|FHWA > Bridge > Tunnels > Technical Manual for Design and Construction of Road Tunnels - Civil Elements|
Technical Manual for Design and Construction of Road Tunnels - Civil Elements
Chapter 13 - Seismic Considerations
Tunnels, in general, have performed better during earthquakes than have above ground structures such as bridges and buildings. Tunnel structures are constrained by the surrounding ground and, in general, can not be excited independent of the ground or be subject to strong vibratory amplification, such as the inertial response of a bridge structure during earthquakes. Another factor contributing to the reduced tunnel damage is that the amplitude of seismic ground motion tends to reduce with depth below the ground surface. Adequate design and construction of seismic resistant tunnel structures, however, should never be overlooked, as moderate to major damage has been experienced by many tunnels during earthquakes, as summarized by Dowding and Rozen (1978), Owen and Scholl (1981), Sharma and Judd (1991), and Power et al. (1998), among others. The greatest incidence of severe damage has been associated with large ground displacements due to ground failure, i.e., fault rupture through a tunnel, landsliding (especially at tunnel portals), and soil liquefaction. Ground shaking in the absence of ground failure has produced a lower incidence and degree of damage in general, but has resulted in moderate to major damage to some tunnels in recent earthquakes. The most recent reminder of seismic risk to underground structures under the ground shaking effect is the damage and near collapse at the Daikai and Nagata subway stations (Kobe Rapid Transit Railway) during the 1995 Kobe Earthquake in Japan . Near-surface rectangular cut-and-cover tunnels and immersed tube tunnels in soil have also been vulnerable to transient seismic lateral ground displacements, which tend to cause racking of a tunnel over its height and increased lateral pressures on the tunnel walls. Their seismic performance could be vital, particularly when they comprise important components of a critical transportation system (e.g., a transit system) to which little redundancy exists.
The general procedure for seismic design and analysis of tunnel structures should be based primarily on the ground deformation approach (as opposed to the inertial force approach); i.e., the structures should be designed to accommodate the deformations imposed by the ground. The analysis of the structure response can be conducted first by ignoring the stiffness of the structure, leading to a conservative estimate of the ground deformations. This simplified procedure is generally applicable for structures embedded in rock or very stiff/dense soil. In cases where the structure is stiff relative to the surrounding soil, the effect of soil-structure interaction must be taken into consideration. Other critical conditions that warrant special seismic considerations include cases where a tunnel intersects or meets another tunnel (e.g., tunnel junction or tunnel/cross-passage interface) or a different structure (such as a ventilation building). Under these special conditions, the tunnel structure may be restrained from moving at the junction point due to the stiffness of the adjoining structure, thereby inducing stress concentrations at the critical section. Complex numerical methods are generally required for cases such as these where the complex nature of the seismic soil-structure interaction system exists.
13.2 Determination of Seismic Environment
13.2.1 Earthquake Fundamental
General: Earthquakes are produced by abrupt relative movements on fractures or fracture zones in the earth's crust. These fractures or fracture zones are termed earthquake faults. The mechanism of fault movement is elastic rebound from the sudden release of built-up strain energy in the crust. The built-up strain energy accumulates in the earth's crust through the relative movement of large, essentially intact pieces of the earth's crust called tectonic plates. This relief of strain energy, commonly called fault rupture, takes place along the rupture zone. When fault rupture occurs, the strained rock rebounds elastically. This rebound produces vibrations that pass through the earth crust and along the earth's surface, generating the ground motions that are the source of most damage attributable to earthquakes. If the fault along which the rupture occurs propagates upward to the ground surface and the surface is uncovered by sediments, the relative movement may manifest itself as surface rupture. Surface ruptures are also a source of earthquake damage to constructed facilities including tunnels.
The major tectonic plates of the earth's crust are shown in Figure 13-1 (modified from Park, 1983). There are also numerous smaller, minor plates not shown on this figure. Earthquakes also occur in the interior of the plates, although with a much lower frequency than at plate boundaries.
Figure 13-1 Major Tectonic Plates and Their Approximate Direction of Movement.
For the continental United States, the principal tectonic plate boundary is along the western coast of the continent, where the North American Plate and the Pacific Plate are in contact. In California, the boundary between these plates is a transform fault wherein the relative movement is generally one of lateral slippage of one plate past the other. Elsewhere along the west coast (e.g., off the coast of Oregon, Washington, and Alaska ), the plate boundary is a subduction zone wherein one plate dives (subducts) beneath the other plate. In the western interior of the United States, adjacent to the western edge of the American Plate, there may be subplates that have formed as a result of subcrustal flow. Earthquake sources in Utah and Montana may be attributable to such subplate sources. Earthquake source areas in the central and eastern United States and along the Saint Lawrence Valley are within the American Plate and are considered to be intraplate source zones. The mechanisms generating earthquakes in these intraplate zones are poorly understood, but may be related to relief of locked-in stresses from ancient tectonic movements, crustal rebound from the ice ages, re-adjustment of stress in the interior of the plate due to boundary loads, sediment load such as the Mississippi River basin, or other unrecognized mechanisms. Earthquakes in Hawaii are believed to be associated with an isolated plume of molten rock from the mantle referred to as a hot spot.
The intensity and impact of earthquakes may be as great or greater in the plate interiors as they are at the active plate boundaries. The differences between plate boundary and intraplate earthquakes is in their geographic spread and the frequency of occurrence. Earthquake activity is much greater along the plate boundaries than in the plate interior. However, ground motions from intraplate earthquakes tend to attenuate, or dissipate, much more slowly than those from plate boundary events. Plate boundary faults are relatively longer than those in the plate interior and tend to be associated with a smaller stress drop (the stress drop is the sudden reduction of stress across the fault plane during rupture), longer duration of shaking, and a more frequent rate of earthquake occurrence.
Fault Movements: Faults are created when the stresses within geologic materials exceed the ability of those materials to withstand the stresses. Most faults that exist today are the result of tectonic activity that occurred in earlier geological times. These faults are usually non-seismogenic (i.e. incapable of generating earthquakes, or inactive). However, faults related to past tectonism may be reactivated by present-day tectonism in seismically active areas and can also be activated by anthropogenic (man-made) activities such as impoundment of a reservoir by a dam or injection of fluids (e.g. waste liquids) deep into the subsurface. The maximum size of an earthquake on an anthropogenically reactivated fault is a subject of some controversy, but earthquakes as large as moment magnitude 6.5 have been attributed to reservoir impoundment.
Not all faults along which relative movement is occurring are a source of earthquakes. Some faults may be surfaces along which relative movement is occurring at a slow, relatively continuous rate, with an insufficient stress drop to cause an earthquake. Such movement is called fault creep. Fault creep may occur along a shallow fault, where the low overburden stress on the fault results in a relatively low threshold stress for initiating displacement along the fault. Alternatively, a creeping fault may be at depth in soft and/or ductile materials that deform plastically. Also, there may be a lack of frictional resistance or asperities (non-uniformities) along the fault plane, allowing steady creep and the associated release of the strain energy along the fault. Fault creep may also prevail where phenomena such as magma intrusion or growing salt domes activate small shallow faults in soft sediments. Faults generated by extraction of fluids (e.g., oil or water in southern California ), which causes ground settlement and thus activates faults near the surface may also result in fault creep. Faults activated by other non-tectonic mechanisms, e.g. faults generated by gravity slides that take place in thick, unconsolidated sediments, could also produce fault creep.
Active faults that extend into crystalline bedrock are generally capable of building up the strain energy needed to produce, upon rupture, earthquakes strong enough to affect transportation facilities. Fault ruptures may propagate from the crystalline bedrock to the ground surface and produce ground rupture. Fault ruptures which propagate to the surface in a relatively narrow zone of deformation that can be traced back to the causative fault in crystalline rock are sometimes referred to as primary fault ruptures. Fault ruptures may also propagate to the surface in diffuse, distributed zones of deformation which cannot be traced directly back to the basement rock. In this case, the surface deformation may be referred to as secondary fault rupture.
Whether or not a fault has the potential to produce earthquakes is usually judged by the recency of previous fault movements. If a fault has propagated to the ground surface, evidence of faulting is usually found in geomorphic features associated with fault rupture (e.g., relative displacement of geologically young sediments). For faults that do not propagate to the ground surface, geomorphic evidence of previous earthquakes may be more subdued and more difficult to evaluate (e.g., near surface folding in sediments or evidence of liquefaction or slumping generated by the earthquakes). If a fault has undergone relative displacement in relatively recent geologic time (within the time frame of the current tectonic setting), it is reasonable to assume that this fault has the potential to move again. If the fault moved in the distant geologic past, during the time of a different tectonic stress regime, and if the fault has not moved in recent (Holocene) time (generally the past 11,000 years), it may be considered inactive. For some very important and critical facilities, such as those whose design is governed by the US Nuclear Regulatory Commission (NRC), a timeframe much longer than the 11,000-yr criterion has been used. In accordance with the US NRC regulations a fault is defined as "capable" (as opposed to "active") if it has shown activity within the past 35,000 years or longer.
Geomorphic evidence of fault movement cannot always be dated. In practice, if a fault displaces the base of unconsolidated alluvium, glacial deposits, or surficial soils, then the fault is likely to be active. Also, if there is micro-seismic activity associated with the fault, the fault may be judged as active and capable of generating earthquakes. Microearthquakes occurring within basement rocks at depths of 7 to 20 km may be indicative of the potential for large earthquakes. Microearthquakes occurring at depths of 1 to 3 km are not necessarily indicative of the potential for large, damaging earthquake events. In the absence of geomorphic, tectonic, or historical evidence of large damaging earthquakes, shallow microtremors may simply indicate a potential for small or moderate seismic events. Shallow microearthquakes of magnitude 3 or less may also sometimes be associated with mining or other non-seismogenic mechanisms. If there is no geomorphic evidence of recent seismic activity and there is no microseismic activity in the area, then the fault may be inactive and not capable of generating earthquakes.
In some instances, fault rupture may be confined to the subsurface with no relative displacement at the ground surface due to the fault movement. Subsurface faulting without primary fault rupture at the ground surface is characteristic of almost all but the largest magnitude earthquakes in the central and eastern United States . Due to the rarity of large magnitude intraplate events, geological processes may erase surface manifestations of major earthquakes in these areas. Therefore, intraplate seismic source zones often must be evaluated using instrumental seismicity and paleoseismicity studies. This is particularly true if the intraplate sources are covered by a thick mantle of sediments, as in the New Madrid, Tennessee, and Charleston, South Carolina, intraplate seismic zones. Instrumental recording of small magnitude events can be particularly effective in defining seismic source zones.
Essentially all of the active faults with surface fault traces in the United States are shallow crustal faults west of the Rocky Mountains . However, not all shallow crustal faults west of the Rocky Mountains have surface fault traces. Several recent significant earthquakes along the Pacific Coast plate boundary (e.g., the 1987 Whittier Narrows earthquake and the 1994 Northridge earthquake) were due to rupture of thrust (compressional) faults that did not break the ground surface, termed blind thrust faults.
A long fault, like the San Andreas Fault in California or the Wasatch Fault in Utah, typically will not move along its entire length at any one time. Such faults typically move in portions, one segment at a time. An immobile (or "locked") segment, a segment which has remained stationary while the adjacent segments of the fault have moved, is a strong candidate for the next episode of movement.
Type of Faults: Faults may be broadly classified according to their mode, or style of relative movement. The principal modes of relative displacement are illustrated in Figure 13-2 and are described subsequently.
Figure 13-2 Types of Fault Movement
Strike Slip Faults: Faults along which relative movement is essentially horizontal (i.e., the opposite sides of the fault slide past each other laterally), are called strike slip faults. Strike slip faults are often essentially linear (or planar) features. Strike slip faults that are not fairly linear may produce complex surface features. The San Andreas fault is a strike slip fault that is essentially a north-south linear feature over most of its length. Strike slip faults may sometimes be aligned in en-echelon fashion wherein individual sub-parallel segments are aligned along a linear trend. En-echelon strike slip faulting is sometimes accompanied by step over zones where fault displacement is transferred from adjacent strike slip faults. Ground rupture patterns within these zones may be particularly complex.
Dip Slip Faults: Faults in which the deformation is perpendicular to the fault plane may occur due to either normal (extensional) or reverse (compressional) motion. These faults are referred to as dip slip faults. Reverse faults are also referred to as thrust faults. Dip slip faults may produce multiple fractures within rather wide and irregular fault zones.
Other Special Cases: Faults that show both strike slip and dip slip displacement may be referred to as oblique slip faults.
Earthquake Magnitude: Earthquake magnitude, M, is a measure of the energy released by an earthquake. A variety of different earthquake magnitude scales exist. The differences among these scales is attributable to the earthquake characteristic used to quantify the energy content. Characteristics used to quantify earthquake energy content include the local intensity of ground motions, the body waves generated by the earthquake, and the surface waves generated by the earthquake. In the eastern United States, earthquake magnitude is commonly measured as a (short period) body wave magnitude, mb. However, the (long period) body wave magnitude, mB, scale is also sometimes used in the central and eastern United States . In California, earthquake magnitude is often measured as a local (Richter) magnitude, ML, or surface wave magnitude, Ms. The Japan Meteorological Agency Magnitude (MJMA) scale is commonly used in Japan .
Due to limitations in the ability of some recording instruments to measure values above a certain amplitude, some of these magnitude scales tend to reach an asymptotic upper limit. To correct this, the moment magnitude, Mw, scale was developed by seismologists (Hanks and Kanamori, 1979). The moment magnitude of an earthquake is a measure of the kinetic energy released by the earthquake. Mw is proportional to the seismic moment, defined as a product of the material rigidity, fault rupture area, and the average dislocation of the rupture surface. Moment magnitude has been proposed as a unifying, consistent magnitude measure of earthquake energy content. Figure 13-3 (Heaton, et al., 1986) provides a comparison of the various other magnitude scales with the moment magnitude scale.
Hypocenter and Epicenter and Site-to-Source Distance: The hypocenter (focus) of an earthquake is the point from which the seismic waves first emanate. Conceptually, it may be considered as the point on a fault plane where the slip responsible for an earthquake was initiated. The epicenter is a point on the ground surface directly above the hypocenter. Figure 13-4 shows the relationship between the hypocenter, epicenter, fault plane, and rupture zone of an earthquake. Figure 13-4 also shows the definition of the strike and dip angles of the fault plane.
The horizontal distance between the site of interest to the epicenter is termed epicentral distance, RE, and is commonly used in the eastern United States . The distance between the site and the hypocenter (more widely used in the western United States) is termed hypocentral distance, RH.
Figure 13-3 Comparison of Earthquake Magnitude Scales (Heaton, et al., 1986)
Figure 13-4 Definition of Basic Fault Geometry Including Hypocenter and Epicenter
13.2.2 Ground Motion Hazard Analysis
For the seismic design of underground tunnel facilities, one of the main tasks is to define the design earthquake(s) and the corresponding ground motion levels and other associated seismic hazards. The process by which design ground motion parameters are established for a seismic analysis is termed the seismic hazard analysis. Seismic hazard analyses generally involve the following steps:
Identification of seismic sources includes establishing the type of fault and its geographic location, depth, size, and orientation. Seismic source identification may also include specification of a random seismic source to accommodate earthquakes not associated with any known fault. Evaluation of the seismic potential of an identified source involves evaluation of the earthquake magnitude (or range of magnitudes) that the source can generate and, often times, the expected rate of occurrence of events of these magnitudes.
Identification of capable seismic sources together with evaluation of the seismic potential of each capable source may be referred to as seismic source characterization. Once the seismic sources are characterized, the intensity of ground motions at the project site from these sources must be characterized. There are three general ways by which the intensity of ground motions at a project site is assessed in practice. They are, in order of complexity: (1) use of existing hazard analysis results published by credible agencies such as US Geological Survey (USGS) and some State agencies; (2) project-specific and site-specific deterministic seismic hazard evaluation; and (3) project-specific and site-specific probabilistic seismic hazard evaluation. Which particular approach is adopted may depend on the importance and complexity of the project and may be dictated by regulatory agencies.
The choice of the design ground motion level, whether based upon probabilistic or deterministic analysis, cannot be considered separately from the level of performance specified for the design event. Sometimes, facilities may be designed for multiple performance levels, with a different ground motion level assigned to each performance level, a practice referred to as performance based design. Common performance levels used in design of transportation facilities include protection of life safety and maintenance of function after the event. A safety level design earthquake criterion is routinely employed in seismic design. Keeping a facility functional after a large earthquake adds another requirement to that of simply maintaining life safety, and is typically required for critical facilities.
The collapse of a modern transportation tunnel (particularly for mass transit purpose) during or after a major seismic event could have catastrophic effects as well as profound social and economical impacts. It is typical therefore for modern and critical transportation tunnels to be designed to withstand seismic ground motions with a return period of 2,500 years, (corresponding to 2 % probability of exceedance in 50 years, or 3% probability of exceedance in 75 years). In addition, to avoid lengthy down time and to minimize costly repairs, a modern and critical transportation tunnel is often required to withstand a more frequent earthquake (i.e., a lower level earthquake) with minimal damage. The tunnel should be capable of being put immediately back in service after inspection following this lower level design earthquake. In the high seismic areas, this lower level earthquake is generally defined to have a 50% probability of probability of exceedance 75 years, corresponding to a 108-year return period. In the eastern United States, where earthquake occurrence is much less frequent, the lower level design earthquake for modern and critical transportation tunnels is generally defined at a higher return period such as 500 years.
Use Of Existing Hazard Analysis Results: Information used for seismic source characterization can often be obtained from publications of the United States Geological Survey (USGS), or various state agencies. These published results are often used because they provide credibility for the designer and may give the engineer a feeling of security. However, if there is significant lag time between development and publication, the published hazard results may not incorporate recent developments on local or regional seismicity. Furthermore, there are situations where published hazard results may be inadequate and require site-specific seismic hazard evaluation. These situations may include: (1) the design earthquake levels (e.g., in terms of return period) are different than those assumed in the published results, (2) for sites located within 6 miles of an active surface or shallow fault where near-field effect is considered important, and (3) the published hazard results fail to incorporate recent major developments on local or regional seismicity.
Seismic hazard maps that include spectral acceleration values at various spectral periods have been developed by USGS under the National Earthquake Hazard Reduction Program (NEHRP). Map values for peak and spectral accelerations with a probability of being exceeded of 2 percent, 5 percent, and 10 percent in 50 years (corresponding approximately to 2,500-yr, 1,000-yr, and 500-yr return period, respectively) can be recovered in tabular form. Figure 13-5 below shows an example of the national ground motion hazard maps in terms of peak ground acceleration (in Site Class B - Soft Rock Site) for an event of 2% probability of exceedance in 50 Years (i.e., 2,500-yr Return Period). In addition, USGS also provides information (e.g., the de-aggregated hazard) that can be used to estimate the representative "magnitude and distance" for a site in the continental United States.
Figure 13-5 National Ground Motion Hazard Map by USGS (2002) - Peak Ground Acceleration with 2% Probability of Exceedance in 50 Years (2,500-yr Return Period) - for Site Class B, Soft Rock
The Deterministic Hazard Analysis Approach: In a deterministic seismic hazard analysis, the seismologist performing the analysis first identifies the capable seismic sources and assigns a maximum magnitude to each source. Then, the intensity of shaking at the site from each capable source is calculated and the design earthquake is identified based on the source capable of causing the greatest damage. The steps in a deterministic seismic hazard analysis are as follows:
The deterministic analysis approach provides a framework for the evaluation of worst-case scenarios at a site. It provides little information about the likelihood or frequency of occurrence of the governing earthquake. If such information is required, a probabilistic analysis approach should be used to better define the seismic ground motion hazard.
The Probabilistic Hazard Analysis Approach: A probabilistic seismic hazard analysis incorporates the likelihood of a fault rupturing and the distribution of earthquake magnitudes associated with fault rupture into the assessment of the intensity of the design ground motion at a site. The objective of a probabilistic seismic hazard analysis is to compute, for a given exposure time, the probability of exceedance corresponding to various levels of a ground motion parameter (e.g., the probability of exceeding a peak ground acceleration of 0.2 g in a 100-year period). The ground motion parameter may be either a peak value (e.g., peak ground acceleration) or a response spectra ordinate associated with the strong ground motion at the site. The probabilistic value of the design parameter incorporates both the uncertainty of the attenuation of strong ground motions and the randomness of earthquake occurrences. A probabilistic seismic hazard analysis usually includes the following steps, as illustrated in Figure 13-6 :
Figure 13-6 General Procedure for Probabilistic Seismic Hazard Analysis
13.2.3 Ground Motion Parameters
Once the design earthquake events are defined, design ground motion parameters are required to characterize the design earthquake events. Various types of ground motion parameters may be required depending on the type of analysis method used in the design. In general, ground motions can be characterized by three translational components (e.g., longitudinal, transverse, and vertical with respect to the tunnel axis). The various types of common ground motion parameters are described in the following paragraphs.
Peak Ground Motion Parameters: Peak ground acceleration (PGA), particularly in the horizontal direction, is the most common index of the intensity of strong ground motion at a site. Peak ground velocity (PGV) and peak ground displacement (PGD) are also used in some engineering analyses to characterize the damage potential of ground motions. For seismic design and analysis of underground structures including tunnels, the PGV is as important as the PGA because ground strains (or the differential displacement between two points in the ground) can be estimated using the PGV. PGA values are generally available from published hazard results such as those from the USGS hazard study. Attenuation relations are also generally available for estimating PGA values. However, there has been little information in the past for estimating the PGV values. Previous studies have attempted to correlate the PGV with PGA by establishing PGV-to-PGA ratios (as a function of earthquake magnitudes, site soil conditions, and source-to-site distance in some cases). However, these correlations were derived primarily from ground motion database in the Western United States (WUS) and failed to account for the different ground motion characteristics in the Central and Eastern United States (CEUS). Recent study (NCHRP-12-70, 2008) has found that PGV is strongly correlated with the spectral acceleration at 1.0 second (S1). Using published strong motion data, regression analysis was conducted and the following correlation has been recommended for design purposes.
PGV is in in/sec
The development of the PGV-S1 correlation is based on an extensive earthquake database established from recorded accelerograms representative of both rock and soil sites for the WUS and CEUS. The earthquake magnitude was found to play only a small role and is not included in the correlation in developing Equations 13-1 and 13-2. Equation 13-1 is based on the mean plus one standard deviation from the regression analysis (i.e., 1.46 x the median value) for conservatism.
Design Response Spectra: Response spectra represent the response of a damped single degree of freedom system to ground motion. Design response spectra including the consideration of soil site effects can be established using code-specified procedures such as those specified in the NEHRP (National Earthquake Hazards Reduction Program) publications or the new AASHTO LRFD Guide Specifications using the appropriate design earthquake parameters consistent with the desirable design earthquake hazard levels (refer to discussions in Section 13.2.2). Figure 13-7 illustrates schematically the construction of design response spectra using the NEHRP procedure. The terms and parameters used in Figure 13-7 are documented in details in NEHRP 12-70 (2008) and in AASHTO LRFD Bridge Design Specifications (2008 Interim Provisions). Alternatively, project-specific and site-specific hazard analysis can also be performed to derive the design response spectra. Site-specific dynamic soil response analysis can also be performed to study the effects of the local soil/site conditions (site effects).
Figure 13-7 Design Response Spectra Constructed Using the NEHRP Procedure.
It should be noted that while the design response spectra are commonly used for the seismic design and analysis of above-ground structures such as bridges and buildings, they are not as useful in the seismic evaluation for underground structure. This is because response spectra are more relevant for evaluating the inertial response effect of above-ground structures while for underground structures, ground strains or ground displacements are the governing factor. Nevertheless, design response spectra effectively establish the ground motion shaking intensity level and can be used for deriving other ground motion parameters that are useful and relevant for underground structures. For example, using the design spectral acceleration at 1.0 sec (SD1), PGV can be estimated using the empirical correlation discussed above (Equation 13-1). In addition, design response spectra can also be used as the target spectra for generating the design ground motion time histories which in turn can be used in seismic analysis for underground structures if more refined numerical analysis is required.
Ground Motion Time histories and Spatially Varying Ground Motion Effects: The developed time histories should match the target design response spectra and have characteristics that are representative of the seismic environment of the site and the local site conditions. Characteristics of the seismic environment of the site to be considered in selecting time-histories include: tectonic environment (e.g., subduction zone; shallow crustal faults in WUS or similar crustal environment; CEUS or similar crustal environment); earthquake magnitude; type of faulting (e.g., strike-slip; reverse; normal); seismic-source-to-site distance; local site conditions; and design or expected ground-motion characteristics (e.g., design response spectrum; duration of strong shaking; and special ground-motion characteristics such as near-fault characteristics).
It is desirable to select time-histories that have been recorded under conditions similar to the seismic conditions (as described above) at the site, but compromises are usually required because of the multiple attributes of the seismic environment and the limited data bank of recorded time-histories. Selection of time-histories having similar earthquake magnitudes and distances, within reasonable ranges, are especially important parameters because they have a strong influence on response spectral content, response spectral shape, duration of strong shaking, and near-source ground-motion characteristics.
For long structures such as tunnels, different ground motions may be encountered by different parts of the structure. Thus, it is sometime necessary for the tunnel to be evaluated for the spatially varying ground motions effects, particularly when the longitudinal response of the tunnel is of concern (refer to discussions in Section 13.5.2). In this case the differential displacements and force buildup along the length of the tunnel could be induced due to the spatially varying ground motion effects. In deriving the spatially varying ground motion time histories, as a minimum the following factors should be taken into considerations:
Ground Motion Parameters Attenuation with Depth: The ground motions parameters discussed above are typically established at ground surface. Tunnels, however, are generally constructed at some depth below the ground surface. For seismic evaluation of the tunnel structure, the ground motion parameters should be derived at the elevation of the tunnel. Because ground motions generally decrease with depth below the ground surface, these parameters generally have lower values than estimated for ground surface motions (e.g., Chang et al., 1986). The ratios of ground motion values at tunnel depths to those at the ground surface may be taken as the ratios summarized in Table 13-1 unless lower values are justified based on site-specific assessments.
For more accurate assessment of the ground motion parameters at depth, site-specific dynamic site response analysis should be performed to account for detailed subsurface conditions and site geometry. Results from the dynamic site response analysis would provide various aspects of ground motion parameters as a function of depth (in a one-dimensional site response analysis) or as a function of spatial coordinates (in a two- or three-dimensional site response analysis).
13.3 Factors that Influence Tunnel Seismic Performance
The main factors influencing tunnel seismic performance generally can be summarized as (1) seismic hazard, (2) geologic conditions, and (3) tunnel design, construction, and condition. Each of these factors is briefly described in the following sections.
13.3.1 Seismic Hazard
In a broad sense, earthquake effects on underground tunnel structures can be grouped into two categories: (1) ground shaking, and (2) ground failure. Based on tunnel performance records during past earthquakes, the damaging effects of ground failure on tunnels are significantly greater than the ground shaking effects.
Ground Shaking: Ground shaking refers to the vibration of the ground produced by seismic waves propagating through the earth's crust. The area experiencing this shaking may cover hundreds of square miles in the vicinity of the fault rupture. The intensity of the shaking attenuates with distance from the fault rupture. Ground shaking motions are composed of two different types of seismic waves, each with two sub-types, described as follows:
As the ground is deformed by the traveling waves, any tunnel structure in the ground will also be deformed, since tunnel structures are constrained by the surrounding medium (soil or rock). As long as the ground (i.e., the surrounding medium) is stable, the structures cannot move independently of the ground. Therefore, the design and analysis of underground structures is based on ground deformations/strains rather than ground acceleration values. If the magnitude of ground deformation during earthquakes is small, the seismic effect on tunnels is negligible. For example, there is generally little concern for tunnel sections constructed in reasonably competent rock because the seismically induced deformations/strains in rock are generally very small, except when shear/fault zones are encountered or when there are large loosened rock pieces behind the lining. In loose or soft soil deposits, on the other hand, the soil deformation developed during the design earthquake(s) should be estimated and used for the structure's design and analysis. In general the potential effects of ground shaking range from minor cracking of a concrete liner to collapse of the liner and major caving of geologic materials into the tunnel.
Ground Failure: Ground failure broadly includes various types of ground instability such as fault rupture, tectonic uplift and subsidence, landsliding, and soil liquefaction. Each of these hazards may be potentially catastrophic to tunnel structures, although the damages are usually localized. Design of a tunnel structure against ground instability problems is often possible, although the cost may be high.
If an active fault crosses the tunnel alignment, there is a hazard of direct shearing displacement through the tunnel in the event of a moderate to large magnitude earthquake. Such displacements may range from a few inches to greater than ten feet and, in many cases, may be concentrated in a narrow zone along the fault. Fault rupture can and has had very damaging effects on tunnels. Tectonic uplift and subsidence can have similar damaging effects to fault rupture, if the uplift/subsidence movements cause sufficient differential deformation of the tunnel.
Landsliding through a tunnel, whether statically or seismically induced, can result in large, concentrated shearing displacements and either full or partial collapse of tunnel cross sections. Landslide potential is greatest when a preexisting landslide mass intersects the tunnel. A statically stable landslide mass may be activated by earthquake shaking. The hazard of landsliding is usually greatest in shallower parts of a tunnel alignment and at tunnel portals.
For tunnels located in soils below the groundwater table, there could be a potential for liquefaction if loose to medium-dense cohesionless soils (sands, silts, gravels) are adjacent to the tunnel. Potential effects of liquefaction of soils adjacent to a tunnel include: (a) increased lateral pressures on the lining or walls of the tunnel, which could lead to failure of the lining or walls depending on their design; (b) flotation or sinking of a tunnel embedded in liquefied soil, depending on the relative weight of the tunnel and the soils replaced by the tunnel; and (c) lateral displacements of a tunnel if there is a free face toward which liquefied soil can move and/or if the tunnel is constructed below sloping ground.
13.3.2 Geologic Conditions
Other unfavorable geologic conditions could lead to unsatisfactory seismic tunnel performance unless recognized and adequately accounted for in the tunnel design and construction. Unfavorable geologic conditions include: soft soils; rocks with weak planes intersecting a tunnel, such as shear zones or well developed weak bedding planes and well developed joint sets that are open or filled with weathered and decomposed rock; failures encountered during tunnel construction that may have further weakened the geologic formations adjacent to a tunnel (e.g., cave-ins or running ground leaving incompletely filled voids or loosened rock behind a lining; squeezing ground with relatively low static factor of safety against lining collapse); and adjacent geologic units having major contrasts in stiffness that can lead to stress concentrations or differential displacement.
13.3.3 Tunnel Design, Construction, and Condition
Elements of tunnel design, construction, and condition that may influence tunnel seismic behavior include:
In evaluating an existing tunnel in the screening stage or in a more detailed evaluation, or in designing retrofit measures, it is important to obtain as complete information as possible on the tunnel design, construction, and condition and the geologic conditions along the tunnel alignment. To obtain this information, the design and evaluation team should review the design drawings and design studies, as-built drawings, construction records as contained in the construction engineer daily reports and any special reports, maintenance and inspection records, and geologic and geotechnical reports and maps. Special inspections and investigations may be needed to adequately depict the existing conditions and determine reasons for any distress to the tunnel.
13.4 Seismic Performance and Screening Guidelines of Tunnels
13.4.1 Screening Guidelines Applicable to All Types of Tunnels
There are certain conditions that would clearly indicate a potentially significant seismic risk to a bored tunnel, cut-and-cover tunnel, or submerged tube and thus require more detailed evaluations. These conditions include:
In addition to the above, detailed seismic evaluations should also be conducted for tunnels that are considered lifeline structures (important and critical structures) that must be usable or remain open to traffic immediately after the earthquake. Transit tunnels in metropolitan areas are often considered as critical/lifeline structures and, therefore, warrant detailed seismic evaluations.
13.4.2 Additional Screening Guidelines for Bored Tunnels
If the above conditions do not exist, then the risk to a bored tunnel is a function of the tunnel design and construction, the characteristics of the geologic media, and the level of ground shaking. In this section, additional screening guidelines are presented considering these factors and empirical observations of tunnel performance during earthquakes.
It should be noted that although not as damaging as ground failure effects, ground shaking effect alone (i.e., in the absence of ground failure) has resulted in moderate to major damage to many tunnels in earthquakes. Figure 13-8 shows a highway tunnel experiencing lining falling off from tunnel crown under the ground shaking effect during the 2004 Niigata Earthquake in Japan. In another incident, the 1999 Koceali Earthquake in Turkey caused the collapse of two tunnels (the Bolu Tunnels) constructed using NATM method (15 m arch high and 16 m wide). At the time of the earthquake, the collapsed section of the tunnel had been stabilized with steel rib, shotcrete, and anchors.
Figure 13-8 Highway Tunnel Lining Falling from Tunnel Crown - 2004 Niigata Earthquake, Japan.
Figure 13-9 presents a summary of empirical observations of the effects of seismic ground shaking on the performance of bored/mined tunnels. The figure is from the study by Power et al. (1998), which updates earlier presentations of tunnel performance data by Dowding and Rozen (1978), Owen and Scholl (1981), and Sharma and Judd (1991). The data are for damage due only to shaking; damage that was definitely or probably attributed to fault rupture, landsliding, and liquefaction is not included. The data are for bored/mined tunnels only; data for cut-and-cover tunnels and submerged tubes are not included in Figure 13-9.
Figure 13-9 Summary of Observed Bored/Mined Tunnel Damage under Ground Shaking Effects (Power et al., 1998)
Figure 13-9 incorporates observations for 192 tunnels from ten moderate to large magnitude earthquakes (moment magnitude MW 6.6 to 8.4) in California, Japan, and Alaska . Ninety-four of the observations are from the moment magnitude MW 6.9 1995 Kobe, Japan, earthquake. This earthquake produced by far the most observations for moderate to high levels of shaking (estimated peak ground accelerations, PGA, at ground surface above the tunnels in the range of about 0.4 g to 0.6 g for the Kobe data). Peak ground accelerations in Figure 13-9 are estimated for actual or hypothetical outcropping rock conditions at ground surface above the tunnel. Other observations are from moderate to large (MW 6.7 to 8.4) earthquakes in California and Japan . Figure 13-9 shows the level of damage induced in tunnels with different types of linings subjected to the indicated levels of ground shaking. Damage was categorized into four states: none for no observable damage; slight for minor cracking and spalling; moderate for major cracking and spalling, falling of pieces of lining and rocks; and heavy for major cave-ins, blockage, and collapse. The figure indicates the following trends:
The trends in Figure 13-9 can be used as one guide in assessing the need for further evaluations of the effects of ground shaking on bored/mined tunnels.
13.4.3 Additional Screening Guidelines for Cut-and-Cover Tunnels
Reporting on the seismic performance of shallow cut-and-cover box-like tunnels has been relatively poor in comparison to the performance of bored/mined tunnels. This was especially evident during the 1995 Kobe, Japan, earthquake (O'Rourke and Shiba, 1997; Power et al., 1998). Figure 13-10 and Figure 13-11 show the damage to the center columns of the cut-and-cover tunnels running between Daikai and Nagata Stations during the 1995 Kobe Earthquake.
Figure 13-10 Fracture at Base of Columns of Cut-and-Cover Tunnel between Daikai and Nagata Stations - 1995 Kobe Earthquake, Japan
Figure 13-11 Shear Failure at Top of Columns of Cut-and-Cover Tunnel Between Daikai and Nagata Stations - 1995 Kobe Earthquake, Japan
The 1995 Kobe Earthquake also caused a major collapse of the Daikai subway station which was constructed by cut-and-cover method without specific seismic design provisions. The schematic drawing shown in Figure 13-12 (Iida et al., 1996) shows the collapse experienced by the center columns of the station, which was accompanied by the collapse of the ceiling slab and the settlement of the soil cover by more than 2.5 m.
Figure 13-12 Daikai Subway Station Collapse - 1995 Kobe Earthquake, Japan
The relatively poor performance of cut-and-cover tunnels under the ground shaking effect may reflect: (1) relatively softer near-surface geologic materials surrounding these types of structures as compared to the harder materials that often surround bored tunnels at greater depths; (2) higher levels of acceleration at and near the ground surface than at depth (due to tendencies for vibratory ground motions to reduce with depth below the ground surface); and (3) vulnerability of these box-like structures to seismically induced racking deformations of the box cross section (Refer to Figure 13-13 in Section 13.5), unless specifically designed to accommodate these racking deformations. Cut-and-cover tunnels in soil tend to be more vulnerable than those excavated into rock because of the larger soil shear deformations causing the tunnel racking. Tunnels in soft soil may be especially vulnerable. The most important determinant in assessing whether more detailed seismic evaluations of cut-and-cover tunnels are required is whether the original design considered loadings and deformations consistent with the seismic environment and geologic conditions, and especially, whether racking behavior was taken into account in the seismic analysis, design, and detailing of the structure.
13.4.4 Additional Screening Guidelines for Immersed Tubes
Submerged tubes are particularly susceptible to permanent ground movements during seismic shaking. Tubes are typically located at shallow depths and in soft or loose soils. Liquefaction of loose cohesionless soils may cause settlement, uplift (flotation), or lateral spreading. Earthquake shaking may also cause permanent displacement of soft clay soils on sloping ground. Joints connecting tube segments must accommodate the relative displacement of adjacent segments while maintaining a watertight seal. Generally, submerged tubes can be screened out from more detailed evaluations if the original design appropriately considered and analyzed the potential for ground failure modes and if joints have been carefully designed to achieve water tightness.