In the late 90s, researchers at the St. Anthony Falls Laboratory at the University of Minnesota built an enormous tank intended to make many millions of years whiz by in an afternoon. Dubbed Jurassic Tank, the device isn’t exactly the time machine it might at first sound like. It’s basically a 40-by-20-foot tank filled with gravel, water, and sludge—hardly the sleek, exciting time machine depicted by science fiction writers and cinematographers. But appearances often deceive, because for geologists, it’s the next best thing to actual time travel. That’s because the tank can simulate in hours the development of sedimentary rock formations that took hundreds of millions of years to form. Now researchers at the University of Texas at Austin are teaming up with the creators of Jurassic Tank to produce computer simulations of the tank’s results, hoping to move the tank and its experiments into the realm of the virtual world.

Geologic time fast-forwarded Formally known as the Experimental EarthScape Facility, Jurassic Tank was devised to help scientists develop hypotheses of how sedimentary rock was formed, as well as test existing hypotheses. The tank is so critical because trying to observe the natural development of sedimentary rock in real life is pointless, since the process is so slow that centuries wouldn’t allow for even a glimpse of progress.

While there are tools based on similar concepts in use by scientists elsewhere, the EarthScape facility is a true first in terms of its sheer size and sophistication. Geologists can configure the tank to mimic specific geologic formations, such as a riverbed, and can then apply the equivalent of many millions of years of the specific conditions required to form and shape sedimentary rocks, such as flowing water or flooding.

To make this sophisticated configuration possible, the tank is constructed with an elaborate honeycomb structure at its base. A layer of gravel is poured into the tank, filling each hexagon of the honeycomb, and the gravel is then covered with a rubber skin. To create the specific topographies required for an experiment, researchers can control the gravel over each individual hexagon. When needed, the gravel is flushed out with water jets to create an unsupported spot, which causes the rubber skin to sag in that particular area. By flushing the gravel from multiple, adjacent hexagons it is possible to form low terrain, valleys, peaks, or other formations required for a particular experiment.

Researchers then systematically release sediment into the tank in the form of coal dust and white quartz dust to produce both black and white layers. Once the experiment is completed—which can be anywhere from a number of hours to a few days—the tank is drained. Geologists then slice the resulting formation to study the black and white layers, in hopes of better understanding the natural phenomenon they’ve simulated.

These experiments are far more than a pure academic pursuit of knowledge. The information they produce can hold great value to commercial ventures, such as petroleum and mining companies.

From sediment to silicon chips JT_99_iso_1680x1050 While it’s easier in many ways to deal with the tank’s miniature version of a specific topography than to tackle the real thing in nature, researchers still face the question of where to look for the most interesting and most revealing formations. In other words, where—and exactly how—should they slice through the tank’s sediment formation? To find the answers to this question, researchers approached the Texas Advanced Computing Center (TACC), at the University of Texas in Austin.

TACC is a supercomputing resource that offers high-performance computing to partners throughout Texas and across the country. Among the center’s specialties are data analysis and visualization. Paul Navratil, a research assistant at TACC pursuing his PhD in computer science, was given the project of creating a visualization from the data that geologists collected during one of their experiments using the tank.

“The Geologists wanted us to first come up with something that replicated the volume of the tank and to see what the data showed. They wanted to determine where interesting features were located by looking at the visualization. In the case of the data we were using, that meant finding where channels were forming,” said Navratil.

The data provided was from a tank simulation that studied an alluvial basis, where a river emptied into a lake or ocean. The data was collected over a 10-day period, representing 40 different time steps. Measurements included basin depth, the surface, and the delta.

EnSight takes the tank into virtual territory From the measurements provided in the geologists’ data, Navratil was able to extrapolate how the layers changed over time. The resulting animation created with EnSight, from CEI (Apex, NC), which covers all 40 time-steps, is a “time-lapse” or “flipbook” animation.

Looking at the experiment over time clearly shows large regions of change, and these “hot spots” can be color-coded to reveal where large quantities of sediment are deposited. The images can also be sliced as needed, to show cross-sections of different areas virtually, just as researchers do with the actual sediment from the tank.

To create the visualization, Navratil aggregated the depth readings to create a pseudo-volume, and then used EnSight Gold’s isosurface slice capacity to examine how the depth changed over time in various parts of the simulation. Z-axis slices in the animation show how formations at that depth evolved.

EnSight Gold, from CEI, was chosen for the project not only because of its visualization capabilities but because loading the extensive geologic data promised to be relatively simple with this software.

“You always have to do that massaging step, and it can be better or worse depending on how pliable the end product is. We found EnSight was really very easy. There wasn’t any problem loading the data, because the formats that EnSight supports are really straightforward,” said Navratil.

In working towards the final animation, Navratil found EnSight’s flexibility very helpful, particularly given the large dataset involved. “As I got more familiar with the data and found other aspects I wanted to explore, or needed to expand different fields, it was easy to do, which isn’t always the case. With other tools I might have had to re-import the data—which could take 20 minutes per import—but with EnSight I could just associate the new data with the existing file structure, and it worked perfectly.”

Now that the project has produced results and has shown that EnSight simulations can provide useful information from Jurassic Tank data, researchers are waiting to see what direction their work will take them next. With added data, such as water flow measurements, simulations could be made even more complex, and include computational fluid dynamics calculations. One day Jurassic Tank itself could go the way of the dinosaurs, so to speak, and a supercomputer could eventually take over new geologic experimentation.