Autumn 1995· Vol. 59· No. 2

The quake struck in the pre-dawn darkness of Jan. 17, 1995. In 20 seconds of violence that registered 6.9 on the moment magnitude scale (7.2 on the Richter Scale), virtually every part of the Japanese port city of Kobe suffered major damage. Survivors picked their way through a nightmare landscape of ruins and rubble. Fires fed by ruptured gas mains and spread by chill winter winds raged out of control. Fire fighters, even if they reached the scene of fires through rubble strewn and damaged streets, often could do little because of damage to the city water system.

The death toll soared toward 5,000. More than 25,000 persons were reported injured, about 300,000 homeless. Some 50,000 buildings were destroyed. Property damage estimates ranged from $30 billion to well over $100 billion. This 20 seconds of terror shattered complacency about which seismic zones are dangerous, about "earthquake resistance," and about the adequacy of recovery plans.

Out of such tragedies much can be learned. Instead of shaker tables in a lab, these "laboratories in the field" show how design and construction technologies perform when the real world starts to rock and roll. However, we have not always been as ready to learn from these events as we are today.

After the great San Francisco earthquake of 1906, the city was rebuilt in almost identical fashion. It was not until after the devastating Santa Barbara earthquake of 1925 that engineers began to include earthquake design provisions in building codes. It took another 20 years for similar provisions to be included in highway bridge design. And not until after the 1971 San Fernando earthquake were earthquake design criteria toughened and a seismic retrofit program initiated.

Since 1971, the Federal Highway Administration (FHWA) has worked to create new sets of earthquake design criteria, codes, guidelines and specifications using the results of actual earthquake investigations and the research and experience of the FHWA's Turner-Fairbank Highway Research Center, the California Department of Transportation (Caltrans), the National Center for Earthquake Engineering Research, the State University of New York at Buffalo, the University of California Earthquake Engineering Research Center, and other organizations, including those in Japan engaged in seismic research.

These investigations of earthquake disasters provide invaluable information for developing new technologies to promote structural resistance to earthquakes. The Jan. 17, 1994 earthquake in the Northridge section of San Fernando Valley in Los Angeles, California, and now, exactly one year later, the Jan. 17, 1995 earthquake in Kobe, Japan are prime examples of such test beds for evaluating U.S. and Japanese seismic design codes.

We have a long working relationship with the Japanese Government. This includes formal agreements and arrangements with the Japanese Ministry of Construction (MOC) and Public Works Research Institute. Reflecting our close cooperation with our Japanese counterparts, they have come to the United States after each of our major earthquakes San Fernando in 1971, Loma Prieta in 1989,and Northridge in 1994 and we have also investigated their earthquake sites.

The February 1995 full-scale investigation of the Kobe disaster was sponsored by the 27-year-old U.S.-Japan Natural Resources Development Program (UJNR) Panel on Wind and Seismic Effects. This joint effort was a prime example of the ongoing cooperation between the United States and Japan.

Many U.S. government agencies participated in organizing the investigating teams,including the National Institute of Standards and Technology, Department of Housing and Urban Development, Army Corps of Engineers, Geological Survey, National Science Foundation, and others a whole list of players that have an interest in public safety.

The authors joined 15 other engineers ingoing to Kobe to investigate the damage. As chief of the FHWA Structures Division, Jim Cooper headed the transportation team that included in addition to Dr. Ian Buckle Dr. Michael Whitney, structural engineer in the FHWA Office of Engineering, and Li-Hong Shen from Caltrans.

Four other teams investigated the seismological and geological effects of the earthquake,building damage, lifeline performance, and effects of fire. We also joined a number of investigators who had preceded us to Japan during the January to March timeframe.

Our transportation team visited 15 bridge sites in 3-1/2 days. We were escorted by officials from the Hanshin Public Expressway Corporation (HPEC), the Ministry of Construction (MOC), and the Japan Highway Public Corporation (JHPC). The two public corporations are quasi-public, quasi-private agencies that run, operate, and maintain toll roads in Japan. The Japanese bureaucracy is somewhat segmented. The MOC is responsible for highways, while the Ministry of Transportation is responsible for rail and mass transit, for which our team also had responsibility. Our team worked primarily with MOC.

What the Kobe "Surprise" Has Taught Us

Although Japan is very highly seismic reactive, there are areas of lower seismicity. The Japanese had previously identified the Kobe-Osaka region as an area of comparatively low seismic vulnerability and of low seismic intensity should an earthquake occur. Therefore, bridge and building designs were not held to the same high standards as more earthquake-prone areas such as Tokyo. Furthermore, many of the structures were built in the 1960s before the development of new design criteria derived from the 1971 San Fernando quake.

To bridge engineers and owners of bridges in the central and eastern United States, this particular earthquake is perhaps of even greater significance than recent earthquakes in California, such as Loma Prieta in 1989 and Northridge in 1994. One reason for this is that in the Japanese case, the possibility of an earthquake larger than the design earthquake was considered unlikely. Only nominal attention had been given to the problem, and then only for structures designed since 1990. This difference between the maximum credible earthquake and the design earthquake is clearly very large for this region of Japan a situation that exists in the United States to a greater degree in the eastern and central states than in the West.

A second reason that this quake is of greater interest to the central and eastern United States is that the predominant type of bridge in Japan is the steel girder superstructure (simple and/or continuous spans) supported by bearings on concrete columns and foundations. This class of bridge is found throughout the central and eastern United States, whereas bridges in California tend to be concrete box girders with monolithic bents and integral abutments, especially in shorter bridges. It follows that bridges in the central and eastern United States are more like those in Japan than those in California. That is why the performance of Japanese bridges is of particular relevance to states east of the Sierra Nevada Mountains.

The port cities of Kobe and Osaka are about 35 kilometers apart, about the distance from San Francisco to San Jose. The population is concentrated in a narrow east-west band of flat land between the mountains and Osaka Bay. It is a classic corridor situation.

A number of factors made this area particularly vulnerable to earthquake damage: the concentration of people, the construction of transportation routes one above the other or side-by-side in the narrow Osaka-Kobe corridor, poor soil conditions, and the fact that many of the structures were designed and constructed in the 1960s.

Because Japan is land poor, much use is made of land reclaimed from the sea. The new Kansai International Airport, for example, is built on a 1.75-km by 3.75-km man-made island using fill cut from a nearby mountain top, transported to the waters edge by conveyor belt and then barged 4 km to the offshore construction site. Because the Japanese used the latest technology in land reclamation in terms of fill compaction and consolidation, this island held up very well with no reported structural damage to the air traffic control tower. It is noteworthy that the older Osaka Airport, on the other hand, experienced cracking of a runway, cracked shear walls, cracked glass, and an inoperative elevator in the control tower.

Port Island, site of the Port of Kobe, the third largest port facility in the world, did not fare as well as Kansai Airport. Port Island was completed in 1983 using 20-year-old technology to build it up from the floor of Osaka Bay. The earthquake caused massive land subsidence, liquefaction and lateral spreading, which disabled most of the dockside gantry cranes and rendered the port inoperable. Japanese engineers estimate it will take about three years to completely restore it.

An interesting contrast to Port Island is the adjacent Rokko Island completed just two years ago, in 1993. Although it too suffered subsidence, the island as a whole performed much better. Here is an interesting test case of two reclaimed areas, built next to each other with different technologies. Each responded differently to the same earthquake.

The Highway Bridges of Kobe

As a general rule, the closer you get to water, the worse soil conditions are for supporting structures of any kind. Because main transport routes in the Kobe-Osaka area closely parallel the shoreline of Osaka Bay, many bridge pilings and bents were displaced, causing catastrophic damage to their supported structures.

In the Kobe quake, damage to highway bridges was both widespread and catastrophic. Typical damage included shear and flexural failures in nonductile concrete columns, flexural and buckling failures in steel columns, unseated girders due to bearing failures, and gross foundation movements due to liquefaction. In addition, pounding between spans occurred and approach fills behind abutment walls settled. Some special structures also suffered atypical damage, such as excessive rotation of a skew bridge on pin-ended columns and the loss of seismic energy dissipators in a cable-stayed bridge, because of the failure of a wind shoe, a device to control bridge sway.

Most of the distress was confined to older structures built more than 30 years ago before the introduction of modern seismic codes. The poor performance of these older bridges and elevated expressways confirms previous lessons learned in California and elsewhere about the pressing need to retrofit the existing inventory of deficient bridges.

However, some new bridges also suffered serious damage, which implies a need to re-evaluate the design loads and procedures in the current design specifications. As noted above, the peak ground accelerations due to this "rare" earthquake were considerably higher than the seismic coefficients used in design. This is a telling example of a low-probability, but highconsequence, event causing serious distress to modern construction.

From a highway bridge owner's viewpoint, these circumstances might best be addressed through multi-level performance criteria. In the United States, dual -level criteria have only recently been developed, and then only for critically important, long-span structures and a privately owned transportation corridor in California. The Kobe experience suggests that use of dual-level criteria should perhaps be extended in the United States. In doing so, it makes sense to clearly state the performance under both small and large, rare and frequent earthquakes and to identify design strategies that will satisfy these criteria.

This view of the Rokko Island bridge from the lower deck of the arch looking back at the approach span shows the 3-meter lateral offset and 1.5-meter vertical displacement, caused by the bridge bearing failure.

The closure of three major expressways in the Hanshin region had a major societal and economic impact on communities in the region. The estimated total repair cost for the Hanshin Expressway Public Corporation is $4.6 billion. Even though this is only about 4 percent of the direct costs for the earthquake (estimated at $130 billion on Feb. 17, 1995), it is still a very large sum. Furthermore, it does not include the loss of toll revenue while the network is down (estimated at $2 million per day), nor does it include the indirect losses that are expected to exceed repair costs by the time the system is fully operational late in 1995. The simultaneous closure of other transportation routes in the region, primarily the rail lines, further aggravated the situation and paralyzed the region. Even emergency access and relief teams were forced to use surface streets. The interdependency of these lifelines, especially when collocated in narrow corridors, certainly deserves further study.

This illustrates the intermodal impact of earthquake damage. In modern industrial societies, there is an intricate interdependency between various modes of transportation. If the interconnections between sea, air, and land transportation systems are disrupted, or the individual systems themselves suffer damage, severe hardship results. In particular, this cutting of the main east-west surface transportation corridor between Kobe and Osaka greatly impacted business, trade and the quality of life, with serious short- and long -term economic consequences for the entire region and for the nation as a whole due to loss of trade through a major port.

The Key Importance of Bridge Bearings

The damage to bridge columns in the Kobe quake was spectacular, but the biggest lesson was the high incidence of bearing failure. When bearings gave way, spans dropped or rotated in plane so that highways didn't line up. We must look for structural fuses and links, mechanisms designed to allow some deformation or controlled failure, to lessen failures in substructures. There is a need for better damping devices and isolation bearing designs to decouple energy and motion.

This new isolation bearing technology is radically different from previous approaches to the problems of the seismic design of bridges. It has great potential for both new construction and retrofitting. However, these measures for taking the load off columns present another problem displacement. More research on both isolation bearings and displacement control is needed.

In the United States, the claims of 14 manufacturers of various newly developed isolation bearing devices are so far mostly unsubstantiated. This is being remedied by FHWA sponsorship of an evaluation of seismic isolation bearing performance and behavior through a panel of experts from Caltrans and the Highway Innovative Technology Evaluation Center (HITEC), a private research organization of the Civil Engineering Research Foundation (CERF). Results of these evaluations will be available in late 1996.

Summary of Lessons Learned

(1) Large earthquakes can be very destructive in terms of lives lost, injuries sustained, business losses, construction costs and social disruption. The closure of arterial highways and bridges affects emergency relief and business recovery and can have a major economic impact on a region and its ability to survive such a disaster.

(2) Large damaging earthquakes can occur in areas considered to have, on average, only moderate exposure to seismic hazards.

(3) Capacity design procedures, ductile details and generous seat widths are necessary to prevent catastrophic collapse during large earthquakes.

(4) Minimum connective forces need to be enforced for all seismic zones unless such connections can be shown to be fully protected by acceptable yielding of the substructures. Redundancy in connection detailing is particularly important for essential structures. Alternative load paths are necessary if the primary load path fails due to unforeseen circumstances.

(5) Critically important structures must be designed to a higher level of performance than that provided by current specifications, if full service is to be maintained after a large earthquake. Multi-level performance criteria and corresponding design strategies are necessary for important bridges.

(6) Retrofit measures reduce damage, but inappropriate use and/or installation can defeat their purpose and perhaps trigger a collapse. This applies particularly to couplers (restrainers).

(7) Lateral spreading due to liquefaction can lead to span collapse even in modern structures with massive foundations (caissons) and well-engineered fills.

(8) Premature failure of some bearings appear to have reduced the seismic loads in their supporting substructures by uncoupling the superstructure from its supports. This fuse-like action may have saved a number of spans from collapse and columns from failure in shear and flexure.

(9) Accelerations in isolated superstructures are less than in conventional structures.

(10) Skewed bridges are susceptible to in-plane rotation leading to large displacements at their supports and possible unseating of girders in the acute corners.

Opportunities for Needed Research

A number of research needs and opportunities have arisen as a consequence of this earthquake. These include correlation studies on those bridges with strong motion records to calibrate theoretical models against actual behavior. As far as can be determined at this time, records are available for two bridges. These are the Higashi-Kobe cable-stayed bridge (with seismic energy dissipators) and the base-isolated Matsunohama bridge. Valuable insight into the performance of these two important types of bridges would be derived from such a study.

A detailed database needs to be developed on the performance of steel bearings and earthquake couplers (restrainers), followed by case studies on selected bridges in which these devices failed. Results from such a study would help determine the need for a review of current U.S. design criteria for connection forces and restrainers. Analysis is needed for structural response due to large foundation movements during liquefaction and lateral spreading and the development of optimal mitigation measures to minimize structural impacts. Case studies could include the Nishinomiya-ko and Shukugawa bridges.

Case studies on those bridges that had been retrofitted before the earthquake would help determine effectiveness of retrofit measures. Known measures for which performance could be evaluated include earthquake couplers, steel jackets, and viscous dampers. Multiple-level performance criteria needs to be developed to define expected damage and serviceability states for various earthquake scenarios, as well as the corresponding design strategies/structural options necessary to satisfy these criteria.

Concrete columns in single- and multi-column bents in Japanese bridges have very large cross sections and fall outside the range of U.S. test data. Although it is unlikely that U.S. practice will adopt similar-sized members for short- and medium-span bridges in the near future, columns of this size and these reinforcement ratios are encountered in some long-span structures. Correlation of assessment methods and limit states with field data are rare, and this opportunity to study failure mechanisms in large concrete columns under combined axial and shear loads is probably unique.

The relatively large number of damaged steel bridge substructures in the Kobe area provides an opportunity to calibrate proposed new methods for the seismic design of steel bridge columns. In particular, their capacity for ductile actions and the identification of their limit states should be investigated with the objective of refining the R-factors in U.S. bridge codes for these components.

Clear Lessons for U.S. Planners

In the United States, it's not just California and Alaska that are at risk. Thirty-nine states have significant seismicity. But lack of public awareness of the earthquake potential outside California has caused inattention to implementing improved design criteria and to the need for retrofitting important existing bridges with the technology derived from on-the-scene investigations of earthquake damage.

Japan is seismically active, but there are areas of lower seismicity just as there are in the United States. This is important. In reality, some 39 states are considered seismically active and have the potential of generating significant damage. We think of California, we think of Alaska, but it's a lot more than that. There is an active fault zone in Missouri, Kentucky, Illinois, and Tennessee that has significant potential for widespread damage and disruption. The St. Lawrence Seaway; St. Lawrence Valley; upstate New York; Charleston, S.C. all these areas have the capability of generating very significant earthquakes. It's just a question of when.

Part of the difficulty in the United States is in creating an awareness of the seismic vulnerabilities in those areas outside of California. California knows they have a problem. Other states are coming around to appreciate the fact that they could also have a problem. But although they know they should prepare for possible seismic events, questions remain. How much retrofit? Which bridges should be selected for retrofit? What is the actual risk? Funds are limited. Risks and costs must be weighed.

The formula is to identify the hazard, analyze the vulnerability of selected priority structures to that hazard, and then to fix those structures. Cost must be balanced against risk. Easy to say, but almost impossible to do perfectly. Identification of structures in most need of fixing (retrofitting) is difficult because various sectors of the population have differing priorities. A bridge that is important to me may be of little value to you.

In evaluating which bridges should be retrofitted and to what extent, states must evaluate the importance of the bridge to the community as a whole. Where does it go? What are the intermodal links that it affects? Are there redundant routes? What is the level of inconvenience to the community if the bridge becomes unusable?

Since it is impossible to design or retrofit a structure to be "earthquake proof" to be totally safe near the epicenter of a large quake the philosophy is to design for "life safety," a philosophy that can accept damage and closures, but avoids the structural collapse that causes loss of life. Serviceability, although important, is always secondary to life safety.

The lessons of the 1971 San Fernando earthquake led to seismic retrofitting that is, going back and evaluating structures vulnerable to earthquakes and determining how to strengthen them in a cost-effective manner. Some fixes that have been developed are simple. Some are complex. In the San Fernando quake, those bridge spans that were simply supported or that had discontinuities simply fell off during the shaking. To counter this, restrainers or couplers were developed to basically tie the spans together. However, we found in subsequent earthquakes that when everything shakes together, couplers just transmit the destructive force somewhere else. If you prevent failure in one place, it may occur in another. If it doesn't fail in the weakest spot, it fails in the next weakest place or is transferred to the columns. Therefore, attention next turned to retrofitting columns and foundations, the next weakest links.

Highway engineers have to trace the paths through which the shaking forces will be applied and then devise fixes that properly distribute those forces. We started with joint restrainers or couplers. Those are very cheap. But in the next big earthquake that came along we saw that the failures were forced someplace else. So next, we looked at bents and columns. Their retrofit is more

expensive. Next, came foundations, footings and pilings, which are even more expensive. Somewhere along the line, you have to balance the importance of the structure, its vulnerability, the nature of the hazards, the severity of expected shaking, and the costs of seismic protection. The goal is to come up with fixes that balance all these factors. Easy to say, but almost impossible to do realistically.

Retrofit Now, Design for the Future

The most cost-effective, long-term program for mitigating earthquake damage to highway bridges is to ensure that new construction is designed even beyond existing specifications in accordance with the latest technology derived from studies of real-world disasters. This is more cost-effective than retrofitting, which can be very expensive.

We at FHWA are interested in the national perspective. Before 1971, earthquake design criteria were very simple, basic and fundamental. The usual extent of design guidance was "consider earthquake forces." Since 1971, Caltrans and FHWA have done extensive research, investigated real-world aftermaths of earthquakes at home and abroad, and updated specifications. The current American Association of State Highway and Transportation Officials (AASHTO) design specifications, based on initial research completed in the late 1970s, were formally adopted in 1983 but were not mandated until 1990, after the Loma Prieta earthquake. Nonetheless, the technology is advancing. There has been significant progress.

During the 10-year period from 1971 to 1981, the San Fernando experience was applied. The importance of the recent earthquakes in California, Japan, and other places is that we have been able to see how the older technology has performed, evaluating its successes and failures. Where research results have been implemented, we have seen positive results in terms of better performance. After 1981, more knowledge was gained so that structures designed after 1982 performed very well in recent large earthquakes.

We have acquired a track record from field and laboratory research by seeing how actual structures and new technology have played out when subjected to real -world earthquakes. That is an improvement over experimental laboratory results. There have been failures, but a lot of improvements have worked well. While there have been shortcomings, we have seen much better performance overall.

Another point to consider is that there are a large number of short, stiff bridges in the United States. It wasn't until the late 1970s that we started designing for ductility (that is, flexible "forgiving" designs) following the San Fernando experience. The difficulty now is in trying to decide how much seismic protection we can provide through new design and, more critically, how much retrofitting should be done. There are some 577,000 highway bridges in the United States. They can't all be retrofitted, certainly not to the same degree. So there must be differing design levels. Not every bridge can be, or needs to be, designed to California standards.

We have made considerable progress since 1971. We have better technologies today, and the lessons learned in Northridge in 1994 and Kobe in 1995 will boost our confidence. Through AASHTO, we are working to develop improved technologies and better design guidance for both new construction and retrofitting.

As an aid to the retrofitting effort, FHWA released in May 1995, Publication No. FHWA-RD-94-052, Seismic Retrofitting Manual for Highway Bridges , prepared by

Finally, FHWA is sponsoring and funding a "National Training Course for the Seismic Design of Highway Bridges," a course that will become available in January 1996. If the subject of this article has been of interest to you, watch for announcements about the one-day national video conference that will launch this three- to four-month FHWA course for state and private-sector bridge designers.

James D. Cooper is the chief of the Structures Division in the Office of Engineering and Highway Operations Research and Development at FHWA's TurnerFairbank Highway Research Center in McLean, Va. He received his bachelor's and master's degrees in civil engineering from Syracuse University. He is a licensed professional engineer in the District of Columbia.

Ian Buckle is deputy director of the National Center for Earthquake Engineering Research and professor of civil engineering at the State University of New York at Buffalo. He received his undergraduate degree and doctorate in civil engineering from the University of Auckland, New Zealand.

TFHRC Home | FHWA Home | Feedback United States Department of Transportation - Federal Highway Administration