Anda di halaman 1dari 10

Proceedings of Indian Geotechnical Conference December 22-24,2013, Roorkee

FULLY INSTRUMENTED FULL-SCALE EMBEDDED CANTILEVER SHEET PILE RETAINING WALL: PLAXIS FE MODELING AND INTERPRETATION
A. Dey Assistant Professor, IIT Guwahati, arindam.dey@iitg.ac.in C. Rainieri CEO, S2X, Termoli, Molise, Italy, carlo.rainieri@gmail.com G. Fabbrocino Professor, University of Molise, Termoli, Molise, Italy, giovanni.fabbrocino@gmail.com F. S. de Magistris Professor, University of Molise, Termoli, Molise, Italy, filippouni1@gmail.com
ABSTRACT: This article reports the PLAXIS Finite Element (FE) modeling of a fully instrumented full-scale embedded cantilever sheet pile retaining wall. The basic model developed on the basis of preliminary soil stratigraphy was then refined based on the experimental results from dynamic tests. The final FE model, formed through a model updation process based on an inverse approach, aids in understanding the soil-structure interaction mechanisms. A superposition principle has been used to obtain the response of the overall system from the decoupled responses of different domains. The achieved paves way for future investigations about the seismic behavior of the system.

INTRODUCTION Deep excavations, bridge abutments or harborquays are usually supported by means of rigid or flexible retaining walls, such as gravity, cantilever or embedded walls. Their behavior under static conditions has been investigated in detail over the years; although, their dynamic response and the soil-structure interaction mechanisms are not thoroughly investigated. When subjected to a dynamic excitation, the different components of the retaining system exhibit complex and interdependent responses, which can be roughly categorized as (a) the dynamic interaction between the wall, the retained backfill and the soil in front of the wall, and (b) the dynamic interaction of the underlying soil with the sustained structure. The above complexities are significantly augmented in the presence of material and/or geometric nonlinearities [1]. Depending on the expected material behavior of the retained soil and the possible mode of wall displacement, design of retaining walls subjected to dynamic excitations is generally carried out by the following two approaches: (a) The classical Mononobe-Okabe approach, where the dynamic earth pressures are computed from pseudo-static limit equilibrium, considering the yielding of walls due to the plastic behavior of the retained soil [2].

However, this type of analysis fails to address the evidences of rocking, bearing and dynamic effects, commonly encountered in the case of seismic excitation. (b) The Subgrade Reaction analysis, which allows the computation of elastic and plastic deformations of the soil. However, the reliability of the obtained results largely depends on the accuracy of the chosen stress-strain relation of the springs used to model the soil [3]. These methods, though commonly used, fail to provide satisfactory prediction of the overall response of a retaining structure subjected to a dynamic excitation. Parameters influencing the behavior of the system, such as damping, natural frequencies of the system, phase differences and amplification effects within the backfill, remain unaccounted in the above theories. A pseudo-dynamic approach, based on the assumption of a finite speed of elastic shear-wave propagation, results in more accurate prediction of the behavior of the retaining system. Such an analysis reveals that the distribution of earth pressure and the point of application of the dynamic thrust are primarily governed by the dynamic properties of the backfill [4]. This approach, although commonly used to analyze gravity walls, may be unreliable for flexible embedded walls. In fact, it is very difficult to define the distributions of active and passive earth

A. Dey, C. Rainieri, G. Fabbrocino, F. S. de Magistris

pressures for sheet-pile pile walls. Moreover, the inertia force and the bottom tom friction at the wall, the major parameters for the analysis of a gravity wall, are not contributory parameters for an embedded retaining wall. As a solution, dynamic analyses, taking into account soil-structure structure interaction, by means of finite-difference e or finite finite-element methods, represent the most effective techniques to predict the response of a flexible retaining wall. Dynamic analyses are carried out by applying dynamic excitations at the base of the domain. Depending on the intensity of the input e excitation relative to the elastic limit, the structure is modeled as linear or nonlinear. The soil is idealized by either an equivalent linear or an effective stress model, depending on the expected strain level in the soil deposit during the induced dynam dynamic motion. However, to achieve a reliable prediction of the wall behavior, an extensive soil characterization through in-situ situ and laboratory investigations, a proper constitutive model for soil and a precise definition of the dynamic excitation are required. d. Information provided by Structural Health Monitoring (SHM) systems can enhance the level of knowledge about retaining systems, and useful hints for their accurate numerical modeling can be obtained. Both operational and earthquake data are relevant to enhance nhance numerical models and obtain more accurate predictions and improve the current trend of practice and seismic design procedures. After a literature review about the use of dynamic analysis for different types of retaining walls, attention is herein focused ocused on the dynamic response of an embedded retaining wall. Various pertinent issues referring to the main uncertainties and steps in the development of the finite element model, and its progressive refinement based on monitoring data are discussed. Attention ntion is also focused on the soil-structure structure interaction mechanisms, the flexible embedded retaining wall being acting as the boundary between non-uniform non geometry sections. The key role for the enhancement of the numerical model played by ambient vibration measurements is described in detail, leading to the development of an updated model, representative of the behavior of the system

in operation and ready to be further enhanced to explore the response of embedded retaining walls during and after an earthquake. STRUCTURAL AND GEOTE GEOTECHNICAL CHARACTERIZATION OF THE WALL The monitored reinforced concrete embedded cantilever retaining wall is a part of the new Student House at University of Molise in Campobasso, approximately 200 km SE of Rome. It is an embedded sheet pile wall, composed by two alignments of adjacent but not contiguous piles. Basic geometric data and the schematic of the arrangement of the piles for the reported wall are depicted in Fig. 1 and 2 respectively respectively, while a complete description is reported orted elsewhere [ [5]. Top beam Instrumented Piles 6.4 m 4.2 m

Pile + Plate for Student House Fig. 1 Pictorial view of cantilever sheet pile wall for the Student House University of Molise

Fig. 2 Schematic of the arrangement of piles in the sheet pile wall system

Fully instrumented full-scale scale embedded cantilever cantil sheet pile retaining wall: Plaxis FE modeling and interpretation

The area where the wall is located is characterized by a medium to high seismic hazard hazard. From a geological point of view the site is characterized by deposits of varicolored scaly stiff clays, with alternate beds of limestone, calcareous marls and sandy materials, with supplementary presence of calcarenite ite and fragments of San Bartolomeos flysch. The top soil is demarcated as a remolded debris cover of anthropogenic origin, the thickness varying between 4 and 10 m. The geological map of the area is shown in Fig. 3.

counts as obtained from boreholes S5 and S6 respectively. Down-hole hole tests carried out in the two boreholes allowed the evaluation of the primary and shear wave velocities (Vp, Vs) of the different strata along the borehole. This information leads to estimates of the Poissons ratio, modulus of elasticity and s shear modulus under dynamic conditions according to the standard expressions:
=
0.5 V p Vs

(V

Vs

;E =

.V p2

(1 + )(1 2 ) ; G =
1

E 2 (1 + )

(1)

The variation of the shear wave velocity with depth of the borehole is reported in Figures 7 and 8 respectively.

Fig. 3 Geological map and section of o the Vazzieri locality The geotechnical characterization of the site is primarily based on two borehole investigations (S5 and S6) carried out on either side of the location of the retaining wall. Fig. 4 shows the schematic location of the boreholes with h respect to the retaining wall on a contour plot of the nearby vicinity. The arrow indicates the general direction of the slope in the locality of the investigated area. The borehole investigations have been carried out up to a depth of 30 m from the ground grou level. Stratigraphic Column Extractions, Standard Penetration Tests (SPT) and Down-Hole Hole tests (DH) have been executed. The extracted samples have been subjected to laboratory investigations to determine the physical characteristics. The strength parameters ers of the soil samples were determined by triaxial and direct shear tests. Figs 5 and 6 show the different soil layers as identified according to the stratigraphical characteristics and the t SPT blow

Fig. 4 Location of investigative tive boreholes in the site As indicated in Fig. 4, , since the boreholes are located along the general direction of the slope in the area, there exists a grade difference of about 8 m at the ground level between the two. The excavation of the borehole S6 started s from an absolute height of about 676 m, while that of borehole S5 commenced from nearly 668 m. The boreholes were separated by a longitudinal distance of ~35 m. Examination of the boreholes by setting up a difference of 8 m, identical to the difference differe in the ground level during the beginning of their excavation, and inferences from stratigraphy

A. Dey, C. Rainieri, G. Fabbrocino, F. S. de Magistris

columns, SPT blow counts and shear wave profiles led to recognize the presence of soil layers with nearly identical characteristics but at a certain level difference.

which is going to be described next in the paper. Based on the above mentioned laboratory and in situ tests, a simplified geotechnical model was adopted and used to set a numerical model of the retaining system.

Fig. 7 Shear wave velocity profile obtained along the depth of borehole S5 Fig. 5 Strata characteristics and SPT results of borehole S5

Fig. 8 Shear wave velocity profile obtained along the depth of borehole S6 BASIC FE NUMERICAL MODEL MEASUREMENT INTERPRETATION TATION Fig. 6 Strata characteristics and SPT results of borehole S6 Further exploration into the results facilitated to decide that there is a level difference of about 3 m between the similar strata in two boreholes. Considering this level difference and the distance between the boreholes, an inclined stratigraphy with an approximate slope (1V:10H) has been adopted for the geometry of the numerical model, FOR

A Finite Element (FE) model of the embedded retaining wall has been set using PLAXIS 2D: Dynamics Module [6] in order to evaluate its dynamic properties in linear elastic conditions, which can be referred to as representative of the system in operation (i.e. when it is subjected to ambient vibrations). Based on the results of stratigraphical columns from borehol borehole investigations, SPT blow counts and the shear wave profiles, a simplified geotechnical model

Fully instrumented full-scale scale embedded cantilever cantil sheet pile retaining wall: Plaxis FE modeling and interpretation

with inclined strata (1V:10H) has been adopted for the embedded retaining wall-soil soil system, as depicted in Fig. 9. . The dynamic behavior of the system under operational tional conditions has been investigated by modeling the soil as linear elastic (LE) in compliance with the observed low amplitude of ambient vibrations. The elastic modulus adopted for each stratum has been evaluated on the basis of the velocity profiles resulting esulting from the DH tests. Average values of the elastic parameters E and as obtained from S5 and S6 boreholes have been used for setting of this model. The embedded retaining wall has been modeled by a plate element of finite thickness. The stiffness of the piles-beam beam system has been evaluated by considering a 1 m large strip of the wall, including two piles (Fig. 2). ). Since, the in-situ in sheet pile wall is comprised of two alignments of non-contiguous contiguous piles, the axial and bending stiffness of the overall all system lies in between the stiffness of a single and the aggregated value of two contiguous piles connected by a top beam. The scaled down values adopted are: Axial stiffness (EA) = 3.08x107 kN/m, and Bending stiffness (EI) = 4.76x106 kN-m2/m. The equivalent valent thickness of the plate element (deq) is evaluated by the following expression: deq=(12*EI/EA), and a value of 1.362 m had been adopted for the same in the model. A no-slip slip debonding condition has been assumed at the soil-wall wall interface. In compliance complian with the high depth of the water table resulting from the geotechnical investigations, no interaction with the water has been considered. Table 1 summarizes the adopted soil and wall parameters for the basic model. Table 1 Material properties of soil adopted a in the basic FE model (L= left side of the wall), (R= right side of the wall)
Soil Layer A B C D E0 (105 kPa) 3.2 4.1 14.5 26.5 G0 (105 kPa) 1.1 1.4 5.0 9.2 0.432 0.426 0.438 0.433 (kN/m3) 18.0 19.0 19.5 20.0 Vs (m/s) 246.2 271.6 503.9 673.3 HL-HR (m) 8-3 3-3 5-5 10-10

This provides the basic guidance to choose the appropriate numerical parameters without jeopardizing the reliability and accuracy of the predictions. Suggestion about analysis parameter settings is provided in [6].

Fig. 9 Geometry and meshing of basic FE model Considering a series of dynamic analyses of vertical propagation of S-waves waves in a homogeneous layer using Plaxis 2D v. 8.4 4 Dynamic module, a suitable calibration technique is suggested in [7, 8]. It takes into account the influence of boundary conditions, mesh distributions and damping parameters on the response of the system. The width of the model domain significantly affects the quality of the computed response. A domain having low width-to-height height ratio does not fulfill the modeling assumption of semi-infinite soil and, as a consequence, the reflection of the propagating waves from the boundaries adulterate the response in the region of interest. Hence, to minimize such effects, , a domain aspect ratio (ratio of total domain width to the average height) greater than 40 is chosen. In order to facilitate wave absorption at the vertical boundaries of the model, absorbent boundaries have been used. Their effect is maximized by setting the relaxation tion coefficients as follows: C1=1 and C2=0.25 [9]. Meshing also has a significant effect on the accuracy of the computed response. The criteria reported in [9] have been adopted in the present study to determine the mesh size and a suitable table refined meshing scheme has been

Before starting the model calibration, calibration an optimization of the model parameters has been achieved in order to obtain in its best functionality.

A. Dey, C. Rainieri, G. Fabbrocino, F. S. de Magistris

accordingly set. Fig. 9 portrays the adopted meshing for the mentioned numerical model. ] affect the Numerical damping parameters [10] amplification of the response at resonances but the shape of the amplification function is not essentially modified. An implicit Newmark scheme governs the numerical time-integration integration and numerical damping is specified by Newmark damping parameters. An Undamped Newmark Scheme, also known as average acceleration scheme, has been considered dered (so that the predicted frequency spectra suffer minimal effect from numerical damping) with the following parameters: N=0.25 and N=0.5. Rayleigh damping is used to simulate material damping under plane-strain strain conditions. Considering a 1% damping on the overall system, Rayleigh damping parameters have been set as follows: R=0.293 and R=0.293, both for the soil and the structure. For the present study a Gaussian white noise of duration 1 hour and an impulse load 0.01 sec long have been applied as input input, both having a sampling frequency of 100 Hz. These types of excitation ion have been selected for convenience in compliance with the main purpose of the dynamic analyses, namely the extraction of the fundamental dynamic properties of the system from the simulated model responses. In fact, the selected input signals are characterized terized by a flat spectrum over the bandwidth of interest and, as such, allow an effective and reliable extraction of the dynamic properties of the system even with output-only only modal identification techniques [11]. Such techniques can be, therefore, conveniently conven used if the purpose of the analysis is a comparison in terms of dynamic properties only, neglecting the aspect of the frequency response function. The excitation is applied as propagating shear waves generated by base shaking. Dynamic analysis of the basic model is carried out to determine the acceleration response of the retaining wall at different locations. s. It has been subsequently analyzed in order to extract the dynamic properties of the modeled ed system and to compare them with the corresponding experimental

estimates obtained from dynamic measurements provided by the SHM system installed on the wall . The values of the first two resonant frequencies of the preliminary numerical model of the wall were equal to 4.7 Hz and 6.7 Hz, respectively. As a consequence, a refinement of the FE model (Fig. 10) to more closely match the experimental results, as enumerated in Table 2, , was needed. The process for minimization of the scatter between experimental observations and numerical predictions is described in detail in the next section.

Fig. 10 Geometry and meshing of first refinement of basic FE model Table 2 Modal identification result and correlation with the basic FE model after the first refinement
Model Basic 1st refine Mode I II I II fexp (Hz) 3.68 7.23 3.68 7.23 fFEM M (Hz) 4.7 6.7 3.64 7.46 Scatter (%) 21.7 -7.33 -1.1 3.1

MODEL REFINEMENT: RESULTS Uncertainty Identification Highly ighly sophisticated FE modeling techniques are already available for integrated structural structuralgeotechnical analysis. However, application a of these procedures to practical problems often reveals discrepancies between numerical predictions and experimental investigations. Modification of the associated parameters and modeling assumptions can definitely be carried out

Fully instrumented full-scale embedded cantilever sheet pile retaining wall: Plaxis FE modeling and interpretation

in order to minimize such differences in results and satisfy practical requirements. A classical approach in this direction would be a trial-and-error scheme; however, when it is mostly governed by the random changes in the contributory factors, it is highly tedious, time-consuming and often based on intuitions rather than justified by a mathematical and logical approach. As alternatives, computational procedures have been developed over time in order to provide some robust methodologies to achieve the same in a regularized manner. In particular, modal data (natural frequencies and mode shapes) extracted from measured response data are being largely used for model parameter adjustments, especially in the field of structural engineering [12]. In this section, the main uncertainties affecting the numerical model of the retaining system are identified and the process for the minimization in a regularized manner of the scatter with the reference experimental results is described. In particular, the model refinement process has been carried out by optimizing the correlation with the experimental results. The frequency scatter:
f = f i , FEM fi ,exp fi ,exp 100, i = 1, 2

(2)

where, fi,FEM and fi,exp are the numerical and experimental value of the natural frequency of the ith mode, respectively) has been minimized and the mode shape correlation has been optimized. The latter has been quantified by the MAC index computed for the corresponding modes:
MAC { i , FEM } , i ,exp

})
{ } {
T i , FEM

({

i , FEM

i , FEM

} { } }) ({ } { })
i ,exp T i ,exp i ,exp

(2)

where, { i , FEM } and { i ,exp } are the numerical and experimental mode shape vectors for the ith mode, respectively and the superscript T denotes transpose. As modeling uncertainties are concerned, it is worth noting that the above-mentioned boreholes, S5 and S6, each 30 m deep, though allow for a

reasonable identification of the soil stratigraphy, do not provide enough insight about the location and profile of the bedrock. However, geometry and thickness of the bottom layer have significant influence on the dynamic properties of the model. Since most of the model parameters are known from in-situ and laboratory tests on soil, the indirect identification of the bedrock depth and profile has been selected as a target of the refinement process. Sensitivity analyses have shown that the results are fairly sensitive to these parameters, which are, at the same time, the most uncertain, as pointed out also by the results of SPT tests. In fact, SPT blow-counts of boreholes S5 and S6 reveal a refusal of penetration at 20 m and 26 m, but, in spite, the borehole is excavated further until 30 m. Hence, it is very difficult to adjudge the soil at this depth to be or not as a part of the bedrock. Moreover, results from the two borehole investigations are actually not sufficient to describe the global profile of the substrata along the direction orthogonal to the wall; as a consequence the boreholes provide only a glimpse of the local stratigraphy. On the contrary, the dynamic response of the retaining wall is definitely affected by the substrata profile. For further clarification about soil geometry, borehole investigations in the nearby vicinity of the study-area have been also inspected. They seem to show the presence of hard soils on the excavated side at a depth of about 15 m, as depicted in Fig. 11, but there is no evidence of the presence of the bedrock. Even less information is available about the soil behind the wall, due to a lack of investigative explorations. The only instance is provided by the already mentioned borehole S6. Since there is a lack of information and a significant grade difference between the stratigraphical profile of the soils in the back and in front of the wall, it is quite difficult to justifiably verify the depth of the bedrock, in particular on the backfill side, or adjudge the irregularity in the profile of the bedrock across the transverse direction of the wall according only to the results of in-situ investigations. As a result, the refinement process had the identification of the depth and the profile of the bedrock towards the backfill soil as a primary

A. Dey, C. Rainieri, G. Fabbrocino, F. S. de Magistris

concern. . After that, minor refinements have been obtained by properly setting the elastic parameters of the two upper layers on the backfill side of the wall. In fact, inspection of Figures 7 and 8 points out also that the shear wave profile from borehole S5 is more scattered than the same from S6, especially within 16 m from the ground level. Taking into account nt that the rotation of the wall due to excavation also implies a slight stiffness reduction of the soil in the upper part of the backfill side of the wall, the effect of the change of stiffness of the different layers in the ranges obtained from the DH tests sts in S6 has been also investigated, , leading to a finer refinement of the numerical model.

the influence of the two domains domains, the spectral responses of the free-field field motions in each of the domains are studied and compared to the overall response of the system. Considering the basic model, trials with different thickness of the bottombottom most layer are carried out in order to get a better insight into the issue. Table 3 enumerates the responses of the study. It can be observed that a rough superposition does hold, , the dynamic response of the overall system being definitely guided by the natural frequencies of the soil domains on either ither side sides of the wall. The fundamental frequencies of the two soil domains form the first two resonant frequencies of the overall response of the system. This observation provides useful suggestions about how to carry out the model refinement. . Hence, the overall response of the system, which is a coupled problem, can be decoupled by controlling the dynamic response of the two soil domains, separately treated. Table 3 Dynamic response of the overall system and free-field field responses of soil domains for different thickness of the bottom-most most layer
Natural frequencies of vibration (Hz) Thickness of soil layer D Overall system measured on the wall 5.0, 8.0 4.6, 6.6 4.1, 5.2 Free-field response of soil behind the wall 5.1, 12.9 4.7, 10.8 4.0, 8.2 Free-field response of soil in-front of the wall 8.0, 16.7 6.6, 13.6 5.2, 11.0

Fig. 11 Conglomerated plot of SPT blow blow-counts from the nearby boreholes oles towards the soil in front of the wall Model Optimization from Analysis of the Interaction of Responses between the Geological Formations The analysis of the model geometry reveals that the retaining wall actually represents a boundary between the two halves of the system. It separates the geological formation in two domains having a significant grade difference of about 8 m. The two domains of the system, , if separately analyzed, show different dynamic properties. . The retaining wall, located at the boundary, depicts influences from both the domains, roughly represented by a superposition of the frequency responses s of the two previously mentioned domains. . In order to study

4 10 20

Table 2 revealed significant scatter between the experimentally obtained frequencies and the numerical values from the basic model. Taking into account the previously stated concept of decoupled problem, adjustments in the bed thickness have been carried out to improve the model correlation with the experimental data. Fig 10 shows the geometry of the refined model, which is characterized by y thicker bed at the left of the wall and a comparatively thinner bed on its right, in agreement with the presence of hard soil recognized during the in-situ situ investigations. The connection between the two beds governs the actual dynamic superposition effec effect on the wall, and has been chosen by trial and guided by the

Fully instrumented full-scale scale embedded cantilever cantil sheet pile retaining wall: Plaxis FE modeling and interpretation

frequency scatter. The meshing of the adopted model is portrayed in Fig. 10 . Table 4 enumerates the material properties of the best model after the first refinement.

values were located on the anti anti-diagonal of the 2by-2 2 MAC matrix between the experimental and numerical mode shapes. The observed inversion was depending on the chosen geometry of the model. The initial choice was guided by the presence of hard soil on the excavated side at a depth of about 15 m. However, the same values of the fundamental frequencies can be obtained also with a different choice of the thickness of the last layer on the two sides of the wall. For the previous ous model, t the top three layers of the model slopes from left to right, indicating a direction of movement of the geological formation. However, although the base layer too slopes in the same direction, the upward sloping connection inhibits the movement of f the geological formation from left to right. Taking into account this physical insight of the problem, the idea of the decoupled problem and the previous correlation results results, another refinement to the model was carried out wherein the thickness of the ba base layer on the left side is less in comparison to the right side side, and hence, the connection in the base layer is downward sloping from left to right, ensuring compatibility in the geological movements movements. Fig. 12 provides the geometry of the model and the adopted meshing of the second and best refinement. Table 4 provides the material properties adopted in the model after the second refinement. Table 5 provides the comparison of the dynamic properties obtained from the model after the second refinement and the e experimental magnitudes. The scatter in terms of frequencies has been substantially reduced with respect to the basic model (Table 2) ) and the correlation has been improved also in comparison to the first refined model (Table 2). ). In particular, the results reported in Table 2 and Table 5 represents the best correlation as provided by a sensitivity analysis with respect to the elastic properties of the two upper layers in the backfill side of the wall. A little stiffness decrease with respect to the corresponding layers on the excavated side was required to optimize the correlation. orrelation. After the second refinement, the mode inversion was no more

Fig. 12 Geometry and meshing of second and best refinement of basic FE model Table 4 Material properties adopted in the FE model after the first and second refinement (L= left side of the wall), (R= right side de of the wall)
Soil Layer E0L (105 kPa) E0R (105 (kN/m3) kPa) First Refinement 3.188 4.086 14.51 26.49 0.432 0.426 0.438 0.433 18.00 19.03 19.47 19.98 HL (m) HR (m)

A B C D A B C D

1.704 2.930 14.86 26.49 1.565 3.482 14.86 26.49

8 3 5 20

3 3 5 4 3 3 5 35

Second and Best Refinement 3.188 0.432 18.00 8 4.086 0.426 19.03 3 14.51 0.438 19.47 5 26.49 0.433 19.98 19

Table 2 provides a comparison ison between the dynamic properties obtained from the experimental investigations and the numerical predictions. The refined model has, therefore, provided results characterized by reduced frequency scatter, and as such it could be referred to as fairly representative of the actual behavior of the system in operational conditions. However, scrutiny of mode shapes revealed a reversal of the modes obtained from the numerical model l with respect to those provided by ambient vibration tests. In fact, the highest MAC

A. Dey, C. Rainieri, G. Fabbrocino, F. S. de Magistris

observed and a high correlation between numerical and experimental mode shapes, pointed out by MAC values very close to 1 (Table 5), has been obtained at the end of the model updating. Table 5 Modal identification results and correlation with the basic FE model after the second refinement
Mode I II fexp (Hz) 3.68 7.23 fFEM (Hz) 3.65 7.32 Scatter (%) -0.81 1.24

CONCLUSIONS The dynamic response of embedded retaining walls is one of the least investigated aspects in earthquake geotechnical engineering. However, the recent seismic events in urban areas have dredged out the necessity to enhance the knowledge about the dynamic behavior of such structures. In the present paper the refinement of a FE model, representative of the dynamic behavior of a real retaining system in operational conditions has been addressed. Based on an extensive geotechnical investigation a basic model has been set and its dynamic response has been investigated. The main sources of uncertainty have been identified and the numerical model has been refined by optimizing the correlation between experimental and numerical estimates of the dynamic properties of the retaining system. The obtained model can be, therefore, referred to as representative of the real system in its operational conditions. A superposition principle has been also validated. The illustrated results are certainly promising and encouraging. The model can be further extended to study the response of the retaining system under seismic excitation through the adoption of more advanced soil constitutive models and taking also advantage of measurements of the wall response to eventual ground motions occurring in the area. REFERENCES
1. Psarropoulos, P.N., Klonaris, G. and Gazetas,

flexible retaining walls, Soil Dynamics and Earthquake Engineering, 25(7-10), 795-809. 2. Visone, C. and Santucci de Magistris, F. (2008), A review of design methods for retaining structures under seismic loadings, ReLUIS, Italy. 3. Veletsos, A.S. and Younan, A.H. (1997), Dynamic response of cantilever retaining walls, ASCE Journal of Geotechical and Geoenvironmental Enginering, 123(2), 161172. 4. Steedman, R.S. (1998), Seismic design of retaining walls, In: Proceedings of Institution of Civil Engineers: Geotechical Engineering, 121, 12-22.
5. Dey, A., di Tullio, M., Lanzano, G., Rainieri, C., C. Laorenza, C., Fabbrocino, G. and de Magistris, F. S. (2011), A full scale instrumented embedded retaining wall: First interpretation of the measurements using numerical tools, 5th International Conference on Earthquake Geotechnical Engineering, Santiago, Chile, 1-12. 6. Brinkgreve, R.B. (2002), Plaxis 2D: Reference,

7.

8.

9.

10. 11.

12.

G. (2005), Seismic earth pressures on rigid and

Scientific and Dynamics Manuals, Lisse: Balkema. Visone, C., Bilotta, E. and Santucci de Magistris, F. (2008), Remarks on site response analysis by using Plaxis dynamic module, Plaxis Practice Bulletin, 14-18. Visone, C., Bilotta, E. and Santucci de Magistris, F. (2010), One-dimensional ground response as a preliminary tool for dynamic analyses in Earthquake Geotechnical Engineering, Journal of Earthquake Engineering, 14(1), 131 162. Lysmer, J. and Kuhlemeyer, R.L. (1969), Finite dynamic model for infinite media, Journal of Engineering Mechanics Division ASCE, 95(EM4), 859-877. LUSAS (2000), Theory Manual, FEA Ltd, United Kingdom. Rainieri, C. (2008), Operational Modal Analysis for seismic protection of structures, Ph.D. Thesis, Naples, Italy. Friswell, M.I. and Mottershed, J.E. (1995), Finite Element Model Updating in Structural Dynamics, Kluwer Academic Publishers, Dordrecht, The Netherlands.

Anda mungkin juga menyukai