The Verrazano-Narrows Bridge is the longest suspension bridge in North America with a main span of 4,260 ft and side spans of 1,215 ft each for a total suspended length of 6,690 ft. It connects the Staten Island Expressway to the west and the Brooklyn-Queens Expressway, Gowanus Expressway and Belt Parkway to the east linking the New York City boroughs of Staten Island with Brooklyn. The two-level bridge carries 12 lanes of traffic, six on the upper level and six on the lower level. The upper level has been open to traffic since 1964 and the lower level since 1969. While the main span of the Verrazano-Narrows is only 60 ft longer than the earlier Golden Gate Bridge, the load carried between its towers is 70% greater than the original design of the six-lane Golden Gate Bridge. The bridge construction was completed in slightly less than five years, six weeks ahead of schedule, on Nov. 21, 1964.
The seismic study included field investigation and determination of site conditions; determination of seismic hazard and site response calculations; ambient vibration monitoring, under normal conditions and during the 2004 New York City Marathon; development of project-specific design criteria based on performance goals and similar projects; development of a global model of the bridge and calibration of that model; seismic analysis and assessment of the seismic vulnerability of the structure including seismic evaluation of a future orthotropic deck; and development—to a conceptual level—of necessary retrofit measures. The project was guided by a peer review panel chaired by professor Frieder Seible who was joined by Dr. Po Lam of Earth Mechanics and by Stoyan Stoyanoff of RWDI.
Following a review and evaluation of the existing geotechnical data, a field subsurface investigation and laboratory testing program were conducted. The field investigation for this study was limited to drilling four borings and performing P-S Logging in three of the borings. Two were performed over water at each pier location, and two were performed on land at each anchorage location.
A generalized subsurface profile, as interpreted from the existing borings and the current borings, was developed longitudinally along the length of the bridge, from the Staten Island anchorage to the Brooklyn anchorage. The subsurface profile consisted of fill; medium-dense to dense sand and gravel; clay and silt; medium-stiff to stiff clay and silt; very dense sand, gravel and cobbles; hard clay and silt; and bedrock (extremely fractured; extremely weathered).
The presence of soil overburden modifies the amplitude and frequency content of the ground motion in a complex manner, depending primarily on the dynamic characteristics of the soil layers over bedrock with respect to the characteristics of the input rock motions. To derive site-specific design soil acceleration spectra for use in the seismic analysis of the bridge structures, the free-field site response analysis was performed. Acceleration time histories (for 2,500-year and 500-year return periods) developed from NYCDOT’s 1998 and 2002 studies were used as design rock input motions in the site response analysis based on the site stratigraphy and static soil and rock properties determined in the field investigations. The dynamic soil and rock properties included the small-strain shear wave velocity and damping ratio, nonlinear relationships between shear modulus and damping ratio versus dynamic shear strains and unit weight.
A field study of bridge motion was conducted under the supervision of professors Andrew Smyth and Rene B. Testa, both of Columbia University’s Department of Civil Engineering and Engineering Mechanics, to characterize the bridge’s overall dynamic behavior for subsequent verification of analytical model representation of the bridge. Acceleration data were collected using a deployment of nine triaxial force balance accelerometers and portable data loggers. These accelerometers were placed at locations such as truss panel points on the suspended structure so that local member and substructure motions were minimized. Very low frequency displacements, such as those caused by wind loading, were monitored using two GPS displacement monitors near midspan, plus a third at a location off the bridge for use as a reference point.
The total number of instruments recording data at any given time was constant, but the locations of individual instruments was varied on different days, so that a wide range of global vibration modes and frequencies could be identified. Ambient excitation was recorded in half-hour windows during various representative times of day (i.e., morning and evening rush hours, midday and nighttime).
Post-processing of measured data involved the generation of power spectra and transfer functions for use in system identification of a total of 20 natural frequencies and modes of the bridge structure. The frequency domain decomposition technique was used to determine the corresponding natural modes for vertical, torsional and lateral motion. Very good correlation was found with the dynamic properties of the analytical model discussed later. Damping was estimated and presented in the dominant modes from -3dB bandwidth.
The project team developed performance criteria to rate the performance of the Verrazano-Narrows Bridge when subjected to seismic excitation at the design level. Taking inspiration from the design of seismic retrofit measures on other major bridges, such as the Golden Gate Bridge, the criteria were based on two seismic hazard levels and performance objectives for various components involving evaluation of component capacity, as compared with the demand, to create a measure of component damage.
The NYCDOT Seismic Criteria Guidelines (1998) set broad goals for retrofit of major NYC bridges. The Verrazano-Narrows Bridge is considered a “critical bridge” in the region, and, according to the NYCDOT guidelines:
. . . must continue to function as a part of the lifeline, social/survival network and as an important link for civil defense, police, fire department and/or public health agencies to respond to a disaster situation after the seismic event.
According to AASHTO specifications, a critical bridge should be “useable by emergency vehicles and for security/defense purposes immediately after a large earthquake (e.g., a 2,500-year return period event). Based on this and other resource documents, performance requirements were established, such that, if necessary, the bridge should be retrofitted to resist two earthquake hazard levels: A functional evaluation earthquake (FEE) is a ground motion with a 10% probability of exceedance in 50 years, which corresponds to a return period of about 500 years, while a safety evaluation earthquake (SEE) is a ground motion with a 2% probability of exceedance in 50 years, which corresponds to a return period of about 2,500 years.
Based on the above material, the performance criteria for the Verrazano-Narrows Bridge for an FEE event were:
- No collapse;
- No damage to primary members;
- Minimal, repairable damage to other members and components; so that
- All traffic should be able to access the bridge within a few hours (with no reduction in speed), after inspection.
Based on the importance of the Verrazano-Narrows Bridge as a critical and nonredundant transportation link in New York City, the performance criteria for the Verrazano-Narrows Bridge for an SEE event were:
- No collapse;
- Minimal damage (or less) to primary members;
- Limited, repairable damage (or less) to other members and components; so that
- All traffic should be able to access the bridge within 24 hours, after inspection and minor repairs. Vehicles may have to travel at reduced speeds if the expansion joints are damaged.
Analytical models of the Verrazano-Narrows Bridge were developed for the purpose of seismic investigation of the bridge. The dynamic characteristics of the analytical models, including natural frequencies, mode shapes and mass participation factors, were determined. The frequencies and mode shapes of the developed analytical models were compared with those from ambient vibration measurements and found to have good correlation, indicating that the models are a close representation of the current bridge state.
The developed global analytical model, consisting of 12,029 joints, 14,092 space frame members and 3,575 finite elements, was subjected to the ground motion time histories for each hazard level. To provide an accurate representation of the static and dynamic behavior of the bridge in the global analytical model, all major structural components of the bridge were explicitly modeled using ADINA, a general-purpose program that is the Caltrans standard for nonlinear time history analysis.
The force and deformation responses of structural components were compared with the strength and elastic/ductility limits established in the performance criteria to determine which components were vulnerable to damage (and to what extent) under each hazard level. This calculation of relative damage was performed by determining the ratio of demand-to-capacity (or D/C ratio) for each member behaving elastically, or nearly elastically, under each of the considered loading combinations. A D/C value of unity or less implies that no damage should be expected in the member.
Performing in the long haul
There do not appear to have been any prior studies reported in the literature that determined the equivalent live loads or excitation frequencies for marathons. To determine the needed frequencies, fastest, average and a range of slower running speeds were estimated using video documentation. The NYC Marathon runners on the Verrazano-Narrows Bridge were videotaped from two cameras located near the top of the Brooklyn tower so that large areas of the runners were visible in the video. The videography was continuous for the race, and a continuous record of time was indicated.
Monitoring of actual deflection on a large bridge like the Verrazano is not straightforward and not often done. With GPS technology, using a fixed dish and one attached to the structure, it was possible to determine the deflection of the dish location on the structure relative to the location of the fixed dish in three dimensions. The accuracy of the measurement depends on the number of satellites viewed by the dishes, but it is possible to distinguish to as low as 1-2 mm in the lateral direction.
Measurements were made at two points on the bridge during the NYC Marathon while runners were on the bridge. The GPS units were located near midspan on the main span, one on each side. The placement of GPS units on opposite sides was to determine if there was any torsion induced in the span due to the unequal marathon loading—while both sides of the upper level are used by the marathon, only one half of the lower level is used, allowing one side to remain available to emergency vehicles.
The measurements made during the 2004 marathon found a maximum vertical deformation at midspan of approximately 250 mm. Based on the observed density and speed of the runners, and an assumed weight of 160 lb per runner, the static load varied from 0.4 to 7.9 psf. The speed varied from 8.3 to 15 ft/second and the frequency (steps/second) varied from 2.52 to 2.90. The maximum static load 7.9 psf can be expressed as a continuous load of 585 lb per foot of bridge on the upper deck (both bounds) and 292 lb per linear foot on the lower deck (one bound), well below the 4,800 lb per linear foot used by O.H. Amman. Professors Thomas Murray and Mehdi Setareh from Virginia Polytechnic University assisted in the field testing and evaluations of the marathon and determined that there were no adverse physiologic effects on the runners due to bridge motions during the marathon.