Perspectives on Earthquake Geotechnical Engineering

Features of the 2011 earthquake in Japan are characterized by predominance of the ground failure due to liquefaction and scour of the ground caused by Tsunami. Unprecedented long duration of the shaking combined with large aftershocks have generated the worst situations resulting in the extensive damage due to liquefaction over the Tokyo Bay and the downstream plain areas of the Tone River 300 ~ 400 km distant away from the epicentral area. In this paper, focus is placed on the characteristic features in the occurrence of liquefaction and consequent damage in the area of the downstream reaches of the Tone River.

The seismic response of shallow soft surface layers is of concern in developing the seismic section of the National Building Code of Canada for 2015. The response of such layers is studied using recorded data from the 2011 Tohoku earthquake. Fourteen sites have been studied in detail so far that had soft, shallow surface layers. At each site two records were available, one at the bottom of the borehole at a depth of the order of 100–500 m and one on the surface. Site response analyses were conducted to determine the ground motion at the top of the rock underlying the surface layer, so that the amplification of the soil layer itself relative to the rock surface could be determined. These analyses were conducted using the program SHAKE. The properties in SHAKE were calibrated to get a good match between the measured and computed surface motions for the recorded input at the bottom of the rock.

Based on a worldwide database of strong ground motion records in different well constrained site conditions, a new soil classification is proposed for EC8 with the associated amplification factors and the normalized response spectra for 5 % damping. Amplification factors and normalized response spectra are also proposed for the current EC8 classification. Then we perform a detailed seismic risk assessment for the city of Thessaloniki using (a) the capacity spectrum method, (b) the Uniform Hazard Spectrum (UHS) for rock conditions compiled applying the SHARE approach, (c) the current and the new site amplification factors to evaluate the site specific demand spectra, (d) the detailed inventory of the Thessaloniki building stock to select the appropriate capacity and fragility curves for each building typology, and (e) available functions to estimate from the physical building damages the causalities and the economic losses. The main conclusions are summarized as following: (a) EC8 site classification, amplification factors and normalized response spectra should be revised. (b) The current methods for the seismic risk assessment are not robust enough; they need serious upgrade in all their components, and yet the uncertainties will remain quite high.

Probabilistic seismic hazard analysis (PSHA) is now the standard approach for developing earthquake ground motions for seismic design because it takes into account the important sources of uncertainty and variability associated with ground motion prediction. PSHA typically does not account for local site response but rather provides a uniform hazard spectrum that is used as the input into site response analyses. This approach does not account rigorously for the uncertainties in the site response analysis, and therefore the resulting surface motion is associated with an unknown hazard level. Seismic hazard assessments for critical facilities, such as nuclear power plants, require that ground motions be defined for specified hazard levels and also include the effects of local site conditions. To achieve this objective, the uncertainty and variability in the site amplification must be quantified and incorporated into the PSHA. This paper describes the convolution approach used to incorporate site response into PSHA. The main components used in the convolution approach are provided, and the required site characterization and site response analyses are discussed. The convolution approach is illustrated through examples from recent projects.

Most of the widely used seismic soil liquefaction triggering methods propose cyclic resistance ratio (CRR) values valid at the reference normal effective stress (σ′_v,0) of one atmosphere and zero static shear stress (τ_st,0) states. Then, a series of correction factors are applied on this reference CRR, for the purpose of assessing the variability due to normal effective and static shear stress states (i.e. K_σ and K_α corrections) acting on the horizontal plane. In the literature, a number of relationships suggested to be used as part of liquefaction triggering methodologies. However, the presence of a wide range of correction factors, some of which with even contradicting trends, suggests that more research needs to be performed to reduce this uncertainty. Additionally, these stress correction factors are treated as being strain-independent and are applied disjointedly to CSR or CRR. The main motivation of this on-going study is defined as to develop a strain-dependent semi-empirical framework to assess combined effects of i) σ′_v,0, ii) τ_st,0 acting on the plane, where cyclic shear stresses either produce iii) shear stress reversal or not. For this purpose, cyclic simple shear tests were performed on laboratory reconstituted sand samples. Additionally, cyclic test data were compiled from the available literature. On the basis of probabilistic assessment of this data, a unified correction scheme, which incorporates the interdependent effects of both overburden and static shear stresses along with the degree of cyclic shear stress reversal, has been developed.

Seismic site amplification formula was developed for virtual surface arrays to be consistent with vertical arrays using a number of KiK-net records during recent 8 destructive earthquakes. A correlation between peak spectrum amplification and S-wave velocity ratio (base V _s /surface V _s ) improved much better if the surface V _s was evaluated from fundamental mode frequency combined with a thickness of equivalent surface layer in which peak amplification is exerted, rather than using the conventional V _s30. Also found was that soil nonlinearity effect during strong earthquakes has only a marginal effect on the surface array amplification observed and evaluated by the formula using small-strain V_s-values. The theoretical basis for the minor nonlinear effect has been discussed by conducting a simple study on a two-layered system with strain-dependent soil nonlinearity. The proposed amplification formula seems also applicable with a satisfactory accuracy to sites where soil liquefaction is involved.

Few small earthquakes with local magnitude slightly larger than M_L = 4 were recorded by geotechnical downhole arrays that have been recently deployed in the west side of Istanbul. Same events were also recorded by Istanbul Rapid Response Network (IRRN) which comprises of 55 surface strong motion stations in the European side of Istanbul. The strongest one of these earthquakes took place on 12/3/2008 in Çınarcık with local magnitude of M_L = 4.8. Even though the observed PGAs were not exceeding 0.01 g, an effort is made to model the recorded response at the downhole array sites as well as the at the IRRN stations using the acceleration records obtained by the deepest sensors, i.e. on the engineering bedrock, at the downhole array sites as input bedrock motions. 1D equivalent linear site response analysis that is generally adopted for site-specific response analysis is used for modelling. Observations from the recorded response and results from 1D modelling of ground response have yielded in general good agreement between the observed and recorded soil response at the station sites.

In this study, a centrifuge model tests and effective stress analyses are performed on a breakwater subject to Tsunami such as those seriously damaged during 2011 East Japan Earthquake (Magnitude 9.0). The centrifuge model tests at a scale of 1/200 are performed to simulate the failure of a breakwater subject to Tsunami. With the effective stress analyses, this study demonstrates the importance of the mechanism of failure in the rubble mound due to seepage flow of pore water in addition to the force of Tsunami action.

It is important to analyze embankment dam behavior from the past lessons learned. It is noticed that modern embankment dams withstand the design earthquake without significant damages. In spite of this scenario it is important to prevent the occurrence of incidents and accidents of embankment dams under earthquakes and so a deep understanding of the triggering factors is needed.

Well documented case histories from different regions of the world related with embankment dam behaviour were carefully selected and are discussed.

The design and the analysis of dam stability under seismic conditions are addressed. The new trend for performance basis design is to consider 2 levels of seismic actions and to analyse the situation when the limit of force balance is exceeded for high intensity ground motions associated with a very rare seismic event.

The reservoir triggered seismicity (RTS) is linked to dams higher than about 100 m or to large reservoirs (capacity greater than 500 × 10⁶ m³), rate of reservoir filling and to new dams of smaller size located in tectonically sensitive areas.

Dam monitoring and inspections of dams are presented. Experience has shown that the rational and systematic control of dam safety should consist of several tasks: (i) regular instrumentation measurements; (ii) data validation; (iii) data storage; (iv) visual inspections; (v) safety evaluation; (vi) corrective actions.

The risks associated with dam projects are discussed. The potential risk associated with dams consists of structural components and socio-economic components. The structural components of potential risk depend mostly on storage capacity and on the height of the dam, as the potential downstream consequences are proportional to the mentioned values. Socio-economic risks can be expressed by a number of persons who need to be evacuated in case of danger and by potential downstream damage.

It is important to develop new ways of thinking and strategies to address the future challenges.

Nonlinear deformation analyses (NDAs) are used for evaluating the potential effects of liquefaction on the seismic performance of embankment dams. NDAs provide insights on potential failure modes and serve as a basis for generating fragility curves for risk analyses. Confidence in these applications of NDAs may be significantly improved by establishing calibration and analysis protocols that reduce undesirable sources of variability between analysis results obtained by different analysts or different models. A key protocol for improving the quality of NDAs is the calibration and validation of a constitutive model using single-element simulations with the types of loading paths potentially important to the structure's response. These simulations provide a basis for evaluating the strengths and weaknesses of the constitutive models and for identifying likely causes of differences between different modeling approaches. Examples of single-element simulations are presented for several constitutive models and loading paths to illustrate some of the important features for NDAs of embankment dams affected by liquefaction. The benefits of NDA studies also depend on the numerical procedures and results being sufficiently documented so that knowledgeable users can critique and interpret the results. Insights from parametric analyses should be adequately demonstrated so that members of the project team can exercise judgments regarding their practical implications. Recommendations on NDA documentation requirements are presented, along with an example of how such documentation can be important for the interpretation of NDA results. Critical examinations of NDA models and modeling procedures are necessary for identifying weaknesses, fostering improvements, and increasing confidence in their use for evaluating the seismic performance of embankment dams affected by liquefaction.

Soil liquefaction has been responsible for extensive damage to buildings, bridges, pipelines and other critical infrastructure in many past earthquakes. The need to predict the potential for initiation of liquefaction and of its various effects has led to greatly improved understanding of the mechanics of the liquefaction process. Most of the efforts made toward evaluating liquefaction hazards have been directed toward various aspects of ground failure, including post-liquefaction settlement, lateral spreading, and the development of flow slides. The occurrence of liquefaction, however, can also affect ground surface motions, and hence the seismic response of structures founded at or near the ground surface. This paper reviews the process of liquefaction and the manner in which its occurrence is typically detected. The hydraulic conditions required for the manifestation of surface effects are reviewed, and limitations in the inference of liquefaction triggering from the presence or absence of such effects are discussed. The effects of liquefaction triggering on soil stiffness, as they pertain to ground response, are described and discussed. Procedures for identification of the timing of liquefaction triggering are reviewed, and illustrated using a procedure based on the Stockwell transform.

The effects of post-triggering response on ground surface motions are interpreted in terms of response spectra. The abilities of four advanced site response programs to compute ground motions in liquefiable soils are examined and evaluated from the standpoints of pre- and post-triggering response. The programs were applied to the Wildlife case history, and to a hypothetical soil profile containing 7 m of liquefiable soil. The programs all predicted consistent response up to the point of triggering, but their predictions of post-triggering response were highly variable. The implications of post-triggering response for the evaluation of ground surface response spectra are discussed.

The gigantic earthquake in 2011 caused significant damage in lifeline in the Tokyo Metropolitan area. In particular, the damage was significant in recent artificial islands where liquefaction affected embedded sewage pipelines profoundly. The encountered problem is that the entire subsoil liquefied in addition to loose backfill soils, and that the damage of branch lines introduced liquefied sand into trunk sewage lines, leading to difficult problem of sand clogging. Moreover, the same problem is expected to occur in other areas where strong earthquakes are expected in near future. The present paper addresses the ongoing model tests by which a variety of mitigation measures for sewage pipelines are examined. In the regions where future earthquake is expected, it is not possible to excavate pipes and reconstruct backfills now because of financial limitations. To cope with this situation, less expensive measures such as mechanical constraint, partial injection of grout, or limited installation of drainage measures are studied.

On February 27, 2010, at 3:34 a.m. local time, a large earthquake of Magnitude 8.8 hit the Central-South region of Chile. A significant number of aftershocks followed the initial quake, being the most important of Magnitude 6.2, which occurred 20 min after the main shock. The 2010 Chile Earthquake has been identified as a thrust-faulting type that occurred on the interface between the Nazca and South American plates, at an average depth of 35 km. Seismologists have reported that the rupture zone covered a rectangular area of approximately 550 km by 170 km. High horizontal peak ground accelerations were recorded on soil deposits, being the maximum one 0.94 g at Angol City, located towards South of the rupture zone. To the north of the rupture zone, the horizontal peak ground accelerations (PGA) was recorded in Melipilla City, reaching a value of 0.78 g. Several of the available records show a ground motion that exceeds 2 min of duration, which may explain the significant amount of liquefied sites. Field observations have shown that the earthquake triggered liquefaction in more than 170 different sites, covering a north-south distance of about 950 km, which approximately corresponds to twice of the length of the rupture zone. Liquefaction phenomenon induced damages to the road infrastructure, railroads system, buildings and houses. Liquefaction-induced ground failure displaced and distorted pile foundations of piers impacting seriously the operation of several ports. Especially interesting is the case of Juan Pablo II Bridge, where significant differential settlements were observed, but almost non transversal lateral displacements took place.

Silty sand liquefaction has been of great interest of research in geotechnical earthquake engineering, especially for the silty sand containing a large amount of non-plastic fines. This article presents a case study on non-plastic silty sand liquefaction. A new sampling technique, Gel-Push sampler, that could retrieve undisturbed non-plastic silty sand and results of series dynamic triaxial tests those investigated influences of fines contents, void ratio, as well as sampling disturbance are reported. Research progress presented here is hoped to be helpful in understanding mechanism as well as consequences of non-plastic silty sand liquefaction.

The case history of a seismic retrofitting of a piled foundation is reported. To analyze and to empathize the importance of the different phases of the seismic design process, in the paper the approach adopted for the structural upgrading of a six storey reinforced concrete frame resisting building in the Saline district of Augusta (Italy), damaged by the Sicilian earthquake (M_L = 5.4) of 13th December 1990, is described and discussed. For evaluating the possibility to repair the building, an investigation on soil, structure and foundation was carried out. As building was founded on piles, for the seismic retrofitting of the structure, the response of the piles subjected to earthquake loading has been studied. Vertical and horizontal loading tests on single pile and pile groups were conducted to test the integrity of the piles after the seismic event. To extrapolate the behaviour of the piled foundation under the design seismic actions, a numerical analysis of inertial and kinematic effects have been performed. The site was well investigated by in-situ and laboratory tests. To estimate dynamic stress-strain geotechnical characteristics of soils resonant column tests were also performed. Loading tests showed that the seismic event of December 13th 1990 have not damaged the effectiveness of the soil-pile system. Nevertheless, as the numerical analysis showed that the existing foundation was enable to carry on design seismic actions, the structural upgrading and seismic retrofitting of the piled foundation required the enlargement of the foundation and new bored piles.

The main findings are summarized of a systematic research effort regarding the response of pile foundations in laterally spreading soils. The incentive of the research were findings from properly scaled (with regard to the pore fluid) centrifuge experiments suggesting that severe soil dilation may occur at the upper part of the pile, as a result of large soil-pile relative movement, causing soil pressures to significantly increase. In essence, this observation negates current design practice which univocally accepts that soil liquefaction drastically reduces seismic demands. To further explore this new evidence, a 3D nonlinear numerical methodology was developed and tested against the aforementioned experiments. On this basis valuable feedback was first gained with regard to current numerical techniques, with most important the need of a new type of tied-node boundaries for the simulation of submerged infinite slopes. Comparative analyses, with the old and the new boundaries, revealed that the former (which also reflect the kinematic response of the laminar box containers employed in model tests) can significantly underestimate soil pressures imposed to the foundation. In the sequel, the numerical methodology was applied parametrically (for various soil, pile and excitation characteristics) and a new set of multivariable relationships was statistically established for the practical estimation of ultimate soil pressures applied to the pile. Compared to existing empirical relationships, the new ones can capture the aforementioned dilation phenomena, which may develop at the upper part of the pile in the case of relatively low permeability soils (e.g. fine-grained, silty sands) and have an overall detrimental effect on pile response.

The use of Swedish weight sounding tests has been examined by the authors for a wide range of stability problems of natural and reclaimed soil deposits during earthquakes. The present study is aimed at integrating these studies into one framework and at facilitating the use of Swedish weight sounding tests for future earthquake reconnaissance investigations. The example case history is given to prove the usefulness of subsurface soil profiling by means of Swedish weight sounding tests, and the procedures for determining the undrained shear strength, liquefaction strength and post-liquefaction settlement are examined and described in detail. Swedish weight sounding tests prove to be useful in analyzing the field performance of natural and reclaimed soil deposits during earthquakes.