Flash Center building simulation computer code, validating results

Three-dimensional computer simulations of the turbulent mixing dynamics of fluids. These simulations provide density snapshots of a single plume of dense fluid (shown in red) descending into light fluid (shown in yellow). The simulations are of increasing resolution from left to right.

The seemingly simple act of stirring cream into a cup of coffee presents a scientifically complicated problem in fluid dynamics and mixing, as does the turbulent mix and flow of gases that trigger exploding stars.

Unfortunately for scientists in the University’s Center for Astrophysical Thermonuclear Flashes, checking the accuracy of a numerical model that involves pouring a cup of coffee is much simpler than running the same test for an X-ray burst, a classical nova or a type Ia supernova. This checking process, which demonstrates that models and simulations accurately describe nature, is called “validation.”

“If you aren’t reasonably sure that your simulations can meaningfully describe nature, then you’re really wasting your time,” said Alan Calder, a Research Scientist in the Flash Center. “It’s not necessarily that the codes have a bug. It’s just that how the codes do the physics and the physics that are included in the codes don’t necessarily mesh with what is really needed.”

Validation, along with building a computer code that simulates exploding stars, has been a significant part of the Flash Center’s effort since it was established in 1997. Theorists and experimentalists in the center also carefully examine the issues involved with modeling scientifically important physical processes. With a new $22 million contract with the U.S. Department of Energy, the center’s researchers will be able to devote more effort to the problem over the next five years. As a result of this increased effort, Calder and 17 of his colleagues have written a validation paper, which will appear in the November issue of the Astrophysical Journal Supplement.

“Validation is almost a new scientific subject in astrophysics,” said Robert Rosner, the William Wrather Distinguished Service Professor in Astronomy & Astrophysics, and a co-author of the paper. “If you’re trying to look for papers and books, it is amazing how little there is. In fact, Alan is the chief author of probably the first thorough-going validation paper on astrophysics.”

The validation of the Flash Center’s simulation code has implications for institutions across the country and around the globe for national security and for a variety of fundamental scientific questions.

The Flash Center has received nearly 200 requests for its simulation code from various institutions. Outside the United States, the heaviest users reside in Germany, Italy, Japan, Poland and Norway. The heaviest users domestically include various national laboratories; schools in the University of California system; the University of Texas, Austin; the University of Illinois, Urbana-Champaign; and the State University of New York, Stony Brook.

“The Flash code is a tool that can be used for all sorts of problems, ranging from astrophysical calculations to simulations of laboratory experiments,” Rosner said. “We want this to be a community code. It’s used by lots of people, not just here at Chicago.” Because the DOE must ensure the reliability of the nation’s stockpile of nuclear weapons, it has a vested interest in the simulations. “Nuclear weapons are very fragile objects,” Rosner said. “They’re known to age in destructive ways, hopefully in predicted ways.”

This computer simulation mimics the turbulent mixing dynamics of a supernova. Colors indicate fluid density in the region of mixed fluid, with the highest being green and the lowest being red.

In the scientific realm, astrophysicists used type Ia supernovae as astronomical measuring devices, which led them to conclude that the universe will expand forever. X-ray bursts provide information about the basic characteristics of neutron stars, which are the remnants of stars that have become supernovae. Classical novae help scientists calculate the abundance of certain elements in the universe and understand the dynamics of aging stars that closely orbit each other in a binary system.

The journal paper explains two tests that compared the simulation code with experimental data. In one case, the code agreed well with the experimental results, while in the other case, it did not. These results highlight both the need for and difficulty of conducting validation tests, which must assess error and uncertainty in theory, experiment and computation, wrote Calder and his colleagues.

The simulation codes tested well against results produced in laser-driven shock experiments conducted at the University of Rochester’s Omega laser facility. These experiments, designed to replicate the collapsing core of a supernova, involve laser blasting a tiny capsule that consists of three layers of decreasing density. The capsule becomes compressed as vaporized material jets from its surface.

The Flash team found that its simulations did not agree with the findings of the Rayleigh-Taylor experiments conducted at Lawrence Livermore National Laboratory’s Linear Electric Motor. These experiments demonstrate how, under gravitational pressure, a heavier fluid interacts with an underlying light fluid. The experiments are named for the Rayleigh-Taylor instability, a process that leads to the turbulent mixing of fluids in supernovae and on Earth. The disagreement indicates significant uncertainty in the initial conditions or other aspects of either the experiments or the models and is the subject of ongoing research by both theorists and experimentalists.

“These tests served to increase our understanding of the physics relevant to the problems of interest, to improve our simulation techniques and to build confidence in our results,” wrote Calder and his co-authors.