This article is reproduced with the permission of New Scientist for exclusive use by Nova users.

Disaster machines: Simulating nature's fury
01 September 2007
NewScientist.com news service

Enlarge All shook up

Enlarge Making waves

Hurricane

The "Wall of Wind" is the kind of experiment any 10-year-old would love to try. Point a pair of 2.5-metre fans at an ageing house. Crank them with some serious muscle - 500-horsepower race car engines - so they blow air at 190 kilometres per hour, the force of a category-3 hurricane. Then add water to mimic the horizontal rain of a hurricane in full fury and watch what happens.

"We immediately blew off the door, then asphalt shingles, then windows started breaking," recalls Stephen Leatherman of Florida International University in Miami. "The wind blew through the house, blew the windows out, then blew stuff out the back windows." As it passed through the house, the wind also pulled down ceiling tiles and scattered them on the floor.

When the roar died away, Leatherman says, awed onlookers asked: "Wow, why did you do that?" As he explains, the goal was to discover the typical weak points in homes so that buildings can be bolstered to withstand hurricane-force winds. It's a vital consideration in Miami, where two years ago category-2 hurricane Wilma caused $16 billion of damage. All the same, the neighbours of the test house were wary. "They thought we were mad scientists going to blow their houses down," Leatherman says.

Since wind tunnels can only hold toy-sized replicas of buildings, these scaled-down tests can't clearly identify what causes structural damage to real homes. To learn how building materials fail, you need a full-scale tempest. That's why the International Hurricane Research Center, which Leatherman heads, built the Wall of Wind and tested it on an actual home - the first full-scale experiment of its kind.

Testing roofs is particularly important. Hurricane Wilma blew off or damaged hundreds of thousands of roofing tiles, and each flying tile can cause serious damage to other structures. Leatherman is studying new ways to bond the tiles and protect them. One answer might be to strap a water-filled "mattress" to the roof before the storm arrives. Another problem comes from soffits, decorative projections with vents at the edge of a house that allow air to cool the roof space. Hurricane winds blow rain horizontally through the vents and soak the inside even if the roof stays on. Tests will show which designs are best at keeping water out while allowing ventilation in good weather.

Powerful as the Wall of Wind is, it can't yet generate a broad enough blast to fully surround a house. To do that, Bermuda-based Renaissance Reinsurance Holdings, which specialises in disaster reinsurance, has paid to develop a larger six-fan version that has already produced winds up to 210 kilometres per hour. The full set-up, which should be ready in October, will be housed indoors with special lighting and an array of high-speed video cameras and gauges to monitor wind speed and air pressure. It will include a huge turntable that can be rotated to expose a house to wind from all sides, and will tackle a number of problems, including how best to secure roofing tiles and boards.

That already sounds like a full agenda, but with $2 million in funding from the state of Florida, Leatherman is also planning a 24-fan array that should be up and running in a year. By redesigning the fans to produce smoother airflow, he aims to reach 265 kilometres per hour to simulate category-5 winds hitting a two-storey house. Leatherman calls it "mad scientist work", but his goal is entirely rational: making houses solid enough that future hurricanes will huff and puff in vain.

Jeff Hecht

Earthquake

A TALL building is probably the last place you want to be in an earthquake. That's why, in the heart of quake country, researchers at the University of California, San Diego, (UCSD) are shaking things up with some new gear: the world's only large-scale outdoor quake platform. Located 15 kilometres east of UCSD's main campus, it can shake a 20-metre-tall, 20-tonne structure from side to side as if it were a toy - all in the name of safety. Researchers hope that this machine will tell them how to build seismically safer structures.

Completed in 2004, the Large High Performance Outdoor Shake Table is a 93-square-metre steel platform attached to a pair of hydraulic actuators on either side. These are fed by high-pressure fluid pumped from a building nearby, and guided by software to precisely control the platform's motion (see Diagram).

About a dozen other large shake tables are in use around the world, but none - not even the biggest, Japan's E-Defense near Kobe - can handle structures as tall as those tested here. "The benefit is the huge load that we can put on it. You can build right on top of it," says Al Clark, an engineer with Minnesota-based MTS Systems, which helped build the system.

Last year, principal investigators José Restrepo, Joel Conte and Enrique Luco of UCSD performed a ground-breaking series of tests on the platform. Together with construction industry partners, they built a full-scale slice of a seven-storey building on the shake table. They used much less steel-reinforced concrete than California's building codes require, to test whether such a design might offer the same or better safety performance for a fraction of the cost.

They then hit the structure with increasingly intense "quakes" by replicating ground motions recorded during the 1971 San Fernando and 1994 Northridge earthquakes in California. All the while, more than 600 sensors in the building measured stresses, strains and accelerations in different parts of the structure. The researchers concluded that walls with half the stipulated amount of reinforced concrete perform better and more predictably during shaking. "This test clearly showed newer ways to design buildings to resist earthquakes with minimal structural damage, and with significant savings in construction costs," Restrepo says.

Meanwhile, a team at the San Diego Supercomputer Center at UCSD has been developing computer visualisations of the tests by incorporating the sensor data into a 3D model based on construction drawings. This allows researchers to zoom in on a piece of the building as well as watch the whole thing during the test - an easier and more intuitive way to home in on the data they are looking for. "It's a fundamentally different way of doing analysis," says Amit Chourasia, a visualisation scientist at the Supercomputer Center.

The table team is now proposing to modify the platform to make it shake in two or, ideally, three dimensions. Restrepo says it could then recreate historical ground motions more accurately and allow researchers to test more complex structures such as bridges, nuclear waste casks and chemical plants. In the shorter term, tests in the coming year will focus on veneer walls, brick structures and a three-storey building made entirely of precast concrete.

Despite increasingly powerful computer models, Restrepo believes that large shake tables will remain important in seismic engineering research. "Would you fly in an untested, but fully computationally designed 787?" he asks. "I always find that testing teaches more than one new lesson. Everything becomes so obvious after the test, but not before."

Kimm Groshong

Fire

FIRES killed 3675 Americans in 2005 - more than all natural disasters combined. Yet until recently, fire investigators hoping to reconstruct fire scenes were out of luck if they wanted somewhere to scientifically test their hypotheses. All that changed in 2003 with the opening of the Fire Research Laboratory in Beltsville, Maryland. Run by the US Bureau of Alcohol, Tobacco, Firearms and Explosives, it is the first facility dedicated to aiding criminal fire investigations.

Imagine a cavernous space so big that engineers can reconstruct a three-room apartment or even a two-storey office building, all beneath the world's largest stainless-steel calorimetry hood for measuring the heat output of fires. The larger of the lab's two "burn rooms" is 17 metres tall and covers 1570 square metres. Here engineers study ignition methods, the causes of electrical fires, the speed at which items burn and the way flammable liquids affect a fire's spread.

The hoods help measure how fast a test fire releases heat, which determines its power. Engineers use this to predict whether a fire would burn in a wastebasket, say, or spread to a chair nearby and ultimately engulf a room. They can also measure the flow rates of smoke and noxious gases rising into the hoods. That allows them to calculate the compounds' concentrations at various locations and estimate how long a person might survive in the environment.

When recreating a fire, the engineers and craftsmen are faithful to the original right down to the furnishings. The total amount of combustible material is crucial. If a room had bundles of laundry tossed on the floor, it is carefully replicated. They can choose from a vast toolbox of measuring instruments: they might position a "thermocouple tree" where a body was found, for instance, to take periodic temperature readings at various heights. This helps them estimate how long the victim remained conscious.

Before a test blaze, engineers don protective gear including masks. They can watch the fire's progress on a bank of monitors away from the test structure. Video cameras record the scene inside and out, while an infrared camera peers through the smoke-filled interior to capture images of the flames. Firefighters in full gear stand by while measurements are collected, before dousing the blaze. As they extinguish the fire, water can hit the floor at 7500 litres per minute. It collects in trench-like drains before being purified and recirculated.

Experiments like these have disproved myths such as the notion that fires burn only upwards, or that blackened burn patterns on the floor indicate liquid was poured there to start a fire deliberately. We know now that if there is enough ventilation inside a structure, a fire can reach flashover - the point at which all flammable materials present simultaneously ignite - and leave any manner of burn marks.

Fire simulations work hand in glove with other methods. Imagine a hotel room fire, for instance. "If people died from carbon monoxide poisoning 95 feet away in a different room," suggests David Sheppard, a senior fire research engineer at the lab, "you want to know how quickly the gases are being generated from the fire and how quickly they'll travel down the hall." Even the Fire Research Lab can't yet test a corridor that long, but they can model enough of the scene to provide data that, combined with models, could speak volumes about what went on in the fire.

One lesson learned: much modern furniture burns faster than traditional types. So leave a building the moment a smoke detector sounds, or it could be too late.

Sue Russell

Tsunami

STEP into the hangar-sized building housing the Tsunami Wave Basin at Oregon State University in Corvallis, and strange things happen to your sense of scale. The water-filled tank that dominates the space is seriously big: 49 metres long by 26.5 wide by 2.1 deep. It is the largest and most sophisticated wave tank in the world, and the first dedicated to tsunamis, yet it's only a drop in the bucket compared with the 400-kilometre-wide super-waves it aspires to simulate.

At one end of the tank, 29 2-metre-high paddles powered by electric pistons make waves. Aided by custom-made software, human operators can play them like piano hammers, directing waves with uncanny precision. Starting with the paddles on the sides and moving in turn towards the middle sends a wave surging towards the centre of the tank's other end. Twenty wave gauges, four velocity transducers, three microphones, 10 underwater cameras and six surface cameras record each wave's height, period, speed and direction from many viewpoints.

These are all familiar technologies in wave research, but what distinguishes the tsunami basin, part of the Hinsdale Wave Research Laboratory, is its scale and ability to generate waves in 3D, says lab director Daniel Cox. That enables it to simulate a tsunami wave's lateral motion as well as its head-on forces, the better to approximate its notorious complexity.

A major question is exactly what impact tsunamis can have on coastal structures and sediment. So in July, researchers built a miniature model of a coastal town along a sloping "beach" at the edge of the basin. They are now setting up experiments to measure the resulting forces as the water hits the shore, and to test whether buildings of certain shapes, such as cylinders, might be better than others for withstanding a tsunami.

Harry Yeh, a researcher at the lab, admits that the basin simply isn't large enough to closely mimic real tsunamis, but it is still vital, he says, because tsunamis are so rare and unpredictable, making field measurements all but impossible. Numerical models can't yet fill the gap, because researchers need more physical data to design good models. "This is the best we can do," says Yeh.

Not for long, according to researchers at University College London and UK-based hydraulics consultancy HR Wallingford. In July, UCL quake researcher Tiziana Rossetto announced plans to complete a tsunami machine by next year that will undertake the first "realistic large-scale experiments".

This wave generator, the brainchild of William Allsop of Wallingford, will use technology from tidal simulators developed in the 1970s. It is essentially a large tank fitted with air pumps that push water into a flume fitted with sensors to gauge the wave, then suck it back into the tank, just as tsunamis surge forward, draw back and then slam the shore - three or four times in the case of the 2004 Indian Ocean tsunami. What's more, says Allsop, the waves will have long periods: 45 to 90 seconds, the scaled version of real tsunamis' periods of 5 to 15 minutes.

The US and UK teams may dispute what is most important in simulating tsunamis - precision, control and 3D versus long periods and multiple, reversing waves - but the fact that both are making progress bodes well for coastal communities seeking ways to withstand killer waves.

Eric Scigliano

From issue 2619 of New Scientist magazine, 01 September 2007, page 40-43

For the latest from New Scientiist visit www.newscientist.com