Embed Size (px)
Citation preview
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/330534361
Significance of Soil-Structure Interaction in Seismic Response of Buildings
Article · January 2019
CITATION
1READS
1,103
3 authors:
Some of the authors of this publication are also working on these related projects:
Post Earthquake Re-construction Project View project
Multidisciplinary approach to understand earthquake risk and resilience View project
Naveed Anwar
Asian Institute of Technology
142 PUBLICATIONS 48 CITATIONS
SEE PROFILE
Abinayaa Uthayakumar
Asian Institute of Technology
3 PUBLICATIONS 1 CITATION
SEE PROFILE
Fawad Najam
National University of Sciences and Technology
58 PUBLICATIONS 23 CITATIONS
SEE PROFILE
All content following this page was uploaded by Naveed Anwar on 22 January 2019.
The user has requested enhancement of the downloaded file.
In current design office practice, a commonly used modelling assumption is that the base of abuilding structure can be idealised with fixed support conditions, thereby neglecting any effectsfrom soil-structure interaction (SSI). Various recent studies, however, have shown that the explicitconsideration of SSI effects in seismic analysis of buildings structures may significantly affect thepredicted seismic demands and resulting structural performance. This study addresses some keyissues and practices in the area of SSI and its effects on the dynamic response and seismic performanceof buildings. It is also intended to demonstrate the significance of considering SSI effects in structuralmodelling and analysis while providing key insights into practical applications in real projects.Using a forty storey example building, the effect of considering SSI on the predicted seismicperformance is demonstrated. For the purpose of comparison, five detailed computer models (onewithout considering any SSI effects, two models with SSI modelled using indirect approach, andtwo models with SSI modelled using direct approach) of the example building were constructed andsubjected to various input ground motions. It is observed that depending upon the modelling approachused, the consideration of SSI effects may affect the predicted seismic performance in varyingdegrees. Moreover, the direct modelling approach presented in this study may provide improvedresults compared to various approximate methods.
Keywords: soil-structure interaction; inertial interaction; kinematic interaction; high-rise buildings;direct approach; substructure approach; soil modelling.
1. INTRODUCTION
One of the challenging aspects in structural modelling of building structures is the accurate idealisationof below-grade components including foundation, soil and their interaction. The characteristics andmechanical properties of medium on which a structure is founded may significantly influence theseismic responses of structure. If the structure is supported on a solid rock, the seismic behaviourof the structure will be similar to a fixed-base structure subjected to free field motion. However, ifthe soil is deformable, the structure responds to the soil dynamics and in return, the soil also mayrespond to the dynamics of the structure - a phenomenon referred to as the soil-structure interaction(SSI). This phenomenon may ultimately cause the response of structure to be governed by theinteraction of soil and structure as well as the characteristics of input ground motions.
SIGNIFICANCE OF SOIL-STRUCTURE INTERACTION IN SEISMIC RESPONSEOF BUILDINGS
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
Naveed Anwar1, Abinayaa Uthayakumar2, Fawad Ahmed Najam3
1 Executive Director, AIT Solutions, Asian Institute of Technology, Bangkok, Thailand, Ph. +66819206569, Email: [email protected] Structural Engineer, AIT Solutions, Asian Institute of Technology, Bangkok, Thailand, Ph. +66614251071, Email: [email protected] Assistant Professor, National University of Sciences and Technology (NUST), Islamabad, Pakistan, Ph. +923345192533,
Email: [email protected].
ABSTRACT
43
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), 201944
Naveed Anwar is an Executive Director at Asian Institute of Technology (AIT) Solutions,Thailand. He is also the Director of Asian Center for Engineering Computations and Software(ACECOMS) and heads the Habitech Building Technology Center at AIT, Thailand. Hisprofessional carrier spans over 35 years. He is an author of over one hundred publicationsincluding books, conferences and journal papers.
The consideration of SSI effects in modelling and analysis of structures is a complex and tedioustask which requires significant expertise and skills. Moreover, the problem is more complicated dueto several possible scenarios of foundation embedment in soil. Consider, for example, the case offoundation for a simple reinforced concrete (RC) column. Figure 1 shows some of the practicalconfigurations of columns embedded in soil. Depending on the size of columns, type of soil andpresence of lateral loads, the column may become effectively fixed or flexible against rotation(Figure 1(a)). In case of columns or piers in waterways, the soil erosion, silting and scour may keepchanging the SSI effects with time. Another factor which may make the problem more complex isthe amount of restraint or fixity provided by back fill and compacted soils. Similarly, consider case�c� where the column size below the soil level is larger than the main column. So, the effect of largercross-section as well as the larger restraint due to soil needs to be taken into account. At the sametime because of the larger column near the footing, it is not immediately clear as to which columndimensions should be considered in determining the effective length for simplified modellingapproaches. In this case, one can use the concept of inverse stiffness (or flexibility) to determinethe stiffness of the variable column. Case (d) is further complicated by the presence of a concretefloor which is not rigidly connected to the column but nevertheless provides significant lateralrestraint. The compact and well constrained compact filling under the floor provide further restraintwhich is numerically difficult to evaluate. One simplified option in this case can be to consider thecolumn as hinge-supported. However, this hinged condition should also be considered in modellingof the frame to obtain consistent and compatible moment in the column. An even more complexcase is that of a column supported on a pile cap and pile foundation. If the piles are driven into softtop layer or are exposed above the firm soil layers, then the column-pile cap-pile-soil system willinteract together to determine the effective bending length or effective length of the whole system.The solution to such problems is to carry out a P-D analysis of the complete system or to estimatethe effective length from a first order analysis. This example shows the complexities involved inthe practical cases of considering SSI effects even for the case of a simple column foundation.
Abinayaa Uthayakumar is a Structural Engineer at Asian Institute of Technology (AIT) Solutions,Thailand. She received her Masters in structural engineering from AIT, Thailand. Her researchinterests include soil-structure interaction, seismic performance assessment of buildings andgeotechnical earthquake engineering.
Fawad Ahmed Najam is an Assistant Professor at National University of Sciences and Technology(NUST), Pakistan. He received his PhD from Asian Institute of Technology (AIT), Thailand.His research interests include earthquake engineering, structural dynamics, seismic performanceevaluation of tall buildings, and seismic hazard and vulnerability assessment.
Figure 1. Levels of complexity involved considering SSI effects-various practical casesand configurations of RC columns embedded in soil.
45NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
2. EFFECTS OF SSI ON DYNAMIC RESPONSE OF BUILDINGS
There are two types of interactions which can influence the response and performance of structuresdue to soil-foundation-structure interaction. These are referred to as the inertial interaction andkinematic interaction. The inertial interaction is mainly related to the base shear and base momentand may result in the elongation of natural period and additional damping effects. The most importantparameter controlling the inertial SSI effects is the structure-to-soil stiffness ratio as shown in Eq. (1).
where h is the height of the centre of mass of building in first vibration mode shape, Vs is shearwave velocity and T is the fundamental period corresponding to fixed-base structural model. If thestructure-soil stiffness ratio is greater than 0.1, a strong inertial interaction may exist. This ratio istypically low for high-rise buildings [1].
Figure 2(a) shows the typical variation of natural period (normalised to fixed-base period) andfoundation damping ratio (normalised to fixed-base damping ratio) with structure-soil stiffness ratiofor different values of h/B (where h is structure height and B is half-width of foundation). It can beseen that the foundation damping increases with structure-soil stiffness and decreases with increasingvalue of h/B, indicating that energy dissipation is more through foundation's lateral movement thanrocking. The natural period also elongates significantly with increasing structure-soil stiffness ratio.Figure 2(b) illustrates the effect of inertial interaction on base shear for elastic response of structures.It can be observed that for positive slope the base shear tends to increase and for negative slope thebase shear decreases. Hence, the long-period structures tend to have reduced base shear due to theeffect of inertial interaction.
The kinematic interaction, on the other hand, deals with the ground motion and deems the differencebetween free-field motion (motion unaffected by wave scattering or any structural vibration signature)and foundation input ground motion (the theoretical motion at base slab if the structure and foundationhad no mass). The difference is due to structure�s configurations such as the rigidity and size of thebase (base-slab averaging), and the depth of foundation below ground (embedment). Kinematicinteraction models are often interpreted as transfer functions of free-field motions to foundationinput motions [1].
3. SSI EFFECTS-LOW-RISE VERSUS HIGH-RISE BUILDINGS
The inertial and kinematic interactions may not affect the seismic performances of low- and high-rise buildings in the same manner. For the low-rise buildings, the most important parameter governingtheir dynamic response is the natural period for fundamental mode of vibration. The design baseshear is a function of natural period in the corresponding direction (Figure 2(b)). For the purposeof simplified structural design, the design codes often provide smooth response spectrum curvesto estimate the design base shear coefficient as shown in Figure 3 [2]. The seismic loading in termsof spectral acceleration or base shear coefficient (base shear normalised to the total seismic weightof the structure) can be directly determined from this response spectrum corresponding to the naturalperiods of vibration of any structure.
To demonstrate the effect of different soil properties, the results of two example buildings (a fivestorey low-rise building and a forty storey high-rise building) supported on a soft soil, stiff soil androck (from NEHRP soil profile type classification) is compared with that of a fixed-base condition.
Figure 2. Response of square footing: (a) period lengthening and foundation dampingversus structure-soil stiffness; (b) effects on spectral acceleration.
Structure-soil stiffness ratio = h/VST (1)
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 201946
Figure 3. Effect of soil type on natural periods and resulting base shear coefficient.
The effect on time period and associated alteration in base shear coefficient is considered and isshown in Figure 3. It is observed that due to the elongation in natural period, the base shearcoefficient is reduced by twenty percent compared to the fixed-base condition for the low-risebuilding. However, there is not much reduction in base shear coefficient for high-rise building. Thecomparison of natural periods and base shear coefficients for other soil types is tabulated in Table 1T/T is the ratio of flexible-base period to fixed-base period, while C/C is the corresponding ratiosof base shear coefficients). It can be seen that rock base is resulting in a periods almost equal tofixed-base condition. However, for other soil types, there is an elongation in natural period (resultingin reduction in base shear coefficient) in varying degrees. Therefore, for low-rise buildings, theconsideration of SSI is often regarded as beneficial due to time period elongation and the associatedreduction in design base shear. As mentioned earlier, the design engineers make use of this phenomenonin retrofitting and design of low-rise buildings [1]. However, studies such as by Maheshwari andSarkar [3] and Saez et al. [4] urge the importance of including SSI for low-rise buildings due to itspotential influence on overall seismic responses.
4. CONSIDERATION OF SSI EFFECTS-EXISTING RESEARCH AND CURRENTPRACTICE
In current design office practice, the most widely used approach is to idealise the structure�s basewith fixed support condition, while ignoring any base flexibility. The base is then subjected to free-field ground motion records for the purpose of dynamic analysis. Similarly, in case of high-risebuildings having deep subterranean levels, the SSI effects along the deep basement walls are generallyignored and the ground motion records are applied at fixed base. However, several recent studieshave shown that the explicit consideration of SSI effects in seismic analysis of buildings structuresmay significantly affect the predicted seismic demands and resulting structural performance. As aresult of this recent realisation, the inclusion of SSI is gradually trending as a part of the performance-based evaluation in retrofitting of existing buildings.
The 1971 San Fernando earthquake first caught the attention towards SSI effects. Since then, severalstudies have been conducted in light of understanding the effects of SSI and its importance onseismic responses of various types of structures. For example, Turek et al. [5] and Fatahi et al. [6]conducted laboratory experiments including shake table tests, ambient vibration tests and micro-tremor tests to study building behaviour due to SSI. Various techniques considering equivalent-linear soil behaviour were employed by Naeim et al. [7], Li et al. [8], Ellison et al. [9] and Lu et al.[10] to study the effect of SSI on behaviour of high-rise buildings. Naeim et al. [7] observed thatthough inclusion of SSI may not result in a significant increase in natural time period, it does
Table 1. Effect of different soil types on fundamental natural periods and base shearcoefficients of example low-rise and high-rise buildings
Soil type
Soft
Stiff
Rock
Fixed-base
Shear wave
velocity,
m/sec (ft/sec)
100 (328)
300 (984)
1000 (3281)
-
Low-rise building
Natural
period
(sec)
0.581
0.504
0.494
0.494
T/T
1.18
1.02
1
1
~C/C
0.8
0.95
1
1
~High-rise building
Natural
period
(sec)
4.601
4.558
4.482
4.424
T/T
1.04
1.03
1.01
1
~C/C
0.96
0.97
1
1
~
~ ~
47NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
significantly affect the vertical distribution of inter storey drift. Ellison et al. [9] considered thenonlinear soil behaviour but focused on the interaction between a deep, long embedded trainboxand the adjacent deep basement of a tall building. This study emphasised the need to consider SSIin dense, urban areas as the adjacent embedded trainbox had a significant effect on the demandsof basement walls.
There exists a gap between the state of existing knowledge and the state of practice in the area of SSIand its consideration for seismic analysis of high-rise building structures. Beside the need of improvedunderstanding, more skills and expertise, an important reason for this gap can be the unavailabilityof appropriate practical recommendations and guidelines by standards and codes [11, 12]. For high-rise building structures with basements, the consideration of SSI can be an important factor in seismicperformance as it can substantially affect their seismic responses, as observed by [7] Naeim et al. andEllison et al. [9]. Given the modern computational capabilities, it is favourable to implement SSI inseismic design of high-rise buildings. Nevertheless, an ambiguous question is whether adoptingtedious approaches of modelling SSI actually produce significantly more accurate results.
5. MODELLING APPROACHES FOR SSI-CHALLENGES AND OPPORTUNITIES
Currently, the modelling approaches to account for the SSI effects in the analysis of buildingstructures can be classified under two categories. The first approach, referred to as the direct approach,represents the soil continuum surrounding the building foundation with three-dimensional finiteelements. This continuum is truncated at a certain distance from foundation by using special absorbingboundary conditions. The second approach, sometimes referred to as the simplified substructureapproach, represents the surrounding soil with a series of springs and dashpots. The mechanicalproperties, modelling parameters and recommendations for using this approach are available invarious guidelines including NIST [1] and PEER [13]. Since the direct approach is based on anexplicit finite element modelling of surrounding soil, it is considered a relatively more accuratemethod than the simplified substructure approach to consider SSI effects and is widely adopted inSSI research studies.
Although several guidelines and recommendations for modelling SSI effects using simplifiedapproach are available [1, 13], it is seldom considered in practical applications. Even when the SSIis implemented, it is often limited to the modelling of vertical foundation springs. Among the projectsthat included SSI, greater applications are generally found in retrofitting of existing low-rise buildingsas a part of the performance-based assessment than in the design of new buildings. One major reasonis that even simplified modelling of SSI requires significant expertise and skills which most of thepracticing designers may not be able to develop under conventional capacity building trainings andprograms. The consideration of SSI effects in practice also requires a close collaboration betweenthe structural and geotechnical engineers. However, in many projects, the geotechnical engineersare generally not a part of the design team meetings that are set up by the architect and projectowner. Also, the consideration of SSI effects may significantly increase the analysis and post-processing time, which the practicing engineers may sometimes not afford. Considering the modellingphase of SSI, existing studies [7] reported several difficulties which may be encountered whilemodelling the SSI. The detailed finite element modelling of SSI is considered to be tedious. However,several studies have managed to implement and test this approach by excluding various SSI effectssuch as base-slab averaging and by scaling down the complexity involved with soil medium andwave motions such as assuming only linear behaviour of soil, and depth invariable motions, etc.
In this study, while discussing some key issues and practices in the area of SSI and its effects onthe dynamic response and seismic performance of low- to medium-rise buildings, the significanceof considering these effects in high-rise buildings is also demonstrated using an example building.As mentioned earlier (and is also demonstrated in Figure 3), the base shear coefficient for high-rise building supported on soft soil is almost same as that of a fixed-base condition. Similarly,Table 1 showed that the maximum reduction in base shear coefficient is only about four percent.Due to such negligible effects on natural periods, generally it is believed that the SSI effects arenot significant for high-rise buildings. However, as stated earlier, several recent studies have indicatedthe potential influence of SSI on seismic responses of high-rise buildings. The case of high-risebuildings will therefore be evaluated here in detail using a case study building. For this purpose, aforty storey RC core wall building is selected and subjected to detailed analysis. The effect ofconsidering SSI on the predicted seismic performance of this case study building is demonstrated.The objective is to study the effect of SSI on a high-rise case study building structure by comparingseismic responses of the model without SSI and the code-based models with SSI. For simplicity,only the linear behaviour of the structure is considered in this study.
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 201948
6. SSI EFFECTS IN HIGH-RISE BUILDINGS-A CASE STUDY
6.1 Site Characteristics and Modelling
An existing forty storey case study building with RC core wall is selected for the detailed performanceevaluation with and without considering the SSI effects. The building has four basement levels andis resting on a mat foundation (Figure 4: Model 1). The soil profile chosen is located in MakatiCity, Manila, Philippines. The soil profile at the site of building (shown in Figure 4) is derived fromthe actual soil profile studies in Philippines. Figure 4 shows that the profile consists of two layers(soft-to-medium stiff clay up to 15 m (49 ft) and weathered rock up to 40 m (130 ft). For the detailedperformance evaluation, a total of five detailed three-dimensional (3D) finite element models weredeveloped. The baseline model (designated as Model 1) represents the typical fixed support idealisationwhich commonly used in design practice (completely ignoring any SSI effects). Generally, thismodel is recommended to be used for the service-level performance evaluation [13]. The modelsdesignated as �2A� and �2B� are constructed as per the recommendations by PEER [13] and NIST[1] for considering the SSI effects on seismic performance of high-rise building structures. Themodel 2A considered SSI at foundation level only yet accounting for the cumulative stiffness ofthe embedded foundation. The model 2B is also known as �bathtub� model. The major differencebetween the two is that the latter considered SSI along basement walls (Figure 4). This modeldoesn�t consider depth-variable motions along the embedment, hence avoiding multi-supportexcitation. For both the models the spring and dashpots coefficients were determined as per therecommendations of NIST [1].
In the models designated as �3A� and �3B�, the soil is explicitly modelled using detailed finite elementmodelling. In Model 3A, the structure is assumed to be bounded perfectly with the soil with nointerface elements. While for Model 3B, the nonlinear compression-only interface elements wereintroduced between structure and soil (Figure 5). For the modelling of soil, an equivalent linear3D model is used in this study. For this purpose, the modulus reduction curves and equivalentdamping curves for individual soil layers are required. To develop the equivalent linear soil model,ground response analysis is also required. The soil profile at the site of case study building wasmodelled in the geotechnical software DEEPSOIL V6.1 [14] with bedrock assumed to be rigid. Forthe soft layer, the widely used reference curve by Vucetic and Dobry [15] was employed. For theweathered rock layer, the expression proposed by Schnabel [16] (1973) was used. The effectiveshear strain was assumed as sixty five percent and the method for complex shear modulus wasspecified as recommended.
One dimensional (1D) equivalent linear ground response analysis was performed for various inputground motion records. The reduced shear modulus and damping values were then calculated. Thelateral boundaries were chosen as three times the foundation width from the centre of the structure[17]. The dimensions of the soil medium were 300×275×40 m (985×900×130 ft). The infinite soilmedium was bounded by viscous boundary in all four lateral directions by placing viscous dampersalong the boundaries in normal and shear directions [18, 19]. The horizontal boundary at the bottomwas fixed at bedrock as it is regarded as the most suitable way to represent bedrock. Also, the elasticmodulus at 40 m (130 ft) depth was more than ten times the elastic modulus of the top layer. Thesoil elements were represented by eight node brick elements with three degrees of freedom. Thechosen dimensions of the soil medium and dampers were verified by comparing the time historyplot and response spectrum at free surface and at the top of second layer of the model with anothermodel with half-width that is ten times the width of the foundation with free lateral boundary(Figure 6). Also, two types of damping elements (linear damper and nonlinear Maxwell damper)
Figure 4. Characteristics and Vs30 profile of soil.
Note: 1 m = 3.28 ft; 1 m/sec = 3.28 ft/sec
49NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
Figure 5. Detailed finite models of case study building.
were used along the boundary. After comparison, the linear dampers were found to be computationallyefficient and less expensive compared to nonlinear dampers and therefore, were chosen for furtherdetailed analysis. The comparison of time histories (five hundred thirty one ground motion) andresponse spectra at the free surface and at top of layer 2 obtained from DEEPSOIL and SAP2000is shown in Figure 7. A good match was seen for all selected ground motions, implying that theuse of 1D results in the 3D FE models is adequate.
6.2 Selection of Ground Motions and Analysis Procedures
In this study, three ground motions were selected from a detailed ground motion development studyconducted for the site of case study building. These ground motions are in the form of groundacceleration histories recorded from three different earthquake events generated by seismic sourceswith three different faulting mechanisms (Table 2).
Figure 6. Comparison of response spectra and time histories between models with x10width, linear damper and NL damper.
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 201950
Figure 7. Comparison of response spectra and acceleration time histories of fivehundred thirty one ground motions in X direction.
Table 2. Ground motions used in this study
Earthquake event
1989 Loma Prieta
1999 Duzce, Turkey
1985 Michoacan, Mexico
Earthquake
ID
531
LGP
UNIO
Faulting
mechanism
Reverse Oblique
Strike slip
Subduction
Magnitude
(Mw)
6.93
7.14
8.1
Vs30, m/sec/site class
(ft/sec)/site class
478 (1568)
660 (2165)
Rock
For the response history analysis of models developed in this study, three types of earthquakemotions are required: (a) free-field ground motion for Model 1; (b) foundation input motion forModels 2A and 2B; and (c) bedrock motion for Models 3A and 3B. The response spectrum of thefoundation input motion differs from that of the free-field motion. With the use of transfer functions[1], the modified response spectrum was developed. The free-field ground motions were then scaledto the modified response spectrum to obtain the foundation input motions. The ground motions wereconverted to bedrock motions using DEEPSOIL software by the deconvolution process [20]. In thisprocess, the input motion was specified at top of Layer 1, and since the bedrock was assumed asrigid, the �within� bedrock motion was obtained. For the models except Model 3B, the LinearResponse History was performed. For Model 3B, Nonlinear Response History Analysis (NLRHA)procedure was performed since the interface elements were nonlinear.
6.3 Results and Discussion
It is observed that as expected the consideration of soil-structure interaction (SSI) did not result ina significant increase in the vibration time periods of the selected case study high-rise building. Thenatural period of Model 1 was 4.4 sec whereas for the models with SSI it was about 4.6 sec. Theconsideration of SSI effects using simplified (substructure) approaches had a moderate effect on thestorey displacements and storey drifts. The other considered responses such as storey shear, storeymoment, roof acceleration were also not significantly affected by the consideration of SSI effects.Both of the substructure approaches produced comparable results, suggesting that the considerationof SSI at the base is an important factor compared to consideration at the basement walls.
The detailed comparison of seismic demands (between the models considering SSI using simplifiedapproaches (Models 2A and 2B) and those using the direct continuum approaches (Models 3A and 3B))
51NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
is performed. The latter is widely regarded as the most realistic technique to model SSI and hence,mostly employed in SSI related research studies. The direct approach is generally not used to studythe effect of SSI on tall buildings. A relatively recent study carried out by Ellison et al. [9] haveattempted to use the direct method, however, the research interest was restricted to the performanceevaluation of basement alone.
Figure 8 compares the storey displacement in X and Y directions reported by the four models foreach of the three ground motions. A moderate difference in storey displacement can be seen fromthe plots for all three ground motions. Similar observation can be made from the storey driftcomparison. Although, some deviation can be seen for UNIO ground motion, the difference is notsignificant. As mentioned before, Naeim et al. [7] observed that storey drift significantly variesalong the height of the structure especially in subterranean levels compared to fixed-base models.However, the difference was not observed to be significant in this study in both X and Y directionsof the case study building. One possible reason could be the inconsideration of nonlinearity in thecase study building (only linear elastic behaviour of the building components was considered).
Figures 9 and 10 shows increased shear and moment demands along the height of the building incase of direct modelling approach to account for SSI effects. To compare the deviation with thebaseline model, the shear and moment values of the models with SSI were normalised with respectto Model 1 as presented in Figure 11. It can be seen that shear differs along the height whereasmoment differs mostly along higher stories only. To further look into the deviation of shear, floor19 was chosen as it experienced higher shear and shear response history was compared as shownin Figure 12. The comparison shows a higher shear throughout the excitation in direct modellingapproach compared to the simplified procedure. To the best of the authors� knowledge, this observationis new. Figure 13 presents the comparison of roof acceleration experienced by the building due todifferent modelling techniques for considering SSI effects. The comparison between the results ofmodel 2A and Model 3A were compared here as an example. For all three ground motions, the finiteelement direct based models (for SSI) had higher roof acceleration than the substructure-basedapproximate models in both X and Y directions.
Figure 8. Comparison of storey displacement for models 2A, 2B, 3A and 3B.
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 201952
Figure 9. Comparison of storey shear for models 2A, 2B, 3A and 3B.
Figure 10. Comparison of storey moment for models 2A, 2B, 3A and 3B.
53NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 2019
Figure 11. Comparison of storey shear and storey moment for models 2A, 2B, 3A and3B (normalised to Model 1).
Figure 12. Shear response history at 19th floor.
Figure 13. Comparison of roof acceleration in X direction of case study building.
N. Anwar et al.
NED UNIVERSITY JOURNAL OF RESEARCH-SPECIAL ISSUE ON FIRST SOUTH ASIA CONFERENCE ON EARTHQUAKE ENGINEERING (SACEE'19), Vol. 1, 201954
7. CONCLUSIONS
Some key issues and practices in the area of SSI and its effects on the dynamic response and seismicperformance of buildings are discussed in this study. The significance of considering SSI effectsin structural modelling and analysis is demonstrated while providing key insights into practicalapplications in real projects. After discussing the common design office practices in the area of SSIand its effects on seismic performance of low- to medium-rise buildings, the significance ofconsidering these effects in high-rise buildings is also evaluated using a forty storey examplebuilding. For this purpose, the analysis results obtained from the response history analysis of fivedetailed computer models (one without considering any SSI effects, two models with SSI asrecommended in the literature and two models using direct approach) were compared. It is observedthat the consideration of SSI effects using simplified (substructure) approaches had a moderateeffect on the storey displacements and storey drifts of the case study building. However, the effecton storey shears, storey moments and roof accelerations were not significant under various inputground motions. Also, a reasonable difference in predicted responses were observed between thesubstructure and direct modelling approaches with the latter producing higher shear and momentdemands. It is therefore proposed that the dynamic behaviour of a structure can be more accuratelystudied by considering SSI effects and 3D modelling of surrounding soil instead of idealising thebase of structure with rigidly fixed support condition.
REFERENCES
[1] National Institute of Standards and Technology (NIST). Soil-structure Interaction for BuildingStructures (NIST GCR 12-917-21). Department of Commerce, Washington, DC, 2012.
[2] Murty CVR, Goswami R, Vijayanarayanan AR, Mehta VV. Some Concepts in EarthquakeBehaviour of Buildings. Gujarat State Disaster Management Authority, Government of Gujarat, 2012.
[3] Maheshwari B, Sarkar R. Seismic Behaviour of Soil-Pile-Structure Interaction in LiquefiableSoils: Parametric Study. Int J Geomech 2011;11(4):335-347.
[4] Saez E, Lopez-Caballero F, Razavi A. Inelastic Dynamic Soil-structure Interaction Effects onMoment-resisting Frame Buildings. Eng Struc 2013;51:166-177.
[5] Turek M, Ventura CE, Thibert K. (2008). Dynamic Characteristics of Tall Buildings in Vancouver,Canada. In: ASCE Structures Congress. Vancouver, British Columbia: 2008.
[6] Fatahi B, Hokmabadi AS, Samali B. Seismic Performance Based Design for Tall BuildingsConsidering Soil-Pile-Structure Interaction. Geotechnical Special Publication (240 GSP),ASCE 2014. p. 333-342.
[7] Naeim F, Tileylioglu S, Alimoradi A, Stewart J. The Real Deal on Soil-Foundation-StructureInteraction (SFSI). In: 2010 Annual Convention of Structural Engineers Association of California.San Diego: 2010.
[8] Li M, Lu X, Lu X, Ye L. In?uence of Soil-structure Interaction on Seismic Collapse Resistance.J Rock Mech Geotech Eng 2014;6(5):477-485.
[9] Ellison KC, Masroor AM, Almufti I, Willford M, O'Riordan N. Structure-Soil-StructureInteraction Analysis in a Highly Seismic, Dense Urban Regeneration Zone. In: 6th InternationalConference on Earthquake Geotechnical Engineering. Christchurch, New Zealand: 2015.
[10] Lu X, Chen B, Li P, Chen Y. Numerical Analysis of Tall Buildings Considering Dynamic Soil-Structure Interaction. J Asia Arch Build Eng 2003;2(1):1-8.
[11] Khoueiry DH, Khouri MF. Integrating Soil-Structure Interaction Along Basement Walls inStructural Analysis Programs. Int J Develop Res 2015;5(4):4061-4068.
[12] Badry P, Satyam N. Seismic Soil Structure Interaction Analysis for Asymmetrical BuildingsSupported on Piled Raft for the 2015 Nepal Earthquake. J Asia Earth Sci 2017;133:102-113.
[13] Pacific Earthquake Engineering Research Center (PEER). Guidelines for Performance-BasedSeismic Design of Tall Buildings. TBI Guidelines Working Group, University of California,Berkeley, California, USA.
[14] University of Illinois. DEEPSOIL v6.1. University of Illinois at Urbana-Champaign, 2016(accessed on 8 January 2018). Available from http://deepsoil.cee.illinois.edu/
[15] Vucetic M, Dobry R. Effect of Soil Plasticity on Seismic Response. ASCE J Getech Eng Div1991;117(1):89-107.
[16] Schnabel P, Lysmer J, Seed H. SHAKE: A Computer Program for Earthquake Response Analysisof Horizontally Layered Sites. Berkeley: Earthquake Engineering Research Center, 1973.
[17] Ghosh S, Wilson EL. Dynamic Stress Analysis of Axi-symmetric Structures under ArbitraryLoading. Report No. EERC 69-10, University of California, Berkeley, 1969.
[18] Kourossis G, Verlinden O, Conti C. Finite-Dynamic Model for Infinite Media: CorrectedSolution of Viscous Boundary Efficiency. J Eng Mech 2011;137(7):509-511.
[19] Tabatabaiefar S, Fatahi B, Samali B. Seismic Behavior of Building Frames ConsideringDynamic Soil-Structure Interaction. Int J Geomech 2013;13(4):409-420.
[20] University of Illinois. DEEPSOIL V4. University of Illinois at Urbana-Champaign, USA, 2011.
View publication statsView publication stats
