Last year, Discover presented a gallery of 6 images generated by scientists using supercomputers to do leading-edge research in various fields. After lots of positive feedback, we gathered together another bunch of supercomputer simulations.The sim shown here could help minimize damage in future earthquakes. Carnegie Mellon University engineer Jacobo Bielak's simulations provide architects and regulators with reliable, localized estimates of the shaking intensity that buildings would have to withstand during earthquakes of varying strength. In this mock earthquake, created at the Pittsburgh Supercomputing Center, a 4.4-magnitude tremor strikes three miles beneath a valley in northern Greece. The colors represent the amplitude of the ground's horizontal motion six seconds into the quake: The soft sands of the valley, shown in blue and red, shake more violently than do the surrounding solid rock, pictured in green and yellow. "It's like shaking a bowl of Jell-O," Bielak says. "The soft Jell-O will always vibrate harder than the solid bowl." The model has proven so accurate that engineers designing ITER, an experimental nuclear fusion facility, have requested simulations of the building site in southern France.
Exploding stars have shaped every part of our world: They created the iron and other heavy elements in our bodies, and one of them may have provided the shove that initiated the formation of our solar system. But studying how a supernova works is exceedingly difficult because the brilliant nuclear flash that destroys the star obscures the underlying details. So astrophysicist Thomas Janka of the Max Planck Institute in Germany created a digital supernova in his computer. The simulation allowed him to watch the core of a dying star collapse and then rebound before flying apart in a fantastic explosion. The image here shows the spread of searing plasma just one second after the explosion. "We put in physics that we think we understand, and we try to create the explosion and follow it to the point where we can compare it with observations," Janka says. Equations allow him to model everything from the mixing of gases to the behavior of neutrinos, enigmatic particles that are released en masse from the stellar core just before detonation. "We believe neutrinos are the crucial ingredient," he says.
A passageway called the circle of Willis opens up to several major arteries at the base of the brain, ensuring a steady flow of oxygen. The sheer volume of blood moving through makes this region susceptible to aneurysms, weaknesses in arterial walls that can lead to stroke and debilitating brain damage. Leopold Grinberg, an applied mathematician at Brown University, developed a fluid dynamics model and ran it at the Pittsburgh Supercomputing Center to learn more about blood flow at the circle of Willis. He and his team use CT and MRI scans of aneurysm-afflicted arteries to create patient-specific simulations. Then they release virtual blood into the model and examine how differences in circulation affect aneurysms. They can even digitally remove the aneurysm to see how the flow changes. "We're working backward to see the flow that led to the problem in the first place," he says. In this image, Grinberg visualizes the large-scale circulation through the entire region, but he can also zoom in to view the journey of individual blood cells.
The human body is a complex electrical network: Nerve cells shuttle signals from the brain, and pulses in the heart cause its muscle cells to expand and contract. Cells would never receive these electrical dispatches without special proteins like the one depicted above. The porelike protein forms a potassium channel, an electrical entryway into the cell that toggles open or shut depending on the outside voltage. In this simulation, an electrical pulse (the rainbow-colored pattern on the right) sweeps toward the potassium channel; amino acid chains that make up the protein appear as colorful ribbons, while the stringy background represents the cell's fatty membrane. Biophysicist Benoit Roux of the University of Chicago recently used the world's fastest supercomputer -- Jaguar, at Oak Ridge National Laboratory in Tennessee -- to see how each of the protein's 350,000 atoms would react to such electrical pulses. Such studies could have important medical implications: Faulty electric signaling contributes to heart arrhythmias and may increase the risk of Alzheimer's disease, too.
It may look like a satellite photo, but this Earth portrait comes from a climate simulation run on the supercomputer at Oak Ridge. The image charts water vapor in the atmosphere, with highest concentrations in white; note the cyclone approaching India. To emphasize human susceptibility to climate, Oak Ridge visualization specialist Jamison Daniel shaded the landmasses to reflect population density (with orange and red denoting the highest concentrations). Such models are intended to bridge the gap between detailed, short-term weather forecasting and broader, long-range climate modeling. These models might be able to peer up to 50 years ahead and "show regional events, like a heat wave in India, rather than just global trends, like higher temperatures," says Kate Evans, a scientist at the lab.
At Sandia National Laboratories in Livermore, California, Jacqueline Chen and Chun Sang Yoo use supercomputers to simulate burning fuel. The image at right depicts a cold jet of ethylene (a hydrocarbon similar to that found in automotive gasoline) combusting in hot air. By controlling the speed of the ethylene jet, Chen and Yoo alter how rapidly the fuel burns, a measure of its efficiency. (The goal is to optimize the speed. Think of a birthday candle: Blow on it gently and the flame intensifies, but blow too hard and the flame goes out.) In this model, blue and green represent formaldehyde, a by-product of ethylene ignition. Red represents hydroxyl radicals, markers of flame. The technology can also depict alternative fuels like ethanol and biobutanol, paving the way for greener internal combustion engines in years to come.