to the negative effect of the low vertical stiffness of the bearings on the possible rocking motion, there is another disadvantage of these bearings which is the occurrence of large lateral deformations of the rubber bearings due to dead load effects.

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

3 DYNAMIC TESTING OF THE PRIMARY SCHOOL "J. H. PESTALOZZI" Dynamic testing of the main (isolated) school building was performed by application of two independent testing methods: Ambient vibration method- This method was used for fast obtaining of preliminary results on the dynamic characteristics - resonant frequencies and mode shapes of vibration of the tested structure. The vertical vibration of the structure obtained by this method was used as a referent one since the vibrators used for the forced vibration tests cannot induce vertical displacements. To perform the tests for definition of the dynamic characteristics of the considered structure, an ambient vibration equipment consisting basically of sensors, signal amplifier and recorder was used. Four Range seismometers, Model SS-1 produced by Kinemetrics from USA were used as sensors. The natural frequency of the seismometer is 1 Hz, while the damping is 0.7 of the critical one. The signal adjuster model SC-1 was used for signal amplification and simultaneous control of the four seismometers. There was a possibility of simultaneous or independent recording at the outputs and each channel provided a nominal increase of 100.000 times. The records were processed in situ by a Spectrum Analyzer, Model 3582A. Forced vibration method - This method was applied to define some characteristic vibrations of the structure in both horizontal directions as well as the torsional one. According to its function, the equipment used for these dynamic tests is divided into two parts:

Equipment for forced vibration generation, and Equipment for recording of the dynamic response of the structure.

To generate forced vibrations on the structure within a frequency range of 0.5 Hz to 9.0 Hz, two vibrators of the type of GSV-101 produced by Geotronix - USA were used. Each of the vibrators was provided with an electric motor with a power of 1.1 kW. The intensity of generated force per vibration exciter was limited to 24.5 kN, depending on the mass of the baskets of the vibration exciter and the square of the frequency by which they rotate.To induce torsional vibration of the structure, the vibration exciters worked with a phase difference of () radians. The dynamic response of the structure was recorded by six accelerometers type A4-0.25 produced by Statham - USA. These accelerometers can record only the horizontal component of the structural response. The measuring range of these accelerometers is 0.25 in a frequency range of 0 to 16 Hz. To record the vertical component of the structural response, three accelerometers type Kistler 8710A50M1 were used in combination with a system for signal amplification type Kistler 5134. The measuring range of these accelerometers is 50g in a frequency range of 1 Hz to 5000 Hz. The signal recording was done by the same spectral analyzer that was used for the ambient vibration tests. To define the damping coefficients, two methods were used: (1) half-power bandwidth (frequency-response) method and (2) logarithmic decrement (time-response) method. 4 EXPERIMENTAL RESULTS The dynamic characteristics of the building were obtained for two characteristic states: the state of presence of reinforced concrete plates covering the seismic gap separating the structure from its surrounding (state I) and the state of reinforced concrete plates being removed from the seismic gap and free motion of the isolated building in all directions (state II). These two states were considered to determine the role of these plates in dynamic behaviour of the structure, i.e., whether they act as fixation of the structure.

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

Fig. 2 displays the plan of the second storey of the structure with marked measuring points and position of vibration exciters by which it was excited during the forced vibration tests.

T. 1

T .2

T. 3

T. 4

T. 5

T. C T. 7 8.45 3.58 8.39 8.39

3.58 61.20

8.39

8.39

3.58

8.45

S
Figure 2. Plan of the second storey of the tested structure with presentation of measuring points and position of the two vibration exciters.

4.1 Results from dynamic testing of the structure state I Results from the tests performed by applying the ambient vibration method: To define the resonant frequencies of the structure in both orthogonal directions - transverse (N-S) and longitudinal (EW), and the vertical one, gauges were placed in the central part of the structure, at the level of the second storey, at point "C" (Fig. 2). Gauges were placed at the ends of the structure (measuring points 1 and 5) and also at the level of the second storey to define the resonant frequencies of the torsional vibration. The Fourier amplitude spectra corresponding to the torsional vibrations of the structure were obtained by combining the signals from both gauges. Fig. 3 shows the Fourier amplitude spectra recorded on the structure in the corresponding directions of vibrations from which the resonant frequencies were defined. For such defined resonant frequencies of the structure, the corresponding mode shapes were obtained . However, these and the mode shapes obtained by other measurements are not presented in this paper due to lack of space. Results from tests performed by using the forced vibration method: In the process of testing, the resonant frequencies of the structure were first of all defined for both orthogonal directions (N- S) and (E-W) and then for torsion. The structural response to the generated excitation, while defining the resonant frequencies for the N-S and E-W directions, was recorded at several measuring points. The recorded responses at the points located in the midst of the structure (point "C" or point "3") were considered referent. The referent point for definition of the torsional resonant frequencies was point "5" on the east side of the structure. The frequency curves presented in Fig 4 were established based on the recorded amplitudes of structural response to the generated excitations. For such defined resonant frequencies, the corresponding mode shapes of vibration were recorded.

11.42

GSV-101

T. 6

. .. .

GSV-101

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

160

Amplitude (mV)

N-S
80 40 2.0
f = 3.76 Hz f = 6.56 Hz f = 6.08 Hz

Amplitude (mV)

f = 1.60 Hz

40 30 20 10 f = 1.60 Hz 2.0 4.0 6.0 8.0 10.0

f = 2.40 Hz f = 2.72 Hz f = 4.72 Hz f = 5.84 Hz

120

E-W

4.0

6.0

8.0

10.0

Frequency (Hz)
40 160

Frequency (Hz)

Amplitude (mV)

Amplitude (mV)

30 20 10
f = 2.48 Hz

Torsion

120 80 40

f = 1.60 Hzf = 2.72 Hz f = 3.76 Hz f = 6.08 Hz

Vertical

2.0

4.0

6.0

8.0

10.0

2.0

4.0

6.0

8.0

10.0

Frequency (Hz)

Frequency (Hz)

Figure 3. Fourier amplitude spectra recorded on the tested structure (ambient, state I).

300.00

f = 1.24 Hz = 6.6 %
250.00

N-S

240.00

S4+L2 Displacement (microns) 200.00

f = 1.92 Hz = 9.7 %

torsion
S4+L2

Displacement (microns)

200.00

f = 2.48 Hz 160.00

150.00

100.00

120.00
50.00

0.00

80.00 1.00 1.50 2.00 2.50 Frequency (Hz) 3.00 3.50 1.60 2.00 2.40 2.80 Frequency (Hz) 3.20

150.00 f = 2.48 Hz = 3.8 % Displacement (microns)

E-W

S4+L2

100.00

f = 2.00 Hz 50.00

0.00 1.00 1.50 2.00 2.50 Frequency (Hz) 3.00

Figure 4. Resonant curves of the structural response (force excitation, state I).

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

4.2 Results from dynamic testing of the structure state II In this phase of testing of the structure, all the reinforced-concrete plates covering the seismic gap were removed. In this way, free motion of the structure in all the directions was enabled. The procedure, the number of tests and the number of measuring points and their locations for determination of the dynamic characteristics was identical to that applied in the case when the seismic gap was covered with the plates. Hence the results obtained for both cases could directly be compared. Results from Dynamic Tests Performed by Applying the Ambient Vibration Method: Fig. 5 presents the Fourier amplitude spectra recorded on the structure in the corresponding directions of vibration. In this case also, the mode shapes of the characteristic resonant frequencies were measured. Results from tests performed by applying the forced vibration method: Based on the recorded amplitudes of the response of the structure to the generated excitations, the corresponding resonant
80 80

Amplitude (mV)

60 40 20

Amplitude (mV)

N-S
f = 0.96 Hz

f = 1.28 Hz

E-W

60 40 20
f = 2.64 Hz f = 3.60 Hz

f = 2.24 Hz f = 5.52 Hz f = 3.60 Hz

2.0

4.0

6.0

8.0

10.0

2.0

4.0

6.0

8.0

10.0

Frequency (Hz)
80 160
f = 1.04Hz

Frequency (Hz) Torsion

Amplitude (mV)

60 40 20

Amplitude (mV)

Vertical
f = 2.56 Hz

120 80 40

f =3.5Hz f =5.44Hz

f = 3.60 Hz f = 4.18 Hz f = 6.16 Hz

2.0

4.0

6.0

8.0

10.0

2.0

4.0

6.0

8.0

10.0

Frequency (Hz)

Frequency (Hz)

Figure 5. Furier amplitude spectra recorded on the isolated part of the building (ambient, state II).

frequency curves were established. Certain tests performed for definition of the same frequency were carried out by different excitation forces for the purpose of evaluating the relation between modification of the resonant frequency and the excitation intensity. Due to the flexibility of the isolators, some of the mode shapes could quite well be excited and even visually observed. Therefore, apart from using accelerometers which is the usual practice in performing these tests, displacemement-meters (LVDT-s) were also used. The frequency curves of the displacement response of the structure are presented in Fig. 6. These were used to directly define the resonant frequencies. The mode shapes for the case of an open seismic gap were find also.

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3
1000.00 f = 0.76 Hz 800.00 Displacement (microns) = 3.8 %

N-S
S4+L4 Displacement (microns)

1600.00

f = 0.84 Hz = 5.4 %

torsion
S4+L4

1200.00

600.00

400.00 f = 2.08 Hz 200.00

800.00

400.00

0.00 0.50 1.00 1.50 2.00 Frequency (Hz) 2.50

0.00 0.50 1.00 1.50 2.00 Frequency (Hz) 2.50

400.00 f = 0.88 Hz = 7.3 % Displacement (microns) 300.00

E-W
S4+L4

200.00

100.00

0.00 0.40 0.80 1.20 1.60 Frequency (Hz) 2.00 2.40

Figure 6. Resonant curves of the structural response (force excitation, state II).

4.3 Mode shapes and dampings Based on the data obtained from the performed testing of the structure, the following is concluded: The values of the frequencies of the fundamental vibration modes in certain directions for state I are: for transverse direction (N-S): fI1f = 1.24 Hz (forced vibration tests) and fI1a = 1.6 Hz (ambient vibration tests); for torsion: fI2f = 1.92 Hz (forced vibration tests) and fI2a = 2.48 Hz (ambient vibration tests); for longitudinal direction (E-W): fI3f = 2.48 Hz (forced vibration tests) and fI3a = 2.40 Hz (ambient vibration tests) and for vertical direction: fI4a = 2.72 Hz (ambient vibration tests). The values of the frequencies of the fundamental vibration modes in certain directions when the plates covering the seismic gap are removed (state II) are: for transverse direction (N-S) = fII1f = 0.76 Hz (forced vibration tests) and fII1a = 0.96 Hz (ambient vibration tests); for torsion: fII2f = 0.84 Hz (forced vibration tests) and fII2a = 1.04 Hz (ambient vibration tests); for the longitudinal direction (E-W): fII3f = 0.88 Hz (forced vibration tests) and fII3a = 1.28 Hz (ambient vibration tests) and for vertical direction: fII5a = 2.56 Hz (ambient vibration tests). Apart from the fundamental vibration modes, some of the higher modes were recorded for state II. The frequency of fII4f = 2.08 Hz represents the second natural frequency of the structure in the (N-S) direction, while f II6f= 3.20 represents the second natural frequency at torsional excitation. The second natural frequency of the structure in the E-W direction is very close (almost equal) to the corresponding frequency at torsional excitation amounting to f = 3.20 Hz, i.e., f II7f= 3.28 Hz.

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

The recorded higher frequencies of f II 8f = 3.44 Hz, f II9f= 5.28 Hz, fII12f = 5.92 Hz in the transverse direction and fII10f = 5.28 in the longitudinal direction, i.e., fII11f = 5.52 Hz at torsional vibrations are associated with some of the higher mode shapes of vibration of the structure. The frequencies and the corresponding damping measured by applying both methods are presented in Table 1.
Table 1. Presentation of the resonant frequencies and damping of the fundamental modes of the structure . Mode No. Type Ambient f (Hz) 1.6 2.4 2.48 2.72 State I Forced vibration f Damping (%) (Hz) HP * LD** 1.24 4.0-6.6 5.9 2.48 3.8 3.6 1.92 9.7 Ambient f (Hz) 0.96 1.28 1.04 2.56 State II Forced vibration f Damp ing (%) (Hz) HP* LD** 0.76 3.8 3.5 0.88 7.3 6.3 0.84 5.4 3.0 2.08 2.9 3.00 2.0 2.5 3.28 5.5 3.0

1 first (N-S) 2 first (E-W) 3 torsion 4 vertical 5 second (N-S) 6 sec.torsion 7 second (E-W) * Half-power bandwidth method.

**

Logarithmic decrement method..

Analyzing the recorded mode shapes, it can be said that the translatory components dominate over the first two modes of vibration of the school building (f1fII = 0.76 Hz, f2fII = 0.88 Hz, State II). Characteristic for the second vibration modes in E-W direction (f4fII = 2.08 Hz) and N-S direction (f7fII = 3.28 Hz) is the rocking motion which is more pronounced for the mode with a frequency of f4fII = 2.08 Hz. Such vibration modes occur as a result of the low vertical stiffness of the rubber bearings. 5 DISCUSSIONS AND CONCLUSIONS First of all it may be said that the assumption about the rocking motion of the school building proved to be true by the recorded mode shapes. The ratio between the vertical and the horizontal frequency indicates that the vertical stiffness of the isolators is only 3 times greater than their horizontal stiffness. It can therefore be said that the vertical stiffness is very low, having in mind that new isolators can possess more than 1000 times higher vertical stiffness than the horizontal one. Since the fundamental modes for the two horizontal directions are not purely translatory (there are rotation components), their frequencies are not identical. There occurs a difference of 18% in the values of their frequencies. The obtained results on damping presented in Table 1 point to similar damping obtained by both methods. The damping values (State II) lead to the conclusion that the internal damping of the bearing is small since the coefficients of viscous damping for the fundamental vibration modes are 3.5%, 3% and 6.3% of the critical one in the N-S and E-W directions as well as torsion respectively. Comparing the values of resonant frequencies for the fundamental vibration modes (Table 1) recorded in both the considered cases (the structure with and without the covering plates), it is concluded that the plates have a considerable effect in both orthogonal horizontal directions and rotation of the structure about the vertical axis. It can also be concluded that the stiffness characteristics are considerably altered since the mass of the structure is constant. For instance, the increase in stiffness due to the presence of the plates in the N-S direction is 39%. In the E-W direction, it is 65%, while the torsional stiffness is increased for 56%. Only the frequencies dominating in the Fourier amplitude spectra referring to the vertical direction of vibration are not deviated from the original ones (from f Ia4= 2.72 Hz to f IIa4 = 2.56), which is a difference of about 6%. Hence, it is 8

11th European Conference on Earthquake Engineering 1998 Balkema, Rotterdam, ISBN 90 5410 982 3

concluded that the reinforced concrete plates over the seismic gap of the structure stiffen the structure in the horizontal direction only and do not have any effect in the vertical direction. Finally it may be concluded that the results obtained from these investigations shall contribute a lot to the achieving of a greater accuracy in preparing the mathematical models to be formulated for analysis of the school building. This particularly refers to the mathematical model for State I which cannot be formulated without such measurements. The measured dampings are also necessary for the formulation of the models for State I and State II. ACKNOWLEDGMENT This publication is based on the investigations sponsored by the Macedonian US. Joint Fund in cooperation with the Macedonian Ministry of Science and US - NIST under project number 056/600. The authors of this paper and indebted to these two institutions for the correct evaluation of the necessity for these investigations. REFERECES
Bridgestone corporation, 1991. Rubber technology for seismic isolation. Proc., 11th international conf. on struc. mech. in reactor technology: 45-56, Tokyo, Japan. Garevki, M. & Kelly, M. J. 1996. Evaluation of the proper functioning of the rubber isolators of the primary school Pestalozzi in Skopje under strong earthquake. U.S.- Macedonian science and technology programproject proposal. Garevki, M., Kelly, M. J. & Bojadjiev, M., in prep. Evaluation of the proper functioning of the rubber isolators of the primary school Pestalozzi in Skopje under strong earthquake : part I- Full-scale dnamic teting of the first rubber base-isolated building in the world . Report. Izumi, M., 1988. Base isolation and passive seismic response control. Proc., 9th world conference on earthquake engineering, vol. VII: 386-396,Tokyo-Kyoto, Japan,. Kelly, J.M., 1981. The Development of base isolation for the seismic protection of structures. Report No. UCB/EERC-81/01, Earthquake engineering research center, University of California, Berkeley, California. Kelly, J.M., 1988. Base isolation in Japan. Report No. UCB/EERC-88/20, Earthquake engineering research center, University of California, Berkeley, California. Kelly, J.M., 1993. Implementation of base isolation in the United States. Proc., 6th symposium on seismic, shock and vibration isolation, ASME pressure vessel and piping conference: 159-170, Denver, Colorado. Staudacher, K., 1982. The Swiss full base isolation system (3D) for extreme earthquake saifty of structures. Proc. of the 1982 convention of the structural engineerin asotiation of Californija, Sacaramento, California.