INPUT GROUND MOTION SELECTION FOR XIAOWAN HIGH ARCH DAM

Transcription

1 3 th World Conerence on Earthquake Engineering Vancouver, B.C., Canada August -6, 24 Paper No INPUT GROUND MOTION LECTION FOR XIAOWAN HIGH ARCH DAM CHEN HOUQUN, LI MIN 2, ZHANG BAIYAN 3 SUMMARY In this paper some special eatures o seismic input or design o large dams in China as well as the police decision, procedure and method o the input ground motion selecting or an extra high arch dam as Xiaowan dam with a height o 292m in area o high seismicity were presented. The site-speciic design response spectrum based on the scenario earthquake with maximum probability and design eective peak acceleration (EPA) by seismic hazard evaluation was described. Finally, a new method o generating an artiicial accelerogram compatible with object design response spectrum by using time domain method was involved. The result o selecting input ground motion parameters or Xiaowan project was illustrated. Key words: large dam; ground motion; seismic hazard evaluation; design response spectrum; artiicial accelerogram. INTRODUCTION As a common accepted principle, it is well known that or seismic design o a structure, the input ground motion should be consistent with the method o analysis o seismic eects and dynamic China Institute o Water Resources and Hydropower Research (IWHR). 2 Chegongzhuang West Rd., Beijing 44,China. chenhq@iwhr.com Tel: (86-) IWHR. leemin@iwhr.com Tel: (86-) IWHR. zhangby@iwhr.com Tel: (86-)

2 resistances o materials. The selecting input ground motion parameters are the important premise or seismic saety evaluation o structures, particularly, or the Xiaowan arch dam with a height o 292m and a design seismic intensity o Ⅸ degrees. Due to the recent rapid progress o computer s capabilities and numerical analysis techniques, to evaluate the saety o dams by simulating their behavior during an earthquake using dynamic analysis method becomes more widespread. To correspond such situation, the peak ground acceleration and response spectrum, sometimes the duration is included are considered as more critical ground motion parameters or seismic design. Since earthquakes are natural phenomena, inherently random primarily due to the tremendous uncertainties in the estimation o the ground behavior during an earthquake. It may be addressed probabilistically based on seismic hazard evaluation, which methodology is now widely used and initially developed by Cornell (968). The outcome o the standard probabilistic analysis is an estimate o the annual probability o exceedance o peak ground acceleration at an engineering site due to known and postulated potential seismic sources. Based on the hazard curve, the consistent site-speciic design response spectra and acceleration time histories can be developed. Principally, this is also the typical methodology used or input ground motion selecting o large dams in China. However, some special eatures o input ground motion selecting or large dams in China are described in this paper united in the typical example o Xiaowan arch dam. SOME SPECIAL FEATURES OF INPUT GROUND MOTION LECTING FOR LARGE DAMS IN CHINA Considering the act that any accident o serious damage o a high dam with huge reservoir can inlict grave secondary catastrophe upon surrounding communities, in China the seismic design o class,2 and 3 dams in seismic zones shall be regulated by the Speciications or seismic design o hydraulic structures (hereinater abbreviated as Speciications). It is a compulsory Speciications issued by Ministries. The earthquake ortiication classiication depends on the dam class related mainly to the probable ailure consequences and also the basic seismic peak acceleration o dam site. A new Seismic ground motion parameter zonation map o China was issued in August 2 and was compiled on the basis o probabilistic method o seismic hazard evaluation. In the map the peak ground acceleration ratings corresponding to a % probability o exceedance in 5 years at smooth surace o ordinary still soil. The map is used to meet ortiication requirement against earthquake or ordinary engineering projects, including small to medium dams. However, or dam o earthquake ortiication class a site-speciic seismic hazard evaluation shall be carried out according to the Code or seismic saety evaluation o engineering site as required in the Speciications. In this case a 2% probability o exceedance in years (with an associated return period o 495 years) is stipulated. The police decision or seismic design o large dam is described in the Speciications as ollows: Dams designed base on this Speciications will be able to withstand the design ground motion; i some local damages occur, they will still be serviceable ater ordinary treatment. At present a dual perormance criteria to address the saety and serviceability o dams through Maximum Credible Earthquake (MCE) and Operating Basis Earthquake (OBE) respectively is

3 prevalent in many agencies. However, it has not been accepted by dam engineering in China, as it has remarkable deects both in deining the MCE more ambiguously and in establishing the unquantized ailure criteria. It is in act very unlikely that a dam that cannot operate any more ater the OBE will ulill the MCE requirements. Furthermore, the objectives and requirements or seismic design o large dams in China are quite similar to that or the OBE while its unitary design seismic input is a match or the MCE. Obviously, the Speciications in China actually set rather strict demands on the design seismic input or large dams. Some urther details are explained in the ensuing section or selecting o design ground motion parameters or Xiaowan arch dam. DESIGN EFFECTIVE PEAK ACCELERATION BAD ON SPECIFIC ISMIC HAZARD EVALUATION OF DAM SITE The seismic hazard evaluation o dam site in China basically ollows the approach traditionally used in international engineering practice. However, two distinctive characteristics are worthy o note. First, the recurrence rate o events is speciied not or each potential seismic source but or seismic belt with suicient seismicity data. For more objectively relect the non-uniorm distribution o the recurrence rate o seismic belt both in time and space or dierent magnitude interval, a special distribution unction ( i, m j ) as weighted coeicient is involved to speciy the weighted coeicient o the recurrence rate or magnitude interval m j at i th seismic source. The magnitude interval is usually taken as.5. The unction( i, m j ) is determined by comprehensive estimation method based on the reliability o source identiication, seismicity pattern, and the researches on the medium to long term seismic prediction. Secondly, the seismic hazard evaluation or Xiaowan arch dam site is based on eective peak acceleration (EPA) instead o the current peak ground acceleration (PGA). As commonly recognized, the PGA has a lack o predictability in the near-ield which is more important or high dam in severe seismic area. Also, it oten occurs at high requencies, which is o little engineering signiicance to large dams. By analyzing 45 accelerograms with M 4.5 recorded at rock sites in the western United States, an average normalized spectral with its peak at period.2 second and a ampliication actor o 2.5 is obtained or events with dierent magnitude and distance intervals. Thereore, the EPA is deined as the spectral acceleration at period.2 second divided by an average ampliication actor o 2.5. Then, an attenuation relationship or EPA is derived based on the regression model o Abrahamson (997) [] as ollows: 2 lg( EPA ) = M.52M.357lg( R exp(.3396M )) () Where the unit o EPA is gal, and the variance σ in the regression model is taken as a constant o.242 or convenience. The distribution map o potential seismic sources and the hazard curve ater uncertainty rectiication based on a logarithmic normal distribution assumption are shown in the Figure and 2 respectively. The design probabilistic level or Xiaowan arch dam has been decided as % in 6 years with an

4 associated return period o 5695 years. From the hazard curve the corresponding design EPA is 3.6 gal..e+.e- annuual probability o exceedance.e-2.e-3.e-4 σ =.242.E-5 Fig. Distribution map o potential seismic sources or Xiaowan dam site.e EPA (g) Fig.2 Hazard curve or Xiaowan arch dam SITE-SPECIFIC DESIGN RESPON SPECTRUM FROM HAZARD- CONSISTENT SCENARIO EARTHQUAKE WITH MAXIMUM PROBABILITY For seismic design the so-called equal-hazard spectra are requently used. Upon speciying the design probability o exceedance, ordinates are read o rom each spectral hazard curve with dierent period and are plotted against period to orm the spectrum with the same probability o exceedance at each period. The equal-hazard spectrum is the result o many earthquakes with dierent magnitudes and locations. As a result, it does not relect the physical characteristics o spectrum o a real earthquake associated with the actual magnitude and distance. Especially, as shown by practice, its spectral value near the period o sec. is usually artiicially ampliied while it is o signiicant importance or high arch dam like Xiaowan. That is why a hybrid method combining both probabilistic and deterministic approaches o selecting speciic scenario earthquake to determine the site-speciic design response spectrum was provided or Xiaowan project. The conception o scenario earthquake was proposed by McGuire(98,995) [2], Ishikawa and Kameda (988) [3], Chapman(995) [4] and others. As an important premise, the peak ground acceleration at dam site o a scenario earthquake must be consistent with the value corresponding to the peak acceleration with design ortiication probability o exceedance. However, in most cases, either the hazard-consistent magnitude and distance proposed by Ishikawa and Kameda, or the modal earthquake proposed by Chapman, the abovementioned presupposed requirement couldn t be satisied. Actually, it is always only a ew potential seismic sources deined by seismic active aults having contribution to the design peak acceleration with ortiication probabilistic level or large dams as required in the Speciications. Among them, to cause the design EPA at dam site, the earthquake located nearest to dam site along the major ault generally has a

5 maximum contribution to the hazard curve, as usually a uniorm space distribution within the source is assumed. Fig.3 shows the seismic resources contributing the design EPA at Xiaowan dam site. A scenario earthquake o magnitude m =6.3 and distance =3km orm source No.3 with maximum probability was determined to generate the site-speciic design response spectrum (with damping ratio o.5) or Xiaowan arch dam as shown in Fig β 2.5 M=6.3 Δ =3.km h=4.8km.5 Fig.3 Seismic sources contributing design EPA at Xiaowan dam site T (second) Fig.4 Site-speciic design spectrurn consistent with EPA or Xiaowan arch dam The annual probability o exceedance or the design acceleration response spectrum can be calculated according ollowing ormula: P( m, A ) = v n i= A i l, mi l, mi e β ( m m ) e β ( mi m ) th Where v is the recurrent rate o the i source with scenario earthquake; A i, A are the source areas associated with the l th magnitude interval m i and that o m within the l th seismic source respectively; l, m i is the spatial distribution unction o l th th source or m j magnitude interval; m is the lower limit magnitude; β is the coeicient in the Gutenberg-Richter ormula o seismicity. Assuming both the design peak acceleration and the normalized response spectrum are statistically independent random variables, the probability o the acceleration response spectrum should be the product o the probabilities o those two variables. Obviously, the probability o acceleration response spectrum provided by ormula (2) has to be smaller than that o the peak acceleration, but never be equal-hazard. (2) DEVELOPMENT FOR DESIGN ACCELEROGRAMS As in the seismic design o high arch dam like Xiaowan the nonlinear response analyses are required. So, development o the design accelerograms is necessary. A more reliable and eicient approach to generate the design artiicial accelerogram compatible with the design response

6 spectrum through variation arithmetic in time domain was developed or seismic design o Xiaowan arch dam [5]. In this approach the variation o the artiicial accelerogram δ a can be expressed as ollows, while taking the response o single degree o reedom system with requency ω l and damping ratio ξ to unit impulse h ( t τ ) as interpolation unction with coeicient j b M N ( t) = b h ( t t) j= l= δ a (3) Where t is the time with maximum response, M, N are the numbers o damping ratio ξ j and spectral requency ω l respectively. Then the integration equations related the variations o artiicial accelerogram δ a and response spectrum δ Sik can be translated into a set o algebraic equations or solving b as ollows: M N j= l= b tik h ( t τ ) h ( t τ ) dτ = δs (4) ik ik In order to improve the integration calculation with higher eiciency and less computer memory a segment ally matching technique was used. Finally ater solving b or δ a ( t) the artiicial accelerogram can be easy acquired as a( t) = a ( t) + δa ( t). The above-mentioned procedures were repeated to satisy convergence precision. So, high convergence precision and high eiciency o the proposed method where ully proved by practice. The synthetic normalized accelerogram and spectrum match were plotted in Fig.5 and Fig.6, respectively. ik 3 Acceleration(g) time( sec. ) Fig.5 Synthetic normalized accelerogram β Object Spectrum Calculation Spectrum T( s) Fig.6 Spectrum match CONCLUSION The input ground motion or Xiaowan arch dam o 292m high located in the area with high seismicity was carried out on the basis o a comprehensive study.. Instead o the traditional PGA the EPA was selected as more reasonable major input parameter or hazard evaluation o dam site with consideration o the spatial distribution unction l, m i or dierent magnitude interval o each source within a seismic belt.

7 2. Not the prevalent dual perormance criteria, but a unique ortiication criterion with a probability o exceedance o 2% in years was accepted. 3. The requently used equal-hazard spectrum was replaced by a more actual site-speciic design spectrum rom hazard-consistent scenario earthquake along the major ault with maximum probability. 4. A more eicient approach with high convergence precision or generating design artiicial accelerogram compatible with design response spectrum by using time domain method but not the conventional requency domain method was proposed. REFERENCES. N.A.Abrahamson, W.J.Silva, Empirical Response Spectral Attenuation Relations or Shallow Crustal Earthquake, Seismological Research Letters, Vol.68, No., R. K. McGuire, Probabilistic Seismic Hazard Analysis and Design Earthquake: Closing the Loop, Bull. Seism. Soc. Am., Vol.85, No.5,995, Y. Ishikawa, H. Kameda, Probability Based Determination o Speciic Scenario Earthquake, th Proc. 4 International Conerence on Seismic Zonation, Vol., M. C. Chapman, A Probabilistic Approach to Ground-Motion Selection or Engineering Design, Bull. Seism. Soc. Am., Vol.85, No.3, Zhang Boyan, Chen Houqun, Hu Xiao, Zhu Liwu, Generation o Artiicial Earthquake Time Histories compatible with Object Response Spectra by Using Time Domain Method, Earthquake Engineering and Engineering Vibration, Vol.8, No. 998,47-55.