You are on page 1of 15

DUCTILITY DESIGN OF FOUNDATIONS OF HIGHWAY BRIDGE ABUTMENTS Masahiro SHIRATO , Jiro FUKUI , and Junichi KOSEKI

Abstract Ductility design of foundations of abutments against severe earthquakes has been newly introduced in the Specications for Highway Bridges in Japan. This paper describes the summary of the ductility design method as well as its background. One of the features of this method is the adoption of a new procedure to evaluate seismic active earth pressure that is applicable up to high seismic loads based on the modied Mononobe-Okabe method. We also conduct back-analyses of the case histories of abutments including foundations damaged in the past earthquakes following the ductility design method proposed here. The results from the back-analyses conrm the applicability of the ductility design method. Introduction Right after the 1995 Hyogo-ken Nanbu earthquake (Kobe earthquake) in Japan, the Specications for Highway Bridges and commentaries were revised in 1996[1, 2]. The specication has introduced ductility design of foundations of piers as standard seismic design against severe and rare scale earthquakes. It should be noted that the aseismic design of piers based on the ductility design itself was not new, but had been already introduced before the 1995 Hyogo-ken Nanbu earthquake against large earthquakes which correspond to type I motions of level 2 earthquake motions dened in detail later. Nevertheless, abutments and their foundations have been exception in applying the ductility design even in the 1996 version of the specication, and the specication still required to check that they behave elastically against level 1 earthquake motions which are small to middle (or frequent scale) earthquakes as it had demanded in the past. The reasons for this judgment are the following. First, because of the lack of knowledge on a proper evaluation method of their performances, including the seismic active earth pressure at high seismic loads, it is dicult to establish any verication methods of seismic performance of abutments and their foundations against severe earthquakes. Second, there has been no case history in Japan where the damage to abutments or their foundations causes unseating of girders. Third, a bridge abutment is a structure resisting against the earth pressure exerted from the backll at any moment, and thus it would be just pushed forward by the seismic
Research Engineer, Foundation Engineering Research Team, Public Works Research Institute Team Leader, ditto. Associate Professor, Institute of Industrial Science, the University of Tokyo

earth pressure even if it suers residual displacements during earthquakes, which would not directly result in the unseating of supporting girders. However it is needed to develop an aseismic design method of abutments and their foundations against large earthquakes, which controls their damage within an acceptable extent to recover the service of road network as early as possible after the earthquakes. After publishing our 1996 version of the specications, we have constructed a database of past earthquake damage to abutments or abutment foundations in Japan, and have studied on the assessment methods which can distinguish the dierent damage extents for the case histories in our database. Separately, Koseki et al.[4] have recently proposed an evaluation method of active earth pressure at high seismic loads, and it enabled us to establish a computational method to assess seismic performance of abutment foundations. Consequently, we could newly introduce the ductility design of abutment foundations into the latest version of the Specications for Highway Bridges revised in March 2002. The design method is based on the preceding frame work of the ductility design method of bridge pier foundations, while some modications are implemented to it in order to reect the above-mentioned peculiar mechanical characteristics of foundations of abutments. This paper describes the verication procedure and the process of the development of the new ductility design method of foundations of abutments. Fundamentals of seismic design of abutments and their foundations A two stage aseismic design procedure is applied on the basis of the combinations of the level 1 and level 2 earthquake motions and corresponding requirements of structural performance. The level 1 earthquake motions are likely to strike a structure once or twice during the expected service period of the structure. Their peak amplitudes are small to medium, and are around 0.2 G at the ground surface. On the other hand, the level 2 earthquake motions are extremely strong, but very unlikely to strike a structure during its service period. The level 2 earthquake motions include two types of motions. One is called type I motions which are generated at plate boundaries in the ocean. Their peak amplitudes at the ground level are smaller than the other one although the type I motions have longer durations. Their peak amplitudes are about 0.3 G to 0.4 G at the ground surface. The other motions are inland strike type motions called type II which are produced by a fault located near the site; for example the 1995 Hyogo-ken Nanbu earthquake is categorized in this type. The type II motions have high intensities but short duration. Their peak amplitudes are set at 0.6 G to 0.8 G at the ground surface, based on the acceleration records on the ground surface observed in the 1995 Hyogo-ken Nanbu earthquake. In the new aseismic design procedure, all abutments and their foundations are veried against the level 1 earthquake motions so that they would cause no structural damage. The abutment and its foundation shall remain in the elastic response region.

Next their seismic performance against the level 2 earthquake motions is checked. However this process is not necessarily applied to all highway abutments and their foundations. For only the foundations to be constructed on/in liqueable ground, the verication against the level 2 earthquake motions shall be conducted. In this case, the foundation shall have the necessary strength and ductility to fulll the structural requirements as follows: the damage to the foundation shall be limited within a level where it can be repaired with reasonable cost and shall not cause a state where rescue operation of the bridge is not available. These requirements refer to those for pier foundations which undergo eects of liquefaction. Consequently, the ductility design against the level 2 earthquake motions is applied only to deep foundations in liqueable ground. For abutment bodies above their foundations, verication may be omitted against the level 2 earthquake motions even in the new specication as well as in the former specications. This is because an abutment which satises the performance requirement for the level 1 earthquake motions is considered to posses enough strength and ductility against the level 2 ones. This can be also supposed by the past damage case histories. It should be noted that some structural details need to be specied in order to supplement skipping the assessment of the seismic performance against the level 2 earthquake motions. The new specication requires to arrange lateral conning reinforcements in the abutment body so as to improve its ductility. It also demands to include the same amount of longitudinal reinforcements on the front side of the cross section of the abutment body as those on its back side in order to prevent excessive bending failure and diagonal tensile failure caused by contacting of the abutment with the deck. Survey of case histories of damage to abutments We made a list of damaged abutments including foundations from the past damage reports of recent large earthquakes in Japan. We selected typical cases from the list as is summarized in Table 1. Table 1 exhibits contents of damage, height of abutment from the bottom of pile cap, bearing condition, and equivalent thickness of liqueable soil strata, which is explained below, for each case. Note that we here picked up only case histories damaged by the type II earthquake motions to simplify the comparison among these cases, because the behavior of abutments, foundations, and sub-soil layers would highly depend on characteristics of seismic motions, e.g. the number of cycles, duration time, and the intensity of earthquakes. All abutments that were nally picked up are supported by group-pile foundation. Note also that pile foundations are employed for about a half of the highway bridge foundations in Japan, in particular those are constructed in deep soft or deep loose soil layers. We judged the rank of damage based on the contents of damage for each case. In

Table 1 List of past damage case histories of abutments and computed nonlinear response of their pile foundations
Case Earthquake* Rank of damage** Damage Height of abutment (m) 9.8 bearing condition Equivalent thickness of liqueable soil strata HE (m) 6.8 Response ductility factor

D E F G

1 1 1 1

3 2 2 2

I J

1 1

2 2

K L

1 1

1 1

- spalling of concrete around anchor bolts of bearings - movement of abutment (10 cm) - tilting of abutment, shear collapse of wingwall - slumping of backll (1.0 m) - tilting of abutment - cracks in parapet wall - outward movement of foundation (11 cm at the top of the foundation) - outward movement of abutment - slumping of backll - slumping of backll (20 cm) - collapse of bearings - damage of abutment - cracks in the surface of approach road - cracks in parapet wall - bump at the connection part between abutment and backll - cracks of abutment - collapse of side-blocks of bearings - exchange of bearings in repair works - slight slumping of backll - slumping of backll - shrinkage of spacing of expansion joint - excessive movement of bearing - bump in backll - a little spalling of concrete of abutment - shrinkage of spacing of expansion joint - cracks in parapet wall - bump in backll - shrinkage of spacing of expansion joint

move

3.2

move

19.7

4.2

3.3

move

18

not yield

5.5 8.5 8.8 10.3

x move move move

13.4 3 3.2 6.9

1.8 not yield 1.4 1.1

move

6.5 12.3

x x

1 0

1.2 2.9

6.6 7.5

x move

11.9 4.7

1.9 1

0.5

1.3

* 1: the 1995 Hyogo-ken Nanbu earthquake, 2: the 2000 Western Tottori earthquake. ** Refer to Table 2

doing so, we set a four-tier criterion for the damage rank emphasizing whether the service of the bridge was available or not right after the earthquake. The larger the number of the rank, the more disastrous the damage. The relation among the ranks, serviceability and repairability, and details of damage is displayed in Table 2. The serviceability indicates the eect of the damage to the abutment or its foundation on the service or function of the bridge right after the earthquake, and the repairability represents whether the complete recover of the abutment and its foundation was possible by repair work with reasonable cost or not. Equivalent thickness of liqueable soil strata H E is dened by
+ H2 H E = H1

Table 2 Categorization of the degree of damage


Rank of damage Degree of damage Serviceability 1 slight Fully operational 2 medium to large Operational with some restrictions w.r.t weight of vehicles and speed limit Possible with minor repair works - slumping of backll No operation temporarily while doing emergency countermeasure works** Possible with major repair works - horizontal movement or rotation of abutment - excessive slumping of backll - collapse of parapet wall 3 4 severe Impossible

Repairability Typical damage contents

Easy* - shrinkage of spacing of expansion joint - cracks of parapet wall

- cracks of structural members

Impossible (Reconstruction) - excessive horizontal movement or excessive rotation of abutment - collapse of structural members

* e.g., with xing slight cracks ** e.g., operational with some restrictions after constructing temporary bents

H1 = 1.5HFL1 + 1.0HFL2 + 0.5HFL3 (0 m z 10 m), H2 = 1.0HFL1 + 0.5HFL2 (10 m z 20 m), where z is the depth of soil layer from the ground level, H 1 is the thickness of liqueable soil strata estimated for strata in between z = 0 m and z = 10 m, and H 2 is the one estimated for strata in between z = 10 m and z = 20 m, where the origin of the ground level GL is set at the bottom of pile cap. H FL1 is the sum of the thicknesses of the soil strata for which the values of factor of safety against liquefaction F L are assessed to be less than or equal to 0.6, HFL2 is the sum of those with 0.6 < F L 0.8, and HFL3 is the sum of those with F L > 0.8. This index has been proposed as one of indices in a past guideline for the assessment of earthquake resistance of existing bridges. The equivalent thickness involves the eects that the shallower the location of the liqueable soil stratum is, the larger the impact of the liquefaction of that stratum is on the horizontal bearing capacity of the foundation against loads acting on its top, and that the smaller the value of F L is, the larger the loss of soil stiness is. In this study, we estimated the values of F L by following the procedures for the type II seismic motions described in the currently eective specication[5].

An example of the cases with the damage rank 4 is shown below. The case B in Table 1 is the abutments of Higashi Uozaki bridge which collapsed by the 1995 Hyogo-ken

constraint by contact with deck

tensile force
residual horizontal displacement

earth pressure

loss of subgrade reaction liqueed soil strata


Photo 1 Failure of abutments of Higashi Uozaki bridge[3] Figure 1 Schematic mechanism of damage to abutments of Higashi Uozaki bridge

Nanbu earthquake[3]. This bridge passed over a canal, and had a total length of 64.8 m with three spans. The abutments were cantilever-type walls with pile foundations consisting of H-shaped steel piles. Photo 1 shows the failure of the A1 abutment. The A1 abutment rotated backward and large cracks widespread on the front face of the abutment. Besides, in the A2 abutment wall, shear collapse with widely opening cracks on the front face was also observed. Based on the observation that the backll of the abutments slumped, we estimated that the abutments suered residual horizontal displacement. The subsoil layers consist of loose or soft soils, namely slime, sand, silty sand, and clay with SPT-N blow counts of zero to six till 14.7 m deep from the river bed. Sand boils caused by liquefaction were observed, and the revetment suered outward lateral movement in the direction to the canal. As is illustrated in Figure 1, the backward rotation of the abutment accompanied with a lot of cracks on its front face may have occurred as follows: 1) the liquefaction of the subsoil layers resulted in the shortage of horizontal subgrade reaction to support the foundation; 2) the foundation was pushed outward by the earth pressure acting on the back face of the abutment and the inertial force of the abutment itself; 3) the decks constrained the above lateral movement at the top of the abutment; 4) the foundation and the lower part of the abutment continued to move laterally with the above constraint, and a large tensile force by bending was mobilized at the front face of the abutment. The cause of damage to A abutment in Table 1 which is categorized into the rank 4 is

also the lateral movement of foundation. It was possibly triggered by loss of horizontal subgrade reaction induced by liquefaction of subsoil layers. Figure 2 shows the relation between the damage rank and the equivalent thickness of liqueable soil strata for the abutments listed in Table 1. Although no unique relation except for some positive correlation can be found, this gure implies that there is a possibility of severe damage (the rank 4) if the equivalent thickness of liqueable soil strata exceeds ve meters. These case histories suggest that liquefaction of subsoil layers is the primary cause of damage to abutments and their foundations. Based on the damage mechanism estimated above, it is expected that we can prevent abutments from severe damage as long as the possible displacement of the foundation of an abutment during an earthquake is well-controlled even though liquefaction of the subsoil layers may occur. Therefore the ductility design of abutments is applied to only their foundations, and the specication requires us to carry out it when the subsoil layers are considered to be liqueable. Seismic loads employed in the ductility design method of foundations of abutments In the ductility design method, eects of earthquakes are modeled with pseudostatic loads, namely seismic active earth pressure acting on the back face of the backll located on footing, inertial force acting on the backll, abutment itself (wall), and footing, and horizontal reaction from deck as shown in Figure 3. It would be reasonable to evaluate the intensities of these pseudo-static loads based on the characteristics of their dynamic responses considering the eects of liquefaction. However the quantitative evaluation of these eects is not well established. Thereby, we set practically the intensities of these seismically induced loads as follows. Note that we do not take the vertical component of seismic motions into account in accordance with the customary practice. The design horizontal seismic coecient for evaluating seismic active earth pressure and the inertial force of backll located on footing is obtained as c A khg , where khg is the design horizontal seismic coecient at ground surface, and c A is the modication factor for the amplication of acceleration in the backll as well as in the adjacent embankment. For simplicity, we assume that c A = 1 here. In Figure 3, khA denotes the design horizontal seismic coecient to determine the inertial forces of abutment and footing. We also assume that khA = cA khg here, because the behavior of abutments during earthquakes would be dominated by that of the backll and embankment. The coecient to be multiplied by horizontal reaction force from deck in Figure 3 represents the eect of phase dierence between the horizontal response of superstructure and that of embankment. However, we adopt that = 1 here for simplicity, and assume the direction of the horizontal reaction force from deck to be forward to result in more severe load conditions for abutment

4 HE = 5 m Rank of damage 3

reaction force from deck khA cA khg

cA khg failure plane

khA earth pressure

1 0 10 20 Equivalent thickness of liqueable soil strata HE (m)

Figure 2 Comparison between equivalent thickness of liqueable soil strata and the Figure 3 Schematic gure of numerical model rank of damage

foundations in practical design. Seismic active earth pressure The Mononobe-Okabe method (denoted herein as M-O method) is well-known and widely accepted in a lot of design specications or manuals as an evaluation method of seismic active earth pressure. Examples of seismic active earth pressure coecients obtained by the M-O method are shown in Figure 4a with dotted lines, where k h is horizontal seismic coecient; K EA is seismic active earth pressure coecient; is the internal friction angle of embankment soil; is the friction angle at the interface between backll located on footing and embankment; and we assume the surface of embankment to be horizontal and the back face of backll on footing to be vertical. Designers usually employ = 30 to 40 degrees as the internal friction angle of embankment soil in the evaluation of earth pressure. These values of are equivalent to residual strength of dense or relatively hard soils in general. As their representative value, = 35 degrees was employed in Figure 4. The levels of khg , the design horizontal seismic coecient, at the ground surface are 0.6 to 0.8 in the ductility design, while it can be seen from Figure 4a that the values of K EA with = 35 degrees cannot be obtained at such levels. Moreover, we examine the ratio of failure zone length L as dened in Figure 4b to height H of abutment from the bottom of footing. From the viewpoint of design practice, the failure zone length L indicates the zone where settlement of road connecting to bridge may occur. The relations between L/H and kh obtained by the M-O method are plotted with dotted lines in Figure 4b. When the value of kh reaches 0.6 to 0.8, the values of L/H predicted by the M-O method with = 35

KEA (a)
Backll 2

Interface

Embankment modied M-O peak = 50 res = 35 failure plane

L/H
4

(b)

L H M-O method = 50 = 35 /2

failure plane 1 resultant force of earth pressure M-O method = 35 = 35 /2 b 0 0


a

c M-O method = 35 = 35 /2 a b modied M-O method peak = 50 , res = 35 0.5 1

present
M-O method = 50 = 35 /2 1

0.5

0 0

kh

kh

Figure 4 Seismic active earth pressure and failure zone of embankment (a, b, c: formation of rst, second, and third failure planes, respectively, in modied M-O method)

degrees become unrealistically large. Now we recall the behavior of dense sand obtained by an element test (plane strain compression test). Firstly the soil element is subjected to the process of strain hardening; and then the element comes to mobilize the peak strength, and a shear band (or failure plane) starts to emerge progressively; after that, strain softening process evolves, accompanying with the development of the shear band; nally the residual strength is mobilized. Since backll on footing of abutment and embankment soils are well-compacted, it would be reasonable to take this behavior into the estimation of seismic active earth pressure. According to the above behavior, one can consider the following mechanism on the formation of failure planes in the embankment soil and the mobilization of active earth pressure: the angle of failure plane is associated with the peak internal friction angle of embankment soil, because the peak strength is mobilized in the soil elements when the shear banding starts, and the value of active earth pressure is associated with the residual internal friction angle, since the postpeak reduction of internal friction angle evolves after the formation of shear band (or along the failure plane that has been already formed). Based on the above hypotheses, the modied Mononobe-Okabe method has been proposed by Koseki et al.[4] Figure 4a also displays an example of the values of K EA calculated by the modied M-O method with dash lines, where we assume the peak internal friction angle peak to be 50 degrees and the residual one res to be 35 degrees; we also assume that the rst failure plain is formed at kh = 0; and we set the values of to be peak /2 until the formation of the

rst failure plane (at kh = 0) and to be res /2 after that, as the value of at the interface between soils has been empirically set to be /2 in applying the original M-O method. In the modied M-O method, after the formation of the rst failure plane, the same failure plane continues to be activated unlike the result from the original M-O method, because the peak strength will be mobilized along any other failure plane in the embankment excluding the rst failure plane along which the residual strength is mobilized. However, when the seismic coecient kh is increased to a certain level, another failure plane will be activated, because it gives the extremum of seismic active earth pressure, and at this level the second failure plane is formed. The third, fourth, and other consecutive failure planes will be formed in a similar manner. The modied M-O method has a couple of advantages in comparison with the original M-O method as can be seen in Figure 4. One is that the modied M-O method can provide active earth pressure coecient at large earthquake loads, even though the values of kh get to 0.6 to 0.8 which are considered in the seismic design against the level 2 earthquakes. Another advantage is that the modied M-O method provides considerably smaller and much realistic size of active failure zone in embankment than the conventional method yields. In not only model test results on retaining structures but also case histories of actual railway retaining walls damaged in the 1995 Hyogo-ken Nanbu earthquake, the observed angle of failure planes formed in backll cannot be explained without introducing the peak strength of soils, as has been reported by Koseki et al.[7, 8] Consequently, the modied M-O method has been introduced into the new specication. We, however, have given some engineering judgments to the original paper by Koseki et al.[4] in order to make the modied M-O method more suitable for practical use. In the modied M-O method, we have to estimate the level of k h at the formation of the rst failure plane. Since, at this moment, no reasonable method to evaluate it has been established, we give zero value to the level of kh at which the rst failure plane emerges. This is because a failure plane can be formed due to a slight displacement of foundation under the working load conditions. Next, we assume that no more failure planes appear after the second failure plane is activated. This is because when the third failure plane appears, the embankment should have already deformed largely, and it is considered that the application of limit equilibrium approaches would be dicult at such a level. Thirdly, although we can derive both the rst and second failure planes, we take only the second failure plane into account in estimating the seismic active earth pressure coecient K EA values for all the values of kh , because the second failure plane would be usually active at the ground acceleration level of 0.6 to 0.8 G corresponding to the level 2 earthquake motions. From the viewpoint of practical design, this assumption is reasonable, since this gives us the values of K EA close to those which have been used in the past design practice at the ground acceleration levels of around 0.2 G corresponding to the level 1 earthquake motions. Hence the dimensions of abutments and their foundations can become similar to

those based on the past design practice. The solid straight line in Figure 4 shows the results obtained by applying these engineering judgments to the results from the modied M-O method. The K EA values can be formulated as a linear function with respect to the horizontal seismic coecient k h . Around the acceleration level of 0.2 G, the present method yields the values of K EA close to those estimated by the original M-O method with res = 35 degrees. Eventually, in the newly revised specication, we have adopted the following equations to evaluate the seismic active earth pressure coecients, KEA = 0.22 + 0.81kh KEA = 0.26 + 0.97kh for dense sand and gravel for dense sandy soil (1) (2)

where the surface of embankment is supposed to be horizontal. In obtaining the above equations, it is assumed that the internal friction angles of dense sand and gravel are peak = 50 degrees and res = 35 degrees, and those of dense sandy soil are peak = 45 degrees and res = 30 degrees. In addition, as mentioned before, it is also assumed that = peak /2 until initiating the rst failure plane and = res /2 after that. Estimation of inelastic response of foundation system It can be easily understood that displacement of the foundation of an abutment would accumulate to one-way direction during an earthquake especially once the foundation reaches yielding, because of the earth pressure acting on backll on footing at any moment. Many experimental studies have revealed this type of behavior. However, it is dicult to predict the progressive movement of the foundation of an abutment since nonlinear hysteresis characteristic of foundation as well as that of seismic active/passive earth pressure is not well-known. In the ductility design method of foundations of abutments, we employ pushover analysis in order to grasp collapse behavior as a global foundation system, and, in addition, we apply so-called energy conservation method to the result from a pushover analysis so as to estimate the possible response ductility factor of the foundation system. As shown in Figure 5, the relation between horizontal seismic coecient k hA and displacement of the top of an abutment is employed for an evaluation of a nonlinear response of the foundation by the energy conservation method. Before carrying out the pushover analysis, we apply vertical dead loads and the active earth pressure that is evaluated at kh = 0 in Eqs. (1) and (2) to the abutment in order to derive the initial displacement of the foundation i . The reasons why we employ the pushover analysis and the energy conservation method are as follows. Firstly it may not be necessary to utilize sophisticated methods,

kh (=acc./g) khA yield point r =

r y

acc.:

g: r : y : r :

horizontal acceleration the acceleration of gravity response ductility factor yield displacement response displacement

r i y initial displacement

Figure 5 Estimation of response ductility factor

since the reliability and accuracy of them in predicting the dynamic behavior of abutment systems including foundations and embankments are not well-known. Secondly we can point out the characteristics of the intensity and duration of the design earthquake motions. When we consider the type II seismic motions which have a few number of cycles with high intensity and short duration, like those of the 1995 Hyogo-ken Nanbu earthquake, the displacement assessed by means of the energy conservation method employing their peak horizontal acceleration would be predominant, and the accumulation of the progressive movement may be omitted. For example, the energy conservation method is valid theoretically for an elasto-plastic response of a single DOF system undergoing an impulse input motion, which is relatively similar to the type II seismic motion. On the other hand, the type I seismic motions have a large number of cycles. However the acceleration intensity of each cycle is less than those of the type II seismic motions as has been explained in the section of Fundamentals of seismic design of abutments and their foundations. Thereby, it is not dicult to achieve enough strength of the foundation of an abutment against the type I earthquake motions. In other words, there are few possibilities of the occurrence of excessive residual displacement against the type I seismic motions with a lot of cycles as long as we assure the response ductility factor of the foundation to be small enough by providing an adequate strength against the type II motions. Based on the above consideration, in the newly revised specication, the verication against the level 2 earthquake motions is required only against the type II earthquake motions. Back-analysis of case histories and determination of allowable ductility factor The allowable degree of damage for foundations of abutments would be essentially the same as that for foundations of piers. With respect to group-pile foundations of piers, the specication recommends that a response ductility factor should not exceed four. This recommendation is based on results from experimental studies on large scale models of group-pile foundations subjected to cyclic loads, such as those conducted by PWRI[9, 10]. However, as has been discussed in the previous section, the behavior of foundations

of abutments is more complicated than that of foundations of piers. Hence, in order to conrm the applicability of the design method and to determine the allowable ductility factor empirically, we conducted back-analyses of case histories on the thirteen abutment foundations as listed in Table 1 following the design method proposed above and compared the results with their actual damage ranks. Modeling of foundation is made based on the relevant specication[6, 9, 10]. A summary of the employed models is illustrated in Figure 3. Nonlinear properties in terms of the bending of piles are considered by evaluating their moment-curvature relations. The resistance of pile against the axial force is modeled as an elastic-perfectly plastic bilinear model by setting its yield points at the ultimate bearing capacity and the ultimate pullout capacity, respectively, against the compression and pull-out forces. The horizontal subgrade reaction is modeled by using a distributed spring having a bilinear property. Moreover, these back-analyses employ the following assumptions. First, we estimate the ground acceleration and the degrees of the reduction of soil stiness caused by liquefaction based on the specication[5] for the type II earthquake motions. Second, for several foundations of which the details of the arrangements of reinforcement bars are unknown, we carry out re-designs based on the specications that were eective when they were constructed. Third, we assume that all the embankments at the back of the abutments are with good quality and well compacted, and hence we set the internal friction angles of the embankment to be 50 degrees for the peak strength and to be 35 degrees for the residual strength. To simplify the back-analyses, we regard the yield point shown in Figure 5 for the relation between horizontal seismic coecient kh and displacement at the top of an abutment as the state when all piles reach yielding or when a compressive reaction force at a pile top reaches its bearing capacity. These criteria are suggested in the specication, and usually yield almost the same point as the one obtained by the so-called log k h -log analysis. Furthermore we assume the kh - curve as an idealized elastic-perfectly plastic bilinear curve as shown in Figure 5. The computed response ductility factors are listed in Table 1, and are compared with the actual damage ranks in Figure 6. Note that two abutments do not reach yielding in the pushover analysis. In such cases, the response ductility factors are set as one in Figure 6. It is seen that a foundation can avoid severe damage categorized as the rank 4 when it is designed to yield the value of response ductility factor less than two or three. In other words, if the foundation of an abutment is given a capacity with the value of response ductility factor being less than two or three and if the foundation consists of well-designed members, it would satisfy the required seismic performance against the design earthquake motions. Although all the foundations analyzed herein are pile foundation, we consider that

Rank of damage

this result can be applicable to any other type of deep foundation. Therefore, the newly revised Specications for Highway Bridges recommend that the limit value of response ductility factor of foundations of abutments should be r = 3.

2 However, we have to realize that some limitations still exist with respect to our proposed assessment method. For instance, the value of response ductility factor 1 for the case C abutment listed in Table 1 1 2 3 4 Response ductility factor is one (i.e. no yielding) although its damage is classied into the rank 3. This abutFigure 6 Comparison between computed rement was damaged in the 2000 Western Totsponse ductility factor and the rank of tori earthquake[11]. It had supported a deck actual damage built over a canal with a simple single span of 17.4 m. The canal had a revetment consisting of a leaning-type retaining wall using concrete blocks. According to the damage survey report[11], it is conrmed that the surrounding ground liqueed and spread laterally, causing a residual displacement of the revetment over 1 m. The equivalent thickness of liqueable soil strata for this case is much larger than that for the other cases, so that we can suppose that this foundation was damaged by the lateral spreading of the ground down to its deeper position. Because the eect of lateral spreading of the ground is not considered in the assessment method employed here, the computed result does not coincide with the actual damage for this case.

Concluding remarks This paper explains the ductility design of abutment foundations of highway bridges and its background, which has been newly introduced in the Specications for Highway Bridges revised in 2002. The design method developed here is a simple and practical one, and its applicability is conrmed based on results from back-analyses of damage case histories. However some limitations still remain to be improved, such as those on the load-combinations, the computation procedure of progressive movement of foundations of abutments, and the safety evaluation method of abutment walls. A study on the damage control design against the lateral spreading of ground induced by its liquefaction is also required. We hope that this paper contributes to the progress of seismic design of abutments and their foundations.

Acknowledgment This research was performed for the revision of the Specications for Highway Bridges in Japan. We particularly acknowledge the Working Group for Seismic Design of Abutments chaired by Dr. M. Saeki set in the Committee for Substructures chaired by Dr. M. Okahara established in the Japan Road Association for their comments and supports to our work. References
[1] Japan Road Association (1996): Specications for Highway Bridges, Part V, Seismic Design, Maruzen, Tokyo (in Japanese, but translated into English and edited by Unjoh, S. and Terayama, T. as the technical memorandum of EED, PWRI., No. 9801, 1998.) [2] Japan Road Association (1996): Specications for Highway Bridges, Part IV, Substructures, Maruzen, Tokyo (in Japanese). [3] Editorial Committee for the Report on the Hanshin-Awaji Earthquake Disaster (1996): Report on the Hanshin-Awaji Earthquake Disaster, Damage to Civil Engineering Structures, Bridge Structures, JSCE, Tokyo (in Japanese). [4] Koseki, J., Tatsuoka, F., Munaf, Y., Tateyama, M., and Kojima, K. (1998): A modied procedure to evaluate active earth pressure at high seismic loads, Special Issue of Soils and Foundations, pp. 209-216. [5] Japan Road Association (2002): Specications for Highway Bridges, Part V, Seismic Design, Maruzen, Tokyo (in Japanese). [6] Japan Road Association (2002): Specications for Highway Bridges, Part IV, Substructures, Maruzen, Tokyo (in Japanese, but the translated into English and edited one will be published in 2003 as a technical memorandum of the Foundation Engineering Research Team, PWRI.) [7] Koseki, J., Munaf, Y., Tateyama, M., Kojima, K., and Horii, K. (1999): Back analysis of case histories and model tests on seismic stability of retaining walls, Eleventh Asian Regional Conf. on Soil Mech. and Geotech. Eng. (ed. Hong et al.), pp. 399-402, Balkema, Rotterdam. [8] Koseki, J., Watanabe, K., Tateyama, M., and Kojima, K. (2001): Seismic earth pressures acting on reinforced-soil and conventional type retaining walls, Landmarks in Earth Reinforcement (ed. Ochiai et al.), Swets & Zeitlinger (Balkema), Vol. 1, pp. 393-398. [9] Nakano, M. and Kimura, Y. (1996): Seismic design of pile foundation, Proc. of the Third US-Japan WS on Seismic Retrot of Bridges, Vol. 3, pp. 343-355. [10] Fukui, J., Kimura, Y., Ishida, M., and Nanazawa, T. (1997): Strength and ductility characteristics of highway bridge foundations, Proc. of 29th UJNR, Technical memorandum of PWRI, No. 3524, pp. 567-582. [11] Sasaki, T., Matsuo, O., Kobayashi, H., Watanabe, T., and Maeda, T. (2001): A back-analysis of the eects of the lateral spreading of the ground on the foundation of an abutment damaged in the 2002 Western Tottori earthquake, Proc. of 56th Annual Conf. of JSCE, III-A, pp. 230231 (in Japanese).

You might also like