Seismic Response of High-Voltage Transformer-Bushing Systems Incorporating Flexural Stiffeners II: Experimental Study Maria Koliou,a) M.EERI, Andre Filiatrault,b) Andrei M. Reinhorn,c) M.EERI

M.EERI,

and

A numerical study conducted on four transformer-bushing models presented in a first companion paper indicated that high-voltage bushings mounted on the cover plates of transformers are more vulnerable to seismic loading than bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field. In this second companion paper, the addition of flexural stiffeners on the transformer cover plates as a means to stiffen the base of bushings and mitigate their seismic vulnerability is investigated experimentally. Shake table testing was conducted on a 230 kV porcelain bushing mounted on a support structure incorporating a flexible cover plate and two stiffener configurations. Test results confirmed that stiffening the cover plates is beneficial to the seismic response of high-voltage bushings. Test results are compared to the predictions of finite element analyses. [DOI: 10.1193/072511EQS185M]

INTRODUCTION The seismic response of high-voltage (at or above 220 kV) bushings is influenced by both their lateral stiffness and the flexural stiffness of the cover plates of the transformer tanks onto which they are mounted on. The results of the numerical study conducted on four transformer-bushing models and presented in a first companion paper (Koliou et al. 2013) indicated that bushings mounted on the cover plates of transformer tanks are more vulnerable to seismic loading than similar bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field during real earthquakes (Ersoy et al. 2008, Filiatrault and Matt 2006, Whittaker et al. 2004). The improved seismic performance of bushings mounted on a rigid base is associated with an increase of their fundamental frequencies causing a reduction of the seismic demand. Moreover, stiffening the transformer cover, reduces the rotational and vertical motion at the base of the bushing (Reinhorn et al.

a)

Graduate Research Assistant, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. b) Professor, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. c) Professor Emeritus of Structural Engineering, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. 1353

Earthquake Spectra, Volume 29, No. 4, pages 1353–1367, November 2013; © 2013, Earthquake Engineering Research Institute

1354

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

2011a) while reducing the bushing response and forces at the mounting interface (Koliou et al. 2013). The results of the numerical study showed also that the introduction of horizontal flexural stiffeners on the cover plates of transformer tanks near the base of the bushings is a viable and simple approach to shift the fundamental frequencies of bushings near that of rigid base condition and, thereby, improve the seismic response of transformer-bushing systems. The stiffeners should have sufficiently high flexural rigidity to maximize the fundamental frequencies of the bushings. This maximum fundamental frequency occurs when the base of the bushing is made locally rigid and is governed by the global flexural flexibility of the tank cover plate and walls (Koliou et al. 2012). In this second companion paper, the seismic response of transformer-bushing systems incorporating flexural stiffeners on the cover plates of transformer tanks is investigated experimentally to confirm the numerical results reported in the first companion paper. The experimental study incorporated two types of testing: (i) system identification testing and (ii) shake table (seismic) testing. Test results are also compared to the predictions of finite element analyses. SYSTEM IDENTIFICATION TESTING DESCRIPTION OF TEST SPECIMEN

The specimen used for the system identification testing consisted of a 230 kV porcelain bushing bolted to a reinforced concrete slab through an embedded adaptor plate. The slab was anchored to the strong floor of the Structural Engineering and Earthquake Simulation Laboratory (SEESL) of the University at Buffalo (UB) to simulate rigid base condition. The bushing structure was 151.4 in. tall while the concrete slab had plan dimensions of 8 ft 8 ft and a thickness of 1 ft. A lumped weight of 25 lbs was added at the top of the bushing to simulate the inertia of the electrical conductors, as required by the IEEE-693 Standard (IEEE 2005) for seismic qualification testing of electrical equipment. Figure 1 shows photographs of the bushing specimen and experimental setup used for the system identification testing. The physical properties of the bushing specimen provided by the manufacturer are summarized in Table 1. More details on the test set-up can be found in Koliou et al. (2012).

TEST PROTOCOL AND INSTRUMENTATION

System identification testing consisted of impact hammer tests to evaluate the natural frequencies and damping characteristics of the bushing specimen mounted on a rigid base. The top of the bushing was impacted with a rubber hammer in two orthogonal directions (north-south and east-west directions in Figure 1). Impact hammer tests were not conducted in the vertical direction since the effects of vertical ground motions were not considered in this study. The dynamic response of the bushing specimen during system identification testing was recorded by 19 instruments, including five accelerometers along the height of the bushing, two linear potentiometers at the top of the bushing and four strain rosettes installed on the mounting flange of the bushing (three strain gauges per rosette) to measure accelerations, displacements and strains, respectively. Details of the instrumentation used for the system identification testing are provided in Koliou et al. (2012).

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1355

Dome

230 kV Porcelain Bushing

Upper Porcelain Unit

Reinforced Concrete Slab Steel Plate at the top of the concrete slab

Sheds Mounting Flange

Lifting Lugs

(b)

(a)

Figure 1. (a) Specimen configuration used for system identification testing and (b) details of 230 kV porcelain bushing specimen.

Table 1. Properties of 230 kV porcelain bushing (Koliou et al. 2012, Reinhorn and Muhammad 2011) Material insulator Voltage capacity Total height Length over mounting flange Length below mounting flange Maximum diameter over mounting flange Maximum diameter below mounting flange Diameter of mounting flange Diameter of bolt pattern on mounting flange Number of bolts on mounting flange Diameter of bolts on mounting flange Total weight Location of center of gravity above flange Upper unit weight Location of upper unit center of gravity Lower unit weight Location of lower unit center of gravity

Porcelain (kV) (in) (in) (in) (in) (in) (in) (in) (in) (in) (lbs) (in) (lbs) (in) (lbs) (in)

230 151.4 91.4 60.0 11.8 10.0 24.0 21 12 1.25 840.0 14.0 447.0 34.0 293.0 28.0

1356

Table 2. base

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

System identification test results for 230 kV porcelain bushing mounted on a rigid

Impact hammer test direction North-south East-west

Fundamental frequency (Hz)

First modal equivalent damping ratio (%)

25.3 25.4

2.3 2.1

TEST RESULTS

The fundamental frequency and first modal equivalent damping ratio for the 230 kV porcelain bushing specimen mounted on a rigid base obtained from the impact hammer tests in each orthogonal direction are shown in Table 2. The fundamental frequency of the bushing specimen was obtained by plotting the Fourier amplitude spectrum from the acceleration time-histories recorded at the top of the bushing during each impact hammer test. The first modal equivalent viscous damping ratio was estimated by the half-power bandwidth method (Bracci et al. 1992). The results for the fundamental frequency and damping ratio are consistent for both orthogonal directions of the bushing. SHAKE TABLE (SEISMIC) TESTING DESCRIPTION OF TEST SPECIMEN

The specimen used for the shake table (seismic) testing consisted of the same 230 kV porcelain bushing described above mounted on a support structure representing the generic cover plate of a transformer tank (Reinhorn et al. 2011b, Kong 2010). As shown in the photograph of Figure 2, the support structure consisted of a rigid frame, a cover plate and an adaptor plate (attached to the cover plate). The rigid frame had dimensions of 8 ft 8 ft 8 ft, while the ¾ in. thick cover plate had plan dimensions of 127 in. 127 in. A steel square tube (TS 5 5 1/2) was used for the four corner columns of the rigid frame. The top of each column was connected to the cover plate by horizontal L5 5 3/4 angles. The four bays of the rigid frame were stiffened by L5 5 3/4 cross bracing welded together at mid-span (Kong 2010). In order to evaluate the addition of flexural stiffeners on the tank cover plate as a means to reduce the bending moment demand at the base of the bushing during earthquake shaking two different sets of steel angles, L8 6 1/2 and L6 4 1/2, were considered. These angle sections were bolted on the cover plate in both perpendicular directions of the frame (two on top of the cover plate in the North-South direction and two underneath it along the east-west direction). Each set of stiffeners was installed in a 4 ft 4 ft square pattern, with each stiffener located at a distance of 2 ft from the center line of the bushing, as shown in the photographs of Figure 3. Detailed drawings of the test specimen used for the shake table tests are provided in Koliou et al. (2012). Shake table testing was conducted for both sets of stiffeners described above as well as for the unstiffened cover plate.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1357

Figure 2. Specimen used for shake table (seismic) testing.

Figure 3. Steel angle stiffeners bolted on the cover plate: (a) On top of the plate in the northsouth direction, (b) underneath the plate in the east-west direction. INSTRUMENTATION

A total of 46 instruments were used to monitor the seismic response of the bushing and of the support structure during shake table testing. The same 19 sensors used for the system identification testing (see above) were installed again on the bushing, seven instruments (three accelerometers and four linear potentiometers) were used to measure the dynamic response of the support structure, 13 accelerometers were placed on the cover plate to record its dynamic response, and seven instruments (three accelerometers and four linear potentiometers) were installed on the shake table in order to measure the achieved input motions. A detailed description of the instrumentation used in the shake table testing is provided in Koliou et al. (2012). It should be noted that the set of strain rosettes installed on the mounting flange of the bushing were calibrated to measure directly the shear forces and bending moments transmitted to the cover plate from the bushing assembly.

1358

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

EARTHQUAKE GROUND MOTIONS

Similar to the numerical studies described in the first companion paper (Koliou et al. 2013), the FEMA P695 Far Field Ground Motion Set (FEMA P695 2009) were considered for seismic testing. This ensemble contains 22 historical motions (2 horizontal components per motion) from all over the world and is considered to be representative of the seismicity in the Western United States. This motion ensemble was scaled to match the IEEE-693, 2% damped, high required response spectrum in a range of frequencies from 2.0 Hz to 30.0 Hz, as described in the first companion paper (Koliou et al. 2013). Information including response spectra for these motions have also been provided in this first companion paper (Koliou et al. 2013). The effects of vertical ground motions were not considered in this study. Vertical input would have a negligible effect on the vertically positioned bushing specimen. To limit the number of seismic tests, a reduced test motion ensemble, consisting of five pairs of ground motions (two components each) out of the 22 P695 motions, was considered for seismic testing (Koliou et al. 2012). The five test ground motions were selected to have similar values as the complete P695 motion set for several selected statistical parameters of interest. The statistical parameters selected were the median, arithmetic mean, geometric mean, standard deviation, maximum and minimum values of a characteristic spectral value in a range of frequencies varying from 10 Hz to 25 Hz. This frequency range was selected to encompass the range of natural frequencies of the bushing specimen mounted on a rigid base (see Table 2 above) and on the support structure (see Table 5 below). The test motion ensemble was also selected to include no more than one motion per historical seismic event. The steps followed to select the test motion ensemble can be summarized as follows: 1.

2.

3. 4.

For each of the 22 P695 motions, the geometric mean of the 2% damped spectral accelerations for the two horizontal components were computed for individual frequencies to obtain a geometric mean response spectrum. The geometric mean of the 2% damped spectral accelerations was calculated for each of the geometric mean spectra across the frequency range of interest to provide a single characteristic spectral value for each motion. Each statistical parameter of interest of the characteristic spectral value was computed across the 22 motions. The 22 motions were listed in ascending order of their characteristic spectral value, and different ensembles of five motions were investigated in order to identify the combination that matched the best its statistical parameter of interest to that of the full ensemble of 22 motions. Note that the geometric mean was selected as the characteristic spectral value, as it provides an orientation-independent measure of earthquake intensity (Boore et al. 2006). A similar selection procedure was used by Sideris et al. (2010).

The final test motion ensemble is presented in Table 3. A comparison of the various statistical parameters of the characteristic spectral value between the full FEMA P695 Far Field Ground Motion Set and the reduced test motion ensemble is provided in Table 4. TESTING PROTOCOL

The shake table (seismic) testing included three phases: in the first phase, the larger flexural stiffeners (L8 6 1/2) were installed on the cover plate of the rigid frame as described earlier.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

Table 3.

1359

Earthquake ground motion subset considered for seismic testing Earthquake event

Test motion

ID.

EQ1 EQ2 EQ3 EQ4 EQ5

12011 12041 12072 12092 12132

Name

Year

Mw

Recording station

Northridge Duzce, Turkey Kobe, Japan Landers Cape Mendocino

1994 1999 1995 1992 1992

6.7 7.1 6.9 7.3 7.0

Beverly Hills-Mulhol Bolu Shin-Osaka Coolwater Rio Dell Overpass

PGA (g)* 0.52 0.82 0.24 0.42 0.55

*Larger component

Table 4. Comparison of statistical parameters of characteristic spectral value between FEMA P695 far field ground motion set and test motion ensemble Values (g) Statistical parameters of characteristic spectral values Median Arithmetic mean Geometric mean Standard deviation Maximum Minimum

FEMA P695 far-field ground motion set (22 historical motions/two horizontal components per motion)

Ensemble of five test ground motions (two horizontal components per motion)

0.909 1.026 0.927 0.495 2.531 0.412

0.924 1.011 0.903 0.529 1.825 0.426

During the second phase, the smaller flexural stiffeners (L6 4 1/2) were bolted to the cover plate. Finally in the third testing phase, all stiffeners were removed and the bushing specimen mounted on the unstiffened cover plate of the rigid frame was tested. Low amplitude white noise tests were also conducted between each seismic test to study the evolution of the fundamental frequency and first modal equivalent damping ratio of the bushing specimen through the testing program. Because the bushing specimen and the support structure remained in the elastic range of their material during all seismic tests, the bushing fundamental frequency and equivalent damping ratio remained essentially constant throughout the testing program. RESULTS OF WHITE NOISE TESTS

The results of the initial white noise tests are shown in Table 5. As expected, the measured fundamental frequency of the bushing specimen increases with the rigidity of the stiffeners installed on the cover plate of the support structure. The fundamental frequencies associated with both stiffener configurations lie between the frequency associated with the unstiffened cover plate and that of the rigid base condition (see Table 2). These results are consistent with the results of the numerical study presented in the first companion paper

1360

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

Table 5. Measured bushing fundamental frequency and first modal equivalent damping ratio during white noise tests Test phase 1: With stiffeners L8 6 1/2 2: With stiffeners L6 4 1/2 3: Without stiffeners

Fundamental frequency (Hz)

First modal equivalent damping ratio ξ (%)

16.1 15.1 10.1

4.1 3.2 1.9

(Koliou et al. 2013). The first modal equivalent damping ratio remains under 5% of critical for all mounting configurations, including the rigid base condition shown in Table 2. Although the increase of damping from ∼2% to ∼4% may account for a decrease of spectral response of up to 15%, for the fundamental frequencies of the bushing assemblies, the contribution of damping increase to the response reduction is less than 10%. RESULTS OF SEISMIC TESTS

The bending moment induced at the base of the bushing specimen was the structural response parameter of interest in this experimental study. For each seismic test, the maximum bending moment at the base of the bushing, M max , was computed according to: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ M max ¼ maxt (1) M 2x ðtÞ þ M 2y ðtÞ EQ-TARGET;temp:intralink-;e1;41;372

where Mx(t) and My(t) are the moments at the base of the bushing specimen at time t recorded by the strain rosettes with respect to the longitudinal (east-west) and transverse (north-south) axis of the support structure, respectively; while maxt is the maximum absolute value over the recorded time-history. For each of the three test phases, experimental cumulative distribution functions (CDF) were calculated for the probability of non-exceeding (PoNE) a prescribed maximum moment at the base of the bushing specimen under the test motion ensemble. The PoNE is estimated by counting the number of test ground motions causing a prescribed value of the maximum bending moment at the base of the bushing specimen not to be exceeded and dividing this number by five (the total number of test ground motions). A lognormal cumulative distribution function (CDF) can then be fitted to the empirical data. The lognormal CDF is defined by the median value (PoNE = 50%) of the maximum bending moment, ^ max and by the dispersion parameter β expressed as the standard deviation of the log of M the values of M max . The resulting CDFs shown in Figure 4 indicate that the maximum bending moments recorded at the base of the bushing specimen reduce with increasing size of flexural stiffeners on the cover plate. The median maximum bending moments for the largest stiffeners (L8 6 1/2) is substantially reduced (30%) compared to that of the median bending moment recorded at the base of the bushing mounted on the unstiffened cover plate. The same trend can also be observed for the relative displacement and the absolute acceleration at the top of the bushing specimen, as shown in Figures 5 and 6, respectively. These

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1361

Probability of Non - Exceedance

1.0 0.9

w/o Stiffeners

0.8

w/ Stiffeners L6x4x1/2

0.7

w/ Stiffeners L8x6x1/2

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0

20.0

40.0

60.0

80.0

100.0

Maximum Bending Moment (kip-in)

Figure 4. Experimental CDF for recorded maximum bending moments at the base of the bushing specimen.

Maximum Relative Displacement at the Top of the Bushing (in)

5.00 4.50

4.54 4.42

w/o Stiffeners 4.25

w/ Stiffeners L6x4x1/2

4.00 3.33

3.50 3.00

w/ Stiffeners L8x6x1/2

3.41 3.17 2.82

2.97 2.91 2.41

2.50

2.17

2.00

1.77 1.66

1.97

1.41

1.50 1.00 0.50 0.00 EQ1

EQ2

EQ3

EQ4

EQ5

Figure 5. Maximum relative displacements measured at the top of the bushing specimen.

Maximum Absolute Acceleration at the Top of the Bushing (g)

3.00

w/o Stiffeners

2.73

w/ Stiffeners L6x4x1/2

2.50 2.18

w/ Stiffeners L8x6x1/2

2.03

1.88

2.00 1.71

1.50

1.49

1.57

1.35 0.98

1.00

0.83

0.82

1.03 0.88

0.45 0.39

0.50 0.00 EQ1

EQ2

EQ3

EQ4

EQ5

Figure 6. Maximum absolute accelerations measured at the top of the bushing specimen.

1362

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

experimental results correlate well with the numerical results presented in the first companion paper (Koliou et al. 2013). NUMERICAL PREDICTIONS OF EXPERIMENTAL RESULTS To predict numerically the experimental results obtained above, three dimensional finite element models of the support structure-bushing system were developed using the commercial structural analysis program SAP 2000 (Computers and Structures 2009). Free vibration analyses and linear dynamic time-history analyses were performed for each tested configuration excited by the five test ground motions ensemble. Analyses were performed with the input motions corresponding to the recorded shake table acceleration time-histories (achieved motions) as well as with the first modal equivalent damping ratios measured from the white noise tests (see Table 5). DESCRIPTION OF NUMERICAL MODELS

Two finite element models were developed based on the dimensions and properties of the test specimens (Fahad 2013). The first model represented the system identification testing and included the 230 kV porcelain bushing bolted to the reinforced concrete slab, as illustrated in Figure 7a. The bushing was modeled by multiple beam elements of appropriate density, geometry and stiffness, assembled in series. The bushing model consisted of three parts: (i) the upper part represented the actual high-voltage bushing, (ii) the central part consisted of a radial array of rigid elements representing the bushing mounting flange connected to the mounting plate, and (iii) the lower part representing the mounting plate, which was modeled as a polyhedron with the same number of surfaces as the number of the radial rigid elements. The concrete slab was modeled with shell elements of appropriate thickness and density.

Figure 7. Finite element models of test specimens: (a) Model for system identification testing and (b) model for shake table (seismic) testing.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1363

The second finite element model represented the specimen used for seismic testing, as shown in Figure 7b. This second model consisted of the rigid frame, cover plate, adaptor plate, and the bushing. The same model of the bushing described above was used again. All the components of the rigid frame were modeled with beam elements of appropriate steel sections. A more detailed view of the finite element model of the rigid frame is presented in Figure 8. Shell elements of appropriate density and thickness were used to model the adaptor plate as well as the cover plate of the rigid frame. Beam elements were used to model the flexural stiffeners (L8 6 1/2 or L6 4 1/2) attached to the cover plate. Details of both finite element models can be found in Koliou et al. (2012). COMPARISON OF NUMERICAL AND EXPERIMENTAL RESULTS

Table 6 compares the fundamental frequencies obtained from the system identification tests with those predicted by the two finite element models. The fundamental frequencies predicted by the model representing the bushing mounted on the support structure agrees well (within 10%) with the fundamental frequencies obtained from the white noise tests.

Figure 8. Isometric view of finite element model of the rigid frame. Table 6.

Comparison between experimental and computed bushing fundamental frequencies Fundamental frequency (Hz)

Bushing configuration Without stiffeners With stiffeners L6 4 1/2 With stiffeners L8 6 1/2 Rigid base

Numerical results Impact hammer tests White noise tests Difference 11.2 14.6 15.2 21.0

– – – 25.3

10.1 15.1 16.1 –

9.7% 3.6% 6.5% 20.5%

1364

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

The fundamental frequency predicted by the model of the bushing mounted on a rigid base, however, is 20% lower than the natural frequency obtained from the impact hammer tests on the bushing bolted to the reinforced concrete slab. Figure 9 compares numerical and experimental cumulative distribution functions (CDF) associated with the probability of non-exceeding (PoNE) a value of maximum bending moment at the base of the bushing. The experimental CDF curves were constructed from the seismic test data using the five test ground motions described above (see Table 3), while the numerical CDF curves were based on the full 22-P695 ground motion set used in the numerical study (Koliou et al. 2013). The comparison in Figure 9 indicates that the finite element models slightly overestimate (3% to 10%) the maximum base moments for all test phases. Note that no seismic test was conducted for the bushing mounted on a rigid base and that only the numerical predictions are shown in Figure 9. A comparison between the measured and computed maximum bending moments at the base of the bushing for all mounting conditions on the support structure and test ground motions is presented in Table 7. The numerical model is able to predict with very good accuracy the maximum bending moments at the base of the bushing specimen for all test ground motions and mounting conditions on the test structure. To quantify further the effect of incorporating flexural stiffeners on the cover plates of the support structure to reduce the maximum bending moments at the base of the bushing, the same efficiency factor, E, defined in the companion numerical study (Koliou et al. 2013) was applied herein to the measured results and numerical predictions of the shake table tests. In the context of this experimental study, this efficiency factor, E, is defined as: M Original M Final 100% (2) E¼ M Original M Rigid EQ-TARGET;temp:intralink-;e2;41;355

where M Final is the median maximum moment at the base of the bushing for the support structure incorporating one of the two stiffener configurations (L8 6 1/2 or L6 4 1/2) on its Probability of Non - Exceedance

1.0 0.9

w/o Stiffeners w/ Stiffeners L6x4x1/2 w/ Stiffeners L8x6x1/2 Rigid Base

0.8 0.7 0.6 0.5

Numerical Results

0.4

Experimental Results

0.3 0.2 0.1 0.0 0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

Maximum Bending Moment (kip-in)

Figure 9. Experimental and computed CDF for maximum bending moments at the base of bushing specimen.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1365

Table 7. Measured (test) and computed (model) maximum bending moments (kip-in) at the base of the bushing specimen for various mounting conditions With stiffeners L6 4 1/2

Without stiffeners

With stiffeners L8 6 1/2

Test Motion

Test

Model

Difference (%)

Test

Model

Difference (%)

Test

Model

Difference (%)

EQ EQ EQ EQ EQ

31.4 37.9 9.7 18.1 23.0

34.5 48.8 10.8 21.0 22.11

9.1 22.3 10.2 13.7 3.8

27.4 26.1 8.7 15.3 20.8

29.1 26.6 9.7 16.7 21.1

6.8 2.0 9.5 8.5 1.6

26.2 20.7 7.9 12.3 15.2

27.5 21.2 7.8 13.6 15.7

4.4 2.1 1.0 9.6 3.4

1 2 3 4 5

cover plate, M Original is the median maximum moment at the base of the bushing for the same support structure but with the unstiffened cover plate, and M Rigid is the median maximum moment at the base of the bushing mounted on a rigid base. For the rigid base condition, only the values of maximum bending moments at the base of the bushing predicted by the numerical model (see Figure 9) were used in Equation 2 since no seismic test was conducted for the rigid base condition. According to Equation 2, a value of E ¼ 0% indicates that the stiffening technique does not improve the response of the original transformer-bushing system, while a value of E ¼ 100% indicates that the stiffened support structure-bushing system achieves the same seismic response as of the bushing mounted on a rigid base.

Efficiency (%)

A comparison of the efficiency factors measured experimentally with those obtained by the numerical model is given in Figure 10 for both stiffener configurations on the cover plate of the support structure. The finite element model overestimates slightly (3% and 12%) the efficient factors derived from the experimental results. Figure 10 also shows that the addition of the heavier stiffeners (L8 6 1/2) yields a value of the efficient factor almost twice as

100 90 80 70 60 50 40 30 20 10 0

w/ Stiffeners L8x6x1/2 w/ Stiffeners L6x4x1/2 74

72

44

Finite Element Model Results

39

Seismic Testing Results

Figure 10. Measured and computed efficiency factors for stiffened bushing on support structure.

1366

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

that of the lighter stiffeners (L6 4 1/2), thereby causing the seismic response of the bushing to approach that of the same bushing mounted on a rigid base. CONCLUSIONS The results of the experimental study conducted on a 230 kV porcelain bushing mounted on a support structure incorporating a cover plate and two stiffener configurations and presented in this second of two companion papers confirmed that bushings mounted on the cover plates of transformer tanks are more vulnerable to seismic loading than similar bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field. The improved seismic performance of bushings mounted on a rigid base is associated with an increase of their fundamental natural frequencies causing a reduction of the seismic demand. The results of the experimental study showed also that the introduction of horizontal stiffeners on cover plates of transformer tanks near the base of the bushings is a viable and simple approach to shift the fundamental frequencies of bushings close to that of rigid base condition and, thereby, improve the seismic response of transformer-bushing systems. Finite element models using commercially available computer software were able to predict with good accuracy the results of the shake table tests. Transformer manufacturers should consider optimizing the selections and locations of horizontal stiffeners on the cover plates of transformer tanks to improve the seismic response of bushings and reduce damage to transformer-bushing systems during earthquakes. ACKNOWLEDGMENTS The authors thank Bonneville Power Administration for its financial support of this project. The support of Dr. Leon Kempner, Principal Structural Engineer at Bonneville Power Administration, in providing important technical information and recommendations, is appreciated. Dr. Anshel Schiff, President at Precision Measurement Instruments, is also acknowledged for serving as external advisor to this research project. Mr. Muhammad Fahad, Ph.D. Candidate in the Department of Civil, Structural, and Environmental Engineering at the University at Buffalo, is gratefully acknowledged for his considerable contribution to the execution of the experimental part of this research and for providing the finite element model of the test specimen. The assistance of the personnel of the Structural Engineering and Earthquake Simulation Laboratory (SEESL) of the State University of New York at Buffalo in the execution of the testing is also greatly appreciated.

REFERENCES Boore, D. M., Watson-Lamprey, J., and Abrahamson, N. A., 2006. Orientation-independent measures of ground motion, Bulletin of the Seismological Society of America 96, 1502–1511. Bracci, J., Reinhorn, A., and Mander, J., 1992. Seismic Resistance of Reinforced Concrete Framed Structures Designed Only for Gravity Loads: Part I - Design and Properties of a One - Third Scale Model Structure, Technical Report NCEER-92-0027, Buffalo, NY, 181 pp. Computers and Structures, Inc., 2009. SAP2000 Advanced V.14.1.0, Berkeley, CA.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1367

Ersoy, S., Feizi, B., Ashrafi, A., and Saadeghvaziri, M. A., 2008. Seismic Evaluation and Rehabilitation of Critical Components of Electrical Power Systems, Technical Report MCEER-08-0011, University at Buffalo, The State University of New York, Buffalo, NY. Fahad, M., 2013. Seismic Evaluation and Qualification of Transformer Bushings, Ph.D. Thesis, University at Buffalo, The State University of New York, Buffalo, NY. Federal Emergency Management Agency (FEMA P695), 2009. Quantification of Building Seismic Performance Factors, FEMA P695, Washington, D.C. Filiatrault, A., and Matt, H., 2006. Seismic response of high-voltage transformer-bushing systems, ASCE Journal of Structural Engineering 132, 287–295. Institute of Electrical and Electronics Engineers (IEEE), 2005. IEEE Recommended Practice for Seismic Design of Substations, IEEE Standard 693, IEEE Power Engineering Society, The Institute of Electrical and Electronics Engineers, Inc., New York, 179 pp. Koliou, M., Filiatrault, A., Reinhorn, A. M., and Oliveto, N., 2012. Seismic Protection of Electrical Transformer Bushing Systems by Stiffening Techniques, MCEER Technical Report –MCEER12-0002, University at Buffalo – the State University of New York, Buffalo, NY. Koliou, M., Filiatrault, A., and Reinhorn, A. M., 2013. Seismic response of high-voltage transformer-bushing systems incorporating flexural stiffeners I: Numerical study, Earthquake Spectra 29. Kong, D., 2010. Protection of High-voltage Electrical Equipment Against Severe Shock and Vibrations, Ph.D. Thesis, University at Buffalo, The State University of New York, Buffalo, NY. Reinhorn, A., Oikonomou, M., Roh, K., Schiff, H., and Kempner, A.L., 2011a. Modeling and Seismic Performance Evaluation of High-voltage Transformers and Bushings, MCEER Technical Report–MCEER-11-0006, University at Buffalo, The State University of New York, Buffalo, NY. Reinhorn, A. M., Filiatrault, A., Muhammad, F., Oliveto, N., Oikonomou, K., and Koliou, M., 2011b. Evaluation of Transformer Bushings: Current and Recommended Practice, MCEER Task Report–MCEER-TR-11-01, University at Buffalo, The State University of New York, Buffalo, NY. Sideris, P., Filiatrault, A., Leclerc, M., and Tremblay, R., 2010. Experimental investigation on the seismic behavior of palletized merchandise in steel storage racks, Earthquake Spectra 26, 209–233. Whittaker, A. S., Fenves, G. L., and Gilani, A. S. J., 2004. Earthquake performance of porcelain transformer bushings, Earthquake Spectra 20, 205–203. (Received 25 July 2011; accepted 25 August 2012)

M.EERI,

and

A numerical study conducted on four transformer-bushing models presented in a first companion paper indicated that high-voltage bushings mounted on the cover plates of transformers are more vulnerable to seismic loading than bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field. In this second companion paper, the addition of flexural stiffeners on the transformer cover plates as a means to stiffen the base of bushings and mitigate their seismic vulnerability is investigated experimentally. Shake table testing was conducted on a 230 kV porcelain bushing mounted on a support structure incorporating a flexible cover plate and two stiffener configurations. Test results confirmed that stiffening the cover plates is beneficial to the seismic response of high-voltage bushings. Test results are compared to the predictions of finite element analyses. [DOI: 10.1193/072511EQS185M]

INTRODUCTION The seismic response of high-voltage (at or above 220 kV) bushings is influenced by both their lateral stiffness and the flexural stiffness of the cover plates of the transformer tanks onto which they are mounted on. The results of the numerical study conducted on four transformer-bushing models and presented in a first companion paper (Koliou et al. 2013) indicated that bushings mounted on the cover plates of transformer tanks are more vulnerable to seismic loading than similar bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field during real earthquakes (Ersoy et al. 2008, Filiatrault and Matt 2006, Whittaker et al. 2004). The improved seismic performance of bushings mounted on a rigid base is associated with an increase of their fundamental frequencies causing a reduction of the seismic demand. Moreover, stiffening the transformer cover, reduces the rotational and vertical motion at the base of the bushing (Reinhorn et al.

a)

Graduate Research Assistant, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. b) Professor, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. c) Professor Emeritus of Structural Engineering, Dept. of Civil, Structural and Environmental Engineering, University at Buffalo, The State University of New York, Buffalo, NY 14260, USA. 1353

Earthquake Spectra, Volume 29, No. 4, pages 1353–1367, November 2013; © 2013, Earthquake Engineering Research Institute

1354

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

2011a) while reducing the bushing response and forces at the mounting interface (Koliou et al. 2013). The results of the numerical study showed also that the introduction of horizontal flexural stiffeners on the cover plates of transformer tanks near the base of the bushings is a viable and simple approach to shift the fundamental frequencies of bushings near that of rigid base condition and, thereby, improve the seismic response of transformer-bushing systems. The stiffeners should have sufficiently high flexural rigidity to maximize the fundamental frequencies of the bushings. This maximum fundamental frequency occurs when the base of the bushing is made locally rigid and is governed by the global flexural flexibility of the tank cover plate and walls (Koliou et al. 2012). In this second companion paper, the seismic response of transformer-bushing systems incorporating flexural stiffeners on the cover plates of transformer tanks is investigated experimentally to confirm the numerical results reported in the first companion paper. The experimental study incorporated two types of testing: (i) system identification testing and (ii) shake table (seismic) testing. Test results are also compared to the predictions of finite element analyses. SYSTEM IDENTIFICATION TESTING DESCRIPTION OF TEST SPECIMEN

The specimen used for the system identification testing consisted of a 230 kV porcelain bushing bolted to a reinforced concrete slab through an embedded adaptor plate. The slab was anchored to the strong floor of the Structural Engineering and Earthquake Simulation Laboratory (SEESL) of the University at Buffalo (UB) to simulate rigid base condition. The bushing structure was 151.4 in. tall while the concrete slab had plan dimensions of 8 ft 8 ft and a thickness of 1 ft. A lumped weight of 25 lbs was added at the top of the bushing to simulate the inertia of the electrical conductors, as required by the IEEE-693 Standard (IEEE 2005) for seismic qualification testing of electrical equipment. Figure 1 shows photographs of the bushing specimen and experimental setup used for the system identification testing. The physical properties of the bushing specimen provided by the manufacturer are summarized in Table 1. More details on the test set-up can be found in Koliou et al. (2012).

TEST PROTOCOL AND INSTRUMENTATION

System identification testing consisted of impact hammer tests to evaluate the natural frequencies and damping characteristics of the bushing specimen mounted on a rigid base. The top of the bushing was impacted with a rubber hammer in two orthogonal directions (north-south and east-west directions in Figure 1). Impact hammer tests were not conducted in the vertical direction since the effects of vertical ground motions were not considered in this study. The dynamic response of the bushing specimen during system identification testing was recorded by 19 instruments, including five accelerometers along the height of the bushing, two linear potentiometers at the top of the bushing and four strain rosettes installed on the mounting flange of the bushing (three strain gauges per rosette) to measure accelerations, displacements and strains, respectively. Details of the instrumentation used for the system identification testing are provided in Koliou et al. (2012).

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1355

Dome

230 kV Porcelain Bushing

Upper Porcelain Unit

Reinforced Concrete Slab Steel Plate at the top of the concrete slab

Sheds Mounting Flange

Lifting Lugs

(b)

(a)

Figure 1. (a) Specimen configuration used for system identification testing and (b) details of 230 kV porcelain bushing specimen.

Table 1. Properties of 230 kV porcelain bushing (Koliou et al. 2012, Reinhorn and Muhammad 2011) Material insulator Voltage capacity Total height Length over mounting flange Length below mounting flange Maximum diameter over mounting flange Maximum diameter below mounting flange Diameter of mounting flange Diameter of bolt pattern on mounting flange Number of bolts on mounting flange Diameter of bolts on mounting flange Total weight Location of center of gravity above flange Upper unit weight Location of upper unit center of gravity Lower unit weight Location of lower unit center of gravity

Porcelain (kV) (in) (in) (in) (in) (in) (in) (in) (in) (in) (lbs) (in) (lbs) (in) (lbs) (in)

230 151.4 91.4 60.0 11.8 10.0 24.0 21 12 1.25 840.0 14.0 447.0 34.0 293.0 28.0

1356

Table 2. base

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

System identification test results for 230 kV porcelain bushing mounted on a rigid

Impact hammer test direction North-south East-west

Fundamental frequency (Hz)

First modal equivalent damping ratio (%)

25.3 25.4

2.3 2.1

TEST RESULTS

The fundamental frequency and first modal equivalent damping ratio for the 230 kV porcelain bushing specimen mounted on a rigid base obtained from the impact hammer tests in each orthogonal direction are shown in Table 2. The fundamental frequency of the bushing specimen was obtained by plotting the Fourier amplitude spectrum from the acceleration time-histories recorded at the top of the bushing during each impact hammer test. The first modal equivalent viscous damping ratio was estimated by the half-power bandwidth method (Bracci et al. 1992). The results for the fundamental frequency and damping ratio are consistent for both orthogonal directions of the bushing. SHAKE TABLE (SEISMIC) TESTING DESCRIPTION OF TEST SPECIMEN

The specimen used for the shake table (seismic) testing consisted of the same 230 kV porcelain bushing described above mounted on a support structure representing the generic cover plate of a transformer tank (Reinhorn et al. 2011b, Kong 2010). As shown in the photograph of Figure 2, the support structure consisted of a rigid frame, a cover plate and an adaptor plate (attached to the cover plate). The rigid frame had dimensions of 8 ft 8 ft 8 ft, while the ¾ in. thick cover plate had plan dimensions of 127 in. 127 in. A steel square tube (TS 5 5 1/2) was used for the four corner columns of the rigid frame. The top of each column was connected to the cover plate by horizontal L5 5 3/4 angles. The four bays of the rigid frame were stiffened by L5 5 3/4 cross bracing welded together at mid-span (Kong 2010). In order to evaluate the addition of flexural stiffeners on the tank cover plate as a means to reduce the bending moment demand at the base of the bushing during earthquake shaking two different sets of steel angles, L8 6 1/2 and L6 4 1/2, were considered. These angle sections were bolted on the cover plate in both perpendicular directions of the frame (two on top of the cover plate in the North-South direction and two underneath it along the east-west direction). Each set of stiffeners was installed in a 4 ft 4 ft square pattern, with each stiffener located at a distance of 2 ft from the center line of the bushing, as shown in the photographs of Figure 3. Detailed drawings of the test specimen used for the shake table tests are provided in Koliou et al. (2012). Shake table testing was conducted for both sets of stiffeners described above as well as for the unstiffened cover plate.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1357

Figure 2. Specimen used for shake table (seismic) testing.

Figure 3. Steel angle stiffeners bolted on the cover plate: (a) On top of the plate in the northsouth direction, (b) underneath the plate in the east-west direction. INSTRUMENTATION

A total of 46 instruments were used to monitor the seismic response of the bushing and of the support structure during shake table testing. The same 19 sensors used for the system identification testing (see above) were installed again on the bushing, seven instruments (three accelerometers and four linear potentiometers) were used to measure the dynamic response of the support structure, 13 accelerometers were placed on the cover plate to record its dynamic response, and seven instruments (three accelerometers and four linear potentiometers) were installed on the shake table in order to measure the achieved input motions. A detailed description of the instrumentation used in the shake table testing is provided in Koliou et al. (2012). It should be noted that the set of strain rosettes installed on the mounting flange of the bushing were calibrated to measure directly the shear forces and bending moments transmitted to the cover plate from the bushing assembly.

1358

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

EARTHQUAKE GROUND MOTIONS

Similar to the numerical studies described in the first companion paper (Koliou et al. 2013), the FEMA P695 Far Field Ground Motion Set (FEMA P695 2009) were considered for seismic testing. This ensemble contains 22 historical motions (2 horizontal components per motion) from all over the world and is considered to be representative of the seismicity in the Western United States. This motion ensemble was scaled to match the IEEE-693, 2% damped, high required response spectrum in a range of frequencies from 2.0 Hz to 30.0 Hz, as described in the first companion paper (Koliou et al. 2013). Information including response spectra for these motions have also been provided in this first companion paper (Koliou et al. 2013). The effects of vertical ground motions were not considered in this study. Vertical input would have a negligible effect on the vertically positioned bushing specimen. To limit the number of seismic tests, a reduced test motion ensemble, consisting of five pairs of ground motions (two components each) out of the 22 P695 motions, was considered for seismic testing (Koliou et al. 2012). The five test ground motions were selected to have similar values as the complete P695 motion set for several selected statistical parameters of interest. The statistical parameters selected were the median, arithmetic mean, geometric mean, standard deviation, maximum and minimum values of a characteristic spectral value in a range of frequencies varying from 10 Hz to 25 Hz. This frequency range was selected to encompass the range of natural frequencies of the bushing specimen mounted on a rigid base (see Table 2 above) and on the support structure (see Table 5 below). The test motion ensemble was also selected to include no more than one motion per historical seismic event. The steps followed to select the test motion ensemble can be summarized as follows: 1.

2.

3. 4.

For each of the 22 P695 motions, the geometric mean of the 2% damped spectral accelerations for the two horizontal components were computed for individual frequencies to obtain a geometric mean response spectrum. The geometric mean of the 2% damped spectral accelerations was calculated for each of the geometric mean spectra across the frequency range of interest to provide a single characteristic spectral value for each motion. Each statistical parameter of interest of the characteristic spectral value was computed across the 22 motions. The 22 motions were listed in ascending order of their characteristic spectral value, and different ensembles of five motions were investigated in order to identify the combination that matched the best its statistical parameter of interest to that of the full ensemble of 22 motions. Note that the geometric mean was selected as the characteristic spectral value, as it provides an orientation-independent measure of earthquake intensity (Boore et al. 2006). A similar selection procedure was used by Sideris et al. (2010).

The final test motion ensemble is presented in Table 3. A comparison of the various statistical parameters of the characteristic spectral value between the full FEMA P695 Far Field Ground Motion Set and the reduced test motion ensemble is provided in Table 4. TESTING PROTOCOL

The shake table (seismic) testing included three phases: in the first phase, the larger flexural stiffeners (L8 6 1/2) were installed on the cover plate of the rigid frame as described earlier.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

Table 3.

1359

Earthquake ground motion subset considered for seismic testing Earthquake event

Test motion

ID.

EQ1 EQ2 EQ3 EQ4 EQ5

12011 12041 12072 12092 12132

Name

Year

Mw

Recording station

Northridge Duzce, Turkey Kobe, Japan Landers Cape Mendocino

1994 1999 1995 1992 1992

6.7 7.1 6.9 7.3 7.0

Beverly Hills-Mulhol Bolu Shin-Osaka Coolwater Rio Dell Overpass

PGA (g)* 0.52 0.82 0.24 0.42 0.55

*Larger component

Table 4. Comparison of statistical parameters of characteristic spectral value between FEMA P695 far field ground motion set and test motion ensemble Values (g) Statistical parameters of characteristic spectral values Median Arithmetic mean Geometric mean Standard deviation Maximum Minimum

FEMA P695 far-field ground motion set (22 historical motions/two horizontal components per motion)

Ensemble of five test ground motions (two horizontal components per motion)

0.909 1.026 0.927 0.495 2.531 0.412

0.924 1.011 0.903 0.529 1.825 0.426

During the second phase, the smaller flexural stiffeners (L6 4 1/2) were bolted to the cover plate. Finally in the third testing phase, all stiffeners were removed and the bushing specimen mounted on the unstiffened cover plate of the rigid frame was tested. Low amplitude white noise tests were also conducted between each seismic test to study the evolution of the fundamental frequency and first modal equivalent damping ratio of the bushing specimen through the testing program. Because the bushing specimen and the support structure remained in the elastic range of their material during all seismic tests, the bushing fundamental frequency and equivalent damping ratio remained essentially constant throughout the testing program. RESULTS OF WHITE NOISE TESTS

The results of the initial white noise tests are shown in Table 5. As expected, the measured fundamental frequency of the bushing specimen increases with the rigidity of the stiffeners installed on the cover plate of the support structure. The fundamental frequencies associated with both stiffener configurations lie between the frequency associated with the unstiffened cover plate and that of the rigid base condition (see Table 2). These results are consistent with the results of the numerical study presented in the first companion paper

1360

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

Table 5. Measured bushing fundamental frequency and first modal equivalent damping ratio during white noise tests Test phase 1: With stiffeners L8 6 1/2 2: With stiffeners L6 4 1/2 3: Without stiffeners

Fundamental frequency (Hz)

First modal equivalent damping ratio ξ (%)

16.1 15.1 10.1

4.1 3.2 1.9

(Koliou et al. 2013). The first modal equivalent damping ratio remains under 5% of critical for all mounting configurations, including the rigid base condition shown in Table 2. Although the increase of damping from ∼2% to ∼4% may account for a decrease of spectral response of up to 15%, for the fundamental frequencies of the bushing assemblies, the contribution of damping increase to the response reduction is less than 10%. RESULTS OF SEISMIC TESTS

The bending moment induced at the base of the bushing specimen was the structural response parameter of interest in this experimental study. For each seismic test, the maximum bending moment at the base of the bushing, M max , was computed according to: qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ M max ¼ maxt (1) M 2x ðtÞ þ M 2y ðtÞ EQ-TARGET;temp:intralink-;e1;41;372

where Mx(t) and My(t) are the moments at the base of the bushing specimen at time t recorded by the strain rosettes with respect to the longitudinal (east-west) and transverse (north-south) axis of the support structure, respectively; while maxt is the maximum absolute value over the recorded time-history. For each of the three test phases, experimental cumulative distribution functions (CDF) were calculated for the probability of non-exceeding (PoNE) a prescribed maximum moment at the base of the bushing specimen under the test motion ensemble. The PoNE is estimated by counting the number of test ground motions causing a prescribed value of the maximum bending moment at the base of the bushing specimen not to be exceeded and dividing this number by five (the total number of test ground motions). A lognormal cumulative distribution function (CDF) can then be fitted to the empirical data. The lognormal CDF is defined by the median value (PoNE = 50%) of the maximum bending moment, ^ max and by the dispersion parameter β expressed as the standard deviation of the log of M the values of M max . The resulting CDFs shown in Figure 4 indicate that the maximum bending moments recorded at the base of the bushing specimen reduce with increasing size of flexural stiffeners on the cover plate. The median maximum bending moments for the largest stiffeners (L8 6 1/2) is substantially reduced (30%) compared to that of the median bending moment recorded at the base of the bushing mounted on the unstiffened cover plate. The same trend can also be observed for the relative displacement and the absolute acceleration at the top of the bushing specimen, as shown in Figures 5 and 6, respectively. These

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1361

Probability of Non - Exceedance

1.0 0.9

w/o Stiffeners

0.8

w/ Stiffeners L6x4x1/2

0.7

w/ Stiffeners L8x6x1/2

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0

20.0

40.0

60.0

80.0

100.0

Maximum Bending Moment (kip-in)

Figure 4. Experimental CDF for recorded maximum bending moments at the base of the bushing specimen.

Maximum Relative Displacement at the Top of the Bushing (in)

5.00 4.50

4.54 4.42

w/o Stiffeners 4.25

w/ Stiffeners L6x4x1/2

4.00 3.33

3.50 3.00

w/ Stiffeners L8x6x1/2

3.41 3.17 2.82

2.97 2.91 2.41

2.50

2.17

2.00

1.77 1.66

1.97

1.41

1.50 1.00 0.50 0.00 EQ1

EQ2

EQ3

EQ4

EQ5

Figure 5. Maximum relative displacements measured at the top of the bushing specimen.

Maximum Absolute Acceleration at the Top of the Bushing (g)

3.00

w/o Stiffeners

2.73

w/ Stiffeners L6x4x1/2

2.50 2.18

w/ Stiffeners L8x6x1/2

2.03

1.88

2.00 1.71

1.50

1.49

1.57

1.35 0.98

1.00

0.83

0.82

1.03 0.88

0.45 0.39

0.50 0.00 EQ1

EQ2

EQ3

EQ4

EQ5

Figure 6. Maximum absolute accelerations measured at the top of the bushing specimen.

1362

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

experimental results correlate well with the numerical results presented in the first companion paper (Koliou et al. 2013). NUMERICAL PREDICTIONS OF EXPERIMENTAL RESULTS To predict numerically the experimental results obtained above, three dimensional finite element models of the support structure-bushing system were developed using the commercial structural analysis program SAP 2000 (Computers and Structures 2009). Free vibration analyses and linear dynamic time-history analyses were performed for each tested configuration excited by the five test ground motions ensemble. Analyses were performed with the input motions corresponding to the recorded shake table acceleration time-histories (achieved motions) as well as with the first modal equivalent damping ratios measured from the white noise tests (see Table 5). DESCRIPTION OF NUMERICAL MODELS

Two finite element models were developed based on the dimensions and properties of the test specimens (Fahad 2013). The first model represented the system identification testing and included the 230 kV porcelain bushing bolted to the reinforced concrete slab, as illustrated in Figure 7a. The bushing was modeled by multiple beam elements of appropriate density, geometry and stiffness, assembled in series. The bushing model consisted of three parts: (i) the upper part represented the actual high-voltage bushing, (ii) the central part consisted of a radial array of rigid elements representing the bushing mounting flange connected to the mounting plate, and (iii) the lower part representing the mounting plate, which was modeled as a polyhedron with the same number of surfaces as the number of the radial rigid elements. The concrete slab was modeled with shell elements of appropriate thickness and density.

Figure 7. Finite element models of test specimens: (a) Model for system identification testing and (b) model for shake table (seismic) testing.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1363

The second finite element model represented the specimen used for seismic testing, as shown in Figure 7b. This second model consisted of the rigid frame, cover plate, adaptor plate, and the bushing. The same model of the bushing described above was used again. All the components of the rigid frame were modeled with beam elements of appropriate steel sections. A more detailed view of the finite element model of the rigid frame is presented in Figure 8. Shell elements of appropriate density and thickness were used to model the adaptor plate as well as the cover plate of the rigid frame. Beam elements were used to model the flexural stiffeners (L8 6 1/2 or L6 4 1/2) attached to the cover plate. Details of both finite element models can be found in Koliou et al. (2012). COMPARISON OF NUMERICAL AND EXPERIMENTAL RESULTS

Table 6 compares the fundamental frequencies obtained from the system identification tests with those predicted by the two finite element models. The fundamental frequencies predicted by the model representing the bushing mounted on the support structure agrees well (within 10%) with the fundamental frequencies obtained from the white noise tests.

Figure 8. Isometric view of finite element model of the rigid frame. Table 6.

Comparison between experimental and computed bushing fundamental frequencies Fundamental frequency (Hz)

Bushing configuration Without stiffeners With stiffeners L6 4 1/2 With stiffeners L8 6 1/2 Rigid base

Numerical results Impact hammer tests White noise tests Difference 11.2 14.6 15.2 21.0

– – – 25.3

10.1 15.1 16.1 –

9.7% 3.6% 6.5% 20.5%

1364

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

The fundamental frequency predicted by the model of the bushing mounted on a rigid base, however, is 20% lower than the natural frequency obtained from the impact hammer tests on the bushing bolted to the reinforced concrete slab. Figure 9 compares numerical and experimental cumulative distribution functions (CDF) associated with the probability of non-exceeding (PoNE) a value of maximum bending moment at the base of the bushing. The experimental CDF curves were constructed from the seismic test data using the five test ground motions described above (see Table 3), while the numerical CDF curves were based on the full 22-P695 ground motion set used in the numerical study (Koliou et al. 2013). The comparison in Figure 9 indicates that the finite element models slightly overestimate (3% to 10%) the maximum base moments for all test phases. Note that no seismic test was conducted for the bushing mounted on a rigid base and that only the numerical predictions are shown in Figure 9. A comparison between the measured and computed maximum bending moments at the base of the bushing for all mounting conditions on the support structure and test ground motions is presented in Table 7. The numerical model is able to predict with very good accuracy the maximum bending moments at the base of the bushing specimen for all test ground motions and mounting conditions on the test structure. To quantify further the effect of incorporating flexural stiffeners on the cover plates of the support structure to reduce the maximum bending moments at the base of the bushing, the same efficiency factor, E, defined in the companion numerical study (Koliou et al. 2013) was applied herein to the measured results and numerical predictions of the shake table tests. In the context of this experimental study, this efficiency factor, E, is defined as: M Original M Final 100% (2) E¼ M Original M Rigid EQ-TARGET;temp:intralink-;e2;41;355

where M Final is the median maximum moment at the base of the bushing for the support structure incorporating one of the two stiffener configurations (L8 6 1/2 or L6 4 1/2) on its Probability of Non - Exceedance

1.0 0.9

w/o Stiffeners w/ Stiffeners L6x4x1/2 w/ Stiffeners L8x6x1/2 Rigid Base

0.8 0.7 0.6 0.5

Numerical Results

0.4

Experimental Results

0.3 0.2 0.1 0.0 0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

Maximum Bending Moment (kip-in)

Figure 9. Experimental and computed CDF for maximum bending moments at the base of bushing specimen.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1365

Table 7. Measured (test) and computed (model) maximum bending moments (kip-in) at the base of the bushing specimen for various mounting conditions With stiffeners L6 4 1/2

Without stiffeners

With stiffeners L8 6 1/2

Test Motion

Test

Model

Difference (%)

Test

Model

Difference (%)

Test

Model

Difference (%)

EQ EQ EQ EQ EQ

31.4 37.9 9.7 18.1 23.0

34.5 48.8 10.8 21.0 22.11

9.1 22.3 10.2 13.7 3.8

27.4 26.1 8.7 15.3 20.8

29.1 26.6 9.7 16.7 21.1

6.8 2.0 9.5 8.5 1.6

26.2 20.7 7.9 12.3 15.2

27.5 21.2 7.8 13.6 15.7

4.4 2.1 1.0 9.6 3.4

1 2 3 4 5

cover plate, M Original is the median maximum moment at the base of the bushing for the same support structure but with the unstiffened cover plate, and M Rigid is the median maximum moment at the base of the bushing mounted on a rigid base. For the rigid base condition, only the values of maximum bending moments at the base of the bushing predicted by the numerical model (see Figure 9) were used in Equation 2 since no seismic test was conducted for the rigid base condition. According to Equation 2, a value of E ¼ 0% indicates that the stiffening technique does not improve the response of the original transformer-bushing system, while a value of E ¼ 100% indicates that the stiffened support structure-bushing system achieves the same seismic response as of the bushing mounted on a rigid base.

Efficiency (%)

A comparison of the efficiency factors measured experimentally with those obtained by the numerical model is given in Figure 10 for both stiffener configurations on the cover plate of the support structure. The finite element model overestimates slightly (3% and 12%) the efficient factors derived from the experimental results. Figure 10 also shows that the addition of the heavier stiffeners (L8 6 1/2) yields a value of the efficient factor almost twice as

100 90 80 70 60 50 40 30 20 10 0

w/ Stiffeners L8x6x1/2 w/ Stiffeners L6x4x1/2 74

72

44

Finite Element Model Results

39

Seismic Testing Results

Figure 10. Measured and computed efficiency factors for stiffened bushing on support structure.

1366

M. KOLIOU, A. FILIATRAULT, AND A. M. REINHORN

that of the lighter stiffeners (L6 4 1/2), thereby causing the seismic response of the bushing to approach that of the same bushing mounted on a rigid base. CONCLUSIONS The results of the experimental study conducted on a 230 kV porcelain bushing mounted on a support structure incorporating a cover plate and two stiffener configurations and presented in this second of two companion papers confirmed that bushings mounted on the cover plates of transformer tanks are more vulnerable to seismic loading than similar bushings mounted on a rigid base. This would explain why the good performance of bushings mounted on a rigid base observed during shake table testing does not correlate well with their performance in the field. The improved seismic performance of bushings mounted on a rigid base is associated with an increase of their fundamental natural frequencies causing a reduction of the seismic demand. The results of the experimental study showed also that the introduction of horizontal stiffeners on cover plates of transformer tanks near the base of the bushings is a viable and simple approach to shift the fundamental frequencies of bushings close to that of rigid base condition and, thereby, improve the seismic response of transformer-bushing systems. Finite element models using commercially available computer software were able to predict with good accuracy the results of the shake table tests. Transformer manufacturers should consider optimizing the selections and locations of horizontal stiffeners on the cover plates of transformer tanks to improve the seismic response of bushings and reduce damage to transformer-bushing systems during earthquakes. ACKNOWLEDGMENTS The authors thank Bonneville Power Administration for its financial support of this project. The support of Dr. Leon Kempner, Principal Structural Engineer at Bonneville Power Administration, in providing important technical information and recommendations, is appreciated. Dr. Anshel Schiff, President at Precision Measurement Instruments, is also acknowledged for serving as external advisor to this research project. Mr. Muhammad Fahad, Ph.D. Candidate in the Department of Civil, Structural, and Environmental Engineering at the University at Buffalo, is gratefully acknowledged for his considerable contribution to the execution of the experimental part of this research and for providing the finite element model of the test specimen. The assistance of the personnel of the Structural Engineering and Earthquake Simulation Laboratory (SEESL) of the State University of New York at Buffalo in the execution of the testing is also greatly appreciated.

REFERENCES Boore, D. M., Watson-Lamprey, J., and Abrahamson, N. A., 2006. Orientation-independent measures of ground motion, Bulletin of the Seismological Society of America 96, 1502–1511. Bracci, J., Reinhorn, A., and Mander, J., 1992. Seismic Resistance of Reinforced Concrete Framed Structures Designed Only for Gravity Loads: Part I - Design and Properties of a One - Third Scale Model Structure, Technical Report NCEER-92-0027, Buffalo, NY, 181 pp. Computers and Structures, Inc., 2009. SAP2000 Advanced V.14.1.0, Berkeley, CA.

SEISMIC RESPONSE OF VOLTAGE TRANSFORMER-BUSHINGS WITH STIFFENERS II: EXPERIMENTAL

1367

Ersoy, S., Feizi, B., Ashrafi, A., and Saadeghvaziri, M. A., 2008. Seismic Evaluation and Rehabilitation of Critical Components of Electrical Power Systems, Technical Report MCEER-08-0011, University at Buffalo, The State University of New York, Buffalo, NY. Fahad, M., 2013. Seismic Evaluation and Qualification of Transformer Bushings, Ph.D. Thesis, University at Buffalo, The State University of New York, Buffalo, NY. Federal Emergency Management Agency (FEMA P695), 2009. Quantification of Building Seismic Performance Factors, FEMA P695, Washington, D.C. Filiatrault, A., and Matt, H., 2006. Seismic response of high-voltage transformer-bushing systems, ASCE Journal of Structural Engineering 132, 287–295. Institute of Electrical and Electronics Engineers (IEEE), 2005. IEEE Recommended Practice for Seismic Design of Substations, IEEE Standard 693, IEEE Power Engineering Society, The Institute of Electrical and Electronics Engineers, Inc., New York, 179 pp. Koliou, M., Filiatrault, A., Reinhorn, A. M., and Oliveto, N., 2012. Seismic Protection of Electrical Transformer Bushing Systems by Stiffening Techniques, MCEER Technical Report –MCEER12-0002, University at Buffalo – the State University of New York, Buffalo, NY. Koliou, M., Filiatrault, A., and Reinhorn, A. M., 2013. Seismic response of high-voltage transformer-bushing systems incorporating flexural stiffeners I: Numerical study, Earthquake Spectra 29. Kong, D., 2010. Protection of High-voltage Electrical Equipment Against Severe Shock and Vibrations, Ph.D. Thesis, University at Buffalo, The State University of New York, Buffalo, NY. Reinhorn, A., Oikonomou, M., Roh, K., Schiff, H., and Kempner, A.L., 2011a. Modeling and Seismic Performance Evaluation of High-voltage Transformers and Bushings, MCEER Technical Report–MCEER-11-0006, University at Buffalo, The State University of New York, Buffalo, NY. Reinhorn, A. M., Filiatrault, A., Muhammad, F., Oliveto, N., Oikonomou, K., and Koliou, M., 2011b. Evaluation of Transformer Bushings: Current and Recommended Practice, MCEER Task Report–MCEER-TR-11-01, University at Buffalo, The State University of New York, Buffalo, NY. Sideris, P., Filiatrault, A., Leclerc, M., and Tremblay, R., 2010. Experimental investigation on the seismic behavior of palletized merchandise in steel storage racks, Earthquake Spectra 26, 209–233. Whittaker, A. S., Fenves, G. L., and Gilani, A. S. J., 2004. Earthquake performance of porcelain transformer bushings, Earthquake Spectra 20, 205–203. (Received 25 July 2011; accepted 25 August 2012)