Multidisciplinary Center for Earthquake Engineering Research logo google logo
navigation bar
bridgesmall.gif (4301 bytes)MCEER/NCEER Bulletin Articles: Research

Recommendations for Improved AASHTO Bridge Design Specifications

The Applied Technology Council recently completed a project to review currently available seismic design criteria and specifications for highway structures worldwide and to provide recommendations for improved national seismic design specifications for highway bridges (ATC-18). These recommendations were made by a Project Engineering Panel, whose members were Ronald Mayes (Chairman and Project Director), Donald Anderson, John Clark, Ian Buckle, John Hom, Richard Nutt, Michael O'Rourke and Charles Thornton. The recommendations embody significant changes to current specifications. For further information, contact Chris Rojahn at ATC, phone: (415) 595-1542 or via email at


While the overall scope of work on Project ATC-18 pertained to all types of highway structures, particular attention was given to bridge structures and foundations. The ATC-18 project team initially focused on a review of current design practice and criteria for new bridge design, as well as the philosophies on which they are based. This involved a review of existing U.S. standards along with the latest codes of Japan, New Zealand, and Europe. Guidelines developed for the Transportation Corridor Agencies (TCA) for Orange County, California, the new AASHTO Load and Resistance Factor Design (LRFD) Bridge Specifications, and information emanating from the ATC-32 project were also reviewed. The final phase of the ATC-18 project focused on the development of recommendations for the future direction of seismic code requirements for bridge structures in the United States. An important aspect of the recommendations is a two-level design approach. The following paragraphs are specific recommendations for future bridge seismic design code development.

Performance Criteria

ATC-18 recommends that the performance criteria to be included in any future code should be approved by a group that includes legislative policy makers. In order for this review to be effective, a significant amount of cost and technical data will need to be developed before the performance criteria can be intelligently discussed by a non-technical panel. The other key requirement for the recommended performance criteria is a specific and unambiguous definition of Important and Ordinary bridges. Future codes may have significantly different design requirements for these two bridge categories.

Two-level Design Approach
The recommended performance criteria for a two-level design approach is shown in Table 1. Both the functional- and safety-evaluation earthquakes should have specific definitions in terms of probabilistic return periods; i.e., 72, 150, or 250 years for a functional-evaluation design event and 950 or 2,475 years for a safety-evaluation event. The expected performance of Ordinary bridges should be similar to those of the typical one-level procedure; i.e., collapse will be prevented, but significant damage may occur. Bridges should be designed so that damage occurs in visible locations.

Table 1: Recommended Seismic Performance Criteria (ATC 1996a)

Ground Motion at Site  Ordinary Bridges  Important Bridges 
Ground Motion 
Service Level-Immediate 
Damage Level- 
Repairable Damage 
Service Level-Immediate 
Damage Level- 
Minimal Damage 
Ground Motion 
Service Level-Limited 
Damage Level- 
Significant Damage 
Service Level-Immediate 
Damage Level- 
Repairable Damage 

One-level Design Approach
The performance criteria for a one-level design approach cannot be as specific as for a two-level design approach, because the performance in the lower level events can only be implied from the design requirements of the upper-level event. (This is consistent with current design philosophies). A suggested philosophy follows:

For a low level earthquake, there should be only minimal damage.

Functional-evaluation ground motion.
This is an event that is determined to have a reasonable (approximately 30 to 50%) probability of occurring during the useful life of a bridge. (Note that this may eventually become a 72-, 150-, or 250-year return period event, depending on the chosen probability of exceedance and the definition of useful life of the bridge.) This definition may also result in more than one design return period since some bridges will have longer useful lives.

Safety-evaluation ground motion.
This is an event with only a small probability of occurring during the useful life of the bridge (i.e., 10% probability of exceedance for a design life of 100 to 250 years. This results in a return period of 950 years or 2,475 years, respectively). Note that current AASHTO specifications use a 10% probability of exceedance in 50 years or 475-year return period event as the design earthquake.

Service Levels

Two definitions of service levels are recommended for the two-level design approach.

Service Level - Immediate.
Full access to normal traffic is available almost immediately (e.g., within hours) following the earthquake. (It may be necessary to allow 24 hours or so for inspection of the bridge.)

Service Level - Limited.
Limited access (reduced lanes, light emergency traffic) is possible within three days of the earthquake. Full service can be restored within months.

Damage Levels
A significant amount of work is required to develop reasonably specific definitions for levels of damage to columns, caissons, abutments and retaining walls. The following definitions are based on the ATC-32 criteria. It is recommended that they be augmented wherever possible with closure time frames, ductility levels, and, if feasible, allowable steel and concrete strain levels.

Minimal damage.
Although minor inelastic response may occur, post-earthquake damage is limited to narrow flexural cracking in concrete. Permanent deformations are not apparent. Criteria for abutments, wing walls, and steel members need to be developed.

Repairable damage.
Inelastic response may occur, resulting in concrete cracking, reinforcement yield, and minor spalling of cover concrete. The extent of damage should be sufficiently limited so that the structure can be essentially restored to its pre-earthquake condition without replacing reinforcement or structural members. Repair should not require closure. Permanent offsets are small and there is no collapse.

Significant damage.
Although there is no collapse, permanent offsets may occur and damage consisting of cracking, reinforcement yield, and major spalling of concrete may require closure to repair. Partial or complete replacement may be required in some cases. Criteria need to be developed for abutments, wing walls, and steel members.

Design Approach

It is recommended that the current AASHTO Seismic Performance Category approach be continued in future codes, since it is a good method of varying design requirements in different seismic zones. It is also recommended that a two-level design approach be adopted at least for Important bridges in higher seismic zones. The lower-level design requirements should be based on elastic design principles to ensure that there is no damage. The upper-level analysis should be deformation-based using nonlinear static (pushover) analysis procedures with strength and stiffness requirements being derived from appropriate nonlinear response spectra.

If a single-level design procedure is adopted for Ordinary bridges, it is recommended that the design approach include a nonlinear static analysis as part of the design procedure.

Seismic Loading

Two-level Design Approach
The upper-level design event is recommended to be a 2,475-year return period event. (Note: It may be necessary to have different return periods for the eastern and western portions of the United States; 2,475 years and 950 years, respectively). The lower-level design event would be in the range of a 72-year to 250-year return period earthquake. The return period would be based on the same probability of exceedance (e.g., 30 to 50%) and may be different for Ordinary and Important bridges because Important bridges would have a longer design life (e.g., 200 years versus 50 years).

One-level Design Approach
It is recommended that a single level design procedure be based on 2,475-year return period risk maps if they are available. (Note: It may be necessary to have different return periods for the eastern and western portions of the United States of 2,475 years and 950 years, respectively).

Site Effects

It is recommended that the 1994 NEHRP (BSSC, 1994) soil factors or their derivatives be the basis for the response spectra used for design.


Current practice is to use a 5% damped response spectra for design. If nonlinear static analysis is adopted for the upper-level design event, then nonlinear spectra will have to be derived from the elastic spectra. The nonlinear spectra will incorporate whatever levels of energy dissipation (hysteretic and viscous) are deemed appropriate. If nonlinear static analysis is not adopted, higher damping levels in combination with elastic response spectra may be appropriate for the upper-level design event.


Two-level Design Approach
Current elastic analysis procedures (equivalent static and multi-modal) are appropriate for the lower-level event. It is envisioned that the lower-level design procedures would use component stiffness values consistent with "little or no" damage. The recommended procedure for the upper-level event should include a nonlinear static analysis. However, this analysis method requires additional development work before it can be used as a standard office procedure.

One-level Design Approach
There is no consensus on the best method of implementing a one-level procedure. It is desirable for nonlinear static analysis to be a part of any analysis requirements. At a minimum, nonlinear static analysis should be required for Important bridges. Current elastic design and analysis procedures may be sufficient for smaller Ordinary bridges. It is possible that the analysis procedure in a one-level design approach could incorporate both the current R-Factor elastic procedure and nonlinear static analysis. Nevertheless, requiring different analysis procedures for different seismic input levels in a two-level approach seems more rational than requiring two analysis procedures for the same seismic input level in a single-level design approach.

Design Forces

Two-level Design Approach

One-level Design Approach

Design Displacements

Design displacement values should be determined from the upper-level design event, using stiffness properties appropriate to the expected level of displacement. These values could be determined directly from a nonlinear static analysis, but in a one-level elastic approach, an iterative analysis procedure may be needed to ensure that the appropriate stiffness values are used. It is recommended that the current minimum seat width requirements remain since it is not considered practical in a design office environment to accurately calculate the relative displacement for all applicable parameters (e.g., surface wave effects). It is also recommended that overall drift limits be incorporated to avoid P-Delta effects on long period structures.

Concrete and Steel Design

It is recommended that capacity design procedures be used to prevent brittle modes of failure in all critical members. Confinement, shear, joint shear, torsion, anchorage, and splice reinforcement requirements will be improved as research progresses over the next several years. It is recommended that the requirements be updated to reflect new research findings.

Foundation Design

Two-level Design Approach
Geotechnical analyses should be conducted for both levels of design acceleration to confirm that response of the soil will not adversely affect the performance of structures supported on or within the soil. These analyses should include assessments of the potential for liquefaction, lateral spreading, and slope instability; dynamic earth pressures on buried walls; soil-structure interaction; and uplift and rocking of the foundation. Since some of the analyses depend on the magnitude of the earthquake causing the design acceleration, care must be used in selecting an earthquake magnitude that is compatible with the seismo-tectonics in the area, both in terms of source mechanism and source distance. Where partial or total liquefaction of saturated, cohesionless soil layers is predicted, the effects of loss in soil strength on the stability of sloping ground, in particular, and the vertical and horizontal bearing capacity of the soil must be addressed to confirm that structures supported on or within the soil can tolerate these effects. The evaluation should also consider the amount and effects of settlement resulting from liquefaction or densification of granular soil and the increased soil downdrag forces on pile foundations resulting from liquefaction or densification of the soil.

It is critical that geotechnical analyses be completed for both levels of design, rather than just the higher level. The implications of soil response at both design levels should be considered in light of bridge performance criteria. In all cases, soil behavior that degrades the structural capacity of the foundations must be prevented at the lower-level event; soil behavior that leads to damage in the upper-level event may be permissible as long as it does not lead to catastrophic failure of the foundation. For pile-supported and spread footing foundations, this generally means that permanent vertical or horizontal foundation movement should not occur during the lower-level event and that movement should be less than a maximum acceptable amount during the upper-level event. For the lower-level event, uncertainties in soil and soil-structure performance should be included in the evaluation by incorporating a factor of safety when estimating soil strength. For the upper-level event, best-estimate soil properties and a factor of safety of 1.0 should be used, given the low probability of this event.

One-level Design Approach
Procedures identified for the two-level design approach should be applicable to the one-level design approach. Given that the design event in the single level approach generally corresponds to the upper-level event in the two-level approach, it is likely that liquefaction, soil spreading, and slope instabilities would be common. These ground failure problems can often be corrected through ground improvement technologies or by avoiding susceptible areas. In many cases, thorough corrective measures would be economically unrealistic. For bridges that are not considered essential, a lower level of ground motion might be appropriate for the geotechnical analyses. It is suggested that this lower level be 50% of the design acceleration. Under this lower acceleration level, large ground movement should not occur.


AASHTO, (1995), Standard Specifications for Highway Bridges, 16th Edition, American Association of State Highway and Transportation Officials.

ATC, (1981), Seismic Design Guidelines for Highway Bridges, Report No. ATC-6, Applied Technology Council, Redwood City. Also published by Federal Highway Administration as FHWA/RD-81/081, Washington, DC

ATC, (1996a), Improved Seismic Design Criteria for California Bridges: Provisional Recommendations, Report No. ATC-32, Applied Technology Council, Redwood City.

ATC, (1996b), Seismic Design Criteria for Highway Structures: Current and Future, Report No. ATC-18, Applied Technology Council, Redwood City.

BSSC, (1994), NEHRP Recommended Provisions for the Development of Seismic Regulations for New Buildings, Building Seismic  Safety Council, Washington, DC


NCEER Bulletin, July 1997, Vol. 11, No. 3

top.gif (1113 bytes)

  Contact Us  |  Acknowledgements   |  Disclaimer  |  Copyright© 2007 by the Research Foundation of the State of New York. All rights reserved.