Standing Up to Earthquakes
For more than fifty years, engineers have been learning—mostly from painful experience—how to design structures that can shake and survive
On September 19,1985, the most disastrous earthquake in North American history struck Mexico City. More than twenty thousand people were killed after a layer of wet clay amplified a distant temblor and set downtown buildings rocking. Hundreds of the buildings collapsed, crushing or trapping their inhabitants. But it could have been far worse. Thousands more buildings stood, including the landmark Latinoamericana tower.
That forty-three-story tower (designed in part by an engineering professor from the United States and built in 1954) has now been through two serious earthquakes with no damage of any consequence. Decades of observation of the effects of earthquakes lie behind the fact that engineers now know how to build structures able to resist quakes so strong they would throw people off their feet. A lot of that engineering development was done in the United States, particularly in California.
California gets practically all the notoriety for earthquakes in the United States, but it doesn’t get all the shocks; states generating major quakes have included Massachusetts, South Carolina, Missouri, Alaska, Nevada, Texas, Utah, Arizona, and Washington. However, the United States has been remarkably fortunate: in all our history, earthquake fatalities have totaled less than two thousand. By contrast, tremors worldwide claimed about three hundred thousand lives in 1976 alone.
Sparse population is largely responsible for the low U.S. death toll. Disaster planners estimate that one powerful quake in a megalopolis would still claim thousands of lives. The San Andreas fault is due to unleash a terrific quake in the Los Angeles area sometime in the next several decades. When it comes, it will be an unforgiving test of how well constructed our new buildings are, and how many dangerous old buildings we’ve removed. It will give us the measure of our earthquake engineering, a science that, for the most part, we have had to learn the hard way.
The United States and Japan are the world leaders in earthquake engineering, but the serious work of preparing for quakes began only in 1933. The famous slip of the San Andreas fault near San Francisco on April 18, 1906, was our best-known earthquake and cost seven hundred lives, but it produced more interest in guarding against post-quake fires than in preventing building collapse.
Many large structures came through the San Francisco quake (8.3 on the Richter Scale) with little or no damage; of fifty-two major buildings downtown (the tallest was nineteen stories), all but six survived. The bigger buildings endured because of designs intended to make them resistant to strong winds—some of their steel frames had been stiffened with masonry fillings. The masonry absorbed a lot of excess energy and Kept the steel from bending.
The biggest destroyer in San Francisco was fire. Flames from heating and cooking devices caught hold in the wreckage of innumerable houses and small buildings. Fire fighters couldn’t keep up because the earthquake had crippled the water system. Hundreds of pipes broke where the distribution system crossed soft, filled-in ground, sapping the whole system’s pressure. Fire engines rapidly emptied cisterns buried under the street intersections, and the fire burned for three days and leveled five hundred city blocks. By 1912 San Francisco had completed a second, independent, high-pressure water system specifically for fire fighting, which it maintains scrupulously to this day. The city also made arrangements to close off pipes broken by earthquakes and installed more cisterns.
Real earthquake engineering first came to the United States in 1933, when a quake hit Southern California and seismographs were in place to record it.
For the next two decades, there was little interest anywhere in this country in preventing earthquake damage to buildings. One engineer hired by the San Francisco chapter of the Economic Society of America even argued that California was peculiarly resistant to earthquakes, because deep sedimentary deposits in the state’s valleys would act like giant shock absorbers. Meanwhile, horrendous death tolls after earthquakes in Italy (1908) and Japan (1923) spurred action in those countries to toughen building requirements. Tokyo slapped a hundred-foot maximum on all buildings after 1923 and kept it for the next forty years.
Because of its regular quakes, Japan provided a brisk business for American designers and engineers interested in the problem. The architect Frank Lloyd Wright became world-famous after his Imperial Hotel survived the 1923 Tokyo earthquake and fire. A few years later American architects and engineers designed the massive Mitsui Bank in Tokyo to resist earthquakes, riots, and fire.
Wright’s hotel went against the popular earthquake-engineering wisdom of the time, which favored simple, strongly braced, heavy buildings built on rock or firm ground. The foundation of the Imperial floated on short piles driven into a shallow layer of firm soil atop a shaky layer of mud. Wright segmented his hotel into twelve sections that could move independently. He tapered the reinforced concrete walls so that the upper sections were much lighter than the bottom ones. Anticipating the fire that would follow any large quake, he insisted on a pond in the courtyard as a fire-fighting reservoir.
But other major buildings besides the Imperial Hotel survived the Tokyo disaster, and no simple lessons emerged. Back in the States, knowledge remained unsettled. The main reason was that no one even knew what the ground did during an earthquake. No recording instruments were in place in quake-prone areas, and, in fact, standard seismographs were too delicate to remain functioning when exposed to shocks strong enough to damage buildings. Although architects and engineers had learned that buildings had their own peculiar frequencies to which they would vibrate when agitated, designers needed to know the frequency of ground waves during powerful quakes, because if the quake’s frequency happened to match, a resonance could be set up. Just as a fine crystal goblet will break when exposed to a critical note, a building might rock gently, then violently, until it collapsed. Lacking the instruments to study ground motion, researchers sent out thousands of postcards to California residents and asked them to report on any quakes. They went to graveyards and recorded the fall of the headstones. One man sent in a drawing of the scratches his gas stove had left on his kitchen floor during a quake; it was one of the best records available.
In 1932 the U.S. Coast and Geodetic Survey began installing “strong-motion” seismographs capable of catching an earthquake’s wild signature. Then, one year later, a near disaster brought real earthquake engineering to the United States. Just before six in the evening on March 10, 1933, a Richter 6.3 earthquake centered in the ocean not far from Long Beach, near Los Angeles, killed 120—most of them struck in the streets by falling stone, brick, and glass. An uproar was caused by evidence that thousands more would have died had the quake come a few hours earlier—when schools were in session.
Of the forty-two major masonry buildings in the Long Beach school system, thirty-eight were unusable after the quake, and nearly half were total losses. In Los Angeles tens of thousands of schoolchildren had to attend classes in tents and bungalows for the next two years, and about a fifth of the schools required demolition. The collapsed schools had typically been cheaply built with walls of unreinforced brick or hollow tile supporting wood roofs. One writer called them “the most shocking collection of deathtraps that ever disgraced a great metropolis.”
Just as hazardous as unreinforced masonry walls were the parapets and stone decorations on buildings. Tremors shook these through roofs or onto the streets, where they exploded like fragmentation bombs. The poor performance by school and commercial buildings prompted the passage of state laws requiring inspectors to oversee school construction. The laws also mandated that new buildings, and schools in particular, had to be capable of handling sideways pushes equaling as much as 10 percent of the total weight. One new Los Angeles school met that requirement with brick walls heavily laced by steel, and floors and a roof of reinforced concrete.
The lateral-load requirement was not a very sophisticated way of preventing damage—an earthquake is not comparable to a giant hand pushing on a wall in a steady way—but it was good enough. In the San Fernando earthquake of 1971, these post-1933 schools would all remain standing.
Several of the strong-motion seismographs installed by the U.S. Coast and Geodetic Survey the year before the Long Beach quake lay close enough to the epicenter to produce usable charts. These tracings, the first ever, showed high accelerations and fast vibrations at the outset, tapering off to one or two seconds between much gentler shocks. Researchers at MlT pounced on these tracings. They rigged a hydraulic-powered platform to reproduce the motions, and tested model buildings on the platform. At a similar Stanford lab, models went onto a heavy wheeled cart that was struck by a half-ton pendulum and shaken by an off-balance fly-wheel to re-create the quake.
In 1934 the same Stanford mechanical engineering laboratory produced a portable real-building shaker, and the Geodetic Survey used it in the ensuing years to measure the resonating frequencies of dozens of actual buildings, dams, and bridges. Bolted to something substantial in these structures, the machine would impart a lurching, back-and-forth motion to the building. The operator would start at a high frequency and work his way down to the lower ranges; with the help of seismographs installed around the structure, he could find its critical frequency. A typical medium-sized building might resonate at one to two seconds per cycle. Sometimes the building shook so much—a few fractions of an inch—that the operator could tell the frequency by the rattling of windows.
This information gave designers feedback on exactly how the height and stiffness of a building affected its frequency (generally speaking, the taller or more flexible a building, the lower its frequency). It also allowed them to check for damage after an earthquake. If the resonating frequency of a building was lower after a quake, damage was the most likely explanation, because cracks to a building’s frame make it more flexible.
The popular press called them “earthquake machines,” but that wasn’t quite the case. They didn’t have enough power to simulate the extreme forces of an earthquake, so still more raw information on actual earthquake ground motion was needed. Seismologists didn’t have long to wait. When a magnitude 7.1 temblor rattled California’s Imperial Valley in May 1940, a strong-motion seismograph was already at work in the southern town of El Centro. The graph showed a maximum sideways acceleration of one-third of gravity. “For many years afterward, it was the strongest motion ever recorded,” says Prof. George Housner of CaItech. “It was treated by some as the maximum possible.” In fact, engineers who thought even the El Centro graphs understated the dangers and tried to design still more stoutly “were fired or had very few jobs,” recalls engineering geologist James Slosson, of Los Angeles.
They were vindicated later. A Parkfield, California, earthquake in 1966 registered one-half of gravity horizontally, and then came the San Fernando quake of 1971 (maximum 1.25 gravities horizontally) and another Imperial Valley quake in 1979 (a peak of 1.75 gravities vertically). These peaks were most likely shock waves amplified by unusual topography, but it was clear that the 1940 El Centro record had been no ceiling at all.
Meanwhile, quakes in Mexico City (1957), Japan (1964), and Alaska (1964) taught that soil conditions could make shock waves much more dangerous than acceleration peaks might indicate. In Mexico City the problem was the soft, wet soil under the business district. It vibrated at just the right frequency to set small towers rocking violently. In Niigata, Japan, vibration made the sandy soil liquefy, and apartment buildings tilted crazily. In Anchorage the quake caused large areas of soil to slump and crack.
Anchorage also revealed that there was a lot of variation in the earthquake resistance of buildings constructed under the same modern codes. One problem was with buildings of different types set immediately adjacent; these rocked at different frequencies and sometimes bashed great holes in one another. Post-quake examinations revealed poor reinforcements, concrete-block walls without cement fillings, and bad mortar joints.
The complete collapse of some modern buildings in Anchorage and serious structural damage to others suggested that frames might need more emergency reserves of strength than the old code had required. This extra margin would come from a concept engineers call “ductility.” Seismic building codes first recognized it in the fifties.
A frame of steel or reinforced concrete can always absorb a limited amount of energy and return to its original shape, like a spring. That’s nonductile behavior. But that frame can absorb a great deal more energy by permanently bending or cracking. If you can design the frame so it will remain standing after that damage, that’s a ductile-frame building.
Two years after the Alaska earthquake, Los Angeles changed its seismic-design code to allow more ductile design and, as a gesture of confidence, removed the thirteen-story height limit. The first tall building to take advantage of this opportunity was the 210-foot Sheraton-Universal Hotel, which used a reinforced-concrete frame. The engineers provided vertical columns stronger than the horizontal beams, inserted weak points halfway down the beams and columns to absorb damage, and put extra-strong connections between them to ensure that the frame wouldn’t fold up under the shocks like a card table.
Patients and staff who occupied the first floor of the main building at Olive View Community Hospital, in Sylmar, California, on February 9,1971, probably owe their lives to a set of concrete columns specifically designed to be ductile. When the San Fernando earthquake hit, the five-story building swayed, started to collapse, and stopped, with the upper floors eighteen inches out of plumb. The credit goes to round concrete cores embedded in the majority of the first-floor columns. The round cores had the usual vertical reinforcing bars, but they also had a spiral wrapping of thick wire that trapped the fractured concrete. The first-floor columns of the hospital’s two-story Mental Health Building next door lacked that spiral reinforcement, and they crumbled as that building crashed down. Fortunately nobody was in it.
Completed just months earlier, both Olive View buildings were total losses and had to be demolished. In fact, the earthquake put all five San Fernando-area medical centers out of action. Even worse, a veterans’ hospital in Sylmar collapsed and killed forty-nine. This poor performance prompted passage of a state law to toughen the quake resistance of new hospitals.
The 1971 San Fernando earthquake also dealt a severe blow to a fifty-year-old concept called the “soft story”—the idea that a flexible floor near the base of a building could absorb ground motion and prevent the top floors from whipping around like a sapling in a storm. Architects liked it because it allowed them to leave out heavy, windowless walls on the first floor and put in bright, street-facing shops. The Olive View buildings had used a soft-story design.
What really turned engineers against the idea, recalls Professor Housner of Caltech, were results from strong-motion seismographs placed in buildings before the 1971 shocks. “For the first time,” he says, “we recorded in many buildings the shaking at the base and the vibration of the building itself at roof, basement, and mid-height. Suddenly the picture became very clear, and a lot of misconceptions were swept away.”
Researchers now believe a flexible lower story can in fact increase the motion of upper floors. Furthermore, according to Henry J. Degenkolb, a San Francisco structural engineer, “in any building with a change of stiffness from floor to floor, you get a concentration of stress.” A weak floor acts like a magnet to attract energy to where it cannot be tolerated.
The 1971 San Fernando quake held still other more painful lessons. All fifteen bridges at the intersection of Interstates 5 and 210 sustained damage ranging from cracks to total collapse. A total of sixty-two bridges needed repair or replacement afterward. The most common problems were nonductile concrete columns and poor or nonexistent ties, which allowed long roadway ramps to separate at the hinges. That prompted the state’s highway department to change the standards for new bridges and examine thirteen thousand pre-1971 bridges. One-tenth of them have recently been strengthened.
Another conclusion from 1971 was that it’s not enough for a building to survive. About half the dollar loss came from nonstructural damage—to facades, interior walls, and equipment—and for some buildings it far exceeded the cost of structural damage. Nearly ten thousand customers lost phone service when tons of automatic switching equipment fell over. In the Los Angeles metropolitan area, the massive counterweights of almost seven hundred elevators shook loose from their guide rails, and some crashed through the roofs of cabs moving in the other direction. The quake knocked the emergency generators at Olive View Hospital from their spring mounts; life support machines in the intensive care unit winked out, and two patients died.
The first solution was to screw everything down more tightly and design automatic locking devices to prevent runaway elevator counterweights. Another approach, called base isolation, aims to reduce damage by mounting entire buildings on rubber feet.
The idea of insulating structures from ground vibrations goes back many years; the Japanese have long known that placing their houses’ foundations on smooth boulders would allow them to slide around harmlessly. John Milne, an English-born professor at Tokyo’s Imperial College of Engineering at the turn of the century, once built a lighthouse on ball bearings. And a 1947 addition to a Sears, Roebuck store in Los Angeles used two layers of roller bearings to separate the old from the new construction.
But anything on rolling bearings has to be restrained from wandering; base-isolation research now concentrates on laminated steel-and-rubber blocks. For several decades heavy rubber mounts have been used to support highway bridge sections, allowing them to safely change length during temperature swings. Rubber pads have also insulated dozens of London buildings from subway-train vibrations. A school built in Skopje, Yugoslavia, in 1969 was the first building to use rubber pads for seismic protection; the Foothills Communities Law and Justice Center in San Bernardino, California, is the only building in this country that relies on them for that purpose. The Foothills building, a four-story courthouse completed this year, sits on ninety-eight round pads made of interleaved rubber and steel sheets. Each pad is the size of a small coffee table.
James Kelly, an engineering professor at the University of California at Berkeley and a co-designer of the Foothills bearings, says the bearings are substitutes for a ductile frame and should provide better protection. He adds, though, “You don’t need it for very tall buildings, and you probably couldn’t use it on them either.” Tall buildings need tension connections to restrain them from capsizing during an earthquake, and rubber bearings don’t provide much tensile strength. Whether they become common on shorter buildings will depend on the performance of the existing few. “Base isolation is like religion—some believe it and some don’t,” says Mete Sozen, an engineering professor at the University of Illinois. “I don’t have that religion.”
Even more visionary are two seismic-protection schemes proposed by American researchers. One plan, by a group of University of Southern California faculty members, proposes using large tanks of compressed air. Jets of the air released from a building’s upper stories would act like thrusters to oppose quake-induced motions. The other idea, called active control, was developed at the State University of New York at Buffalo. It would use computer-controlled hydraulics to tighten or loosen steel cables on selected floors of a building, thereby counteracting oscillations caused by winds or earthquakes. Professor Tsu Soong of SUNY admits that the machinery sounds hard to maintain, but adds, “It can be tested periodically to see that everything works. I don’t think that’s a problem.”
For a long time to come, nothing will beat an actual earthquake for showing what works. Mexico was the severest test ever of modern buildings.
One of the tools used by an earthquake engineer like Soong is a modern shake table. A typical table is supported on a cushion of air while large hydraulic cylinders shove it all over the compass and up and down. Scale models mounted on the platform register what a full-scale building might do in a similar situation. The earthquake lab of the University of California at Berkeley recently finished a five-year project that used its twenty- by twenty-foot shake table to cross-check parallel experiments taking place at an extraordinary facility in Tsukuba, Japan, called the Large Size Structure Laboratory. While Berkeley worked with scale models of both a reinforced concrete and a steel-framed building, the Japanese tested the same two types, but at life size, using hydraulic cylinders anchored to a massive wall a few feet away. Computers controlled the cylinders, forcing the same distortions that an earthquake would cause. Everything took place in ultra-slow motion: the cylinders moved a few fractions of an inch, then paused as the computers recalculated the next step. One 30-second earthquake required ten days to run but provided highly detailed data, says Robert Hanson, a technical coordinator for the American team.
For a long time to come, though, nothing will beat a real earthquake for showing what works and what doesn’t. The 1985 Mexico City disaster provided the severest test ever of large modern buildings. Damage was concentrated in only a small portion of the city’s area, but for buildings having a resonating frequency near two seconds, it was a terrible ordeal. A layer of lake-bed sediments took the fading shocks from a Richter 8.1 quake two hundred miles away and amplified them into a motion that rocked the ground sixteen inches back and forth every two seconds. In less than a minute, more than seven hundred buildings sustained severe damage; about a third of them collapsed.
“I think it has confirmed our understanding of structural response,” says Professor Sozen, who led a seven-member team of U.S. earthquake investigators at Mexico City. “A few changes in the seismic code will come out of it, but nothing dramatic.” His advice for American cities exposed to infrequent but powerful earthquakes: Be prepared to dig out the survivors very quickly. Of the twenty to thirty thousand people killed in the Mexican quake, four out of five died because rescuers couldn’t get to them in time.
Unfortunately, Sozen admits, there’s no likelihood of earthquake-proofing cities like Memphis, Tennessee, to the degree that California cities have been—though that city is exposed to powerful quakes of the type generated at nearby New Madrid, Missouri, in 1811. “It would take billions of dollars and many years,” he says, and the citizens won’t support it because quakes in that region are so rare.
Of course, the urgency with which California has been quakeproofing has led to some mistakes and miscalculations there. Six years after Los Angeles began requiring the seismic buttressing of unreinforced masonry buildings built before the first earthquake codes, it and other California cities face the problem of identifying and strengthening deficient buildings constructed after the codes took effect. Multistory buildings with nonductile concrete frames built as recently as the early seventies may be a problem because standards of the time underestimated the likely earthquake shocks. “Tilt-up” buildings made with concrete slabs and light roofs may also need attention. They’re quick and cheap to erect but require especially strong fastenings in quake-prone areas. “The seismic codes can never keep up with present construction,” says Degenkolb. “The building code changes every three years or so, and it follows current knowledge by about ten years.”
The economics of quakeproofing aside, it’s clear that in the last half-century we’ve learned an enormous amount about how to make our structures stand up to the strongest earthquakes. We may never tame earthquakes, but we’re well on the way toward subduing our buildings—and they are the real killers.
James R. Chiles, a lawyer and writer, lives in Dallas. His article on the lessons of the great blackouts appeared in our Fall 1985 issue, and he wrote about the history of the ball bearing in our Summer 1985 issue.