Earthquake Protection Systems_11 docx

Theymay also take account of variations in ground conditions as explained below.They are based on probabilistic mathematical models of the earthquake sourcemechanism, and the earthquake

Zone C The Calabrian Arc

Zone B The Apennines Zone A The Northem Apennines

Zone D Background Activity

Figure 7.4 Division of earthquake activity in southern Italy into seismic source zones

within the Seismic Source Zone of greater than

or equal to

Zone A Northern Apennines

Zone B Central Apennines

Zone C The Calabrian Arc

Zone D Background Activity

give infinitely large magnitude values, which is unrealistic For each earthquakeregion, there is in effect a limit on the maximum size of earthquakes whichcould occur, deriving from the geological nature of the faulting To deal withthis, various modifications of the Gutenberg–Richter formula have been pro-posed, such as the use of a curved or truncated linear relationship Figure 7.3compares the Gutenberg–Richter formula with an alternative formula14 for thedata for earthquakes in southern Italy, showing that the curved relationship with

a definite upper bound is much more useful for predicting the recurrence ofearthquakes of magnitude greater than 7.0

Time Sequence Analysis

These analyses generate ‘expected’ return periods for events, i.e the average rates

of occurrence of earthquake activity It is clear from most earthquake cataloguesthat earthquake activity does not occur uniformly in time – it is sporadic andunevenly spread over the years An administration responsible for a region or

an organisation working across an area may well be concerned to estimate thenumbers of earthquakes of different sizes likely to occur within that region inany given period of time An example of a typical time sequence of earthquakes

of different magnitudes occurring across a region is given in Figure 7.6.The number of earthquakes occurring within a given time period, e.g

10 years, can be derived from the data for successive time intervals (1900–1910,1901–1911, etc.) as presented in Table 7.1 Analyses like this give the range ofobserved behaviour in the past, indicate confidence limits for any prediction ofaverage activity rates and identify any obvious patterns in the seismicity, such

as cycles of quiescence and activity

Rate of Strain Energy Released

A further alternative way of presenting the recurrence of earthquakes in a region,which derives more directly from an understanding of plate tectonics, is as a plot

of cumulative strain energy released with time If the plate boundaries slip at aconstant rate, it would be expected that energy would be stored in the rocks at

a constant rate, and that over a long period of time, energy would be released

at a rate which, over a period of time, would be roughly constant This type ofanalysis can be a means of indicating at any time whether there is a significantamount of stored energy, and also the size of earthquake which would occur if

it was all released For areas where the history of energy released on a fault isknown, this type of information can be used in the estimation of present-dayhazard, and in the compilation of seismic hazard maps.15

14Based on Gumbel’s extreme value analysis, as proposed by Burton et al (1984).

15Frankel et al (2000).

Malazgirt Earthquake 6,000 killed

Panisler Earthquake 310 killed

Polomor Earthquake 91 killed

Varto Earthquake 2,500 killed

Lice Earthquake 2,400 killed

Table 7.1 Time sequence analysis of earthquake occurrence in a 10-year period (earthquakes magnitudeMs
6.0, recorded in eastern Turkey, 37◦– 41.5◦ N, 38◦– 45◦E,

Probability of N

or more earthquakes

in a 10-year period (%)

The number of earthquakes likely in any particular period for a known average activity rate can be estimated mathematically, assuming a Poisson distribution, see Section 9.9.

Ground motion attenuation relationships give estimates of various parameters ofground motion, as a function of the magnitude and depth of the event and distancefrom the site to the epicentre (or fault rupture), with a known uncertainty Theymay also take account of variations in ground conditions as explained below.They are based on probabilistic mathematical models of the earthquake sourcemechanism, and the earthquake wave transmission process, calibrated by theactual available data from strong motion recordings

Where substantial data is available, e.g in western North America, there is now

a rather good agreement between the various published relationships, examples

of which are shown in Figure 7.7.16 For other areas of lower seismicity (ordata availability) there is greater uncertainty, but it is clear that the attenuationrelationships for parameters of spectral response are very different in differentareas, and that a distinction needs to be made between regions of higher and lowerseismicity Ground motion tends to attenuate faster in areas of high seismicitythan in areas of low seismicity Attenuation relationships suitable for areas subject

to intra-plate earthquakes have also been developed.17

16Atkinson and Boore (1990), Boore et al (1997).

17Dahle et al (1990), Toro et al (1997).

Figure 7.7 The range of published average attenuation relationships for acceleration with distance from an earthquake of magnitude 6.5 in western North America (after Atkinson and Boore 1990)

Figure 7.8 Average EMS intensity attenuation relationships from analysis of isoseismals

of 53 earthquakes, southern Italy, 1900 to present (after Coburn et al 1988)

In areas of lower seismicity where ground motion data is limited, it may bepossible to derive attenuation relationships for macroseismic intensity based onhistorical records Ambraseys18 has derived intensity–attenuation relationshipsfor the low-seismicity north west European area, and also derived appropriatemagnitude–intensity relationships which can predict magnitude from the use ofone or more isoseismal radii Grandori19has given intensity–attenuation relation-ships from Italian earthquakes in terms of the epicentral intensity (I0) Figure 7.8shows intensity–attenuation relationships for southern Italy derived from theanalysis of isoseismals of past earthquakes in the region.20

Using the recurrence relationships and other data relevant to earthquake rence, and the appropriate attenuation relationship for the relevant ground motionparameter, the hazard at any site can be determined This now involves aggregat-ing the effects at that site of earthquakes originating in each relevant source zone

occur-at each of a series of increments of distance from the site, up to the maximumdistance at which the largest possible earthquake can have any significant effect.Appropriate and widely used algorithms for this are available21 and computerprogrammes incorporating these algorithms have been published.22 Since there

is often uncertainty about which of several alternative earthquake occurrencemodels and attenuation relationships is appropriate, hazard maps are often syn-thesised by blending the data from different sources, using weightings for eachsource which are based on expert scientific judgement

The US national seismic hazard maps produced by the US Geological Survey23are amongst the most advanced maps produced to date Separate maps show peakhorizontal ground acceleration and spectral response at 0.2 and 1.0 second periodswith 10%, 5% and 2% probabilities of exceedance in 50 years, correspondingapproximately to recurrence times of 500, 1000 and 2500 years The referencesite conditions for the maps is firm rock with an average shear wave velocity of

760 m/s in the top 30 m

18 Ambraseys (1985).

19Grandori et al (1988).

20Derived as a part of the analysis of site hazard in Campania, Italy (Coburn et al 1988).

21 The algorithm described by Cornell (1968) is commonly used.

22 For example, that of McGuire (1978).

23Frankel et al (2000).

The maps are based on the combination of three components of the seismichazard:

(1) spatially smoothed historical seismicity, assuming that future damaging quakes will occur near areas that have experienced such earthquakes inthe past;

earth-(2) large background source zones based on geological criteria with maximummagnitudes of 6.5 to 7.5 for areas with little historical seismicity; and(3) the hazard from 450 specific fault sources on which geological slip rates(observed or estimated from palaeoseismic data) were used to determineearthquake recurrence rates

Hazard curves were calculated at a site spacing of 0.1◦ for the western UnitedStates and 0.2◦ for the central and eastern United States, a total of 150 000 sites.Several separate attenuation relationships were used and the results combinedwith equal weightings Disaggregation plots for major cities (New York, Chicago,Los Angeles and Seattle) have also been produced to show what proportion

of the total hazard at that location derives from different bands of magnitudeand distance.24 The maps of spectral acceleration at periods of 0.2 seconds and1.0 second with 10% exceedance probability in 50 years are the basis of the maps

of maximum credible earthquake (MCE)25 used in the new 2000 InternationalBuilding Code (Figure 7.9).26

GSHAP was one of the major international achievements of the InternationalDecade for Natural Disaster Reduction (1990–2000) It aimed to produce region-ally coordinated and homogeneous seismic hazard evaluations and regionallyharmonised seismic hazard maps One key output was the world seismic haz-ard map of peak horizontal ground acceleration shown in Plate I.27 This wasproduced by the integration of separate regional maps produced by 10 separategroups, each a collaboration between the major seismological groups active inthe areas

To some extent methods adopted and outputs produced varied from region

to region In Region 3 for example, which covers the 29 countries of centralnorth and north west Europe, the work had as an additional goal the production

of consistent maps to support the seismic zonation needed for application of

24Frankel et al (2000).

25Leyendecker et al (2000).

26 ICBO (2000).

27 GSHAP (1999).

150 100 80 125 150

2 1

2

175 210

207 150

192 175

205

150192

160 150

150 1 210

1 175

166

160

205

+ 38

+ 39

+ 84 125

175

100 100

49

175

150

+ 113

+ 144

2021

150 175

175 150

90 80

100

+ 109

6 5

100

+ 106 +

90 +

115

+ 60

+ 80 70

9

90

100 80

80 90

60

100 540

35

28 + 24 25 30

60 90 + 117

coun-• Definition of a single set of seismic source zones – in all 196 separate sourcezones were distinguished – and estimating characteristic focal depths, upperbound magnitudes and magnitude–recurrence relationships for each zone

• Defining appropriate ground motion attenuation relationships to adopt andweighting coefficients to use where several separate attenuation relationshipswere relevant

• Performing hazard calculations for a grid size of 0.1◦latitude by 0.1◦longitude(except in northern Europe), a total of 59 217 separate points

28Gr¨unthal et al (1999).

The resulting regional map of horizontal peak ground acceleration with an dance probability of 10% in 50 years is shown in Plate II The information shown

excee-on this map can be used directly in design to define a spectral respexcee-onse curve,and will also inform the national maps produced in the National ApplicationDocuments which accompany EC8.29

For the designers or owners of individual buildings, or for urban planners orcity authorities, the issue is how likely a specific site is to experience earthquakeforces of a certain severity Building design codes adopt one of two alternativeprocedures for specifying the geographical distribution of design loads:

(1) seismic zonation or

(2) contour mapping of expected ground motion

Most national codes of practice use the seismic zonation concept The country(or region) covered by the code is divided into a small number (usually no morethan four or five) of separate source zones, within each of which the lateralloading requirement for earthquake-resistant design is constant, and is specified

by a zone coefficient The zone coefficient relates to the expected peak groundacceleration within a predefined return period, but this information does not need

to be known by the designer The Turkish seismic zonation map (Figure 7.10) is

a typical example In this code the zone coefficients are 0.1, 0.06, 0.04 and 0.2for Zones 1, 2, 3 and 4 respectively, corresponding roughly to the peak groundacceleration (as a proportion of the gravitational acceleration g) with a 10%

probability of exceedance in 50 years These coefficients are converted into aresponse spectrum for design using further coefficients for local soil type andbuilding importance

The advantage of this method for specifying design loads is its simplicity fordesigners The zones, although defined from knowledge of regional seismicity,are not given a formal definition in terms of expected ground motion Theirsignificance derives from the use of the zone coefficient in the formulae in theaccompanying code, so they have a semi-legal character, like district boundaries.However, the approach also has disadvantages One disadvantage is that theseismic zonation is coarse, and is unable to take into account the effects of localfeatures such as fault zones Another is that only a single parameter is defined,whereas it is now accepted that at least two independently varying parameters areneeded to take adequate account of the variations in regional seismicity.30 Thesetwo disadvantages are overcome through the use of contour maps such as those

29 CEN (1994), Lubkowski and Duian (2001).

30Leyendecker et al (2000).

accompanying the 2000 International Building Code.31 The code specifies that

the design loading should be that associated with the maximum credible

earth-quake (MCE) at the site Contour maps of the entire United States indicate the

values of two key design parameters to be used to construct the design groundmotion response spectrum at that site: the spectral acceleration values at 0.2 s and1.0 s periods The value of these parameters is derived from the US GeologicalSurvey’s hazard maps which are contour maps of the 0.2 and 1.0 s spectral accel-erations with a 10% probability of exceedance in 50 years, but with modificationsfor some parts of the United States to take account of the effects of known localfaulting on design loads, and with variations for different classes of site defined

by soil conditions.32 Figure 7.9 shows, for example, the MCE ground motionmap of a small part of Western United States for the 0.2 s horizontal spectralacceleration (% ofg), for Site Class B Site effects are discussed in Section 7.4.

Maps such as these represent a considerable step forward in defining appropriatedesign load coefficients and are likely to become the standard approach for futurecodes in other countries

Whichever approach to hazard estimation is used, the influence of site conditionsneeds to be taken into account It has been shown that amplification of peakground acceleration (PGA) by a factor of 5 or more is possible in a particularlyunfavourable site,33 while studies of macroseismic intensity34 indicate that anintensity increment up to three steps on the EMS scale is possible owing toground conditions

The site conditions giving rise to ground motion amplification have alreadybeen discussed in Section 7.2 Variation in subsoil surface geology is the principalcause of variation, though site topography can be a significant factor as well.Ideally, the effect of these factors on the site hazard will have been determined

by a detailed microzoning study as discussed in Section 7.5 Methods based onrecords of micro-tremors can be used to determine relative amplification factorsover an area for comparison with a particular reference site

The type of subsoil condition also affects the shape of the site responsespectrum – on soft sites low ground motion frequencies are often amplified andhigh frequencies filtered out, for instance Thus different amplification factorsmay need to be defined for different frequency ranges The influence of soil

31 ICBO (2000).

32Leyendecker et al (2000).

33Such as Mexico City, see Singh et al (1988).

34 For example, in the USSR, see Medvedev (1965).

Figure 7.11 Influence of soil conditions on average acceleration spectra experienced at

a site (after Seed et al 1974)

conditions on average acceleration spectra is illustrated by Figure 7.11, which isbased on the shape of the spectra for 104 records.35

The effects of ground conditions have been incorporated into some publishedattenuation relationships.36 These attenuation relationships divide ground con-ditions into mainly three types, namely rock, shallow soil and deep soil Theinfluence of ground conditions on peak ground velocity and displacement appears

to be stronger than on peak ground acceleration, and the influence is much greaterfor spectral values at low frequencies than at high frequencies However, verylittle of the data on which such relationships depend is derived from groundmotions strong enough to cause building damage

Microzoning is a developing technique which promises in the future to bring veryimportant benefits for earthquake protection Its aim is to identify and map thevariation in earthquake hazards within a limited area – typically a city or munic-ipality – as a result of variation in ground conditions or other characteristics.Microzoning maps can be used in conjunction with larger-scale hazard map-ping to inform urban land-use planning and decide on allocation of resources forstrengthening existing buildings

35Analysed by Seed et al (1974).

36 Such as those of Joyner and Boore (1981), Campbell (1981) and Toro and McGuire (1987).

The local site conditions which may have an influence on the suitability of thesite for a building or settlement are discussed in Section 7.1 They include:

• soil conditions which can amplify ground motions, either generally or tively in certain frequency ranges;

selec-• susceptibility to liquefaction and other types of ground instability;

• topographical variations which can cause ground motion amplifications;

• sites which could experience large permanent ground deformations such asthose associated with surface or shallow faulting;

• low-lying coastal sites vulnerable to tsunamis

Other characteristics of an urban area which affect the expected consequences of

an earthquake are related to the buildings and their occupants; such characteristics

as highly vulnerable building types, fire potential and social deprivation can beusefully mapped

An ideal microzoning would consist of the determination of relevant groundmotion parameters with specified probability of exceedance for given return peri-ods for all points in the study area, taking account of local effects, and presented

as contour maps In practice, simplified methods are needed to make ing feasible One method is based on micro-tremor survey, another is based oncalculating the response of the site based on subsoil survey data.37

microzon-Micro-tremor Methods

The first type of approach uses instrumental records of micro-tremors and otherlow-amplitude ground motions to determine and plot ground motion amplifica-tion factors across the zone The method38 makes a valuable contribution to anunderstanding of the parts of the zone that can expect the greatest ground motion

in future earthquakes, and its validity has been demonstrated by the close tion found between these areas of high amplification and areas of extreme damage

correla-in several earthquakes.39Figure 7.12 shows a microzoning map of Mexico Cityderived in this way The method can also be applied making use of aftershockmeasurements following a major earthquake However, the method has been crit-icised on the grounds that the characteristics of low-amplitude vibration may be

in many cases quite different from those of damaging ground motion, in wayswhich it would be impossible to assess in advance of an earthquake

37 Vaciago (1989).

38One version is referred to as the Nakamura method (Mucciarelli et al 1996).

39Singh et al (1988).