Behavior of Structures Built on Active Faults Zones

5/25/2018 Behavior of Structures Built on Active Faults Zones

1/26

Seismic Fault-induced Failures, 1-26, 2001 January

1

THE BEHAVIOUR OF STRUCTURESBUILT ON ACTIVE FAULT ZONES:

EXAMPLES FROM THE RECENTEARTHQUAKES OF TURKEY

R. Ulusay1, . Aydan2and M. Hamada3

1Dr. of Eng., Professor, Dept. of Geological Eng., Hacettepe University(06532 Beytepe, Ankara, Turkey, [email protected])

2Dr. of Eng., Professor, Dept. of Marine Civil Eng, Tokai University(Orido, 3-20-1, Shimizu, Shizuoka-Ken, 424-8610, Japan, [email protected])

3

Dr. of Eng., Professor, Dept. of Civil Eng, Waseda University(3-4-1 Okubo Shinjuku-ku, Tokyo, 169-8555, Japan, [email protected])

The 1999 Kocaeli and Dzce earthquakes of Turkey caused loss of many people and severe structural damages tostructures. In addition to poor quality construction and inappropriate construction materials, the damage was caused

partly by the permanent displacement of the ground due to faulting and partly by the liquefaction and lateral spread ofthe ground. The aim of this paper is to make an attempt to describe the general features of the surface ruptures andrelated ground failures and damages to structures, and to discuss several important examples from both earthquakeswhich may be precursory for the earthquake engineering community to develop seismic codes for structures in activefault zones.

Key words:Active fault, earthquake, ground failure, lateral spreading, North Anatolian Fault, seismic code, submarine

landslide, subsidence, structure, surface rupture.

1. INTRODUCTION

The Kocaeli earthquake of August 17, 1999 andthe Dzce earthquake of November 12, 1999occurred in Turkey are the inland earthquakesinvolving surface rupturing of the ground for greatlengths. These two earthquakes causedground-surface failures across a broad area of thenorthwestern part of Turkey. These failures includedzones of ground fissures and extensional cracking,lateral displacements, settlements with verticaldisplacements, compressive deformations andsubmarine landslides. Furthermore, the fault breaksoccurred right beneath various structures. Groundcracks displaced the foundations of houses and

bridges, broke apart sidewalls and streets, deformedrailways, ruptured utility lines and destroyed some

buildings with in a belt. There is no doubt that thisphenomenon is of great concern to the earthquakeengineering community all over the world as it may

be observed in any country having active fault

zones.The seismic design of engineering structures isgenerally carried out by considering the possible

shaking characteristics of the ground duringearthquakes in the region. It is a fact that the

permanent relative displacement of the ground is notconsidered in any seismic code all over the world,except for very long linear structures such as

pipelines. This problem is currently considered to bebeyond the capability of seismic design concept forstructures in the earthquake engineering, although itmust be dealt with somewhat.

After earthquakes, some scientists from variousdisciplines say that it was told before not to buildthe structures in this area since a fault or faults

passed through it. This statement is alwaysencountered by the engineering community all overthe world. If this advice is just followed anystructure should not be built in seismically veryactive countries such as Turkey, Japan, Taiwan andthe Western part of USA while the societies of thesecountries demand better living environments.Therefore, the engineering community must find theways to be able deal with the designing and building

structures in active fault zones rather than justturning their heads other way. The examplesobserved in these two great earthquakes occurred in
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]


2/26

2

Turkey in 1999 taught the authors some importantlessons and also provided some direct answers tothose people who hesitate to suggest constructionsnear fault zones. In this paper, an attempt is made todescribe and discuss several important exampleswhich may be precursory for the earthquakeengineering community to develop seismic design

codes for structures in active fault zones togetherwith responsible scientists for better livingenvironment of their societies who have just to liveon the top of the seismically active belts of theearth.

2. DEFINITION OF FAULTS ANDFAULTING TYPES

The fault is geologically defined as a discontinuity

in geological medium along which a relativedisplacement took place. Faults are broadlyclassified into three main groups (Figure 1) as given

below.(1) Normal faults(2) Thrust faults, and(3) Strike-slip faults.

Figure 1: Diagram showing the main types of fault

Faults may range in length from a few micronsto many kilometers. A fault is geologically definedas active if a relative movement took place in a

period less than two millions years or if slip hasoccurred, variously in historical or Holocene orQuaternary time or earthquake foci are located. Onthe other hand, the service life of engineeringstructures could not generally be greater than100-150 years. Taking this fact into consideration,there were new attempts to define active faults for

engineering purposes by various US governmentagencies defined the active faults as given below(Yeats et al. 1997):

(1) It moved in the last 10000 years (Holocene),(2) It moved in the last 35000 years (in nuclear

power plant site selection)(3) It moved in the last 150000 years,(4) It moved twice in the last 500000 years.However, these diverse definitions, based on the

agencys perception of risk, also lead to confusion.

Since the definitions of active faults are still quitebroad and diverse, the engineering communityexpect more specific and quantitative definitionsconcerning active faults from the earth scientists inorder to be able to develop seismic design codes inactive fault zones.

It is well known that the earths crust is rupturedand contain numerous faults and various kinds ofdiscontinuities, and it is almost impossible to find a

piece of land without faults. During the constructionof structures such as tunnels, dams, power plants,

roadways, railways, power transmission lines,bridges, elevated expressways etc, in most cases, itis almost impossible not to cross a fault or faults.Therefore, one of the most important items is how toidentify which fault segments observed on groundsurface will move or rupture during an earthquake.It is well known that a fault zone may involvevarious kinds of fractures as illustrated in Figure 2and it is a zone having a finite volume (Aydan et al.1997). In other words, it is not a single plane.Furthermore, the faults may have a negative or

positive flower structure, which consists of more or

less symmetrical splays into sub-faults near theintersection of the main fault with the groundsurface, as a result of their trans-tensional ortrans-pressional nature and the reduction of verticalstress near the earth surface as shown in Figure 3.For example, even a fault having a narrow thicknessat depth may cause a quite broad rupture zones andnumerous fractures on the ground surface duringearthquakes. In addition to that, it is impossible tosay the same ground breaks will re-rupture during afuture earthquake in regions with thick alluvialdeposits. Furthermore, the movements of a fault

zone may be diluted if a thick alluvial deposit isfound on the top of the fault as it happened inErzincan and elsewhere in Turkey (Hamada andAydan 1992). The ground ruptures observed in theKocaeli earthquake of August 17, 1999 could not

probably correspond to those of 1719 in engineeringsense while it may broadly correspond to the samefault in a geological sense. Therefore, the size andlocation concepts in earth sciences must have thesame sense in structural design concerning the faultzone in quantitative terms for developing seismiccodes for structures built and/or to be built in activefault zones.


3/26

3

Figure 2: Fractures in a shear zone or fault

Figure 3: Negative and positive flower structuresdue to trans-tension or transpression faulting

3. THE SURFACE RUPTURES OF THE 1999KOCAELAND DZCE EARTHQUAKES

3.1. Kocaeli Earthquake and Surface FaultRupture

The August 17, 1999 Kocaeli earthquake(Ms=7.4) occurred at 03.02 on local time and struckthe eastern Marmara region. Records of themaximum ground acceleration were measured as0.45g. The earthquake lasted 45 seconds andconsisted of several subevents. The earthquake wasfelt over a large area, as far as in east of Ankara thatis about 300 km away. Official estimates place thedeath tolls about 18000 and injuries more than25000, and collapse or heavy damage of more than40000 buildings. Glck, zmit, Yalova andAdapazar are the cities which suffered severestructural damages during the earthquake. Theearthquake also caused considerable damage in the

suburbs of stanbul, approximately 70 km awayfrom the earthquakes epicentre. In addition to poorquality construction and inappropriate constructionmaterials, the damage was caused partly by the

permanent displacement of ground due to faulting,and partly by liquefaction and lateral spreading ofthe ground towards lakes, rivers and sea.

The earthquake originated at a depth of 15kilometers and occurred on the northern strand ofthe North Anatolian Fault Zone (NAFZ) with aright-lateral strike slip movement. The Kocaeliearthquake produced a surface rupture over adistance of 120 km between Glyaka in the east andHersek in the west as shown in Figure 4. BetweenGlck and Deirmendere the surface rupturedisappeared and extended towards west (Hersekdelta) as a near-shore strike-slip fault. Although theHersek segment of the NAFZ cuts the tip of a large

delta plain in the western end of the surface rupturezone, any surface rupture was not observed alongthe shoreline between Hersek delta and YalovaMilitary Airport at the west. However, Z shapedruptures extended in a limited area in the vicinity ofthe airport were evident (Figure 5). These were a setof en echelon open cracks striking E-W. Theseobservations indicated that these fractures might beassociated with the main fault, probably as asecondary faulting, although the relativedisplacements were very small.

From the surveys of offset fence lines, roads and

tree lines, the maximum offset along the surfacerupture was measured about 4.5-5 m (Figure 6).General trend of the surface rupture has anorientation of NE-SW in the east between Glyakaand Akyazand then it appears with a general trendof E-W in the western part of the earthquake region.Width of the primary surface rupture traces variedfrom 5 to 20 m with little or no secondarydeformation. However, broad zones of primary andsecondary deformation up to 200 m wide wereevident. The fault rupture produced typicalexamples of strike-slip offset, including well-formed

moletracks, left-stepping en-echelon fault traces invarious scales, uplift of pressure ridges andsubsidence (Figure 7). Furthermore, theobservations of striations on the existing faultrupture near Arifiye (Adapazar) and Tepetarla(between Adapazar and Kocaeli) imply that thevertical component of the fault motion has a normalfault sense with vertical offsets of 10-20 cm as seenin Figure 8 in these parts of the fault segment. Thismay be a further evidence of trans-tensional natureof the western part of the North Anatolian Fault.

In the western part of the earthquake-affectedregion, between Kavakl (Glck) and Hisareyn, anormal fault striking N45-60W/75-80NE with a


4/26

4

Figure 5: Z cracks observed near Yalova airport

(note the tension gashes - NE-SW indicated by anarrow - which are characteristics of strike-slipfaults)

significant vertical offset was observed. It produceda scarp of 3.5 km in total length and a verticalmovement or a drop up to 2 m and some lateralmovement (Figure 9). The fault initiates somewhereat the sea shore in Kavakland terminates near theD-80 highway in Hisareyn village without apparentconnection with any other earthquake break. This

fault caused rotation of a pylon of a power line nearthe Ford automobile factory construction site about

(a)

(b)

Figure 6: Typical views from right lateral offsetduring Kocaeli earthquake: (a) Western wall of the

Naval Command Center in Glck (slip: 4-5m) and

(b) a poplar tree separated by the surface rupture inTepetarla (slip: 30cm)

Figure 4: Rupture path of the fault during 1999 Kocaeli earthquake


5/26

5

(a)

S

N

(b)

(c)Figure 7: (a) The fault break near the railway nearTepetarla, (b) left-stepping en-echelon fractures inthe surface rupture and (c) moletrack betweenGlyaka and Akyaz in Aksudere valley (KocaeliEarthquake)

Figure 8: The striations on a fault surface nearArifiye

N

zmit Gulf

GLCK

Crack

Vertical offset1.5 m

Normalfault

ZMT GULF

Normal fault

Rupture of 1999 Kocaeli earthquake

Glck

+

-

N

+

+

_

_

0 2(km)

Figure 9: (a) Location and orientation of the normalfault at the east of Glck, and (b) a part ofsubmerged land in Glck town (rearranged fromIshihara et al. 2000)

10 degrees and global subsidence of the basin withlocal submergence of the coastline and inundation ofthe northeastern part of Glck town (Figure 10).Several models are suggested for explaining thecause of the normal faulting. These models may be

broadly divided into the pull-apart model and the

NNNN

SSSS


6/26

6

fracture zone model as illustrated in Figure 11. Tovalidate the pull-apart model one need to find onemore lateral strike-slip fault segment at the Hisareynterminus of the normal fault. Since such a faultcould not be found during both aerial and landsurveying of this normal fault, this model must bedeclined.

(a)

(b)Figure 10: (a) The fault with a normal component atthe south of Ford automobile factory constructionsite, and (b) a view from the same fault passing near

the Glck football stadium and submerged land.

Surface rupture of the 1999 Kocaeli earthquake,which traversed both rural and urban areas,generally occurred across the floors or edges ofalluvial basins, and locally followed the prominentmountain range front to the south of the rupturezone. The direction of fault scarps coincides almostwith that from the fault plane solutions. The faultseems to follow the centre of the depression of theregion rather than the fault segments depicted on the

previous geological maps, due to floweringphenomenon.

Figure 11: Models for explaining the cause ofnormal faulting between Kavakland Hisareyn

3.2. Dzce Earthquake and Surface Fault

RuptureAn earthquake of magnitude 7.2 on the Richterscale occurred at 18.45 on local time on November12, 1999 in Bolu province of Turkey, which wassubjected to big earthquakes in the past, threemonths after the devastating Kocaeli earthquake.The epicentre of this earthquake was about 110 kmeast of the Kocaeli earthquake (see Figure 4) and itoccurred on Dzce fault. Official estimates placedthe death tolls more than 850 and injuries about4950. This earthquake caused severe structuraldamages in Dzce and Bolu cities, and Kaynal

town. It is estimated that about 1350 buildingscollapsed and 7000 residential and industrial

buildings suffered heavy damage. Records of themaximum ground accelerations were 0.513g and0.805g at Dzce and Bolu, respectively. On thecontrary to the Kocaeli earthquake, liquefaction wasnot widely distributed throughout the Dzceearthquake region. Liquefaction fissures and sand

boils were limited on or near the surface faultrupture.

The North Anatolian Fault between Erzincan inthe far east and Bolu splays into two strandswestwards Bolu (Figure 12). The southern strandgoes into the Mudurnu valley and the

SubmergedSubmergedSubmergedSubmergedareaareaareaarea


7/26

7

Dzce-Hendek fault in the north surrounding acrustal block called Almack block. The Dzceearthquake originated at a depth of 14 kilometersand caused a right-lateral strike slip movement onthe Dzce fault with a slight normal faulting

component. The surface rupture was 40 km long.There is an approximately 5 km step over at theeastern tip of the surface rupture resulted from theKocaeli earthquake near Efteni Lake. The surfacerupture, with a general trend of E-W, appeared at thesouth of Glyaka town and terminated at a location

about 2 km far from the west portal of the BoluTunnels (Figure 13).

The maximum lateral offset is about 5 m. As seenfrom Figure 13, the amount of right-lateral offsets,clearly observed on roads, fences and tree lines

(Figure 14), increase in the central part of therupture zone. The average lateral offset is about 3 m.Vertical offsets observed at some locations in thesouthern block are between 0.1 to 1.5 m andindicate that although the earthquake fault exhibitsthe characteristics of a strike-slip fault, it also has a

Bolu tunnels

stanbul - Ankara Highway

ALMACIK BLOCK

ElmalkFault

North

Anatoli

anFau

lt1944

1967

1957

Kaynal

Efteni lakeDzce fault

Bolu

Motorway Alignment

Surface ruptureof the Kocaeli earthquake

0 15(km)

Abant lake

igure 12: Active faults and previous earthquakes between Bolu and Dzce, and the position of theAnkara-stanbul motorway project

DZCE

Glyaka

HacYakup

Glorman

Aydnpnar

. brahim

narl Kutlu

Karaal

BeykyOvapnar

Fndkl Kaynal

Daryeri Hasanbey

Bolu M.TEM

TEM

E-5

E-5

E-5

Elmalk

imir

Lateral offsetin meters

Maximum lateral offset

Bakacak

Surface rupture(November 12, 1999)

0 2 4(km)

N

Motorwayunderconstruction(TEM)

Highway

Viaducts

EfteniLake

Figure 13: Rupture path of the fault during Dzce earthquake and amount of lateral offsets measured atcertain locations


8/26

8

normal component (Figure 15a). The striations withinclinations between 250 and 300 on the surfacerupture confirm the presence of normal component(Figure 15b). Towards west, the surface rupturefollows the southern shore of Efteni Lake. Thenearly vertical displacements reaching up to 3 mobserved at this locality are probably due to sliding

of loose and weathered material forming the steepslopes rather than by faulting.

Figure 14: Typical right lateral offsets from thesurface fault rupture resulted from the 1999 Dzceearthquake

The width of the fault zone varied from 1 m to 50m. Faulting seems to occur in the form of a series of

planes parallel to each other with a general trend ofE-W. Due to left stepping observed at somelocations, pressure ridges were evident (Figure 16).In this zone, in addition to Riedel fractures crossingthe main rupture surface with angles between 250and 600, en-echelon fractures were also welldeveloped. The direction of fault scarps coincidesalmost with those obtained from the fault planesolutions. The surface rupture seems not to followthe depicted fault segment that passes through the

boundary between the alluvial sequence and rock

units at the southern edge of Dzce Plain. It shouldbe remembered that such traces of rupture observedalong the south of the Dzce plain are just the

products of flowering phenomenon during themotion of strike-slip faults as a result of reduction ofvertical stress near the ground surface.

Figure 15: The striations with an inclination of 28from the horizontal on the surface of fault rupture(Kutluky)

Figure 16: A typical view from the surface ruptureand pressure ridges at the eastern part of narl

village (view from east to west)

4. EFFECTS OF FAULT BREAKS ONSTRUCTURES

When large faults produce earthquakes, theygenerally deform the ground surface. Primarysurface faulting, such as 120 km and 40 km longsurface ruptures associated with the 1999 Kocaeliand Dzce earthquakes, respectively, is the direct

effect of movement on an earthquake producingfault.Although the mechanisms that cause ground


9/26

9

failure near fault zones can vary, faults appear to beimportant in localizing ground failure. This sectionis included to demonstrate what fault breaks havedirect impacts on the structures which may be somehelp to earthquake engineering community todevelop seismic design codes for structures in activefault zones together with responsible scientists for

better living environment of their societies whohave just to live on the top of the seismically active

belts of the earth. For the purpose, several importantexamples from both devastating earthquakesoccurred in Turkey in 1999 (Aydan et. al. 1999 and2000) are selected and discussed.

4.1. RoadwaysRoadways are line-like structures and they may

be built directly on the existing ground surface,elevated through viaducts and/or embankments or a

mixture of the both. The fault zones in many parts ofTurkey have generally alluvial deposits on the rockbasement. The D-100 and D-130 highways passingthrough the earthquake affected region in the 1999Kocaeli earthquake were directly built on existingground surface. The D-130 highway was damagednear Baiskele region at the southeast corner of thezmit Gulf where the fault has a relative horizontaldisplacement of 230 cm without any remarkablevertical component (Figure 17a). The damaged partof the roadway was re-surfaced on the same day ofthe earthquake and the roadway was in service in

such a short period of time. Damage to the D-100highway took place at several points near theKocaeli-Sakarya provincial boundary as a result ofsecondary normal faulting associated with the mainlateral strike-slip event (Figure 17b). The largestvertical displacement was about 15-20 cm, theroadway was temporarily re-surfaced and it was inservice within a few hours after the earthquake. Inthis location, some rock slope failures were alsoobserved and one car was hit by fallen debris. Thehighway was in service at this section by re-placingthe fallen debris in a short period of time. The other

sections of the highways were non-damaged in-spiteof severe ground shaking even they were in alluvialareas.

The Trans-European Motorway (TEM) wasdamaged at three different locations by theearthquake fault. The motorway with east and west

bounds having 3 lanes each is slightly elevatedthrough embankments in the earthquake affectedregion. The surface of the motorway was damaged

by rupturing and buckling as seen in Figure 17c.Furthermore, the settlement of roadwayembankments was observed along the 30 km sectionof the motorway between Arifiye and Kseky. Thedamaged surfacing of the motorway was replaced

within three days and was in service again. There-surfacing of the motorway was completed in18days after the earthquake.

(a)

(b)

(c)Figure 17: (a) Damaged and resurfaced D-130highway near Baiskele region at the SE corner ofthe zmit Gulf, (b) damaged D-100 highway, and (c)

buckling of roadway surfacing on the motorway

between Adapazarand Kocaeli


10/26

10

The E-5 (D-100) highway and TEM, connectingAnkara to stanbul, also pass through the regionaffected by the Dzce earthquake. The fault breakobserved in this earthquake crossed the E-5 highwayin Kaynal town between Bolu and Dzce cities(Figure 18a). In addition to these, several countryroads were crossed by the fault break. The surface

of the roadway directly built on existing groundsurface was buckled since the maximumcompression force induced by the ground movementcoincided with the alignment of the highway similarto that of the TEM near Arifiye during the 1999Kocaeli earthquake. The highway was quicklyre-surfaced in a short period of time and wasoperational. A country roadway, shown in Figure18b, was crossed by the fault break and easilylevelled with the dirt available on the site.

(a)

(b)Figure 18: (a) Damaged and repaired location of theE-5 highway in Kaynal and (b) a displaced localroad at Kaynal

From these examples it is clear that some damageto roadways could not be prevented where the

relative movement of the ground occur due tofaulting and non-elevation of roadways in faultzones is a key element to deal with the negative

effects of faulting on roadways. There is a veryimportant engineering decision, that is, thehighways should be directly built over the existingground surface in active fault zones.

4.2. RailwaysRailways are also line-like structures and they

may be built directly on the existing ground surface,elevated through viaducts and/or embankments or amixture of the both. The railways connect stanbulto both Ankara and Adapazar and they have

junctions near Tepetarla and Arifiye. Both electricaland diesel powered passenger and freight trainsoperate on the railways. The railways were also builton the existing ground surface. During the 1999Kocaeli earthquake, the railways were buckled nearTepetarla station where the earthquake fault crossedthe railways at an angle of 50-550with well known

S shape (Figure 19). The fault was EW directionand crossed the rail with an angle of 600. The rightlateral offset was 2.5 m near this station. Thisdirection is almost parallel to the maximumcompression of the ground due to faulting and anapproximately 200 m section of the railways wasdamaged. The ground was also heavily deformed byfaulting for a length of 700 m resulted near theKurtky (close to Arifiye). The railways, however,were operational within three days after theearthquake by replacing buckled rails and damagedtraverses, and the railways were completely repaired

together with the correction of displaced alignmentin 20 days after the earthquake, so that trains beganto travel at their original speed. Particularly thedamage near Kurtky took more time to level theground to the pre-earthquake level.

Figure 19: Buckled railway near Tepetarla station

Although the damage to the railways could notalso be prevented, even they are directly built on theexisting ground surface where the railways must

pass through an active fault zone, the recovery of

railways is very quick.


11/26

11

4.3. Bridges and ViaductsBridges and viaducts are elevated structures. Since

roadways and railways were not elevated structures,no viaducts existed in the region of the fault breaksresulted from the Kocaeli earthquake. The bridgesof highways and railways over Sakarya River wereintact and no direct faulting event occurred at these

localities. However, the YSE bridge 5 km south ofAdapazarspanning over Sakarya River for a serviceroad collapsed as a result of relative motion due tofaulting between its abutments (Figure 20). Alongthe damaged section of the motorway, there wereseveral overpass bridges. Among them, a four spanoverpass bridge at Arifiye junction collapsed as aresult of faulting (Figure 21). The girders collapsedonto the motorway killing more than 10 people in a

bus who were just happened underneath theoverpass bridge at the time of collapse. The fault

rupture passed between the northern abutment andthe adjacent pier. The overpass was designed as asimply supported structure according to themodified AASHTO standards and girders hadelastometric bearings. However, the girders wereconnected to each other through pre-stressed cables.The angle between the motorway and the strike ofthe earthquake fault was approximately 150, whilethe angle between the axis of the overpass bridgeand the strike of the fault was 650. Themeasurements of the relative displacement in thevicinity of the fault ranges between 330 and 450 cm.

Therefore, an average value of 390 cm could beassumed for the relative displacement between thepier and the abutment of the bridge. The girders ofthe bridge rotated and pulled so that the fall of thegirders between the north abutment and the adjacent

pier could not be prevented. However, the collapseof the other girders of the remaining part of the

bridge might have been prevented if the girderswere not connected to each other through the

pre-stressed cables.

Figure 20: A view from a collapsed bridge on

Sakarya River at SE of Adapazar due to faulting(view from south to north)

(a)

(b)Figure 21: Views from the collapsed overpassbridge onto the motorway near Arifiye

In the region affected by the Dzce earthquake,no bridge was directly crossed by the fault break.Although the bridges were very close to the fault

break, there was no damage to those. A section ofthe Trans-European Motorway (TEM) between Boluand Dzce is under construction. The motorwayfollows the northern side of the Dzce basin from

Gmova to the Asarsuyu portal of the BoluTunnels. A 6 km long section of the motorwayselevated through viaducts with a maximum heightof 49 m and a width of 17.5 m. The maximum spanis 39 m and each viaduct has 12 piles with thediameter of 180 cm. The elevated section is dividedinto to large segments, namely East viaducts andWest viaducts (Figure 22a). Although the viaductsexhibited a well performance in the Kocaeliearthquake, the fault break passed through theelevated section of the TEM in Asarsuyu valley atan angle of 20o-30o and caused the rotation andtilting of some of piers during the Dzce earthquake


12/26

12

(Figure 22b, c). Particularly, the pier numbered 45for constructional purposes was rotated as much as12oand some permanent deformation took place. Asa result of that, the simply supported girders of theviaducts were displaced from their supports.Although no girder was fallen off from this elevatedsection of the motorway and piers performed very

well during the earthquake, some of the girder werevery close to fall.

(a)

(b)

(c)Figure 22: (a) A general view of the viaducts ofAnkara-stanbul motorway in Asarsuyu Valley, (b)the surface rupture of the Dzce earthquakeappearing below the viaducts of the motorway,and (c) a view from a rotated pier of the viaducts

The examples mentioned above indicated that it isnot desirable to elevate the roadways in fault zoneswhere permanent deformation of the ground maytake place during earthquakes. If they have to beelevated for some important reasons, such as floodsor landslides, the limitation of roadway inclination,

etc., as being in the case of viaducts in Asarsuyuvalley, it seems better that the girders should bedesigned as redundant structures, and simplysupported structures must be avoided as much as

possible. Furthermore, T-type piers should begenerally avoided as this type piers result intop-heavy situations.

4.4. DamsThe failure of dams has very severe impacts on

the structures as well as on the residential areas in

downstream. Since such failures are not allowablewhatsoever, the site selection of the dams is carriedout with utmost care in dam constructions. As aresult, it is very rare to see a dam failure caused byearthquakes due to active faulting.

The nearest dam was Yuvack dam in the areaaffected by the 1999 Kocaeli earthquake that is anearth-fill dam with a height of 40 m. Although thisdam was at a distance of 10 km to the earthquakeepicentre, no failure was observed. The rock-filldams are also known that they could accommodaterelatively large ground displacements while arch

concrete dams or gravity concrete dams couldrupture even at a relatively small amount ofdisplacements of the foundation rock.

There are several dams either earth-fill or rock-filltype around the epicentre of the Dzce earthquake.The distances of the nearest dams to the earthquakeepicentre are 12 and 21 km and they were built forthe purposes of irrigation and flood control,respectively. Although the authors did not visit thesedams, it was reported by the State Hydraulic Works(DS1999) that the dams were not damaged duringthe earthquake.

The conclusion for dams is that active fault zonesin engineering sense must be avoided as dam sites.If such sites could not be avoided for some reasons,dams should be designed as rock-fill dams so thatsome relative displacement could be accommodated.

4.5. TunnelsSimilar to roadways and railways, tunnels are

also line-like tubular structures. The pastexperience on the performance of tunnelsthrough active fault zones during earthquakes

indicates that the damage is restricted intocertain locations. During the 1999 Kocaeli


13/26

13

earthquake, portal failures of quite limited scalewere observed at the short tunnels of the TEMon the northern side of the zmit Gulf. As thefault breaks did not cross these tunnels, nodamage to tunnels was observed. In addition, arailway tunnel on the southern side of the

Sapanca Lake was not damaged even thoughthis tunnel is within the active fault zone in ageological sense.

The tunnels in the Dzce earthquake region areonly the Bolu Tunnels, which are presently underconstruction as a link of the Anatolian Motorway

between stanbul and Ankara (see Figure 12). Thetunnels have a cross-section of about 170-200square-meter and each tunnel being 16 m wide and11 m high has tree lines. The tunnel portals are atElmalk on the Bolu side and Asarsuyu Valley on theDzce side. The tunnels are about 3400 m long.Figure 23 shows the geological cross-section alongthe tunnel alignment. The rock mass on the Elmalksection is very weak and characterized as squeezingrocks (Aydan and Dalg 1998). Rocks in thissection of the tunnels were geologically subjected tothrust type faulting and folding which took place

before the North Anatolian Fault took place inMiocene. This led to the formation of low-anglefault zones, some of them up to 100 m wide. It isthese fault zones that have created the majordifficulties during tunnelling. During the excavation

of the tunnels, very large deformations of 1000 mmwere observed in phyllites on the Asarsuyu sectionand in siltstone, mudstone and clayey rocks on theElmalk section for about 800 m from the Elmalk

portal.In view of the orientation of the fault break, it is

considered that the fault break should not cross thetunnels although it passed them very closely anddisappears near the Asarsuyu portal in the west. Dueto limited investigation time, the authors could notvisit the Asarsuyu portals. However, it is expected

that the fault break should not appear in the tunnelsin view of the orientation of the fault breaksobserved near the portals in Asarsuyu Valley asshown in Figure 24a. Because of limited access tothe tunnels, only the Elmalk portal (Figure 24b)could be possible to visit. Therefore, it is not knownif there is any damage to the tunnels. Even if there

was any damage to the tunnels due to faulting, thedamage should probably be limited to locationswhere the fault breaks crossed the tunnels. Anothertype of damage to the tunnels may occur at locationswhere heavy squeezing problems were previouslyencountered. Since the loads acting on supportsystems at these locations are already high, theground shaking can impose additional dynamicloads on support systems when the overburden isrelatively shallow. As a result of that, if the supportsystem is insufficient in capacity, some roof and

side-wall collapses may take place.After a certain period of time, it was reported(Unterberger and Brandl 2000) that a collapseoccurred in an unfinished portion of the tunnel inthe fault gauge zone on the Elmalk side where themajor difficulties had previously been experienced.The collapse on the Elmalk side occurred about 50m beyond the end of the completed inner liningsection. Cracks and craters could be observed on thesurface (Figure 25a), although the overburden ismore than 50 m at this location. This section islocated at the transition from fairly good ground into

a major fault zone. Damage can be observed at theinner side drift shotcrete structure. The 30 cm thickshotcrete shell was fractured in the invert and theshoulders (Figure 25b). The causes for this

behaviour still have to be determined. It is alsonoted that accelerations of 0.6 to 0.8g weremeasured at the stations in the vicinity of the site,far in excess of the 0.4g design earthquake. Theseobservations have similarities with the preliminary

Figure 23: A longitudinal section of the Bolu Tunnels (after Dalg 1994)


14/26

14

(a)

(b)Figure 24: Views from the Asarsuyu (a) and

Elmalk (b) portals of the Bolu Tunnels after theDzce earthquake

assessments arisen from the authors about thepossible damage to the tunnel.

As understood from these examples, it is difficultto prevent the rupturing of tunnel linings at locationswhere the fault cause relative movements. If anychange in tunnel's alignment could not be possiblefor some reasons, the current seismic design codesfor tunnels, underground caverns and subways can

be used and seismicity of the region must bere-assessed if necessary, as being in the Dzceearthquake. Nevertheless, this type damage is quitelocalised and it does not cause any danger to theoverall stability of tunnels. Therefore, the currentseismic design concepts for tunnels can be safelyused and design should also be based on dynamicnumerical calculations.

4.6. Power Transmission Lines and TubularStructures

Power transmission lines generally consist of

pylons and power transmission cables. The design

(a)

(b)Figure 25: (a) Surface crater above the collapsed

tunnel on the Elmal

k side, and (b) damages to pilottunnels (after Unterberger and Brandl 2000)

of pylons and cables are generally based on thewind loads resulting from typhoons or hurricanes.Additionally, the possibility of slope failure of the

pylon foundations is considered in the design ofpylons. The cables do not fail during earthquakesunless the pylons are toppled due to faulting,shaking or slope failure.

During the 1999 Kocaeli earthquake, only one

pylon was damaged nearby Ford-Otosan automobilefactory construction site near Glck town. At thissite a normal fault, which is a secondary fault to themain lateral strike-slip faulting event, crossedthrough the foundations of the pylon and its verticalthrow was about 240 cm . One of the foundations ofthe pylon was pulled out of the ground and wasexposed as seen in Figure 10a. Some of its trusselements were slightly buckled. Nevertheless, thedamage to the pylon was quite limited and thisdamage could not hinder its function.

During the Dzce earthquake, the fault break

induced damage to a minor energy transmission lineoccurred near Cevizli village as shown in Figure 26a.


15/26

15

The roadway embankment failure induced by thefault break nearby Efteni Lake caused the steel

pylons sank to the lake and electricity cables weresubsequently cut off. A high voltage energytransmission line, which follows a route fromBakacak to Daryeri Hasanbey village, Kaynalandcontinues to Dzce, passed over the fault break in

Daryeri Hasanbey village (Figure 26b). Althoughnone of pylons straddled the fault break, the fault

break passed by the nearest pylon at a distance of 15m. Neither pylons nor cable were damaged, as longas the pylons were able to stand against earthquakeinduced shaking.

(a)

(b)Figure 26: (a) Damaged to a minor energytransmission line due to fault break inducedembankment failure of a roadway on the shore ofEfteni Lake and (b) views of energy transmissionlines at Kaynal(1999 Dzce earthquake)

Tubular structures may be specifically designatedas petrol and gas pipelines, water pipes and sewagesystems. They can be also classified as line-likestructures. These structures may fail either by

buckling or separation during a faulting event. Threesuch incidents were observed during the 1999Kocaeli earthquake (Figure 27). One of the incidentsinvolved the separation of a ductile iron pipe as aresult of faulting near the collapsed overpass bridge.The second incident took place at the pumpingfacility of the Seka paper-mill plant at Sapanca Lake.The third incident occurred near Tepetarla villagewhere the railways were buckled. Similar type offailures took place in the sewage pipe networks

whenever faulting breaks were observed. Thenatural gas pipelines crossing the zmit Gulf

between Yalova and Pendik were undamaged.

(a)

(b)Figure 27: Damaged pipes during the Kocaeliearthquake (a) Tepetarla and (b) Seka papermill

plant pump


16/26

16

No damage to the wells, pipes and pumps of thewater supply systems Bolu and Dzce cities duringthe Dzce earthquake was reported. Nevertheless,the water supply system was severely hit in Dzceand Kaynal. It is reported that the water supplysystem of Dzce city was heavily damaged as aresult of breaks at pipe joints due to ground shaking

and probably lateral spreading. On the other hand,the damage to the water supply system of Kaynaltown was quite heavy since the water pipes weredamaged by either buckling or breaking as a resultof ground shaking and faulting. The damage toasbestos or plastic pipes was quite heavy. Anexample of such a failure is shown in Figure 28a.Similar failures were also observed in water supplysystem of the villages along the fault break. Figure28b, c shows a buckled and split water pipe and aruptured small diameter plastic pipe in Fndkl

village. The sewage systems of Dzce and Kaynal

were damaged by ground shaking, lateral spreadingand faulting. Nevertheless the damage due tofaulting was quite localised at locations where thefault crossed the sewage pipes as compared withthat of ground shaking.

Ground failure, rather than ground shaking, is theprincipal cause of damage to tubular structures. Thedamage to tubular structures is generally difficult to

prevent during faulting. The brittle sewer pipes tendto fail under much lower strains than water lines, sodamage to sewer lines is considerable more

extensive. Such systems are checked systematicallythrough water flow states at manholes in cities,towns and villages. If there is any blockage, they arerepaired. However, some damage may undergounnoticed unless an indication of damage in theform of ground cave-in takes place or material orsmell leaks out of the ground. Nevertheless, usingthe flexible joints developed originally for pipesagainst the lateral spreading of liquefied ground,some of damage to pipelines, however, may be

prevented in such active fault zones.

4.7. BuildingsThe vast majority of losses to property and lives

caused by an earthquake involve buildings and thepeople inside. Particularly the buildings located onthe fault zones have more risks to damages induced

by fault ruptures. Many buildings along theearthquake faults of the 1999 Kocaeli and Dzceearthquakes behaved in various manners. During thesite investigations of the 1999 earthquakes inTurkey one could see either totally collapsed,severely damaged or intact buildings just on or nextto the traces of the fault breaks. The examples aremany and it is quite difficult to quote all of them.Some of typical examples from both earthquakes are

(a)

(b)

(c)

Figure 28: Views from dameged pipes due to faultbreak: (a) ruptured asbestos water pipe in Kaynal,(b) a buckled and ruptured water pipe, and (c)ruptured plastic pipe in Fndkl village (Dzceearthquake)

given and briefly discussed.The first example from the Kocaeli earthquake is

a two-storey reinforced concrete house with a raftfoundation in Tepetarla village. The fault passed justunderneath the foundation of the southern side ofthe building (Figure 29a). The relative displacement

of the fault break was about 200 cm. Since 70 % ofthe foundation of the building was on the stationary


17/26

17

northern side of the fault and the southern side ofthe fault had a very small amount normalcomponent, no damage to this building wasobserved. In other words, this implies that if thefault just slides beneath a structure with a raftfoundation in the case of strike-slip faulting, nodamage could be caused to the structure, provided

that the structure is strong enough against shaking.A very peculiar behaviour of an apartment

complex consisting of 8 five-storey apartmentblocks was observed in Kullar village as shown inFigure 29b. Seven apartment blocks failed in a

pancake mode while one apartment block remainedself-standing. One of the failed apartment blocks

just crossed by the fault break that has a relativehorizontal displacement of 240 cm and 20-25 cmvertical throw (north side down). The groundsurface was sloping to north. One of two apartment

blocks on the southern side was damaged while theother one collapsed towards east in a pancake modein accordance with the movement of its foundation.The 5 blocks on the northern side were completelycollapsed in a pancake mode towards west inaccordance with the direction of shaking. Except theapartment block over the fault break, the failure of 5

blocks on the northern side of the fault break may beconsidered to be purely due to shaking. Althoughthe intensity of shaking on the southern side of thefault break should be the same, the damagedself-standing apartment block should deserve some

special consideration. The simplest explanation maybe just a slight variation of the ground condition ifthe apartment blocks are assumed to be structurallyidentical. Whatever the reason is, it is of greatinterest that the most vulnerable buildings may alsosurvive within a distance of 5 to 6 m to the fault

break during the in-land earthquakes.Third example from the Kocaeli earthquake is

concerned with the behaviour of apartmentbuildings beneath which a fault break with a relativehorizontal displacement of 450 cm and 50 cmvertical throw passed in Yzbalar district of

Glck town. As seen from Figure 29c, theapartment buildings were severely damaged bytorsion. In spite of severe damage, the apartment

blocks did not totally collapse due to the closeproximity of their foundation resulting in somelateral restrains during the interaction of theirfoundations and the fault break.

In the Kocaeli earthquake, the surface faultrupture traversed not only through rural areas, butalso some of the highly populated urban areas. Onthe contrary, the surface rupture produced by theDzce earthquake generally passed through the ruralareas. Therefore, mostly one- or two-storey villagehouses were affected by the ground failure induced

(a)

(b)

(c)Figure 29: (a) The behaviour of a two-storeyreinforced building at Tepetarla; (b) the collapse ofapartment block at Kullar, and (c) rotated buildingsdue to strike-slip faulting in Glck (Kocaeliearthquake)

by the fault.The first example from the Dzce earthquake is a

one-storey reinforced concrete house with a raftfoundation in Fndklvillage. The fault passed just

underneath the foundation of the building (Figure30a). The relative displacement of the fault break


18/26

18

was more than about 250 cm with a vertical throwof 90 cm. Since more than 50% of foundation of

buildings was on the southern side of the fault,buildings tilted and rotated as seen in the figure.However, no structural damage to this building wasobserved. In other words, the building behaved likea rigid box during the earthquake.

The next example is concerned with thebehaviour of an old one-storey hm (wooden)house on the fault with same displacementcharacteristics in akribrahim village. Figure 30bshows the state of the hm house after theearthquake. The building is distorted as a result ofthe vertical displacement of the fault rather than thestrike-slip faulting. Nevertheless, there is no totalcollapse of the house and diagonal wooden beamsabsorbed the shocks. It seems that pure lateralstrike-slip movement of the fault is not of great

concern as long as the building freely rotates andslides during the faulting motion. The main reasonof the failure is probably due to the vertical throw ofthe fault. The building behind this hm house isalso a one-story reinforced concrete building that

behaved exactly in the same way as the one shownin Figure 30a.

Figure 30c shows a brick masonry house with araft foundation and a continuous concrete slab allaround the structure in Fndkl village. The fault

break just passed beneath the left corner of thismasonry house, and its relative lateral strike-slipwas 320 cm at this location. There was nostructural damage to this brick house from a visual

inspection. In addition to this house, no structuraldamage to a hmhouse was observed at this site ata distance of 3 m from the fault break. Figure 30dshows a two-storey reinforced concrete house nextto the fault in akribrahim village. In spite ofvery close proximity of this house to the fault break,no damage could be deduced from the visualinspection.

Expect Kaynal, it is very rare to see reinforcedconcrete buildings having stories greater than 4 invillages through the fault break passed through. The

reinforced concrete buildings with 4 stories or moreon the alluvial plain collapsed. Figure 31 shows atotally collapsed 5-storey reinforced concrete

building in Kaynal, beneath that the fault breakpassed. Of course, one may easily think of the faultbreak as primary culprit for the collapse of thisbuilding. Considering other totally collapsed similar

(a) (b)

(c) (d)Figure 30: Tilted and rotated one-story reinforced concrete building in Fndkl (a) and one-story hm

building in akribrahim village (b), a brick masonry house in Fndklvillage (c) and two storey RC house

in akribrahim village (Dzce earthquake)


19/26

19

type buildings, which were not directly crossed bythe fault, it should be plausible to put the blamedirectly onto the fault break.

Figure 31: A totally collapsed 5-storey reinforced

concrete building in Kaynal(Dzce earthquake)

Most of the main compounds of mosques in theearthquake affected region were intact or sufferedvery slight damage. The damage was generallyconcentrated at corners as observed in the previousearthquakes of Turkey. The main compounds ofmosques generally have single dome or multiplesemi-spherical domes and they are structurallysymmetric. For this reason, the main compoundsremain intact during shaking. The main compounds

of the mosques either failed or damaged when thefaulting occurred right underneath the structure orhit by the falling minarets. However, some mosqueshaving shops at ground floor, which is historicallyunconventional, failed as a result of weak-floorsituation. Two examples from near faulting areawere observed in Kaynal and Daryeri Hasanbey

village after the Dzce earthquake. The fault breakpassed beneath the corner of the main compound ofa partially damaged mosque in Daryeri Hasanbeyvillage and another mosque collapsed due to thefault brake with 3.3 m dextral slip and 60 cmvertical throw passing beneath its main compound(Figure 32).

As these examples show that the lives can besaved if the buildings are constructed like a rigid

box structures with the use of reinforced concreteshear walls instead of fragile hollow brick wallshaving almost no structural strength.

5. GROUND SUBSIDENCE AND SUBMARINELANDSLIDE DUE TO FAULTING

In addition to liquefaction-induced lateral

spreading with a large extent (Figure 33), a hugescale subsidence occurred along the southeasternshore of the zmit Gulf during the 1999 Kocaeliearthquake. This ground failure was typicallyobserved between Kavakl district of Glck townand Seymen at its east (Figure 34). As previouslydiscussed in Chapter 3.1, the ground subsidencetook place as a result of secondary normal faultingassociated with the main lateral strike-slip event(see Figures 9-10). The normal fault could be tracedfrom Kavaklshore up to Hisareyn village along themain highway where the fault terminated. The

vertical throw of the fault was more than 2 m atseveral locations. The land on the overhangingblock of the fault subsided with a gradual tilttowards the sea with the northern tip submerged intothe sea. In other words, a huge land has moved as awhole towards the sea with lower relief. A largeindustrial plant under construction for Ford Otosan

(a) (b)Figure 32: (a) Partially damaged mosque in Daryeri Hasanbey village due to fault break passing beneath theleft corner of the main compound, and (b) a collapsed mosque in Kaynaldue to fault break passing beneaththe main compound (Dzce earthquake)


20/26

20

was on the overhanging wall of the fault. It wasundamaged except where ground settlement andtrace of the normal fault with a vertical throw of 240cm intersected one corner of the 50.000 m2building(Figure 35). At this location the building (steel rooftrusses on concrete columns with metal cladding)experienced settlements that pulled the columnsover several tens of centimeters. Particularly atKavakl, fine sand was ejected as a result ofliquefaction (see Figure 34b). However, the groundliquefaction is not considered to be the main causeof the subsidence, and this event was mainlyassociated with the ground settlement due to normal

Figure 35: A view of Ford-Otosan factory with the

normal fault scarp

(a) (b)Figure 34: (a) A flooded housing complex at the eastern side of Ford automobile factory construction siteand (b) submerged Kavakldistrict of Glck town due to normal fault induced ground subsidence

(HD)

(VD)

S+ LS

SML

Lateral Spreading (not-to-scale)

Horizantal Displacement (cm)

Vertical Displacement (cm)

Submarine Landslide

Normal Fault

Subsidence and Lateral Spreading

Explanations

92 (HD)

479 (HD)

409 (HD) / 103 (VD)329 (HD) / 43 (VD)

S+ LSS+ LS

SML

60 (HD) / 28 (VD)

Ulal

BaiskeleSeymenMKEYeniky Park

SekaGlckD. Dere

Haldere

KOCAEL

0 250 500

N

(m)

ZMT GULF

Figure 33: Locations of lateral spreading, subsidence and submarine landslides along the southern shore ofzmit Gulf (after Aydan and Ulusay 2000)


21/26

21

faulting and partly lateral spreading induced byliquefaction.

About one hundred shallow-geotechnicalboreholes with depths varying from 3 to 21 m werepreviously drilled by private companies atKavakl-Glck, hsaniye, and Yeniky districts fora sewerage project. Some of these boreholes were

located on the northern block of the normal faultand near to its trace. The records of these boreholeswere employed by the authors for preliminaryassessments of ground conditions. The availabledata indicated that the groundwater table was veryshallow (1 to 5 m) and the ground consisted ofalluvial deposits. At the top there were generallyloose sand, silty sand and sand and gravel mixtures,while clay and silty clay layers appeared at deeperelevations. Soft-to-moderately stiff clay layers withorganic content were occasionally observed in these

boreholes. In spite of presence of limited shearstrength data of these clay layers, their shearstrength parameters seem to be low (c=13-20 kPaand =5-210). On the other hand, the data from five25 m deep boreholes drilled at the southeastern tipof the zmit Gulf near Kullar village indicated thatthe sequence consisted of fill materials 1.5-2 m inthickness at the top, and sandy and gravely clays atmoderate depths. These soils were underlined by

blue-greenish grey clays with organic contentsimilar to those observed in the boreholes betweenKavakl and Yeniky. The presence of shallow

groundwater and sand levels at upper elevationsgenerates condition for to occurrence of liquefactionand confirm the liquefaction phenomena observedalong the coastline. On the other hand, thetopography gently inclined towards the sea and the

presence of clay layers suggest a possible movementon one of the clay layers which acts as a basalsliding surface and on the normal fault plane actingas a rear release surface similar to landslide. Basedon site observations, Ishihara at al (2000), whosuggest a similar mechanism for the groundsubsidence, considered the existence of a layer of

stiff clays or soft rocks at great depths between 50and 100 m. If the piers beneath the Ford automobilefactory reaching to a depth of 20 m and no damageto the plants except ground settlement are taken intoconsideration, depth of the sliding surface seems to

be greater than 20 m. However, it is still discussableat this time to conclude on the depth of this slidingsurface, and further investigation is needed.

Submarine and sea shore slides took mainly alongthe southern shore of the zmit Gulf, namelyDeirmendere, Haldere, Ulal and Karamrselduring the 1999 Kocaeli earthquake (see Figure 33).These slides were mostly associated with reclaimedlands along the shore and it seems that in addition to

ground disturbance near faulting, soil liquefactionhas also some roles in submarine slides. The most

publicised submarine slide took place atDeirmendere where a 5-storey hotel slid into thesea together with two adjacent buildings (Figure 36).The site investigations showed that there weretension cracks extending into the land for a distance

of 75-80 m from the post failure seashore. Thespacing of these tension cracks was quite consistent.The coastal land about 250 m long was choppedaway into the sea (Figure 37).

(a)

(b)

(c)Figure 36: Some views from the shore ofDeirmendere coast: (a) toppled buildings, (b)slided buildings, and (c) coastal land chopped awayinto the sea due to submarine landslide during 1999Kocaeli earthquake


22/26

22

After the Kocaeli earthquake, depth of water wasmeasured by means of ultrasonic device along across-section perpendicular to the coastline ofDeirmendere (Ishihara at al 2000). In addition, the

bathymetry map of the seabed in the vicinity ofDeirmendere was prepared by a private company(Kiper and Arel 2000) during an investigation

program for site selection. Both the bathymetry mapand the outcome of the water depth measurementsfor Deirmendere shoreline are depicted in Figure38. From these figures it is evident that the seabedtopography considerably changed after theearthquake. Abrupt changes in the configurations ofcontour lines at northwest of the sliding land (shown

by dashed arrows in Figure 38a) and heaves at thebottom of the sea (Figure 38b) probably suggest that

the sliding material spreaded about 300-350 m fromthe shoreline. It is also known that the right- lateraloffset reaches to a maximum value (4.5 m) in thisarea as measured on the western wall of the NavyCommandership (see Figure 6a) and the fault breakdisappears near the shore between Glck andDeirmendere. Based on the general trend of thefault break at this locality, it is possible to considerthat the fault break passes very close toDeirmendere coastline under the sea as illustratedin Figure 37. As previously stated by Ishihara et al(2000) and Kiper and Arel (2000) it seems that the

most important factor contributed to this groundfailure occurred in the form of submarine landslide

is the disturbance effect in the fault zone.An attempt was made by the authors to assess the

effect of liquefaction on this failure as asupplementary work to the previous observations.For the purpose, data from four geotechnical

boreholes of 20 m deep drilled after the earthquakenear the slip-away coast line of Deirmendere by a

private form (see Figure 37) were employed andground conditions and liquefaction potential wereassessed. Figure 39, prepared from the borehole logs,illustrates the general soil conditions, position of thegroundwater table and variation of SPT-N valueswith depth. This figure suggests a very shallowgroundwater table ranging between 0.6 and 1.8 mand fully saturated soils. At the top, except fillmaterials 1 to 1.5 m thick, there are generally silty

sand (SM) layers of 3 to 7 m thick above moderatelydense silty sandy gravels (GW-GM). The methodbased on the field performance data (Seed andDeAlba 1986) was employed as a mean ofevaluating the resistance to liquefaction of thealluvial soils observed through these boreholes. Inaddition, liquefaction potential index (Iwasaki et al1984) in order to estimate the severity of theliquefaction degree at the site was also calculatedfor each layer. The results of the liquefactionsusceptibility assessments indicated that the sandlayers appearing at depths between 1.5 and 4.5 m in

boreholes S-1, S-2 and S-4, and between 1.5 and8 m in borehole in S-3 were liquefable. Comparing

0 250 500(m)

Fault

Slip-away areasDeirmendere

NBoreholes

Damagedwalls

Navy Base

IS-1

IS-1

IS-2 IS-3IS-4

ZMT GULF

Figure 37: Locations of fault break, land submerged and geotechnical boreholes drilled after the Kocaeliearthquake (re-arranged from Ishihara et al 2000, and Kiper and Arel 2000)


23/26

23

0 200 400(m)

26000.00 26100.00 26200.00 26300.00 26400.00 26500.00

31100.00

31000.00

30900.00

30800.00

30700.00

30600.00

30500.00

A'

A

DERMENDERE

(a)

Coast line afterthe earthquake

Before theearthquake

After the earthquake

0 100(m) Liquefable sand layer

(b)Figure 38: (a) Bathymetry of Deirmendere coastand vicinity, and (b) configuration of the seabed

before and after the Kocaeli earthquake (rearrangedfrom Kiper and Arel 2000, and Ishihara et al 2000)

the values of liquefaction potential index valueswith the limits stated by Iwasaki et al (1984), it vasconcluded that these sand layer have high-to-very

high liquefaction potential (see Figure 38). Thesepreliminary assessments suggested that liquefactionoccurred along the shoreline of Deirmendere at adepth between 1.5 and 8 m during the 1999 Kocaeliearthquake. If this ground failure is considered onlydue to liquefaction, seabed relief at a location about75 m off the coast should be taken placeapproximately at -8 m. But it is evident from Figure38a that water depth is about at -15 m at thisdistance. Another important issue is that theextension of the fault probably runs at a locationwhere the toe of the failed slope takes place beforethe earthquake. All these suggested that the groundfailure occurred as a submarine landslide mainly

due to faulting induced shaking and liquefactioninduced lateral spreading can also be considered asanother factor contributing to this ground failure.However, it is still premature at this time toconclude the cause of the ground failure alongDeirmendere coast and more detailedinvestigations involving different analysis

techniques are needed to clarify the mechanism.Whereas many buildings in the central part of

Glck suffered severe damage, most of those builton the slump land itself were little damaged, butsettled and submerged due to fault-inducedsubmarine landslides and subsidence of hugeterrains, and partly liquefaction-induced lateralspreading. Therefore, the association of materialsand groundwater depths, and the orientation andextent of seismogenic and secondary faults can beused to identify the areas where site-specific studies

of potential ground failure may be advisable. Atsuch near-shore zones that are candidates to subjectto groundfailure hazards, structures on theshoreline should be avoided.

6.6.6.6. CONCLUSIONS ANDRECOMMENDATIONS

The authors tried to present the behaviour ofdifferent types of structures built in active faultzones during the last two in-land earthquakes

occurred in Turkey, although it is impossible tocover every aspect of this important topic in theearthquake engineering in a short article such as theone presented here. As a conclusion, it must beclearly stated that "it is almost impossible formankind to prevent the damage to structures if afault break happens to be just passing underneaththe structures." Nevertheless, the adverse effects offault breaks on structures can be minimised to acertain level in view of findings from the actualexamples presented in this article as it would beimpossible re-settle huge populations in earthquake

prone countries without faults. On the basis of thesefindings, the following recommendations may beconsidered as guidelines in developing the seismiccodes for structures in active fault zones.

1. Roadways and railways should not be elevatedand they should be constructed on the existingground surface in active fault zones. In doing so, therecovery of roadways and railways to be damaged

by fault breaks in the event of the earthquakeswould become very rapid. As a result, the effectiverescue and rehabilitation works would become

possible.2. It is not desirable to built bridges and viaducts


24/26

24

in fault zones where permanent deformation of the

ground may take place during earthquakes. If theyhave to be constructed for some reasons, such asfloods or landslides, the limitation of roadwayinclination, etc., they should be constructed asredundant structures, and simply supportedstructures should be avoided. Although pre-stressingcan be used to increase the confinement of girdersalong the longitudinal axis of the bridge, thestructures should not be wholly pre-stressed in orderto prevent domino action in the event of failure.Furthermore, T-type piers must not be used as theyconstitute top-heavy situations that are

undesirable in case of faulting as well asshaking caused by earthquakes.

3. Dams should never be built on active faults

defined in engineering sense as they have severeimpacts on structures and residential areas indownstream. For some reasons, if they have to be

built, their height should be kept to a minimum andthe type of dams should be rock-fill.

4. The current seismic design codes for tunnels,subways and underground caverns can be safelyused. As the damage to tunnel linings due to fault

breaks would be localised, there is no need for extraprecautions at those locations. Nevertheless, thecross sections may be kept larger than the requireddesign size at those locations in order to allow extra

space in the event of relative movements.5. The ground failure due to faulting and

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50

SPT-N VALUE

DEPTH(m)

SAND

GRA-

VELLY

SAND

SILTY

SANDY

GRA-

VEL

WITH

COBB-

LES

0.6

0.63

2.45

FL LI

11.

5

NON-LIQUEFABLE

SPTisnot

applicable(>50)

BOREHOLE: IS-1

0.9 GRAVELLY

SANDY

CLAY

SANDY

GRAVEL

0.35

FL LI

20.

9

NON-LIQUEFABLE

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50

SPT-N VALUE

DEPTH(m)

SANDY

GRAVEL

SAND

N.L.

BOREHOLE: IS-2

0.6

BOREHOLE: IS-3

FILL

SANDY

GRAVEL

WITH

COBBLES

0.6

FL LI

17.

3

NON-LIQUEFABLE

GRAVELLY

SAND

SAND-

SILTY

GRAVEL-

LY SAND

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50

SPT-N VALUE

DEPTH(m)

0.9

>50

1.8

BOREHOLE: IS-4

FILL

SANDY

GRAVEL

0.69

FL LI

7.

8

NON-LIQUEFABLE

SILTY

SAND

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50

SPT-N VALUE

DEPTH(m)

GRAVEL-

LY SILTY

CLAY

0.6

Figure 39: Simplified borehole logs illustrating ground conditions, variation of SPT-N values with depth andliquefiable layers at Deirmendere coast line (FL: Factor of safety against liquefaction, LI: Liquefaction

potential index, N.L.: Non-liquefable)


25/26

25

liquefaction, rather than shaking, is the principalcause of damage to tubular structures. Therefore,such structures should be designed with the use offlexible joints to accommodate relativedisplacements with the consideration of likelyrelative displacement of the ground.

6. Damages to buildings on or near fault zones

seem to be minimized if they are designed as rigidbox-like structures with the use of shear walls. Walls,with fragile hollow bricks having no structuralstrength, must not be allowed to use in buildingconstruction. If the solid brick walls are to be used,the walls should be constructed before theconstruction of columns and beams in order to attainthe structural integrity of the structure and theyshould be light, resistant and ductile.

7. Vertical accelerations of in-land earthquakescould be 0.65-0.75 times the maximum horizontal

accelerations. Sometimes they may exceed, even themaximum horizontal acceleration as observed inDzce during the 1999 Kocaeli earthquake. Theeffect of vertical accelerations must be considered instructural design.

8. In recent years, particularly during the 1994Northridge and 1995 Kobe earthquakes, significantnumber of records with large peak accelerations(e.g.~1g) and with long duration pulses have beenacquired in the near field (


26/26

26

Tuncay, E. [1998] "A site investigation ofAdana-Ceyhan earthquake of June 27, 1998",Turkish Earthquake Foundation, TDV/DR006-30, 131 p.

Aydan, ., Ulusay, R., Hasgr, Z. and Takn, B.[1999] "A site investigation of Kocaeliearthquake of August 17, 1999", TurkishEarthquake Foundation, TDV/DR 08-49,180 p.

Aydan, ., Ulusay, R., Kumsar, H. and Tuncay, E.[2000] "Site investigation and engineeringevaluation of the Dzce-Bolu earthquake of

November 12, 1999", Turkish EarthquakeFoundation, TDV/DR 05-21, 307 p.

elebi, M., Toprak, S. and Holzer, T. [2000]"Strong-motion, site-effects and hazardsissues in rebuilding Turkey: in light of the 17August, 1999 earthquake and its afterschoks",

The 1999 zmit and Dzce Earthquakes:Preliminary Results, stanbul TechnicalUniversity, 247-263.

Dalg, S. [1994] "Anadolu otoyolu Bolu Dageiinin mhendislik jeolojisi", DoktoraTezi, stanbul niversitesi, Fen BilimleriEnstits (unpublished PhD thesis; inTurkish).

DS [1999] "Marmara Depremi zel Sayfas: 12Kasm 1999 Tarihli Dzce DepremininDeerlendirilmesi",http://www.dsi.gov.tr/deprem/

Ergnay, O. and Glkan, P. [1993] "Land-useplanning as instrument of earthquakehazard mitigation", In ComprehensiveApproach to Earthquake Disaster

Mitigation, 235-278.Glkan, P., Ycemen, S., Koyiit, A., Doyuran, V.

and Baz, N. [1993] "Probabilisticacceleration map of Turkey", METU, Report

No. 93-01, 156 p.Hamada, M. and Aydan, . [1992] "A report on

the site investigation of the March 13Earthquake of Erzincan, Turkey", ADEP,

Association for Development ofEarthquake Prediction, 86 p.

Ishihara, K. Erken, A. and Kiku, H. [2000]"Geotechnical aspects of the ground damageinduced by the faults", Proceedings of theThird Japan-Turkey Workshop onEarthquake Engineering,stanbul, 1-8.

Iwasaki, T., Arakawa, T. and Tokida, K.I. [1984]"Simplified procedures for assigningliquefaction during earthquakes", Journal ofSoil Dynamics and Earthquake Engineering,3 (1), 49-58.

Kiper, B. and Arel, E. [2000] "Deirmendere(Kocaeli) yerleim amaltemel sondajlarve

jeolojik-jeoteknik inceleme raporu", Vol. 1,46 p (in Turkish).

Seed, H.B. and DeAlba, P. [1986] "Use of SPT andCPT tests for evaluating the liquefactionresistance of sands", In Use of In-situ TestsinGeotechnical Engineering, ASCEGeotechnical Special Publication, 6,281-302.

Ulusay, R. [1999] "Kentleme srecinde yerseimive depreme dayankl yap inasnda

jeolojik-jeoteknik ettlerin nemi ve ilevi",Teknik Klavuz Serisi-6, No:51, TMMOBJeoloji Mhendisleri Odas. 23 p (inTurkish).

Unterberger, W. and Brandl, J. [2000] "The effectsof the recent earthquake in Turkey on theBolu Tunnels",Felsbau, 2, 50-54.

Yeats, R.S., Sieh, K. and Allen, C.R. [1997] "The

Geology of Earthquakes", Oxford UniversityPress, Newyork.
http://www.dsi.gov.tr/deprem/http://www.dsi.gov.tr/deprem/

Documents

Behavior of Structures Built on Active Faults Zones