For years, NASA has been running simulations of asteroid impacts to understand the risks (and likelihoods) of asteroids colliding with Earth. Now, NASA and the European Space Agency (ESA) are preparing for the next, crucial step in planetary defense against asteroid impacts: physically deflecting a real asteroid.

Orbiting between Earth and Mars are the “Didymos” double asteroids, the larger measuring 780 meters and the smaller 160 meters. Following a launch in July 2021, NASA and ESA are planning to deflect the smaller of the double asteroids (Didymos B) around September 2022 through the Asteroid Impact Deflection Assessment, or AIDA. 

To do this, they will employ two spacecrafts: one (NASA’s DART, for “Double Asteroid Redirection Test”) to deflect the asteroid by crashing into it, the other (ESA’s “Hera” mission, launching in 2023) to survey the impact site on the asteroid after the fact.

“DART’s target, Didymos, is an ideal candidate for humankind’s first planetary defence experiment,” said planetary scientist Nancy Chabot of the Johns Hopkins University Applied Physics Laboratory. “It is not on a path to collide with Earth, and therefore poses no current threat to the planet. However, its binary nature enables DART to trial and evaluate the effects of a kinetic impactor.”

While the Didymos asteroids are not on course for collision with Earth, the goal is to ensure that such a mission would succeed in that situation and to better understand the physical dynamics of an asteroid that is impacted in space. A similar mission by the Japanese Aerospace Exploration Agency (JAXA), which dropped a small explosive on an asteroid, produced a larger-than-anticipated crater and unexpected, sand-like behavior, indicating that gravity was dominating the process.

“If gravity is also dominant at Didymos B, even though it is much smaller, we could end up with a much bigger crater than our models and lab-based experiments to date have shown,” said planetary scientist Patrick Michel of the French National Center for Scientific Research. “Ultimately, very little is known about the behaviour of these small bodies during impacts and this could have big consequences for planetary defense.”

Beyond the practical results, the agencies have another, crucial objective with AIDA: validating their internal models. “The key question that remains to be answered is, are the technologies and models that we have good enough to actually work?” Ian Carnelli of the ESA explained to Technology Review.

Harrison Agrusa, a PhD student from the University of Maryland, is part of a team that has been simulating DART’s impact with Didymos B in preparation for the mission. Agrusa conducted his simulations on the “YORP” cluster at the University of Maryland’s Astronomy Center for Theory and Computation. YORP is a heterogeneous system with an aggregate 22 nodes, 544 cores (a mixture of Intel Xeon and AMD Opteron CPUs), 1.12 TB of RAM and around 75 TB of storage.

“[YORP] is a relatively small, heterogeneous system but it was perfect for our application,” Agrusa elaborated. “Most of my simulations just ran on a single node of 32 cores for about one month. This allowed me to integrate the dynamics of the Didymos system for five months of real life time with our gravitational n-body code[.]” 

Agrusa’s simulations modeled the dual asteroids as collections of small spheres or particles, then applied the force of DART’s collision. The researchers have also leveraged another piece of code called GUBAS (and developed by Agrusa’s colleague Alex Davis) that allows them to integrate the motions of the asteroids as polyhedra rather than as collections of particles. “This is a significant speed-up since we don’t have to compute the potential by summing over every single constituent particle at every timestep,” Agrusa said.

Through these simulations, Agrusa and his fellow researchers found that – under certain conditions – DART should be able to produce “libration” (essentially, a distinct wobble) in the orbit of Didymos B around Didymos A. “The interesting thing, depending on where DART hits and how hard, is that we can see a pronounced wobble triggered as a result,” Harrison said. “We’ve compared four different simulation codes to study this post-impact swinging back and forth and seen the same effect recur in all of them, even with conservative estimates of DART’s momentum transfer.”

Since those initial models, the research team is setting his sights higher. While GUBAS can run on a single core, Agrusa’s team is looking into running it on YORP with a much larger parameter space. “I plan to use that code to do roughly ~1000 simulations split over 64 cores on YORP,” Agrusa told HPCwire. “Since these simulations only take a handful of hours of computer time, my 64 cores can run through ~1000 simulations in less than a week.”

Agrusa also recently returned to Lawrence Livermore National Laboratory (LLNL), where he began taking advantage of much more powerful computing resources – namely, the RZTopaz cluster, a Penguin system with 768 nodes, 27,648 Intel Xeon E5 cores, over 98 TB of RAM and a peak performance of 929 teraflops. 

Using RZTopaz, Agrusa is modeling DART’s high-velocity impact using Spheral, a multiphysics hydrodynamics code. For now, Agrusa has been running low-resolution test simulations – but even those, he says, take up to a week of clock time and over 30 TB of disk space. “So you can imagine,” he elaborated, “that when I end up doing the ‘high-resolution’ versions, that I won’t be able to do a whole lot of simulations because I simply don’t have the disk space for it.”

Ideally, the results of Agrusa’s models will line up with the results observed by Hera after DART’s impact, supporting the ability of these modeling approaches to accurately predict the results of deflection attempts in future real-world scenarios. Still, many unknowns remain – among them, the nature of Didymos B’s interior, which might be illuminated by the real-world observed change in libration. (“Measuring this effect will give researchers an important insight into the nature of Didymoon’s interior, constraining our models,” said Agrusa. “However, it is essential to have a spacecraft on location to make such a measurement.”)

While the culmination of this journey is still a few years off, Agrusa has his eyes on the prize.

“When the DART mission ends with its impact in 2022, then my PhD does too,” Agrusa concluded. “We’ll get a first glimpse of the actual shape of Didymoon from DART and the LICIA CubeSat – provided by ASI, the Italian Space Agency – it will deploy before colliding. Then, within a few years Hera will be providing its data, so we can rigorously compare our models to reality.”