U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-11-030
Date: March 2011
Post-Earthquake Reconnaissance Report on Transportation Infrastructure: Impact of the February 27, 2010, Offshore Maule Earthquake in Chile
CHAPTER 7. CONCLUSIONS AND RECOMMENDATIONS
Based on the overall performance of various bridge components, retaining walls, and other structures summarized in chapter 6, conclusions can be drawn about the seismic performance of the transportation infrastructure in Chile during the offshore Maule earthquake. Considering the geologic conditions, earthquake settings, ground motions, and geotechnical and structural performance of the infrastructure, recommendations are made in this chapter for both immediate implementation and future investigation. In addition, research needs are identified for the improvement of seismic design of bridges, walls, and other structures used in transportation facilities. These recommendations may also be applicable for seismic retrofit purposes.
Some of the lessons learned require additional study to specifically assess how such lessons should be implemented. Toward that end, immediate needs are identified to gather more information and confirm assumptions and interpretations made in this report.
Less than 0.15 percent of the bridges in the MOP inventory collapsed or suffered damage that rendered them unusable. Considering the magnitude of the earthquake, the transportation infrastructure in Chile performed relatively well and was able to support postearthquake emergency response and recovery. Even so, improvements can be made to further enhance the seismic performance of the infrastructure both in structural and geotechnical terms.
Many spans of precast prestressed discontinuous girder bridges with continuous decks fell from their supports, probably due to the significant in-plane rotation of their superstructures due to severe shaking. Lateral steel stoppers used to provide both vertical and lateral restraint to the girders were largely unsuccessful due to their inadequate connection detail to cap beams or abutments. Conversely, reinforced concrete shear keys served their design purpose, transferring lateral loads to the substructures. Vertical seismic bars were widely used to restrain the vertical motion of the spans, and they also performed well. Bridge substructures (foundation, column, and cap beam) generally behaved satisfactorily (undamaged cover concrete and no evidence of yielded reinforcing steel) except that several columns suffered shear failure due to ground settlement and lateral spreading. However, the satisfactory performance of the columns generally occurred in bridges with failed superstructure-to-substructure connections, and it is not clear whether this superior performance was due to overdesign of the columns or to the failure of the connections, which limited the forces in the columns to less than yield. All retaining walls exceeded performance expectations.
In past earthquakes, full-depth diaphragms between girders have been very effective in distributing earthquake loads among the girders and to the bearings and substructure below. Bridges with full-depth diaphragms again performed well in Chile, whereas those without diaphragms performed poorly. To minimize the possibility of lateral shear failure and transverse unseating of girders, it is recommended that full-depth diaphragms be installed in all bridges and implemented in all future designs.
Adequate support length is a proven countermeasure to the unseating of bridge spans. It is recommended that generous seat lengths be provided in all new designs based on seismic hazard, soil type, column height, distance between movement joints, and angle of skew. Both the AASHTO specifications and Caltrans' seismic design criteria have examples of such requirements. Existing bridges with inadequate support lengths should be retrofitted with seat extenders or longitudinal restrainers as a matter of priority.
Lateral steel stoppers of girders were typically anchored to their supports with two anchor bolts aligned in the direction of the span. In this case, only one bolt was available to transfer the overturning moment when the seismic load from the girder was applied to the stopper some distance above the cap beam or abutment seat. This situation is quite ineffective. It is recommended that at least four anchor bolts be used to anchor each stopper to the cap beam in a configuration that provides both transverse and span-wise resistance to shear and overturning moments.
Unless specifically designed to fail at lateral load that is less than the capacity of the column below (thus protecting the column), all shear keys (both steel and concrete) should be designed against failure. This can be done in one of two ways: either the key or stopper is designed to resist the unreduced load from an elastic analysis (strength reduction factor = 1.0 or 0.8), or it is designed to resist the maximum shear the column can generate in its fully yielded state (i.e., with fully developed plastic hinges and including overstrength effects). Regardless of the method, the strength of any shear key should not be less than a specified minimum, usually expressed as a percentage of the tributary weight at the key.
Shear keys are not necessary if the bearings are engineered to take the lateral loads and uplift forces. In such cases, the bearings need to have sufficient internal strength to transfer the forces without rupture and remain functional. They also need to be anchored accordingly. If it is uneconomical to provide bearings of this type, shear keys are required. However the bearings should be anchored to the cap beams or abutments to prevent being dislodged during shaking.
Confinement has been demonstrated to be an effective means for preventing shear failure in reinforced concrete columns. It is recommended that columns with inadequate transverse reinforcing steel be retrofitted with external concrete, steel, or polymer jackets to prevent brittle shear failures and to help ensure flexural ductile yielding.
Since retaining walls performed very well in general, new design specifications could be developed to provide a no-analysis seismic design option for internal and external wall stability (i.e., for sliding, eccentricity, and bearing). Such an option might only be applicable within certain limits of ground acceleration, wall height, and surcharge conditions and would require minimum wall details (e.g., engineered connectivity of vertical joints in the face of the wall, especially at the corners; use of well-graded granular backfill; and good coping details on top of the wall).
The minimum wall details developed for the no-analysis option should also be required for walls that do not qualify for this option such as those in high seismic zones and should be considered for inclusion in the AASHTO specifications.
Bridges supported on MSE walls also appeared to have performed well in this earthquake. It is recommended that this cost-effective technology be investigated as an alternative to conventional, CIP, reinforced concrete abutment walls.
Conduct a rigorous study of earthquake ground motion records to look for evidence of a rotational component in ground motion that could explain the in-plane rotation of non skewed symmetric bridges.
The ground motions recorded during the offshore Maule earthquake can be characterized by long duration and multiple pulses, particularly at those stations near Santiago where soils are relatively stiff compared to other regions such as Constitución and Concepción. Although farther from the epicenter of the earthquake, the bridges near Santiago exclusively experienced significant in-plane rotations in their superstructures during ground shaking. The reason for this rotation is unclear, but one possibility is that the stiffer soils near Santiago favored the propagation and amplification of rotational ground motions.
To the authors' knowledge, the occurrence of rotational components in earthquake ground motions and their effect on bridge response and safety has not been previously studied. It is recommended that the ground motions recorded during this earthquake be rigorously analyzed and that case studies be undertaken for the non-skewed bridges that exhibited strong rotational response.
Study the relationship between the strength reduction factor, ductility, and period of vibration during long-duration ground motions with multiple pulses. Also study the influence of these characteristics in the ground motion on the in-plane rotation of non-skewed symmetric bridges. Prepare recommendations for AASHTO consideration regarding long-duration effects and potential changes to response modification factors.
It is known that long duration and multiple pulses can have a significant effect on the inelastic response of a bridge and can directly impact the integrity of the load path, especially if the bridge behaves as a pure elastoplastic system. In these circumstances, the relationship between the strength reduction factor, ductility, and period of vibration may differ from that obtained under more conventional ground motions and may warrant the use of a different set of response modification factors. The unique long-duration and multipulse feature of the ground motion may also affect the rotational response of bridge superstructures. These unique features in the ground motion and the corresponding inelastic response in the structure and approach fills may also explain the in-plane rotation of the superstructures in non-skewed bridges.
Considering the similarities in geologic and tectonic conditions, design specifications, and construction practices between the Chile and the United States, indepth studies on the effect of long-duration ground motions (with and without multiple pulses) on bridge behavior during this earthquake would be of immediate relevance to U.S. practice, particularly in Oregon and Washington. Research outcomes are likely to affect ground motion provisions as well as design requirements for bridges in these states.
Validate the support length requirements in the AASHTO specifications for skewed bridges.
Many spans of skewed bridges were unseated, confirming the sensitivity of this kind of bridge to collapse. The minimum requirements in the AASHTO specifications are based on engineering judgment rather than rigorous analysis, and this earthquake provides the opportunity to check the validity of these requirements, particularly for skew. Research outcomes would likely affect the support length requirements for bridges throughout the United States.
Study the implications of combining the scour and earthquake load cases for coastal bridges located in tsunami inundation zones.
The earthquake-induced tsunami generated significant waves and currents that added both horizontal and vertical loads on some coastal bridges and caused the erosion of soils around bridge foundations. These additional loads may further damage a bridge that has already been damaged by earthquake shaking. Structures may collapse and low-lying bridges may be lifted off their substructures and swept upstream. Although designing for scour is required in U.S. specifications, the combination of the effect of scour and earthquake effects is not required, as in the Chilean code (see appendix B). These combined effects warrant further investigation.
Conduct feasibility studies of bridges without diaphragms for potential application in accelerated bridge construction.
Concrete blocks (or shear keys) on each side of cap beam girders are an alternative to diaphragms for the distribution of transverse loads between girders and from the deck slab to the cap beam. They can effectively prevent the girders from shear failure by reducing the maximum load on each individual girder. If the joint between the top flange of the girder and deck slab can be economically designed to transfer significant moments without distress, this alternative may be an attractive strategy in accelerated bridge construction.
Conduct case studies on the AASHTO global design strategy that permits fusing of the super-to-substructure connection and protection of the substructures and foundations from seismic damage.
Adjusting the strength of the shear keys in a bridge superstructure is an alternative strategy for managing the load path and energy dissipation, thus enabling cost-effective designs that ensure bridge substructures suffer little or no damage. This strategy is permitted in the new AASHTO specifications for seismic design, but no comprehensive study has been done to date to prove its viability. This earthquake provides the opportunity to fill this gap.
Study the trade-off between lower maintenance costs with fewer joints and reduced redundancy for longitudinal seismic loads in continuous bridges.
Jointless bridges are widely favored for their lower corrosion-related maintenance costs and the increase in redundancy for gravity loads. However, the longitudinal distribution of load in these continuous bridges can be a challenge, especially in those cases where the superstructures are not monolithic with their piers (i.e., supported on bearings that allow thermal expansion). In long, multispan bridges, very large loads can be attracted into a few structural elements, and, unless designed for this heavy demand, failure can occur in the super-to-substructure connections at these elements. Studies are required to understand the trade-off between lower maintenance costs (with fewer joints) and the increased demand on a few substructures (i.e., the reduced redundancy for lateral loads).
As previously noted, geotechnical conclusions in this report are based on surficial observations and approximate geometry. Subsurface data and foundation and geometric details are required for most of the sites visited to confirm these preliminary interpretations. These data are a necessary first step towards deriving the most benefit from the investigation of this earthquake. Digitized ground motion records are also required if the more detailed modeling necessary to confirm the preliminary conclusions is to be achieved. Once site-specific data are obtained, future efforts in the geotechnical area should focus on the following:
Develop the ground motion records from this earthquake for use in seismic design when long-duration subduction zone records are needed and add them to databases used by designers for conducting site-specific seismic analyses.
Achieving this goal will involve obtaining subsurface information and other site information where the ground motions were recorded so that they can be deconvolved to base rock motions, cleaned up, baselined, etc. in order to make them most useable. Such digital records would also be extremely useful for design where subduction zone earthquakes must be considered (e.g., the west coast of North America).
Conduct research to better quantify the beneficial effect of three-dimensional geometry issues on abutment performance when liquefaction occurs.
For example, how wide does the fill need to be to eliminate the three-dimensional benefit identified in chapter 5 and chapter 6 such that the use of a two-dimensional slope stability or lateral spreading analysis in the direction toward the bridge is not too conservative? The case histories summarized in this report should be used to calibrate three-dimensional numerical models to investigate this issue more fully.
Conduct research to improve the quantification of lateral and vertical (downdrag) forces on foundations and abutments induced by liquefaction.
Based on the foundation performance observed in this earthquake, liquefaction lateral spreading and downdrag design with regard to their effect on foundations may be too conservative.
Conduct research to improve the estimation of lateral and vertical ground deformation caused by liquefaction using measured deformation data from the 15 bridge sites where liquefaction occurred as a result of this earthquake.
Methods currently available for estimating these deformations are relatively crude. In addition, a reasonable estimate of both vertical and horizontal ground movements due to ground liquefaction and lateral spreading should be made available to practitioners who will implement the displacement-based design methodology to ensure that undesirable structural failure and damage modes of bridges are removed. The fact that most liquefied sites during the Chile earthquake were not treated for improved performance offers a good opportunity to improve the engineering estimate of liquefaction-induced vertical and horizontal ground movements.
Use the liquefaction performance research identified above to develop a strategy from a no-collapse design objective perspective to focus liquefaction mitigation efforts on only the more severe cases.
Judging what should be considered severe may be a challenge and may be very site-specific. Even if such a strategy could be formulated and implemented, thorough site- and project-specific geotechnical analyses would still be needed to assess the severity of the likely effects of liquefaction at a given site.
Conduct research to develop more accurate lateral deformation models for walls under seismic loading considering the likelihood of rotational rather than translational movement.
Since very little evidence of the translational movement of a wall as a rigid body was found, the theoretical approach typically used to perform seismic deformational analysis of walls (Newmark analysis) may not match reality.(14,20) However, the good seismic wall performance observed still provides justification for using a reduced horizontal acceleration coefficient for design, even considering the lack of translational movement.
Conduct research to investigate the effect of high vertical acceleration on the performance of selected bridges and walls.
Vertical accelerations are often ignored in current seismic design practice. The effects of these relatively high vertical accelerations on bridge and wall performance should be further investigated and their implications to design practice determined.