- Email: [email protected]
Evaluation of seismic behavior of soils under nuclear containment structures via dynamic centrifuge test
Nuclear Engineering and Design 277 (2014) 64–75
Contents lists available at ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Evaluation of seismic behavior of soils under nuclear containment structures via dynamic centrifuge test Jeong Gon Ha, Dong-Soo Kim ∗ Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, South Korea
h i g h l i g h t s • A series of dynamic centrifuge tests were performed for NPP structure to investigate the soil–foundation-structure interaction with various soil conditions • • • •
from loose sand to weathered rock. SFSI phenomena for NPP structure were observed directly using experimental method. Effect of the soil stiffness and nonlinear characteristics on SFSI was estimated. There are comparisons of the control motions for seismic design of a NPP structure. Subsoil condition, earthquake intensity and control motion affected to seismic load.
a r t i c l e
i n f o
Article history: Received 11 December 2013 Received in revised form 16 June 2014 Accepted 17 June 2014
a b s t r a c t To evaluate the earthquake loads for the seismic design of a nuclear containment structure, it is necessary to consider the soil–foundation-structure interaction (SFSI) due to their interdependent behavior. Especially, understanding the effects of soil stiffness under the structure and the location of control motion to SFSI are very important. Motivated by these requirements, a series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. The different amplification characteristics in peak-accelerations profile and effects of soil-nonlinearity in response spectrum were observed. The dynamic behaviors were compared between surface of free-field and foundation of the structure for the evaluation of the control motion for seismic design. It was found that dynamic centrifuge test has potentials to estimate the seismic load considering SFSI. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The destruction of nuclear containment structures during earthquakes may cause a catastrophic loss of life, property damage, or disruption to society. Therefore, the safe and reliable seismic design of structures is crucial. To evaluate the earthquake loads for the seismic design of a nuclear containment structure, it is necessary to consider the soil-structure interaction (SSI) due to their interdependent behavior. Recently, the Nuclear Regulatory Commission (NRC) raised the shear wave velocity standard for a general rock outcrop to 2800 m/s (9200 ft/s) (Regulatory Guide 1.208). This means that the region where the soil-structure interaction is considered to occur has been increased
∗ Corresponding author. Tel.: +82 42 350 3619; fax: +82 42 350 7200. E-mail addresses: [email protected] (J.G. Ha), [email protected] (D.-S. Kim). http://dx.doi.org/10.1016/j.nucengdes.2014.06.013 0029-5493/© 2014 Elsevier B.V. All rights reserved.
and requires a more advanced research and design methodology. When estimating the seismic load, because most nuclear structures have been built on hard rock, the local site effects are not considered, and the soil-structure analyses are normally performed in the frequency domain, while assuming that the soil is a linear elastic material (Roesset, 1998). However, the demand for the construction of nuclear power plant structures on deep soil deposits has been increasing. In this soil condition, a seismic wave is trapped, leading to large soil deformations, and the soil shows significant nonlinear characteristics. The seismic behavior of a nuclear containment structure can be greatly altered by these local site conditions. During the evaluation of the earthquake load in the seismic design of nuclear power plant structures, another controversial point is the location of the control motion where the design motion is specified (Roesset, 1998; Avilés and Pérez-Rocha, 1998; Kim and
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Stewart, 2003; Pitilakis et al., 2008). In most cases, the free field surface level-ground motion is selected as the control motion and used at the foundation level, neglecting the kinematic interaction effects (Verma, 2004). However, recent researches about seismic design have referred the importance of considering SSI effects to defer the earthquake input motion (Rayhani and El Naggar, 2008; Yamahara, 1970; Hradilek et al., 1973). Roesset (1998) discussed five possible choices for the design earthquake motion: the free surface of the soil deposit at the site, a hypothetical outcropping of rock, bedrock when there is rock at some finite depth at the site, the elevation of the foundation in the free field, and directly at the foundation. To reliably estimate the seismic load, it is of interest to assess the characteristics and amplitude differences between these potential motions. So far, most of the seismic analyses considering SSI have been performed based on numerical methods. Although numerical methods have been successfully applied to the design of nuclear containment structures, they also contain numerous uncertainties, including nonlinear soil characteristics, an impedance problem, etc. (Roesset, 1998; Finn et al., 1986). Motivated by the need to calibrate numerical analysis tools for SSI analyses, large-scale seismic model tests have been utilized (Hualian, Lotong). Even though a large-scale seismic test (LSST) is the most precise method to evaluate the SSI, it is difficult to use it to perform various parametric studies because of the expense and difficulty in adjusting the test conditions. To complement it, a dynamic centrifuge test that reproduces the field stress condition indoors and provides a relatively cost saving can be used as an alternative. Ghosh and Madabhushi (2006) used a dynamic centrifuge test as a tool to investigate the foundation response of a typical power plant during an earthquake. Ha et al. (2012) successfully simulated the seismic behavior of the Hualien LSST in a centrifuge. The objective of this study was to evaluate the seismic load applied to a nuclear containment structure while considering the soil–foundation-structure interaction (SFSI) in dynamic centrifuge tests. This study focused on two main points: (i) the effect of the soil
65
Table 1 Scaling law for centrifuge modeling. Parameters
Centrifuge model scaling
Length Velocity Acceleration Strain Stress Time (dynamic)
1/N 1 N 1 1 1/N
stiffness and corresponding nonlinear characteristics on SFSI and (ii) the comparison of the control motions for the seismic design of a nuclear containment structure. A series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. The dynamic soil properties of the weathered soil, WR, and sandy soil (SS), in terms of the shear wave velocity profile, modulus reduction, and damping curves, were evaluated using inflight bender element tests and laboratory resonant column tests. With small-scale model structures, the soil and structure responses were measured during a series of shakings in the centrifuge test, and an assessment was made of the importance of considering the soil nonlinearities and determining the control motions when evaluating the earthquake load for nuclear power plants. 2. Centrifuge equipment The basic idea of physical modeling using a centrifuge is to accelerate a reduced-scale geotechnical structure to the appropriate high g-level to simulate the prototype-scale stress field in the model structure. Centrifuge modeling with shaking table equipment provides an excellent opportunity to observe the SSI in a scale model. To make good use of this test method, reasonable scaleddown modeling with a proven scaling law is important. Schofield (1980) derived the scaling law for centrifuge tests. Some of the variables related to the modeling in this study are listed in Table 1.
Fig. 1. Test result for simulation of Hualien large-scale seismic test (Ha et al., 2012).
66
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 2. Typical test layout.
The beam-type centrifuge facility in KAIST was used to perform all the tests reported here in a 50 × g centrifugal acceleration field. This facility has a platform radius of 5 m and a maximum capacity of 240 g-tons (Kim et al., 2013a,b). The model earthquake was generated by a self-balanced electro-hydraulic earthquake simulator mounted on the centrifuge. The base shaking acceleration could be exerted to a maximum value of 0.5 × g in the prototype scale at 50 × g of centrifugal acceleration. The allowable frequency
range of the earthquake loading was from 30 Hz to 300 Hz on the model scale, and the utilization of the self-balanced shaking table for the sinusoidal and real earthquake motion was verified (Lee et al., 2013). During the dynamic centrifuge test, an equivalent shear beam (ESB) model box was used. This box can simulate a wall deformation that is similar to that of soil using the bearing and rubber located between the wall layers (Lee et al., 2013). The model box has
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
internal dimensions of 490 mm × 490 mm × 630 mm and external dimensions of 670 mm × 670 mm × 650 mm in length, width, and height, respectively. 3. Verification of centrifuge testing method SFSI is a very complex problem with numerous factors that influence the seismic behavior. Thus, precisely controlled test conditions and the verification of the dynamic centrifuge test of a nuclear power plant are needed. Based on these requirements, Ha et al. (2012) simulated the Hualien LSST using a dynamic centrifuge test before this research. The Hualien LSST program was an international project to observe real SFSI for a nuclear containment structure and began recording earthquake data from 1990 at a 1/4 scale nuclear site above well-investigated soil. To simulate real earthquake phenomena, the soil, structure, and earthquake motion are the most important things to consider in modeling. More specific details for modeling methods are mentioned in the following sections. Fig. 1 compares the typical acceleration time-histories and frequency properties measured at the foundation and free-field surface during the Chi-Chi earthquake, and those recorded at the corresponding locations in the centrifuge test. The amplification characteristics and time-histories are well matched at the soil layer and the foundation of the structure. This shows the potential to simulate the Hualien LSST using a dynamic centrifuge test and verifying the reliability of the centrifuge test method for SFSI research. Thus, a similar centrifuge testing method was utilized in this study. 4. Model preparation and testing procedure 4.1. Typical layout for test
Stokoe, 1992, 1994), and the soil models for centrifuge tests were composed based on the expected properties from the laboratory test. The representative characteristics of the soil model used for each test are listed in Table 2. The shear wave velocity of weathered soil was approximately 350 m/s. Because it was similar to the gravelly soil in the Hualien LSST simulated by Ha et al. (2012), the same method was utilized for the soil modeling. The compaction of the mixed soils between the weathered soil and small gravel (approximately 2–3 mm) could simulate the target shear wave velocity. The mixture ratios for the weathered soil and small gravel were 4:6 for the weathered soil and 5.5:4.5 for the backfill soil, respectively. Weathered rock is generally used as a bedrock condition for the installation of a nuclear power plant, and its shear wave velocity is around 900 m/s. Although it would have been ideal to use real rock for the modeling, a compacted cemented sand was selected because of the experimental limitations. The centrifuge test was performed three days after the compaction to ensure that the cement had hardened. In the WR, the modulus was too stiff to be affected by the confining pressure. Thus, the FFRC test was performed instead of the RC test to measure the shear wave velocity of the specimen (Kim et al., 1997). The dry sand condition was modeled by the pluviation of the silica sand (Lee et al., 2013). This soil condition was modeled to represent loose sand and soft soil conditions, and the effect of the soil nonlinearity on the SSI could be effectively assessed as increasing the earthquake intensity. In this study, the ground water table was not simulated for simplicity. The RC tests were performed on weathered soil and dry sand. Based on the result of the RC test, the shear wave velocity profiles were found, and the nonlinear soil characteristics, G/Gmax , and damping curves were evaluated. The bender element arrays were buried to measure the in-flight shear wave velocity profiles during
The schematic diagrams for each test are represented in Fig. 2 as the prototype scale. Dynamic centrifuge tests were performed with various subsoil conditions (different stiffness values): weathered soil with sand backfill (WSS), weathered soil with stiff backfill (WSB), sandy soil (SS), and weathered rock (WR). The simulated depths of the weathered soil and SS were 29 m, whereas that of the WR was 20 m. The length and width of the target model soil area were 24.5 m and 24.5 m, respectively. Regardless of the subsoil test condition, the same superstructure was used in each test to see the effect of the soil stiffness on the SSI. Accelerometers and bender elements were used as measuring sensors. The seismic behaviors of the soil, foundation, and structure during earthquake loading could be recorded by these accelerometers. In the case of the weathered soil and WR, the accelerometers were attached to the wall of the ESB box because of the potential damage during the compaction used for the soil model preparation. The bender element was utilized for measuring the shear wave velocity of the soil model at an in-flight state by using a concept similar to that of a field cross-hole test (Kim and Kim, 2010). This means that a shear wave velocity profile of the soil model that was close to the real stress condition could be obtained directly during the centrifuge test. 4.2. Soil modeling SFSI is the main phenomenon investigated in this study, and it was important to make reduced soil models to simulate the various dynamic soil properties. In particular, the shear wave velocity profile, shear modulus reduction (G/Gmax ), and damping curves are serious variables that must be considered in seismic design and analysis. To evaluate dynamic soil properties, laboratory tests such as a resonant column test and FFRC test were performed (Kim and
67
Fig. 3. Shear wave velocity profile for soil models.
68
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Table 2 Information for soil model. Soil condition
Soil modeling material
Depth under structure (m)
Average shear wave velocity (m/s)
Soil density (t/m3 )
Site period (s)
Friction angle (degree)
Weathered soil with backfill (WSB)
Weathered soil + small gravel Weathered soil + small gravel Silica sand
24
350
2.1 (WS)
0.43
42 (WS)
24
350
2.1 (WS)
0.42
42 (WS)
24
225
0.58
Silica sand + cement
15
850
1.7 (SS) 1.9 (SS)
39 (SS) –
Weathered soil with sand (WSS) Sand soil (SS) Weathered rock (WR)
n
(1) Site period : TG = 4
i=1
0.23
Di /VSi .
the centrifuge test, except for the case of the WR, where the shear wave velocity profile was estimated using the FFRC test. Fig. 3 presents the measured in-flight shear wave velocity profiles for the soil models, and the G/Gmax and damping ration with the strain level are depicted in Fig. 4. 4.3. Structure modeling The structural model was the same structure as the one simulated in the Hualien LSST to evaluate the seismic SFSI in the centrifuge. The model structure was constructed from the reducedscale by 50 times using the centrifuge scaling law. The dimensioned schematic of model structure is described in Fig. 5 with model and prototype scale. The centroid location and moment of inertia were also included. The final model structure has a mass of 11.7 kg, which is corresponding to about 1580 tons in corresponding prototype scale. Because of the difficulties involved in constructing a model structure with concrete aggregate and the appropriate reinforcement, aluminum was chosen for the model fabrication. Because of the difference in the damping ratios of concrete and aluminum, there was a possibility that the dynamic response through the structure could have been higher in the centrifuge test compared to the real world. This study focused on evaluating the responses of the soil foundation interaction due to earthquake loadings, instead of the structural response. 4.4. Input earthquake The Chi-Chi earthquake motion that was recorded at Hualien in 1999 was used for the input earthquake motion (Chen and Hsu,
Table 3 Test procedure. Earthquake
ChiChi EQ. (N–S dir.)
Bedrock peak ground acceleration as soil model WSB (g)
WSS (g)
SS (g)
WR (g)
0.0431 0.0707 0.0903 0.1306 0.2847
0.0396 0.0627 0.0925 0.1392 0.3088
0.0435 0.0828 0.1167 0.1818 0.2565
0.0409 0.1013 0.1526 0.2523 0.3377
2009). By maintaining the waveform of the earthquake, but changing the peak acceleration, staged earthquake loadings from low to high amplitudes were applied. Maximum peak acceleration of input motion is about 0.4 × g by the limitation of the earthquake simulator in the centrifuge facility. With site amplifications, the motion in the soil and the foundation was increased up to 0.5 × g showing non-linear behaviors of the soils. The earthquake motions used in this research were passed through bandpass filtering from 30 Hz to 300 Hz on the model scale because of the limitations of the shaking table. The input earthquake motions in the time domain and frequency domain are shown in Fig. 6, and the test stages are listed in Table 3. 5. Test result and discussion 5.1. Peak ground acceleration in soil layer The variations in the amplification ratios of the peak acceleration with depth for four different soil models are shown for two bedrock earthquake intensities of approximately 0.04 × g and 0.3 × g in Fig. 7. Because it is difficult to duplicate the exact peak
Fig. 4. Soil nonlinear characteristics depending on strain level (weathered soil and silica sand in confining pressure of 50 kPa).
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
69
Fig. 5. Model structure schematic diagram (unit: m).
Fig. 6. Input earthquake motion (ChiChi Earthquake, N–S direction).
acceleration of the bedrock motion when using a hydraulic shaking table, comparisons of the seismic responses of each soil model were made using a normalized amplification ratio with the input bedrock motion as the denominator. When a weak earthquake load was applied, the soil layer behaved linearly, and the site period could be determined by the shear wave velocity without a reduction. Fig. 7(a) shows the amplification characteristics at a low seismic intensity. It can be noticed that the ground motion is steadily amplified from the bedrock to the ground surface, except in the case of weathered rock (WR). At the ground surface, the amplitudes of the peak acceleration are in the order of weathered soil with sand backfill (WSS), sandy soil (SS), weathered soil with stiff backfill (WSB), and WR. The maximum amplification at the ground surface is much more than doubled compared to the bedrock motion, except in the case of WR, which is less than 1.5 times because of the high stiffness of WR. The effect of the backfill stiffness on the site amplification was determined based on the amplification result for the upper 5 m. With stiff backfill (WSB), the amplification at the surface was less than that of WSS, showing the advantage of using stiff backfill for less ground motion near the surface. Most of the nuclear containment structures in high seismicity areas are designed using a peak rock outcrop motion of close to or higher than 0.3 × g, which can cause significant damage to an ordinary structure and substantial nonlinear characteristics of the adjacent soil. Above the elastic threshold strain, the shear modulus will decrease and the damping ratio will increase with an increase in the strain amplitude (Kim and Stokoe, 1994), but the effects of soil nonlinearity on SFSI are still uncertain and major challenges in the design of nuclear containment structures (Dutta and Roy, 2002). Thus, it is of great interest to understand the seismic behaviors of a soil–foundation-structure system with respect to the high intensity of earthquake loading (Kobayashi et al., 2002).
70
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 7. Peak ground acceleration profile.
The variations in the amplification ratio profiles of the four different soil models at a high-intensity bedrock motion of about 0.3 × g are depicted in Fig. 7(b). In contrast to the case of low-intensity bedrock motion, there was almost no amplification or sometimes de-amplifications up to a depth of 7.5 m, and some amplifications occurred at depths greater than 7.5 m in the WSB and WSS cases. For SS, the ground motion was steadily amplified, but the amplification ratio was much smaller than the case of low intensity. For WR, the amplification trend at a high earthquake intensity was similar to that at a low intensity. Generally, in the cases of WSB, WSS, and SS, the amplification ratio at a high seismic intensity was greatly reduced compared to the results at a low seismic intensity. These differences were caused by the soil’s nonlinear properties according to the strain level, which caused a decrease in the normalized shear modulus and an increase in the damping ratio of the soil with an increase in the seismic load. The variations in the amplification ratios at the ground surface with the bedrock earthquake intensities are shown in Fig. 8. As seen in Fig. 7, the amplification ratio generally decreases from 3 to 1.5 as the earthquake intensity increases in the cases of WSB, WSS, and SS, but the amplification ratio is small and almost constant in the case of WR. The result of the WSB layer appeared to have a smaller amplification at a shallow depth, similar to the result shown in Fig. 7(a). It is important to mention that the centrifuge modeling can be effectively used to simulate site amplification phenomena considering variations in the soil stiffness and damping characteristics, and it can be directly used in seismic design or for the calibration of a numerical analysis method.
of the site, which are expressed as a site period, along with the dynamic soil properties and bedrock earthquake intensity. The site period is related to the soil stiffness and depth of the bedrock, and the soil stiffness varies with respect to the earthquake intensity because of the nonlinearity of the soil. In the seismic design, the site period of the soil layer and natural period of the structure are serious factors on the SSI and should be reflected. For this purpose, the variations in the ratios of the response spectrum (RRS) between the surface and bedrock motions of four different models are depicted for low and high bedrock intensities in Fig. 9. In the case of a low seismic-intensity earthquake, as shown in Fig. 9(a), the site periods can be clearly distinguished depending on the soil stiffness of the model: 0.52 s for SS, 0.45 s for WSB and WSS, and 0.3 s for WR, showing that the site period decreases as the soil stiffness increases. The site period can be calculated by the depth of the bedrock and the shear wave velocity, assuming a
5.2. Site period and ratio of response spectrum For the reliable seismic design of a nuclear power plant, it is important to identify the variations in the frequency characteristics
Fig. 8. Amplification ratio comparison on soil surface by bedrock PGA increments.
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
7
WSB WSS SS WR
Ratio of Response Spectrum
T = 0.5 sec 6 5 4
3 2 1 0
0
0.2
0.4
Period (s) (a) PGA = 0.04g
0.6
0.8
1
7
Ratio of Response Spectrum
T = 0.5 sec
WSB WSS SS WR
6 5 4
3 2 1 0 0
0.5
1 Period (s) (b) PGA = 0.30g
1.5
2
Fig. 9. Ratio of response spectrum for comparison of site period.
fixed-free boundary condition. The calculated site period in Table 2 matches the experimental one for SS, but is slightly shorter for the cases of weathered soil and WR. The maximum amplification ratios near the site period range from four to six, which are substantially high because of the low damping ratio at small strains. The response spectrum ratios between the surface and bedrock motions for high-intensity bedrock motion are presented in Fig. 9(b). The site periods of the four soil models were lengthened to longer periods, and the maximum amplification ratios were significantly decreased compared to the cases of the low seismic intensity. The soil’s nonlinear characteristics, which caused the modulus decrease and damping increase with an increase in the stain amplitude, were the main reasons for these phenomena. However, the degrees of the period lengthening differed according to the subsoil condition. The lengthening of WR, which was relatively stiff, was approximately 0.1 s, but those of the weathered soil and sandy soils, which were relatively soft, were approximately 0.3 s. In addition, the bell-shape of the graph around the site period in the WR cases was sharper than those in other cases. This occurred because the amount of nonlinear behavior for the soil was small as a result of the smaller increase in the strain level with the seismic load under the stiff soil condition. The variations in the response spectrum with the earthquake intensity and soil stiffness are plotted in Fig. 10, which explains the gradual increase in the site period and decrease in the amplification ratio. 5.3. Comparison of responses between soil and foundation There are major uncertainties in the seismic design of a nuclear power plant, particularly in the evaluation of SFSI induced by inertia
Fig. 10. Ratio of response spectrum by bedrock PGA increments.
71
72
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 11. Different possible locations for specification of input motion during dynamic analysis (Ghosh and Madabhushi, 2006).
and kinematic interactions and soil nonlinearities (Roesset, 1998). In most cases, free-field surface level ground motion is selected as the control motion used at the foundation level, neglecting kinematic interactions (Verma, 2004). This method is usually conservative but sometimes leads to an unsafe design (Gazetas and Mylonakis, 1998). Therefore, it is a critical issue to determine the location of the control motion when estimating the design seismic loads for a nuclear power plant. Ghosh and Madabhushi (2006) discussed five possible locations, as shown in Fig. 11, including the bedrock, rock outcrop, surface free field, foundation level in the free field, and foundation. In this study, the responses at four different locations, including the bedrock, foundation level free field, foundation, and ground surface, were compared with respect to the soil conditions and earthquake intensities. Typical time domain accelerations during earthquake at soil, foundation, and bedrock are compared in Fig. 12 for the case of the SS model. In the time domain, it is difficult to distinguish the acceleration responses at the foundation level between the soil and the foundation, while the amplification phenomena are easily observed from the bedrock to ground level. Thus, the response spectrum and PGA of the measured accelerations, which show the wave characteristics, were utilized for a more effective comparison of the control motions. The variations in the response spectra measured at the bedrock, free surface, and foundation level in the free field and foundation are depicted for various soil conditions and earthquake intensities in Fig. 13. The response of the foundation during earthquakes, which can be considered to be an input motion for the design of a superstructure, includes the effects of both inertia and kinematic interactions, and thus can be considered to be an index for the evaluation of the interaction effects (Ghosh and Madabhushi, 2006). Compared to the bedrock motion, the responses of the free-field surface and foundation level motions, as well as the foundation motion, were amplified, particularly near the site period, as a result of the local site effects, as discussed in the previous section. It is interesting to note that the spectral acceleration of the foundation was close to that of the surface free-field motion near the resonance, but slightly higher than that of the foundation-level free-field motion in most cases. These SFSI phenomena have already been discussed by several researchers through numerical analysis and large scale seismic tests. The amplitude of foundation motion in
the high-frequency (the short period) could be reduced than that of free-field by kinematic interaction, and the inertia effect of the structure causes the amplification in the foundation motion (Sarma and Srbulov, 1996; Pitilakis et al., 2008; Yamahara, 1970; Hradilek
Fig. 12. Time domain acceleration comparison between soil and foundation with bedrock motion (SS, bedrock PGA = 0.04 × g).
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
3
1 0.8
Foundation Spectral Acceleration (g)
Foundation Spectral Acceleration (g)
73
Soil(Surface) Soil(G.L.=5m)
0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
0
2
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (a) WSB
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
0
2
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (b) WSS
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
2
0
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (c) SS
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
(i) PGA = 0.04g
1.5
2
(d) WR
0
0.5
1 Period (s)
(ii) PGA = 0.30g
Fig. 13. Comparison of response spectrum between soils and foundation.
1.5
2
74
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(a) WSB 4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(b) WSS 4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2
6. Conclusions
1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(c) SS 4
The main objective of this study was to evaluate the seismic load applied to a nuclear containment structure while considering the soil–foundation-structure interaction (SFSI) in dynamic centrifuge tests. A series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. From these test result, the main conclusions can be summarized as follow:
Foundtion
3.5
Amplifcation Ratio
below 0.5 s compared to the foundation motion. This may have been the result of the resonance of the sand backfill at a shallow depth. In the case of WR, the surface free-field motion was larger than the foundation motion at both low and high earthquake intensities. The variations in the amplification ratios of the motions at the free-field surface, free-field foundation level, and foundation with the bedrock intensities are plotted for four different soil conditions in Fig. 14. With an increase in the earthquake intensity, the amplification ratio decreases owing to the soil nonlinearity, except in the case of WR, as discussed in the previous session. It can be noticed that the foundation motion is usually higher than the free-field motion at the foundation level as a result of the SFSIs expressed in terms of the inertia interactions. However, it is interesting to note that the free-field ground surface motion is higher than the foundation motion in most cases, which shows that the current practice of using the free-field ground motion in seismic design is conservative in the view of PGA. The difference in amplitude between the surface free-field and foundation motions is higher in the cases with sand backfill (SS and WSS) because the sand layer near the ground surface can cause an amplification of the motion. For the case of WR, which is a very stiff soil condition, the amplification ratio is much smaller than those of the other cases and is less than 1.5 irrespective of the bedrock intensity. The difference in amplitude between the foundation and free-field motions is almost negligible. Therefore, it can be mentioned that the effects of SFSI are significantly affected by the adjacent soil conditions and earthquake intensities. The kinematic interaction effects become larger to seismic motions of soil and foundation as the soil stiffness softer, so the free-field ground surface motion is higher than the foundation motion. But the amplitude of difference reduced by the inertia effects when the earthquake intensity increase. And the SFSI effects are relatively small in the stiff soil condition.
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(d) WR Fig. 14. Amplification ratio comparison by bedrock PGA increments.
et al., 1973). This justifies the current practice of using the free-field surface motion in the estimation of seismic loads, and an increase in the foundation motion due to both the inertia and kinematic interactions could be noticed compared to the foundation level free-field motion. Moreover, it should be noted that the spectral acceleration of the surface free-field motion was more amplified at short periods
(1) In this paper, SFSI for nuclear containment structure under well controlled conditions were observed directly using verified experimental method. (2) The stiffness of soil under the structure highly affected to acceleration amplification characteristics such as less amplification in stiff soil condition. And amplification ratios through the soil layer were generally decreased by the input motion increasing. (3) In soft soil condition like sandy soil and weathered soil, soil nonlinear properties generated the less amplification in soil layer and the period lengthening in site period more than those in rock condition. (4) The performed centrifuge tests results shows that the spectral acceleration of the foundation was close to that of the surface free-field motion near the resonance, but slightly higher than that of the foundation-level free-field motion in most cases. (5) In the view of PGA, the free-field ground surface motion is higher than the foundation motion in most cases by the kinematic interaction effect. But the amplitude of difference reduced by the inertia effects when the earthquake intensity increases. And the SFSI effects are relatively small in the stiff soil condition.
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
These conclusions note that subsoil condition, earthquake intensity and control motion were seriously considered in seismic design for nuclear power plant. And even though there are some limitations of using only one model structure, the reported results show that the dynamic centrifuge tests could well describe the soilfoundation interaction phenomena and provide a reliable database for the calibration of numerical analysis method. Acknowledgements This research was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Ministry of Knowledge Economy (Grant no. 2010T100101048, Dynamic Geo-centrifuge Technology for Seismic Design Loads of Soil - Foundation - NPP Containment System). And this research was also supported by a grant (11 Technology Innovation D02) from Construction Technology Innovation Program funded by Ministry of Land, Infrastructure and Transport of Korean government. References Avilés, J., Pérez-Rocha, L.E., 1998. Effects of foundation embedment during building–soil interaction. Earthq. Eng. Struct. Dyn. 27 (12), 1523–1540. Chen, C.H., Hsu, S.Y., 2009. Seismic Motions at Hualien LSST Arrays During the 1999 Chi-Chi Earthquake. Earthquake Geotechnical Case Histories for Performancebased Design. Taylor & Francis Group, London, pp. 245–257, ISBN 978-0-41580484-4. Dutta, S.C., Roy, R., 2002. A critical review on idealization and modeling for interaction among soil–foundation-structure system. Comput. Struct. 80 (20), 1579–1594. Finn, W.D.L., Ledbetter, R.H., Beratan, L.L., 1986. Seismic soil structure interaction analysis and centrifuge model studies. Nucl. Eng. Des. 94, 53–66. Gazetas, G., Mylonakis, G., 1998. Seismic Soil Structure Interaction: New Evidence and Emerging Issues. Geotechnical Special Publication No. 75., pp. 1119–1174. Ghosh, B., Madabhushi, S.P.G., 2006. Centrifuge modelling of seismic soil structure interaction effects. Nucl. Eng. Des. 237, 887–896. Ha, J.G., Lee, S.H., Kim, S.H., Kim, D.S., 2012. Dynamic centrifuge simulation for soil–foundation–NPP containment system. In: GeoCongress 2012, pp. 2008–2016.
75
Hradilek, P.J., Carriveau, A.R., Saragoni, G.R., Duke, C.M., 1973. Evidence of Soilstructure Interaction in Earthquakes, 2. World Conf. on Earthquake Eng. (Fifth), Rome, Italy, pp. 2076–2079. Kim, D.S., Kim, N.R., Choo, Y.W., Cho, G.C., 2013a. A newly developed state-of-the-art geotechnical centrifuge in Korea. KSCE J. Civil Eng. 17 (1), 77–84. Kim, D.S., Kim, N.R., Choo, Y.W., 2013b. Self-balanced earthquake simulator on centrifuge and dynamic performance verification. KSCE J. Civil Eng. 17 (4), 651–661. Kim, D.S., Kweon, G.C., Lee, K.H., 1997. Alternative method of determining resilient modulus of compacted subgrade soils using free-free resonant column test. Transp. Res. Rec. 1577, 62–69. Kim, D.S., Stokoe, K.H., 1992. Characterization of resilient modulus of compacted subgrade soils using resonant column and torsional shear tests. Transp. Res. Rec. 1369, 89–91. Kim, D.S., Stokoe, K.H., 1994. Torsional motion monitoring-system for small-strain (10−5 to 10−3 percent) soil testing. Geotech. Test. J. 17 (1), 17–26. Kim, N.R., Kim, D.S., 2010. A shear wave velocity tomography system for geotechnical centrifuge testing. Geotech. Test. J. 33 (6), 434–444. Kim, S.H., Stewart, J.P., 2003. Kinematic soil-structure interaction from strong motion recordings. J. Geotech. Geoenviron. Eng. 129 (4), 323–335. Kobayashi, T., Yoshikawa, K., Takaoka, E., Nakazawa, M., Shikama, Y., 2002. Time history nonlinear earthquake response analysis considering materials and geometrical nonlinearity. Nucl. Eng. Des. 212 (1), 145–154. Lee, S.H., Choo, Y.W., Kim, D.S., 2013. Performance of equivalent shear beam (ESB) model container for dynamic geotechnical centrifuge tests. Soil Dyn. Earthq. Eng. 44, 102–114. Nuclear Regulatory Commission, March 2007. A Performance-Based Approach to Define the Site-Specific Earthquake Ground Motion. Regulatory Guide 1.208. Pitilakis, D., Dietz, M., Wood, D.M., Clouteau, D., Modaressi, A., 2008. Numerical simulation of dynamic soil-structure interaction in shaking table testing. Soil Dyn. Earthq. Eng. 28 (6), 453–467. Rayhani, M.H., El Naggar, M.H., 2008. Numerical modeling of seismic response of rigid foundation on soft soil. Int. J. Geomech. 8 (6), 336–346. Roesset, J.M., 1998. Seismic design of nuclear power plant—where are we now? Nucl. Eng. Des. 182, 3–15. Sarma, S.K., Srbulov, M., 1996. A simplified method for prediction of kinematic soil–foundation interaction effects on peak horizontal acceleration of a rigid foundation. Earthq. Eng. Struct. Dyn. 25 (8), 815–836. Schofield, A.N., 1980. Cambridge geotechnical centrifuge operations. Geotechnique 30, 227–268. Verma, U.S.P., 2004. Structural mechanics issues related to design and construction of concrete structures of Indian nuclear power plants. In: National Workshop on Advances in Material and Mechanics of Concrete Structures, pp. 220–225. Yamahara, H., 1970. Ground motions during earthquakes and the input loss of earthquake power to an excitation of buildings. Soil Found. 10 (2), 145–161.
Contents lists available at ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Evaluation of seismic behavior of soils under nuclear containment structures via dynamic centrifuge test Jeong Gon Ha, Dong-Soo Kim ∗ Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, South Korea
h i g h l i g h t s • A series of dynamic centrifuge tests were performed for NPP structure to investigate the soil–foundation-structure interaction with various soil conditions • • • •
from loose sand to weathered rock. SFSI phenomena for NPP structure were observed directly using experimental method. Effect of the soil stiffness and nonlinear characteristics on SFSI was estimated. There are comparisons of the control motions for seismic design of a NPP structure. Subsoil condition, earthquake intensity and control motion affected to seismic load.
a r t i c l e
i n f o
Article history: Received 11 December 2013 Received in revised form 16 June 2014 Accepted 17 June 2014
a b s t r a c t To evaluate the earthquake loads for the seismic design of a nuclear containment structure, it is necessary to consider the soil–foundation-structure interaction (SFSI) due to their interdependent behavior. Especially, understanding the effects of soil stiffness under the structure and the location of control motion to SFSI are very important. Motivated by these requirements, a series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. The different amplification characteristics in peak-accelerations profile and effects of soil-nonlinearity in response spectrum were observed. The dynamic behaviors were compared between surface of free-field and foundation of the structure for the evaluation of the control motion for seismic design. It was found that dynamic centrifuge test has potentials to estimate the seismic load considering SFSI. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The destruction of nuclear containment structures during earthquakes may cause a catastrophic loss of life, property damage, or disruption to society. Therefore, the safe and reliable seismic design of structures is crucial. To evaluate the earthquake loads for the seismic design of a nuclear containment structure, it is necessary to consider the soil-structure interaction (SSI) due to their interdependent behavior. Recently, the Nuclear Regulatory Commission (NRC) raised the shear wave velocity standard for a general rock outcrop to 2800 m/s (9200 ft/s) (Regulatory Guide 1.208). This means that the region where the soil-structure interaction is considered to occur has been increased
∗ Corresponding author. Tel.: +82 42 350 3619; fax: +82 42 350 7200. E-mail addresses: [email protected] (J.G. Ha), [email protected] (D.-S. Kim). http://dx.doi.org/10.1016/j.nucengdes.2014.06.013 0029-5493/© 2014 Elsevier B.V. All rights reserved.
and requires a more advanced research and design methodology. When estimating the seismic load, because most nuclear structures have been built on hard rock, the local site effects are not considered, and the soil-structure analyses are normally performed in the frequency domain, while assuming that the soil is a linear elastic material (Roesset, 1998). However, the demand for the construction of nuclear power plant structures on deep soil deposits has been increasing. In this soil condition, a seismic wave is trapped, leading to large soil deformations, and the soil shows significant nonlinear characteristics. The seismic behavior of a nuclear containment structure can be greatly altered by these local site conditions. During the evaluation of the earthquake load in the seismic design of nuclear power plant structures, another controversial point is the location of the control motion where the design motion is specified (Roesset, 1998; Avilés and Pérez-Rocha, 1998; Kim and
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Stewart, 2003; Pitilakis et al., 2008). In most cases, the free field surface level-ground motion is selected as the control motion and used at the foundation level, neglecting the kinematic interaction effects (Verma, 2004). However, recent researches about seismic design have referred the importance of considering SSI effects to defer the earthquake input motion (Rayhani and El Naggar, 2008; Yamahara, 1970; Hradilek et al., 1973). Roesset (1998) discussed five possible choices for the design earthquake motion: the free surface of the soil deposit at the site, a hypothetical outcropping of rock, bedrock when there is rock at some finite depth at the site, the elevation of the foundation in the free field, and directly at the foundation. To reliably estimate the seismic load, it is of interest to assess the characteristics and amplitude differences between these potential motions. So far, most of the seismic analyses considering SSI have been performed based on numerical methods. Although numerical methods have been successfully applied to the design of nuclear containment structures, they also contain numerous uncertainties, including nonlinear soil characteristics, an impedance problem, etc. (Roesset, 1998; Finn et al., 1986). Motivated by the need to calibrate numerical analysis tools for SSI analyses, large-scale seismic model tests have been utilized (Hualian, Lotong). Even though a large-scale seismic test (LSST) is the most precise method to evaluate the SSI, it is difficult to use it to perform various parametric studies because of the expense and difficulty in adjusting the test conditions. To complement it, a dynamic centrifuge test that reproduces the field stress condition indoors and provides a relatively cost saving can be used as an alternative. Ghosh and Madabhushi (2006) used a dynamic centrifuge test as a tool to investigate the foundation response of a typical power plant during an earthquake. Ha et al. (2012) successfully simulated the seismic behavior of the Hualien LSST in a centrifuge. The objective of this study was to evaluate the seismic load applied to a nuclear containment structure while considering the soil–foundation-structure interaction (SFSI) in dynamic centrifuge tests. This study focused on two main points: (i) the effect of the soil
65
Table 1 Scaling law for centrifuge modeling. Parameters
Centrifuge model scaling
Length Velocity Acceleration Strain Stress Time (dynamic)
1/N 1 N 1 1 1/N
stiffness and corresponding nonlinear characteristics on SFSI and (ii) the comparison of the control motions for the seismic design of a nuclear containment structure. A series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. The dynamic soil properties of the weathered soil, WR, and sandy soil (SS), in terms of the shear wave velocity profile, modulus reduction, and damping curves, were evaluated using inflight bender element tests and laboratory resonant column tests. With small-scale model structures, the soil and structure responses were measured during a series of shakings in the centrifuge test, and an assessment was made of the importance of considering the soil nonlinearities and determining the control motions when evaluating the earthquake load for nuclear power plants. 2. Centrifuge equipment The basic idea of physical modeling using a centrifuge is to accelerate a reduced-scale geotechnical structure to the appropriate high g-level to simulate the prototype-scale stress field in the model structure. Centrifuge modeling with shaking table equipment provides an excellent opportunity to observe the SSI in a scale model. To make good use of this test method, reasonable scaleddown modeling with a proven scaling law is important. Schofield (1980) derived the scaling law for centrifuge tests. Some of the variables related to the modeling in this study are listed in Table 1.
Fig. 1. Test result for simulation of Hualien large-scale seismic test (Ha et al., 2012).
66
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 2. Typical test layout.
The beam-type centrifuge facility in KAIST was used to perform all the tests reported here in a 50 × g centrifugal acceleration field. This facility has a platform radius of 5 m and a maximum capacity of 240 g-tons (Kim et al., 2013a,b). The model earthquake was generated by a self-balanced electro-hydraulic earthquake simulator mounted on the centrifuge. The base shaking acceleration could be exerted to a maximum value of 0.5 × g in the prototype scale at 50 × g of centrifugal acceleration. The allowable frequency
range of the earthquake loading was from 30 Hz to 300 Hz on the model scale, and the utilization of the self-balanced shaking table for the sinusoidal and real earthquake motion was verified (Lee et al., 2013). During the dynamic centrifuge test, an equivalent shear beam (ESB) model box was used. This box can simulate a wall deformation that is similar to that of soil using the bearing and rubber located between the wall layers (Lee et al., 2013). The model box has
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
internal dimensions of 490 mm × 490 mm × 630 mm and external dimensions of 670 mm × 670 mm × 650 mm in length, width, and height, respectively. 3. Verification of centrifuge testing method SFSI is a very complex problem with numerous factors that influence the seismic behavior. Thus, precisely controlled test conditions and the verification of the dynamic centrifuge test of a nuclear power plant are needed. Based on these requirements, Ha et al. (2012) simulated the Hualien LSST using a dynamic centrifuge test before this research. The Hualien LSST program was an international project to observe real SFSI for a nuclear containment structure and began recording earthquake data from 1990 at a 1/4 scale nuclear site above well-investigated soil. To simulate real earthquake phenomena, the soil, structure, and earthquake motion are the most important things to consider in modeling. More specific details for modeling methods are mentioned in the following sections. Fig. 1 compares the typical acceleration time-histories and frequency properties measured at the foundation and free-field surface during the Chi-Chi earthquake, and those recorded at the corresponding locations in the centrifuge test. The amplification characteristics and time-histories are well matched at the soil layer and the foundation of the structure. This shows the potential to simulate the Hualien LSST using a dynamic centrifuge test and verifying the reliability of the centrifuge test method for SFSI research. Thus, a similar centrifuge testing method was utilized in this study. 4. Model preparation and testing procedure 4.1. Typical layout for test
Stokoe, 1992, 1994), and the soil models for centrifuge tests were composed based on the expected properties from the laboratory test. The representative characteristics of the soil model used for each test are listed in Table 2. The shear wave velocity of weathered soil was approximately 350 m/s. Because it was similar to the gravelly soil in the Hualien LSST simulated by Ha et al. (2012), the same method was utilized for the soil modeling. The compaction of the mixed soils between the weathered soil and small gravel (approximately 2–3 mm) could simulate the target shear wave velocity. The mixture ratios for the weathered soil and small gravel were 4:6 for the weathered soil and 5.5:4.5 for the backfill soil, respectively. Weathered rock is generally used as a bedrock condition for the installation of a nuclear power plant, and its shear wave velocity is around 900 m/s. Although it would have been ideal to use real rock for the modeling, a compacted cemented sand was selected because of the experimental limitations. The centrifuge test was performed three days after the compaction to ensure that the cement had hardened. In the WR, the modulus was too stiff to be affected by the confining pressure. Thus, the FFRC test was performed instead of the RC test to measure the shear wave velocity of the specimen (Kim et al., 1997). The dry sand condition was modeled by the pluviation of the silica sand (Lee et al., 2013). This soil condition was modeled to represent loose sand and soft soil conditions, and the effect of the soil nonlinearity on the SSI could be effectively assessed as increasing the earthquake intensity. In this study, the ground water table was not simulated for simplicity. The RC tests were performed on weathered soil and dry sand. Based on the result of the RC test, the shear wave velocity profiles were found, and the nonlinear soil characteristics, G/Gmax , and damping curves were evaluated. The bender element arrays were buried to measure the in-flight shear wave velocity profiles during
The schematic diagrams for each test are represented in Fig. 2 as the prototype scale. Dynamic centrifuge tests were performed with various subsoil conditions (different stiffness values): weathered soil with sand backfill (WSS), weathered soil with stiff backfill (WSB), sandy soil (SS), and weathered rock (WR). The simulated depths of the weathered soil and SS were 29 m, whereas that of the WR was 20 m. The length and width of the target model soil area were 24.5 m and 24.5 m, respectively. Regardless of the subsoil test condition, the same superstructure was used in each test to see the effect of the soil stiffness on the SSI. Accelerometers and bender elements were used as measuring sensors. The seismic behaviors of the soil, foundation, and structure during earthquake loading could be recorded by these accelerometers. In the case of the weathered soil and WR, the accelerometers were attached to the wall of the ESB box because of the potential damage during the compaction used for the soil model preparation. The bender element was utilized for measuring the shear wave velocity of the soil model at an in-flight state by using a concept similar to that of a field cross-hole test (Kim and Kim, 2010). This means that a shear wave velocity profile of the soil model that was close to the real stress condition could be obtained directly during the centrifuge test. 4.2. Soil modeling SFSI is the main phenomenon investigated in this study, and it was important to make reduced soil models to simulate the various dynamic soil properties. In particular, the shear wave velocity profile, shear modulus reduction (G/Gmax ), and damping curves are serious variables that must be considered in seismic design and analysis. To evaluate dynamic soil properties, laboratory tests such as a resonant column test and FFRC test were performed (Kim and
67
Fig. 3. Shear wave velocity profile for soil models.
68
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Table 2 Information for soil model. Soil condition
Soil modeling material
Depth under structure (m)
Average shear wave velocity (m/s)
Soil density (t/m3 )
Site period (s)
Friction angle (degree)
Weathered soil with backfill (WSB)
Weathered soil + small gravel Weathered soil + small gravel Silica sand
24
350
2.1 (WS)
0.43
42 (WS)
24
350
2.1 (WS)
0.42
42 (WS)
24
225
0.58
Silica sand + cement
15
850
1.7 (SS) 1.9 (SS)
39 (SS) –
Weathered soil with sand (WSS) Sand soil (SS) Weathered rock (WR)
n
(1) Site period : TG = 4
i=1
0.23
Di /VSi .
the centrifuge test, except for the case of the WR, where the shear wave velocity profile was estimated using the FFRC test. Fig. 3 presents the measured in-flight shear wave velocity profiles for the soil models, and the G/Gmax and damping ration with the strain level are depicted in Fig. 4. 4.3. Structure modeling The structural model was the same structure as the one simulated in the Hualien LSST to evaluate the seismic SFSI in the centrifuge. The model structure was constructed from the reducedscale by 50 times using the centrifuge scaling law. The dimensioned schematic of model structure is described in Fig. 5 with model and prototype scale. The centroid location and moment of inertia were also included. The final model structure has a mass of 11.7 kg, which is corresponding to about 1580 tons in corresponding prototype scale. Because of the difficulties involved in constructing a model structure with concrete aggregate and the appropriate reinforcement, aluminum was chosen for the model fabrication. Because of the difference in the damping ratios of concrete and aluminum, there was a possibility that the dynamic response through the structure could have been higher in the centrifuge test compared to the real world. This study focused on evaluating the responses of the soil foundation interaction due to earthquake loadings, instead of the structural response. 4.4. Input earthquake The Chi-Chi earthquake motion that was recorded at Hualien in 1999 was used for the input earthquake motion (Chen and Hsu,
Table 3 Test procedure. Earthquake
ChiChi EQ. (N–S dir.)
Bedrock peak ground acceleration as soil model WSB (g)
WSS (g)
SS (g)
WR (g)
0.0431 0.0707 0.0903 0.1306 0.2847
0.0396 0.0627 0.0925 0.1392 0.3088
0.0435 0.0828 0.1167 0.1818 0.2565
0.0409 0.1013 0.1526 0.2523 0.3377
2009). By maintaining the waveform of the earthquake, but changing the peak acceleration, staged earthquake loadings from low to high amplitudes were applied. Maximum peak acceleration of input motion is about 0.4 × g by the limitation of the earthquake simulator in the centrifuge facility. With site amplifications, the motion in the soil and the foundation was increased up to 0.5 × g showing non-linear behaviors of the soils. The earthquake motions used in this research were passed through bandpass filtering from 30 Hz to 300 Hz on the model scale because of the limitations of the shaking table. The input earthquake motions in the time domain and frequency domain are shown in Fig. 6, and the test stages are listed in Table 3. 5. Test result and discussion 5.1. Peak ground acceleration in soil layer The variations in the amplification ratios of the peak acceleration with depth for four different soil models are shown for two bedrock earthquake intensities of approximately 0.04 × g and 0.3 × g in Fig. 7. Because it is difficult to duplicate the exact peak
Fig. 4. Soil nonlinear characteristics depending on strain level (weathered soil and silica sand in confining pressure of 50 kPa).
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
69
Fig. 5. Model structure schematic diagram (unit: m).
Fig. 6. Input earthquake motion (ChiChi Earthquake, N–S direction).
acceleration of the bedrock motion when using a hydraulic shaking table, comparisons of the seismic responses of each soil model were made using a normalized amplification ratio with the input bedrock motion as the denominator. When a weak earthquake load was applied, the soil layer behaved linearly, and the site period could be determined by the shear wave velocity without a reduction. Fig. 7(a) shows the amplification characteristics at a low seismic intensity. It can be noticed that the ground motion is steadily amplified from the bedrock to the ground surface, except in the case of weathered rock (WR). At the ground surface, the amplitudes of the peak acceleration are in the order of weathered soil with sand backfill (WSS), sandy soil (SS), weathered soil with stiff backfill (WSB), and WR. The maximum amplification at the ground surface is much more than doubled compared to the bedrock motion, except in the case of WR, which is less than 1.5 times because of the high stiffness of WR. The effect of the backfill stiffness on the site amplification was determined based on the amplification result for the upper 5 m. With stiff backfill (WSB), the amplification at the surface was less than that of WSS, showing the advantage of using stiff backfill for less ground motion near the surface. Most of the nuclear containment structures in high seismicity areas are designed using a peak rock outcrop motion of close to or higher than 0.3 × g, which can cause significant damage to an ordinary structure and substantial nonlinear characteristics of the adjacent soil. Above the elastic threshold strain, the shear modulus will decrease and the damping ratio will increase with an increase in the strain amplitude (Kim and Stokoe, 1994), but the effects of soil nonlinearity on SFSI are still uncertain and major challenges in the design of nuclear containment structures (Dutta and Roy, 2002). Thus, it is of great interest to understand the seismic behaviors of a soil–foundation-structure system with respect to the high intensity of earthquake loading (Kobayashi et al., 2002).
70
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 7. Peak ground acceleration profile.
The variations in the amplification ratio profiles of the four different soil models at a high-intensity bedrock motion of about 0.3 × g are depicted in Fig. 7(b). In contrast to the case of low-intensity bedrock motion, there was almost no amplification or sometimes de-amplifications up to a depth of 7.5 m, and some amplifications occurred at depths greater than 7.5 m in the WSB and WSS cases. For SS, the ground motion was steadily amplified, but the amplification ratio was much smaller than the case of low intensity. For WR, the amplification trend at a high earthquake intensity was similar to that at a low intensity. Generally, in the cases of WSB, WSS, and SS, the amplification ratio at a high seismic intensity was greatly reduced compared to the results at a low seismic intensity. These differences were caused by the soil’s nonlinear properties according to the strain level, which caused a decrease in the normalized shear modulus and an increase in the damping ratio of the soil with an increase in the seismic load. The variations in the amplification ratios at the ground surface with the bedrock earthquake intensities are shown in Fig. 8. As seen in Fig. 7, the amplification ratio generally decreases from 3 to 1.5 as the earthquake intensity increases in the cases of WSB, WSS, and SS, but the amplification ratio is small and almost constant in the case of WR. The result of the WSB layer appeared to have a smaller amplification at a shallow depth, similar to the result shown in Fig. 7(a). It is important to mention that the centrifuge modeling can be effectively used to simulate site amplification phenomena considering variations in the soil stiffness and damping characteristics, and it can be directly used in seismic design or for the calibration of a numerical analysis method.
of the site, which are expressed as a site period, along with the dynamic soil properties and bedrock earthquake intensity. The site period is related to the soil stiffness and depth of the bedrock, and the soil stiffness varies with respect to the earthquake intensity because of the nonlinearity of the soil. In the seismic design, the site period of the soil layer and natural period of the structure are serious factors on the SSI and should be reflected. For this purpose, the variations in the ratios of the response spectrum (RRS) between the surface and bedrock motions of four different models are depicted for low and high bedrock intensities in Fig. 9. In the case of a low seismic-intensity earthquake, as shown in Fig. 9(a), the site periods can be clearly distinguished depending on the soil stiffness of the model: 0.52 s for SS, 0.45 s for WSB and WSS, and 0.3 s for WR, showing that the site period decreases as the soil stiffness increases. The site period can be calculated by the depth of the bedrock and the shear wave velocity, assuming a
5.2. Site period and ratio of response spectrum For the reliable seismic design of a nuclear power plant, it is important to identify the variations in the frequency characteristics
Fig. 8. Amplification ratio comparison on soil surface by bedrock PGA increments.
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
7
WSB WSS SS WR
Ratio of Response Spectrum
T = 0.5 sec 6 5 4
3 2 1 0
0
0.2
0.4
Period (s) (a) PGA = 0.04g
0.6
0.8
1
7
Ratio of Response Spectrum
T = 0.5 sec
WSB WSS SS WR
6 5 4
3 2 1 0 0
0.5
1 Period (s) (b) PGA = 0.30g
1.5
2
Fig. 9. Ratio of response spectrum for comparison of site period.
fixed-free boundary condition. The calculated site period in Table 2 matches the experimental one for SS, but is slightly shorter for the cases of weathered soil and WR. The maximum amplification ratios near the site period range from four to six, which are substantially high because of the low damping ratio at small strains. The response spectrum ratios between the surface and bedrock motions for high-intensity bedrock motion are presented in Fig. 9(b). The site periods of the four soil models were lengthened to longer periods, and the maximum amplification ratios were significantly decreased compared to the cases of the low seismic intensity. The soil’s nonlinear characteristics, which caused the modulus decrease and damping increase with an increase in the stain amplitude, were the main reasons for these phenomena. However, the degrees of the period lengthening differed according to the subsoil condition. The lengthening of WR, which was relatively stiff, was approximately 0.1 s, but those of the weathered soil and sandy soils, which were relatively soft, were approximately 0.3 s. In addition, the bell-shape of the graph around the site period in the WR cases was sharper than those in other cases. This occurred because the amount of nonlinear behavior for the soil was small as a result of the smaller increase in the strain level with the seismic load under the stiff soil condition. The variations in the response spectrum with the earthquake intensity and soil stiffness are plotted in Fig. 10, which explains the gradual increase in the site period and decrease in the amplification ratio. 5.3. Comparison of responses between soil and foundation There are major uncertainties in the seismic design of a nuclear power plant, particularly in the evaluation of SFSI induced by inertia
Fig. 10. Ratio of response spectrum by bedrock PGA increments.
71
72
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
Fig. 11. Different possible locations for specification of input motion during dynamic analysis (Ghosh and Madabhushi, 2006).
and kinematic interactions and soil nonlinearities (Roesset, 1998). In most cases, free-field surface level ground motion is selected as the control motion used at the foundation level, neglecting kinematic interactions (Verma, 2004). This method is usually conservative but sometimes leads to an unsafe design (Gazetas and Mylonakis, 1998). Therefore, it is a critical issue to determine the location of the control motion when estimating the design seismic loads for a nuclear power plant. Ghosh and Madabhushi (2006) discussed five possible locations, as shown in Fig. 11, including the bedrock, rock outcrop, surface free field, foundation level in the free field, and foundation. In this study, the responses at four different locations, including the bedrock, foundation level free field, foundation, and ground surface, were compared with respect to the soil conditions and earthquake intensities. Typical time domain accelerations during earthquake at soil, foundation, and bedrock are compared in Fig. 12 for the case of the SS model. In the time domain, it is difficult to distinguish the acceleration responses at the foundation level between the soil and the foundation, while the amplification phenomena are easily observed from the bedrock to ground level. Thus, the response spectrum and PGA of the measured accelerations, which show the wave characteristics, were utilized for a more effective comparison of the control motions. The variations in the response spectra measured at the bedrock, free surface, and foundation level in the free field and foundation are depicted for various soil conditions and earthquake intensities in Fig. 13. The response of the foundation during earthquakes, which can be considered to be an input motion for the design of a superstructure, includes the effects of both inertia and kinematic interactions, and thus can be considered to be an index for the evaluation of the interaction effects (Ghosh and Madabhushi, 2006). Compared to the bedrock motion, the responses of the free-field surface and foundation level motions, as well as the foundation motion, were amplified, particularly near the site period, as a result of the local site effects, as discussed in the previous section. It is interesting to note that the spectral acceleration of the foundation was close to that of the surface free-field motion near the resonance, but slightly higher than that of the foundation-level free-field motion in most cases. These SFSI phenomena have already been discussed by several researchers through numerical analysis and large scale seismic tests. The amplitude of foundation motion in
the high-frequency (the short period) could be reduced than that of free-field by kinematic interaction, and the inertia effect of the structure causes the amplification in the foundation motion (Sarma and Srbulov, 1996; Pitilakis et al., 2008; Yamahara, 1970; Hradilek
Fig. 12. Time domain acceleration comparison between soil and foundation with bedrock motion (SS, bedrock PGA = 0.04 × g).
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
3
1 0.8
Foundation Spectral Acceleration (g)
Foundation Spectral Acceleration (g)
73
Soil(Surface) Soil(G.L.=5m)
0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
0
2
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (a) WSB
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
0
2
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (b) WSS
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
1.5
2
0
(i) PGA = 0.04g
0.5
1 Period (s)
1.5
2
(ii) PGA = 0.30g (c) SS
1
3
0.8
Foundation Spectral Acceleration (g)
Spectral Acceleration (g)
Foundation Soil(Surface) Soil(G.L.=5m) 0.6
Bedrock
0.4 0.2 0
2.5
Soil(Surface) Soil(G.L.=5m)
2
Bedrock 1.5 1 0.5 0
0
0.5
1 Period (s)
(i) PGA = 0.04g
1.5
2
(d) WR
0
0.5
1 Period (s)
(ii) PGA = 0.30g
Fig. 13. Comparison of response spectrum between soils and foundation.
1.5
2
74
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(a) WSB 4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(b) WSS 4 Foundtion
Amplifcation Ratio
3.5
Freefield(Surface)
3
Freefield(Foundation)
2.5
2
6. Conclusions
1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(c) SS 4
The main objective of this study was to evaluate the seismic load applied to a nuclear containment structure while considering the soil–foundation-structure interaction (SFSI) in dynamic centrifuge tests. A series of dynamic centrifuge tests were performed with various soil conditions from loose sand to weathered rock (WR), as well as different seismic intensities for the bedrock motion. From these test result, the main conclusions can be summarized as follow:
Foundtion
3.5
Amplifcation Ratio
below 0.5 s compared to the foundation motion. This may have been the result of the resonance of the sand backfill at a shallow depth. In the case of WR, the surface free-field motion was larger than the foundation motion at both low and high earthquake intensities. The variations in the amplification ratios of the motions at the free-field surface, free-field foundation level, and foundation with the bedrock intensities are plotted for four different soil conditions in Fig. 14. With an increase in the earthquake intensity, the amplification ratio decreases owing to the soil nonlinearity, except in the case of WR, as discussed in the previous session. It can be noticed that the foundation motion is usually higher than the free-field motion at the foundation level as a result of the SFSIs expressed in terms of the inertia interactions. However, it is interesting to note that the free-field ground surface motion is higher than the foundation motion in most cases, which shows that the current practice of using the free-field ground motion in seismic design is conservative in the view of PGA. The difference in amplitude between the surface free-field and foundation motions is higher in the cases with sand backfill (SS and WSS) because the sand layer near the ground surface can cause an amplification of the motion. For the case of WR, which is a very stiff soil condition, the amplification ratio is much smaller than those of the other cases and is less than 1.5 irrespective of the bedrock intensity. The difference in amplitude between the foundation and free-field motions is almost negligible. Therefore, it can be mentioned that the effects of SFSI are significantly affected by the adjacent soil conditions and earthquake intensities. The kinematic interaction effects become larger to seismic motions of soil and foundation as the soil stiffness softer, so the free-field ground surface motion is higher than the foundation motion. But the amplitude of difference reduced by the inertia effects when the earthquake intensity increase. And the SFSI effects are relatively small in the stiff soil condition.
Freefield(Surface)
3
Freefield(Foundation)
2.5
2 1.5 1 0.5 0
0
0.05
0.1
0.15 0.2 Bedrock PGA (g)
0.25
0.3
0.35
(d) WR Fig. 14. Amplification ratio comparison by bedrock PGA increments.
et al., 1973). This justifies the current practice of using the free-field surface motion in the estimation of seismic loads, and an increase in the foundation motion due to both the inertia and kinematic interactions could be noticed compared to the foundation level free-field motion. Moreover, it should be noted that the spectral acceleration of the surface free-field motion was more amplified at short periods
(1) In this paper, SFSI for nuclear containment structure under well controlled conditions were observed directly using verified experimental method. (2) The stiffness of soil under the structure highly affected to acceleration amplification characteristics such as less amplification in stiff soil condition. And amplification ratios through the soil layer were generally decreased by the input motion increasing. (3) In soft soil condition like sandy soil and weathered soil, soil nonlinear properties generated the less amplification in soil layer and the period lengthening in site period more than those in rock condition. (4) The performed centrifuge tests results shows that the spectral acceleration of the foundation was close to that of the surface free-field motion near the resonance, but slightly higher than that of the foundation-level free-field motion in most cases. (5) In the view of PGA, the free-field ground surface motion is higher than the foundation motion in most cases by the kinematic interaction effect. But the amplitude of difference reduced by the inertia effects when the earthquake intensity increases. And the SFSI effects are relatively small in the stiff soil condition.
J.G. Ha, D.-S. Kim / Nuclear Engineering and Design 277 (2014) 64–75
These conclusions note that subsoil condition, earthquake intensity and control motion were seriously considered in seismic design for nuclear power plant. And even though there are some limitations of using only one model structure, the reported results show that the dynamic centrifuge tests could well describe the soilfoundation interaction phenomena and provide a reliable database for the calibration of numerical analysis method. Acknowledgements This research was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Ministry of Knowledge Economy (Grant no. 2010T100101048, Dynamic Geo-centrifuge Technology for Seismic Design Loads of Soil - Foundation - NPP Containment System). And this research was also supported by a grant (11 Technology Innovation D02) from Construction Technology Innovation Program funded by Ministry of Land, Infrastructure and Transport of Korean government. References Avilés, J., Pérez-Rocha, L.E., 1998. Effects of foundation embedment during building–soil interaction. Earthq. Eng. Struct. Dyn. 27 (12), 1523–1540. Chen, C.H., Hsu, S.Y., 2009. Seismic Motions at Hualien LSST Arrays During the 1999 Chi-Chi Earthquake. Earthquake Geotechnical Case Histories for Performancebased Design. Taylor & Francis Group, London, pp. 245–257, ISBN 978-0-41580484-4. Dutta, S.C., Roy, R., 2002. A critical review on idealization and modeling for interaction among soil–foundation-structure system. Comput. Struct. 80 (20), 1579–1594. Finn, W.D.L., Ledbetter, R.H., Beratan, L.L., 1986. Seismic soil structure interaction analysis and centrifuge model studies. Nucl. Eng. Des. 94, 53–66. Gazetas, G., Mylonakis, G., 1998. Seismic Soil Structure Interaction: New Evidence and Emerging Issues. Geotechnical Special Publication No. 75., pp. 1119–1174. Ghosh, B., Madabhushi, S.P.G., 2006. Centrifuge modelling of seismic soil structure interaction effects. Nucl. Eng. Des. 237, 887–896. Ha, J.G., Lee, S.H., Kim, S.H., Kim, D.S., 2012. Dynamic centrifuge simulation for soil–foundation–NPP containment system. In: GeoCongress 2012, pp. 2008–2016.
75
Hradilek, P.J., Carriveau, A.R., Saragoni, G.R., Duke, C.M., 1973. Evidence of Soilstructure Interaction in Earthquakes, 2. World Conf. on Earthquake Eng. (Fifth), Rome, Italy, pp. 2076–2079. Kim, D.S., Kim, N.R., Choo, Y.W., Cho, G.C., 2013a. A newly developed state-of-the-art geotechnical centrifuge in Korea. KSCE J. Civil Eng. 17 (1), 77–84. Kim, D.S., Kim, N.R., Choo, Y.W., 2013b. Self-balanced earthquake simulator on centrifuge and dynamic performance verification. KSCE J. Civil Eng. 17 (4), 651–661. Kim, D.S., Kweon, G.C., Lee, K.H., 1997. Alternative method of determining resilient modulus of compacted subgrade soils using free-free resonant column test. Transp. Res. Rec. 1577, 62–69. Kim, D.S., Stokoe, K.H., 1992. Characterization of resilient modulus of compacted subgrade soils using resonant column and torsional shear tests. Transp. Res. Rec. 1369, 89–91. Kim, D.S., Stokoe, K.H., 1994. Torsional motion monitoring-system for small-strain (10−5 to 10−3 percent) soil testing. Geotech. Test. J. 17 (1), 17–26. Kim, N.R., Kim, D.S., 2010. A shear wave velocity tomography system for geotechnical centrifuge testing. Geotech. Test. J. 33 (6), 434–444. Kim, S.H., Stewart, J.P., 2003. Kinematic soil-structure interaction from strong motion recordings. J. Geotech. Geoenviron. Eng. 129 (4), 323–335. Kobayashi, T., Yoshikawa, K., Takaoka, E., Nakazawa, M., Shikama, Y., 2002. Time history nonlinear earthquake response analysis considering materials and geometrical nonlinearity. Nucl. Eng. Des. 212 (1), 145–154. Lee, S.H., Choo, Y.W., Kim, D.S., 2013. Performance of equivalent shear beam (ESB) model container for dynamic geotechnical centrifuge tests. Soil Dyn. Earthq. Eng. 44, 102–114. Nuclear Regulatory Commission, March 2007. A Performance-Based Approach to Define the Site-Specific Earthquake Ground Motion. Regulatory Guide 1.208. Pitilakis, D., Dietz, M., Wood, D.M., Clouteau, D., Modaressi, A., 2008. Numerical simulation of dynamic soil-structure interaction in shaking table testing. Soil Dyn. Earthq. Eng. 28 (6), 453–467. Rayhani, M.H., El Naggar, M.H., 2008. Numerical modeling of seismic response of rigid foundation on soft soil. Int. J. Geomech. 8 (6), 336–346. Roesset, J.M., 1998. Seismic design of nuclear power plant—where are we now? Nucl. Eng. Des. 182, 3–15. Sarma, S.K., Srbulov, M., 1996. A simplified method for prediction of kinematic soil–foundation interaction effects on peak horizontal acceleration of a rigid foundation. Earthq. Eng. Struct. Dyn. 25 (8), 815–836. Schofield, A.N., 1980. Cambridge geotechnical centrifuge operations. Geotechnique 30, 227–268. Verma, U.S.P., 2004. Structural mechanics issues related to design and construction of concrete structures of Indian nuclear power plants. In: National Workshop on Advances in Material and Mechanics of Concrete Structures, pp. 220–225. Yamahara, H., 1970. Ground motions during earthquakes and the input loss of earthquake power to an excitation of buildings. Soil Found. 10 (2), 145–161.