De

Ha

b

a

ARRAA

KB1NF

1

wtscmtstg

wis

s

(

0h

Applied Ocean Research 49 (2015) 72–82

Contents lists available at ScienceDirect

Applied Ocean Research

journal homepage: www.elsevier.com/locate/apor

ynamic responses of two blocks under dynamic loading usingxperimental and numerical studies

.K. Cihana,∗, A. Ergina, K. Cihanb, I. Gulera

Middle East Technical University, Civil Engineering Department, Ocean Engineering Research Center, TurkeyKırıkkale University, Faculty of Engineering, Civil Engineering Department, Turkey

r t i c l e i n f o

rticle history:eceived 13 June 2014eceived in revised form 2 October 2014ccepted 26 November 2014vailable online 25 December 2014

eywords:lock type quay walls

a b s t r a c t

Block type quay walls are one of the most generally used type of gravity quay walls however seismic risksof this kind of structures have not already received the proper amount of attention. In this study, stabilityof block type quay wall which consists of two concrete blocks is investigated experimentally and numer-ically. 1 g shaking table tests are used for experimental study. Model scale is 1/10 and model is placed onrigid bed to ignore damage due to foundation deformation. Two different granular materials (Soil 1 andSoil 2) which have different nominal diameters are used as backfill materials to understand the effectof nominal diameters on structure’s stability. During the experiments accelerations, pore pressures, soil

g shaking table testsumerical modelingriction coefficient

pressures and displacements are measured for two blocks under different cycling loadings. Soil pressuretest results are presented in non-fluctuating and fluctuating components to determine the distributionand application point of the fluctuating component on two blocks. By using experiment results, thefriction coefficients between the rubble-block and block-block are determined and compared with rec-ommended friction coefficients in standards. PLAXIS V8.2 software program is used for numerical studyto determine the material properties.

. Introduction

Block type quay wall is the simplest type of gravity quay wall,hich consists of blocks of concrete or natural stone placed from

he waterside on a foundation including a layer of gravel or crushedtone on top of each other. After placing, the blocks a reinforced con-rete cap is placed as cast in situ. Block walls require much buildingaterial however labor necessity is relatively little. The height of

his structure exceeds 20 m. It is important to have a good filtertructure behind the wall to prevent the leakage of soil. This fil-er structure should involve thick filling of rock fill material with aood filter structure (CUR [1]).

Block type quay wall is one of the most important gravity quayalls which would suffer during earthquakes; however, this truth

s known clearly, seismic risk of this kind of structures have nottudied in depth, yet.

Fig. 1 shows the typical section of block type quay wall.

Blocks maintain their stability through friction between them-

elves and between the bottom block and the seabed. Typical failure

∗ Corresponding author. Tel.: +90 5326636415.E-mail addresses: hkarakus@yukselproje.com.tr (H.K. Cihan), ergin@metu.edu.tr

A. Ergin), kubilaycihan@gmail.com (K. Cihan), iguler@yukselproje.com.tr (I. Guler).

141-1187/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.apor.2014.11.003

© 2014 Elsevier Ltd. All rights reserved.

modes during earthquakes involve seaward displacement, settle-ment, and tilting of blocks.

The evidence of damage to gravity quay walls suggests that(PIANC [2]):

1. most damage to gravity quay walls is often associated with sig-nificant deformation of a soft or liquefiable soil deposit, and,hence, if liquefaction is an issue, implementing appropriateremediation measures against liquefaction may be an effectiveapproach to attaining significantly better seismic performance;

2. most failures of gravity quay walls in practice result from exces-sive deformations, not catastrophic collapses, and, therefore,design methods based on displacements and ultimate stressstates are desirable for defining the comprehensive seismic per-formance; and

3. overturning/collapse of concrete block type walls could occurwhen tilting is excessive, and this type of wall needs carefulconsideration in specifying damage criteria regarding the over-turning/collapse mode.

The heavy damage was observed on coastal structures such as

refineries, petrochemical plants and ports the Eastern MarmaraEarthquake occurred on 17 August 1999 with an Mw = 7.4 andIzmit Bay and north-west Turkey had been seriously affected fromthis earthquake. Especially, earthquake was caused crucial damage

H.K. Cihan et al. / Applied Ocean

me

cCbcad

tAsawtsqbs

btebr(lmaetll

qtasCnuat

m

i

the limitations of this instrument as:

Fig. 1. Typical section of block type wall.

ostly on block type quay walls at Derince Port in Izmit (Yükselt al. [3]).

The design of block type quay walls should be performedonsidering stability, serviceability and safety as well as economy.onventional seismic design methodology is generally used forlock type quay walls. However, this traditional design methodannot provide the required design data and also cannot provideny information about the performance of the structure afterynamic loading (Karakus [4]).

Sumer et al. [5] prepared an inventory including the observa-ions of damage to marine structures caused by liquefaction inugust 17, 1999 Eastern Marmara Earthquake. According to thistudy, backfills behind quay walls and sheet-piled structures werelmost invariably liquefied; quay walls and sheet-piled structuresere displaced seaward; storage tanks near the shoreline were

ilted; there were cases where the seabed settled, and structuresettled and collapsed. Furthermore, in Tuzla Port, the block typeuay wall was displaced seaward by O (40 cm) and backfill settledy O (10 cm). There was no direct evidence of liquefaction (i.e., noand boils) in this area.

Sadrekarimi et al. [6] investigated both static and dynamicehavior of hunchbacked gravity quay wall by using the 1 g shakingable tests for various base accelerations on models with differ-nt subsoil relative densities. The results revealed that (i) negativeack-slope (elevations below the breaking point of the hunch)educes the lateral earth pressure however positive back-slopeelevations above the breaking point of the hunch) increases theateral earth pressure, (ii) relative density of sea bed affected the

ovement of the wall significantly, the wall moved more with largecceleration when the sea bed was softer, (iii) if the model wasxposed to same earthquake again, due to the subsoil densifica-ion less wall movement was observed, (iv) application point of theateral thrust fluctuated within the mid-third of wall’s height (v)arger the height provided safer area behind the wall.

Sadrekarimi [7] studied seismic displacement of broken-backuay walls by shaking table model experiments. Sadrekarimi [7]ried to estimate the sliding displacements of structure by usingn improved sliding block model that incorporates the pseudo-tatic method of Mononobe-Okabe for lateral earth pressures.hakraborty and Choudhury [8,9] study on the stability of a generalo-vertical waterfront retaining wall supporting inclined backfillnder earthquake forces and combined action of the earthquake

nd tsunami forces using limit equilibrium method. The factor ofhe sliding was computed using pseudo-dynamic approach.

There are several studies which are conducted by numerical andodel studies in order to understand the stability of gravity quay

Research 49 (2015) 72–82 73

wall especially for caisson type quay wall under dynamic loading;Towhata et al. [10], Woodward and Griffiths [11], Ghalandarzadehet al. [12], Zeng [13], Madabhushi and Zeng [14], Kim et al. [15,16],Choudhury and Ahmad [17,18], Lee [19], Moghadam et al. [20],Maleki and Mahjoubi [21], Na et al. [22], Tiznado and Roa [23],Torisu et al. [24], Dewoolkar et al. [25].

In this study, a block type quay wall which is composed oftwo blocks are used to understand the dynamic response of thesetype of structures both experimentally and numerically. By using1 g shaking test method, block displacements, accelerations, soilpressures are measured. Additionally, friction coefficients betweenblock-block and block-rubble are determined and compared withthe values given in literature. Since usage of rock fill material issuggested behind the wall (CUR [1]), granular materials (Soil 1and Soil 2) are used as backfill material for the first time in suchtype of experiments. And, experimental study is modeled numeri-cally by using PLAXIS V8.2 software program to define the materialparameters.

2. Experimental set-up

In general, three types of laboratory model studies are availablefor evaluating the dynamic response of structures: the real scaledmodeling test, the centrifuge test and 1 g shaking table test.

Real scaled modeling tests investigations are expensive andrequire the services of a construction contractor in most of thecases. Centrifuge tests can be more reliable than the 1 g tests dueto point of reduced stress level which affected the soil behaviorsignificantly. On the other hand, relatively small model scale is rec-ommended for the centrifuge tests since it affects the soil grainsize.

In literature, disadvantages of 1 g shaking table tests and solu-tions suggested are given as;

i. dilatancy of sand and development of excess pore water pres-sure. This problem can be solved by compacting sand in the modellooser than in the corresponding real-life structure (Torisu et al.[24]).

ii. It is difficult to simulate the stress–strain behavior of granularsoil over a wide range of strain and different confining stresslevels. According to Towhatam (1995), “the density of sand shouldbe reduced in the model scale in order to create a similar typeof stress–strain behavior in the lower confining stress level”. “Thevalue of reduced density is calculated by the formula proposed byGhalandarzadeh [12]” (Moghadam et al. [20]).

ii. The boundary effects formed by the physical modeling mightaffect the responses of the whole model (Moghadam et al. [20]).According to Dewoolkar et al. [25], “If the ratio of backfill length tothe wall height is high enough (over 2), then the boundary has nosignificant effect on the wall structure response”.

iv. Dissipation of excess pore pressure is faster in the model thanthat of prototype when the pore fluid and soil particles in modeland prototype are the same (Yoshimi and Tokimatsu [26]).According to Ghalandarzadeh [12], “Regarding the fast dissipationproblem, occurring in excess pore pressure, the input shaking isrecommended to be applied in a longer duration time”.

The experimental study is carried out by using 1 g shaking table,which is available in laboratory as a part of infrastructure. In thisstudy above given recommendations are considered to overcome

i. granular backfill materials (Soil 1 and Soil 2) are used to reducethe scale effect and significance of pore pressure generation,

74 H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82

one block experiment set-up.

i

avWC

td4aatbF

-

-

-

-

-

mihsaicg

dofK

3 position transducers. In Fig. 4, instruments and their placementsare shown.

In Fig. 5, maximum acceleration measurements (amax) for 4 Hzare presented as an example of acceleration (g) versus time (s) for

Table 1Soil parameters for backfill and foundation.

3 ◦

Fig. 2. General view of

i. backfill length to the wall height are taken as high enough (over2) to reduce the boundary effect.

1 g shaking table tests are carried out at Hydraulics and Coastalnd Harbor Lab., Civil Engineering Faculty at Yıldız Technical Uni-ersity as a part of “Simplified Dynamic Analysis of Block Type Quayall” project sponsored by Scientific and Technological Research

ouncil of Turkey (TUBITAK).A series of 1 g shaking table tests are carried out to inves-

igate the dynamic response of block type quay walls. The oneegree of freedom 1 g shaking table had deck dimensions of00 cm–100 cm–100 cm with a 4 t load capacity. It is driven by

100-kN capacity hydraulic actuator with operator controllingnd PC software. Shaking table is one dimensional in its motion,hus only longitudinal components of accelerations are obtainedy omitting the transverse and vertical components. As shown inig. 2;

The blocks are placed on the shaking table between dummies.Dummies are used to give the side effects from the adjacent blocksas in the prototype conditions.

All the instruments (soil pressure cells, accelerometers, positiontransducers and pore pressure cells) are placed.

Backfill material is placed behind the blocks. Two different gran-ular backfill materials (Soil 1 and Soil 2) which have differentnominal diameters are used as backfill materials

The system is filled with water before starting the experimentsand the absorbers are used to prevent the end effects due toreflections caused by dynamic loading.

As it is known that liquefaction is very important subject, in thisstudy, it is assumed that soil improvement techniques has to beused to improve the existing soil conditions to obtain the satis-factory conditions for backfill and foundation.

The raining system is used to prepare the backfill behind theodel wall to obtain same relative density for each test. Poros-

ty, initial velocity of soil particles, deposition height and fallingeight are the major factors affecting the relative density of theoil particles prepared by raining method. Falling height is chosens 65 cm and is kept constant by lifting the sieve at each stage dur-ng backfilling (Fig. 3). Relative density of the Soil 1 and Soil 2 areomputed between 60% and 70% respectively. Soil properties areiven in Table 1. As seen in Table 1, Soil 2 is finer than Soil 1.

Granular materials are used as backfill material to define the

ynamic response of block type quay walls. In practical, the weightf the granular material which is used as backfill is given 5–100 kgor San Antonia Port, block type quay wall, Chile and 1–50 kg foralamata Port, block type quay wall in PIANC [2] and 3–50 kg for

Fig. 3. Raining system and shaking table.

San Pedro, block type quay wall in CUR [1]. This means that nom-inal diameter of the backfill can be taken as 7 cm < Dn50 < 34 cm.In this study the Dn50 of Soil 1 and Soil 2 are selected as 22 cm,10 cm respectively in prototype. Scale factor of model is deter-mined as 1/10 and the general block dimensions are determinedas 3 m–2 m–2.5 m.

In this study, the law of similitude is developed based on the fac-tor considered as most important in the simulation. Iai [27] deriveda similitude relation with the basic equation governing the equilib-rium and mass balance of the soil skeleton, pore water, pile andsheet pile structures, and external waters such as the sea. The soildensity is kept the same for both the prototype and model to sim-plify scaling parameters in the 1 g model testing. The correspondingscaling of parameters between the prototype and model used in thisexperiment are shown in Table 2.

In 1 g shaking table tests, four different types of instruments areused; 4 soil pressure cells, 3 accelerometers, 2 pore pressure cells,

Soil parameters (�dry) (kN/m ) (�) ( ) Dn50 (cm)

Soil 1 (backfill) 16 40 2.2Soil 2 (backfill) 16 40 1.0Foundation 16 40 2.2

H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82 75

Fig. 4. Blocks, dummies

Table 2Scaling factors in present model.

Items Scaling factorsin general

Scaling factors for the presentmodel (prototype/model)

Length � 10Time �0.5 3.16Acceleration 1 1Displacement �1.5 31.62Force �3 1000Density 1 1

Modulus � 10

Acc 1 Acc 2 Acc 3

-0.3-0.2-0.10.00.10.20.3

Acc

1

[ g ]

-0.4

-0.2

0.0

0.2

0.4

0.6

Acc

2

[ g ]

-0.3-0.2-0.10.00.10.20.30.4

Acc

3

[ g ]

151050

0.26898 g

-0.28055 g0.4996 g

-0.3712 g

-0.27

Fig. 5. Acceleration values of Base (Acc 1), Blo

and equipments.

the accelerometers placed at Base (Acc 1), on Block 2 (Acc 2) andon Block 1 (Acc 3). In Tables 3 and 4, maximum Base and Blocksacceleration values and amplification ratios (Block/Base) are shownfor Soil 1 and Soil 2 with respect to frequencies, respectively as anexample (Karakus [28]).

During the experiments, it is assumed that (i) at a particularinstant, both the backfill soil mass and the retaining wall havebeen assumed to shake simultaneously with the same earthquakeintensities (Chakraborty and Choudhury [8,9]), (ii) however it is

important to consider the pressure due to tsunami wave for design-ing the waterfront retaining wall (Chakraborty and Choudhury[8,9]), the wave and current loads on the wall are not taken into

4035302520s

0.35479 g

263 g

ck 1 (Acc 3) and Block 2 (Acc 2) for 4 Hz.

76 H.K. Cihan et al. / Applied Ocean

Table 3Maximum accelerations at Base (Acc 1), on Block 1 (Acc 3) and on Block 2 (Acc 2)for Soil 1 with respect to frequencies.

Frequency (Hz) Base acceleration (g) Ratios

Block 1/Base Block 2/Base

2 0.07 1 13 0.18 1 14 0.28 1.25 1.795 0.4 1.73 26 0.55 2 2.32

Table 4Maximum accelerations at Base (Acc 1), on Block 1 (Acc 3) and on Block 2 (Acc 2)for Soil 2 with respect to frequencies.

Frequency (Hz) Base acceleration (g) Ratios

Block 1/Base Block 2/Base

4 0.24 1.13 1.45

ctla

ttm

-

-

-

-

5 0.41 1.61 1.706 0.60 2.98 2.79

onsiderations, (iii) water elevation is kept equal on both sides ofhe quay wall, and no tidal changes applied, and (iv) all the actingoads due to mooring, berthing and crane operation and live loadsre not taken into consideration (Karakus [28]).

Systematic and random errors in any experimental study arehe major concerns in achieving the test objectives. It is necessaryo focus on reducing the errors through selection of measurement

ethods and equipment. In this study for each experiment,

the instruments are determined carefully considering the exper-imental requirements (instrumentation limitations are checkedin the planning of the experiment stage),

calibrations of the instruments (soil pressure cells, pore pressurecells, position transducers and accelerometers) are made in theplanning of the experiment stage,

all the measurements are collected electronically and stored incomputer,

to reduce the possible vibration affects on acceleration mea-surements, during the experiments other tests are scheduled atdifferent times.

Pore P 1 Pore P 2

27.427.627.828.028.228.428.628.829.029.229.429.629.830.030.2

Por

e P

1

[ 10^

-3 b

ar ]

9.810.010.210.410.610.811.011.211.411.611.812.012.212.4

Por

e P

2

[ 10^

-3 b

ar ]

20151050

30

27.5084 10^-3 bar

12.3226 10^-3 bar

Fig. 6. Pore pressure values of Pore P1 and Pore

Research 49 (2015) 72–82

- various ground motion sets are used and each ground motionresults are compared to each other,

- soil pressure measurements are checked and compared with theprevious measurements for two blocks for dry and saturated con-ditions before cyclic loadings.

Results of the measurements are compared by repeating theexperiments. Total 20 experiments are carried out in laboratory and8 of them are presented in this manuscript. In case of incompatiblemeasurements, the experiments are repeated after controlling theexperiment set-up.

3. Results and discussion

3.1. The excess pore pressure generation

The excess pore pressure has significant effects on soil pressureand horizontal displacements on structures under dynamic loading.Kim et al. [16] stated that if the excess pore pressure increases, thebackfill soil behaves increasingly like a fluid, thus the mobility ofthe soil increases. Zeng [13] emphasized that excess pore pressurehad a significant effect both on the angle of backfill wedge andhorizontal thrust, thus when excess pore pressure developed in thebackfill, comprehensive numerical procedures should be made tounderstand the response of gravity quay walls.

On the other hand, experimental, numerical and analytical stud-ies showed that when permeability increases, the accumulation ofexcess pore pressure is reduced. In this study, gravel type backfillmaterials (Soil 1 and Soil 2) are used and since gravels are more per-meable, significant excess pore pressures usually do not generatefor this kind of backfill. The excess pore pressures occurred underdynamic loading disappears immediately. Pore pressure values atdifferent location in backfill (Soil 1) for 4 Hz are shown in Fig. 6.As seen in Fig. 6 as an example (Karakus [28]), pore pressure didnot generate in backfill significantly under cyclic loadings. Similarobservation is obtained for Soil 2 due to high permeability.

3.2. Soil pressure

Four soil pressure cells which are placed to different depthsof block type quay walls are used to determine the soil pres-sure variation during cyclic loadings. In Fig. 7, a typical total soil

504540353025s

.0959 10^-3 bar

9.9401 10^-3 bar

P2 located at 28.4 cm and 11.1 cm for 4 Hz.

H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82 77

-0.4-0.2

00.20.40.60.8

11.2

403020100Sa

tura

ted

Soil

Pres

s ure

s (k

pa)

Time (s)

Total Soi l Pres sure

Fluctua�n g C omp onent

Non-Fluctua�ngComponent

Fig. 7. Total soil pressure, non-fluctuating and fluctuating compone

Table 5Frequency and the depth where pressure change takes place for Soil 1.

Frequency (Hz) Depth (cm)

3 254 16

psaSn(ne

aeiadbc0

idTao0

1s

3

awohd

TF

the horizontal displacement measurements increase for Block 1and Block 2, while frequency is increasing for Soil 1 and Soil 2,the vertical displacement measurements and tilting degree alsoincrease while frequency is increasing for Soil 1 and Soil 2,

5 256 17

ressure measurement for 4 Hz frequency is presented. Soil pres-ure cells are placed at 5 cm–15 cm below the top of the Block 2 (SP4nd SP3) and 25 cm–35 cm below the top of the Block 2 (SP2 andP1). Total soil pressures are separated into a fluctuating compo-ent and a non-fluctuating component by using smoothing processMatlab program is used). Total soil pressure and fluctuating andon-fluctuating components are shown together in Fig. 7 as anxample (Karakus [28]).

While non-fluctuating component variation with depth islmost linear, fluctuating component variation changes nonlin-arly. Max. fluctuating components of total soil pressures for Soil 1ncreases until a “certain depth” for 3 Hz, 4 Hz, 5 Hz and 6 Hz. After

certain depth, max. fluctuating components of total soil pressuresecrease. The depth where this change in pressure takes place cane defined as in Table 5. Application point of the max. fluctuatingomponents of total soil pressure for Soil 1 is between 0.40 H and.63 H 3 Hz, 4 Hz, 5 Hz and 6 Hz (H is the structure height).

Max. fluctuating components of total soil pressures for Soil 2ncrease until a “certain depth” for 5 Hz and 6 Hz. After a certainepth, max. fluctuating components of total soil pressures decrease.he depth where this change in pressure takes place can be defineds in Table 6. Application point of the max. fluctuating componentsf total saturated soil pressure for Soil 2 is between 0.375 H and.65 H for 6 Hz and 5 Hz (H is the structure height).

Maximum fluctuating components variation with depth for Soil for 5 and 6 Hz in Fig. 8a and b and for Soil 2 for 6 Hz in Fig. 8c arehown as an example (Karakus [28]).

.3. Displacements and tilting of blocks

Displacement and tilting measurements in the experimentsre defined as described below. Tilting of the block type quayall computed on the assumption that almost no settlement will

ccur at the firm foundation of the structure (Fig. 9). If the seismicorizontal movement of the wall is characterized by the horizontalisplacement at the wall base, �x2, and at the wall top, �x1, then

able 6requency and the depth where pressure change takes place for Soil 1.

Frequency (Hz) Depth (cm)

5 266 15

nts of total saturated soil pressures for SP2 for 4 Hz for Soil 2.

the tilting of the upper block, ˛, is expressed as (Tiznado and Roa[23]):

= tan−1(�x2 − �x1

h

)(1)

where h is the concrete block height.Measured displacements of blocks experimentally and calcu-

lated tilting values of blocks for Soil 1 and Soil 2 are shown inTables 7 and 8 respectively.

According to Tables 7 and 8;

Fig. 8. Maximum fluctuating components variation with depth for Soil 1 and Soil 2.

78 H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82

Fig. 9. Displacement and tilting for two blocks.

Table 7Horizontal displacement measurements and tilting values of blocks for each fre-quency for Soil 1.

Frequency (Hz) Horizontal disp.block 1 (mm)

Horizontal disp.block 2 (mm)

Tilting (◦)

2 0 0 03 0 0.13 04 0.68 1.56 0.25 8.6 14.34 1.326 13.07 51.56 –

Table 8Horizontal displacement measurements and tilting values of blocks for each fre-quency in tests which Soil 2 was used as backfill material.

Frequency (Hz) Horizontal disp.block 1 (mm)

Horizontal disp.block 2 (mm)

Tilting (◦)

4 1.64 3.36 0.39

Table 9Damage level of block(s) for Soil 1 and Soil 2.

Soil type Min. damage Cont. damage

Block (S) Soil 1 (g) Soil 2 Soil 1 Soil 2

Two blocks

The direction of the hydrodynamic pressure is determinedconsidering the worst possible condition with respect to stability of

5 13.32 27.94 3.356 23.32 141.21 –

for 2 Hz, there is no motion for Block 1 and Block 2 for Soil 1,for 3 Hz, there is no motion for Block 1. Block 2, starts to slide forSoil 1,for 5 Hz, sudden increment occurs for horizontal and vertical dis-placement measurements and tilting degree,it is not possible to evaluate tilting degree for 6 Hz due to the big

horizontal displacement measurements. Thus, tilting degree aregiven for 4 Hz and 5 Hz,

0

10

20

30

40

50

d/H*

100

Blo

Blo

0 0.07

ock 2

ock 1

0.14 0.2 8

PGA (BASE)

8 0. 4 0.5

So

55

il 1

Fig. 10. Relative displacement/structure height (d/H

Block 1 0.33 0.28 0.44 0.37Block 2 0.32 0.26 0.43 0.36

the horizontal displacement for Block 2, which is located on Block1, is greater than the horizontal displacement for Block 1 for Soil1 and Soil 2,the horizontal displacement measurements for Soil 2 are greaterthan the horizontal and vertical displacement measurements forSoil 1. However, both backfill material are selected as gravel, Soil2 is finer than Soil 1.

Levels damage of structure is obtained by d/H from experi-mental results. d and H define blocks displacement and heightof block, respectively. Figs. 10 and 11 show the relative displace-ment/structure height (d/H × 100) and tilting versus PGA (Base) forSoil 1 and Soil 2. When experimental results of level damage of thisstudy is compared with the level damage table given PIANC [2],it is seen that minimum and controlled level damage of Soil 2 iscritical than Soil 1 in terms of given damage criteria (Table 9). Itis observed that Soil 2 slumped down toward the structure moreeasily. Thus, Soil 2 can push the blocks more strongly. Moreover,two blocks are placed without any shear key between blocks andSoil 2 can replace the space between the blocks and can increasethe slipping condition between the blocks.

The friction coefficient is one of the key parameter for analyzingof the seismic design of block type quay walls. The stabilities of theblock type quay walls are provided by the friction between blocksand friction at the bottom of the structure.

Static friction coefficients are calculated by evaluating the 1 gshaking table tests results and tilting tests. The accelerations,displacements and soil pressure measurements obtained exper-imentally for Soil 1 and hydrodynamic pressure obtained byWestergaard [29] are used to compute the forces acting on thestructure then by using Coulomb Law the static friction coefficientis calculated.

Coulomb Law governed by Eq. (2):

Ff ≤ �Fn (2)

where, Ff is the force exerted by friction, � is the friction coefficient,Fn is the normal force.

the wall in sliding mode of failure Chakraborty and Choudhury [8,9].As it is seen from Table 10 the calculated static friction coefficients

d/H*

100

0

10

20

30

40

50

0

Block 2

Block 1

0.24

PGA (BAS E

0.41 0

)

.6

Soil 2

× 100) versus PGA (Base) for Soil 1 and Soil 2.

H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82 79

0

1

2

3

4Ro

ta�o

n

Block 2

0 0.28

(BASE)PGA

0.4

Soil 1

Rota�o

n

0

1

2

3

4

0

Block 2

0.24

(BASE)PGA

0.41

Soil 2

Fig. 11. Tilting versus PGA (Ba

Table 10Comparisons of static friction coefficients with the standards.

Surface Tiltingtests

1 g shakingtable tests

Turkish regulations(2008)

OCDI(2009)

aCa

3

waaaasaaidmt

aamanged

a

Shear wave velocity at pref:

TM

Block-rubble 0.55 0.57 0.60 0.60Block-block 0.47 0.47 0.50 0.50

re close to recommended values given in Seismic Specification foroast and Harbor Structures, Railway, Airport Constructions [30]nd OCDI [31].

.4. Stress–strain analysis

Stress–strain analyses are performed by using PLAXIS V8.2 soft-are program. Fifteen node, triangular, 2D plane strain elements

re used in the Finite element (FE) model. In numerical model runsre carried out for 10 s duration in accordance with the model char-cteristics and limitations. Thus, comparisons of the displacementsnd soil pressure results between numerical and experimentaltudies are made for test duration 10 s which corresponding tolmost 30 s duration in prototype. When the experimental resultsre examined, it is seen that within 10 s the representative hor-zontal displacement values are reached close to total horizontalisplacement values occurred in 30 s thus, duration of numericalodel as 10 s is found to be accurate enough to obtain the horizon-

al displacement.In order to simulate the behavior of the soil, a suitable soil model

nd appropriate design material (soil and concrete) parameters aressigned as input parameters in model. Soil is the most complicatedaterial. There are different types of material models which can be

pplied for the solution of geotechnical problems for analytical andumerical analysis. Geoengineering studies can be divided into tworoups; (a) limit state analysis which is used for slope, wall stability,

tc., (b) deformation analysis which is used for retaining deflection,isplacements, etc.

The “Hardening Soil Model” is an advanced nonlinear soil modelnd it is recommended for deformation analysis. For this reason,

able 11aterial properties used in FE analysis.

Symbol Parameters Units

�unsat Unsaturated unit weight kN/m� sat Saturated unit weight kN/mEref50 Reference secant Young’s modulus kN/mErefoed

Reference constraint modulus kN/mErefur Reference unloading–reloading modulus kN/m� Shear strength angle ◦

Dilatancy angle ◦

�ur Poisson rate –

pref Reference stress kN/mm Power for stress level dependency –

se) for Soil 1 and Soil 2.

in this study, hardening soil model (HS) is used for modeling thedynamic behavior of the granular backfill material. The inputparameters of (HS) are (i) internal friction angle (�), (ii) cohesionintercept (c), (iii) soil stiffness, and (iv) dilatancy angle ( ). Soilstiffness is defined by the (Eref50 ), which characterizes the shear

behavior of the soil; (Erefoed

), which mainly controls volumetric

behavior; and (Erefur ), is the unloading–reloading modulus.Additionally, simulating friction between block-block, and to

achieve relative displacement between two blocks, a very thin soillayer is defined as interface. The properties of interface and all othermaterial properties are summarized in Table 11.

In numerical analysis, an acceleration time history obtainedfrom experimental tests for 5 Hz is used as an input motion (Fig. 12).

Numerical distortion of the propagating wave can occur in adynamic analysis as a function of the modeling conditions. Boththe frequency content of the input wave and the wave-speed char-acteristics of the system will affect the numerical accuracy of wavetransmission. Kuhlemeyer and Lysmer [32] showed that for accu-rate representation of wave transmission through a model, thespatial element size, �l, must be smaller than approximately onetenth to one-eighth of the wavelength associated with the highestfrequency component of the input wave, i.e., Arabloueri et al. [33].

(�l)soil ≤�min,soil

10(3)

� is the wavelength corresponding to the maximum frequency f ofinterest. Additionally, to calculate shear velocity of backfill mate-rial, following expression is used:

Grefur = Erefur2(1 + vur)

(4)

Gmax

Gur= 3 (5)

Vs =√Gmax

�(6)

Material type

Backfill Seabed Interface

3 16 20 163 19 22 192 10,000 150,000 60,0002 10,000 148,910 60,0002 30,000 450,000 180,000

40 45 4010 15 00.3 0.3 0.3

2 100 100 1000.5 0.5 0.5

80 H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82

0-4

-2

0

2

4ax [m/s2]

2 4

Dynamic time [s]6 8 10

tion u

3icVa

rdodbs˛

˛

vbmT

Fig. 12. Input mo

In numerical analysis, Eur of backfill material is chosen as0,000 kPa and validated by comparing experimental and numer-

cal results. The wave propagation velocity of backfill material isalculated as 147 m/s. In this case �/10 = VS/10 f = 2.94 m, beingS = 147 m/s and f = 5 Hz. In the analyses of this present work anverage element size is used as 0.12 m.

Damping in a soil layer has a significant influence on itsesponse. Though there are a lot of researches about defining ofamping parameters, a commonly accepted procedure is not devel-ped. PLAXIS V8.2, damping parameters are defined by the Rayleighamping method. The Rayleigh damping coefficients and cane determined from at least two given damping ratios i that corre-pond to two frequencies of vibration, ωi. The relationship between, ˇ, ωi and i can be presented as (PLAXIS Manual [34]).

+ ˇω2i = 2ωii (7)

In this study, to determine damping parameters, acceleration

alue at surface of the backfill is recorded. Amplification ratioetween backfill surface and bottom of the tank is used to deter-ine the first natural frequencies of backfill by using FFT analysis.

he first natural frequency of backfill material is determined as

Fig. 13. Deformed mesh (

Fig. 14. Contours of total displacem

sed in FE model.

5.56 Hz. Second natural frequencies of soil layers are taken 10 Hzbased on assumption that soil response is important up to 10 Hz.Some authors such as Matasovic [35] and Lanzo and Vucetic [36]recommend the use of a constant viscous damping ratio of 1.5–4%.In the present research a value of 3% is used. The calculated and

values for backfill material are 1.25 and 0.0006, respectively.Figs. 13–15 show deformed mesh, total displacements and ver-

tical displacements respectively. It is seen that the significantdisplacements are occurred behind the blocks and this result issimilar to the results of site investigations. Additionally, maximumvertical displacement occured behind the wall is 1.3 cm. In Fig. 16,initial and final backfill surface profiles are shown. This shows thatnumerical calculation results are compitable with experimentalresults.

Soil pressure results obtained by using the 1 g shaking table testsresults and soil pressure results obtained by using the PLAXIS V8.2computer program are given in Fig. 17a–d.

Even in the case of maximum deviation of average soil pressuresbetween experimental and numerical studies for two blocks for5 Hz are 9% for SP1, 4.8% for SP2, 17% for SP3 and 4.3% for SP4 fortwo blocks (Table 12).

scaled up 5 times).

ent for two blocks for Soil 1.

H.K. Cihan et al. / Applied Ocean Research 49 (2015) 72–82 81

Fig. 15. Contours of vertical displacement for two blocks for Soil 1 (maximum vertical displacement behind the wall is 1.3 cm).

Fig. 16. Backfill surface profile before and after test.

Table 12Maximum deviation of average soil pressures between experimental and numericalstudies for two blocks for 5 Hz.

Average soil pressures (kPa)

SP1 SP2 SP3 SP4

e

n

i

ico

TC

a)

b)

c)

02468

1086420Tota

l Sat

urat

ed S

oil

Pres

sure

(kpa

)

Time (s)

SP1

1G

PLAXIS

0

2

4

6

8

1086420Tota

l Sat

urat

ed S

oil

Pres

sure

(kpa

)

Time (s)

SP2

1G

PLAXIS

0

1

2

3

4

1086420Tota

l Sat

urat

ed S

oil

Pres

sure

(kpa

)

Time (s)

SP3

1G

PLAXIS

0

0.5

1

1.5

2

1086420

Tota

l Sat

urat

ed S

oil

Pres

sure

(kpa

)

Time (s)

SP4

1G

PLAXIS

PLAXIS 8.2 5.35 4.30 2.37 0.961 g shaking table tests 4.89 4.10 2.02 0.92Max. deviation (%) 9 4.8 17 4.3

Displacements obtained experimentally and numerically at thend of 10 s are shown in Table 13.

There are some differences between experimental results andumerical results which can be attributed to:

i. the complexity of the system. Four basic elements, namely: rigidblocks, backfill, water, subsoil, are the main parameters forminga system for block type quays. When such a system is subjectedto dynamic loading, extremely complex problem is formed dueto complicated couplings between these elements and it is toodifficult to model this complexity in numerically,

i. PLAXIS V8.2 neglects the hydrodynamic force.

Although, there is no perfect similarity between obtainednstantaneous soil pressures and displacements at the end of 10 s, itan be assumed that average total soil pressures and displacementsbtained from experimental and numerical studies are compatible.

able 13omparisons of displacements results for 5 Hz.

Horizontal displacementresults (mm)

Block 1 Block 2

PLAXIS 8.2 8.0 131 g shaking table tests 6.3 11Max. deviation (%) 27 18

d)

Fig. 17. Comparisons of soil pressure cells (SP1, SP2, SP3, SP4) measurements fortwo blocks.

4. Conclusion

1 g model tests have been performed for two blocks structure. Toignore effect of seabed settlement on total damage, model is placedon rigid bed. Two types of granular soils which have different nom-

inal diameters are used as backfill materials and their effects onstructure stability are investigated. Additionally, numerical analy-sis is performed to determine soil parameters.

8 cean

s

b

d

A

eb

CA1

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

2 H.K. Cihan et al. / Applied O

Based on the experimental and numerical results, some conclu-ions are reached as follows:

a. Finer backfill (Soil 2, Dn50 = 2.2 cm, scale is 1/10) material causesmore damage than coarser material (Soil 1, Dn50 = 1 cm, scale is1/10). Soil 2 slumped down toward the structure more easilyand the space between the blocks and backfill occurring due tothe sliding of the blocks during dynamic loading can be filled bySoil 2. Thus, Soil 2 can push the blocks more strongly. The choiceof the backfill material in case of smaller peak ground accelera-tion (<0.4 g Hz) depends of the cost optimization of the materialhowever in case of regions where the seismic loading is criticalthen the choice of the coarser backfill material is recommended.

. Variation of fluctuating component of total earth pressure alongsoil depth for Soil 1 and Soil 2 is not linear. Application pointof the fluctuating components of total saturated soil pressureis obtained between 0.40 H and 0.63 H for Soil 1 (coarser) and0.375 H and 0.65 H (H is the structure height) for Soil 2 (finer).This result has a practical importance in the seismic design ofblock type quay wall.

c. Friction coefficients between block-block and block-gravel aredetermined as 0.47 and 0.56 from tilting test and 1 g shakingtest, respectively. These values are compatible with “TechnicalSeismic Specifications on Construction of Coastal and HarborStructures, Railways and Airports” [30] and “OCDI” [31].

. Physical model is modeled by using PLAXIS V8.2 software pro-gram. Numerical results are close to experimental results. Designparameters obtained in this study will be helpful for the coastalengineers in the performance based design of block type quaywall under dynamic loads.

cknowledgements

I would like to thank Prof. Dr. Yalc ın Yüksel for supporting myxperimental studies carried out at Hydraulics and Coastal and Har-or Lab., Civil Engineering Faculty at Yıldız Technical University.

I would like to thank to Scientific and Technological Researchouncil of Turkey (TUBITAK) sponsored the “Simplified Dynamicnalysis of Block Type Quay Wall” project (Project Number:11Y006) which form the basis of my thesis.

eferences

[1] Center for Civil Engineering research and Codes (CUR). Quay wall handbook.The Netherlands: CUR; 2005, 717 p., ISBN 0 415 364396.

[2] PIANC. Seismic design guidelines for port structures. Rotterdam: Balkema;2001.

[3] Yüksel Y, Alpar B, Yalc ıner AC, C evik E, Özgüven O, C elikoglu Y. Effects of theEastern Marmara Earthquake on marine structures and coastal areas. Proc InstCivil Eng: Water Marit Eng 2002;156(WM2):147–51.

[4] Karakus H (M.Sc. thesis) New seismic design approaches to block type quaywalls. Ankara: METU; 2007.

[5] Sumer BM, Kaya A, Hansen NEO. Impact of liquefaction on coastal structuresin the 1999 Kocaeli, Turkey Earthquake. In: Proceedings of the twelfth interna-

tional offshore and polar engineering conference. 2002. ISBN 1-880653-58-3(Set); ISSN 1098-6189 (Set).

[6] Sadrekarimi A, Ghalandarzadeh A, Sadrekarimi J. Static and dynamic behav-ior of hunchbacked gravity quay walls. Soil Dyn Earthq Eng 2008;28:99–117.

[[

[

Research 49 (2015) 72–82

[7] Sadrekarimi A. Seismic displacement of broken-back gravity quay walls. JWaterw Port Coast Ocean Eng 2011;137:75–84. © ASCE/March/April.

[8] Chakraborty D, Choudhury D. Stability of non-vertical waterfront retainingwall supporting inclined backfill under earthquake and tsunami. Ocean Eng2014;78:1–10.

[9] Chakraborty D, Choudhury D. Stability of non-vertical waterfront retaining wallsupporting inclined backfill subjected to pseudo-dynamic earthquake forces.Appl Ocean Res 2014;47:174–82.

10] Towhata I, Ghalandarzadeh A, Sundarraj KP, Vargas-Monge W. Dynamic fail-ures of subsoils observed in waterfront areas. Soils Found 1996;1:149–60.

11] Woodward PK, Griffiths DV. Comparison of the pseudo-static and dynamicbehavior of gravity retaining walls. J Geotech Geol Eng 1996;14:269–90.

12] Ghalandarzadeh A, Orita T, Towhata I, Yun F. Shaking table tests on seismicdeformation of gravity quay walls. Soils Found 1998;38(2):115–32 (SpecialIssue on Geotechnical Aspects of the January 17, 1995 Hyogoken-Nambu Earth-quake).

13] Zeng X. Seismic response of gravity quay walls—I: centrifuge modeling. JGeotech Geoenviron Eng 1998;124(5):406–17.

14] Madabhushi SPG, Zeng X. Seismic response of gravity quay walls—II: numericalmodeling. J Geotech Geoenviron Eng 1998;124(5):418–27.

15] Kim SR, Kwon OS, Kim MM. Evaluation of force components acting on gravitytype quay walls during earthquakes. Soil Dyn Earthq Eng 2004;24:853–66.

16] Kim SR, Jang IS, Chung CK, Kim MM. Evaluation of seismic displacements ofquay walls. Soil Dyn Earthq Eng 2005;25:451–9.

17] Choudhury D, Ahmad SM. Design of waterfront retaining wall for the passivecase under earthquake and tsunami. Appl Ocean Res 2007;29:37–44.

18] Choudhury D, Ahmad SM. External stability of waterfront reinforced soil struc-tures under seismic conditions using a pseudo-static approach. Geosynth Int2009;16(1):1–10.

19] Lee CJ. Centrifuge modeling of the behavior of caisson-type quay walls duringearthquakes. Soil Dyn Earthq Eng 2005;25:117–31.

20] Moghadam AM, Ghalandarzadeh A, Towhata I, Moradi M, Ebrahimian B, Hajia-likhani P. Studying the effects of deformable panels on seismic displacementof gravity quay walls. Ocean Eng 2009;36:1129–48.

21] Maleki S, Mahjoubi S. A new approach for estimating the seismic soil pressureon retaining walls. Sci Iran 2010;17(4):273, 284, Sharif University of Technol-ogy.

22] Na UJ, Chaudhuri SR, Shinozuka M. Probabilistic assessment for seismic perfor-mance of port structures. Soil Dyn Earthq Eng 2008;28:147–58.

23] Tiznado F, Roa R. Seismic lateral movement prediction for gravity retainingwalls on granular soils. Soil Dyn Earthq Eng 2011;31:391–400.

24] Torisu SS, Sato J, Towhata I, Honda T. 1-g model tests and hollow cylindri-cal torsional shear experiments on seismic residual displacements of fill damsfrom the view point of seismic performance-based design. Soil Dyn Earthq Eng2010;30:423–37.

25] Dewoolkar MM, Ko HY, Pak RYS. Experimental developments for studying staticand seismic behavior of retaining walls with liquefiable backfills. Soil DynEarthq Eng 2000;19:583–93.

26] Yoshimi Y, Tokimatsu K. Settlement of buildings on saturated sand duringearthquake. Soils Found 1977;17(1):23–38.

27] Iai S. Similitude for shaking table tests on soil–structure–fluid in 1g gravita-tional; 1989.

28] Karakus H (Ph.D. thesis) Experimental and numerical studies on block typequay walls under dynamic loading. Ankara: METU; 2013.

29] Westergaard HM. Water pressures on dams, during earthquake, transactions.Am Soc Civil Eng 1933;92.

30] Seismic specification for coast and harbour structures, railway, airport con-structions. Ankara, Turkey: Ministry of Transportation, Government of Republicof Turkey; 2008.

31] The Overseas Coastal Area Development Institute of Japan (OCDI). Technicalstandards and commentaries for port and harbor facilities in Japan [Goda Y,Tabata T, Yamamoto S, Trans.]. Daikousha Printing Co., Ltd.; 2002.

32] Kuhlemeyer RL, Lysmer J. Finite element method accuracy for wave propaga-tion problems. J Soil Mech Found Div 1973;99(SM5):421–7.

33] Arablouei A, Gharabaghi ARM, Ghalandarzadeh A, Abedi K, Ishibashi I. Effectsof seawater–structure–soil interaction on seismic performance of caisson-typequay wall. Comput Struct 2011;89:2439–59.

34] PLAXIS V8.2 Manual.35] Matasovic N (Ph.D. thesis) Seismic response of composite horizontally-layered

soil deposits. Los Angeles: University of California; 1993.36] Lanzo G, Vucetic M. Effect of soil plasticity on damping ratio at small cyclic

strains. Soils Found 1999;39(4):121–41.