Sometimes the impetus behind large-scale computing endeavors can be surprising. Take the case of nuts and bolts. Given the right context, these everyday objects become a much bigger deal. Like when the context is nuclear missile design.

Every component of a nuclear weapon body must go through a painstaking review process. As an article at Deixis Magazine explains, each part of the structure is made up of numerous subcomponents, and these are held together with screws, nuts, bolts, and other fittings. In terms of structural integrity, joints are an obvious failure point. At Sandia National Laboratories in New Mexico, which conducts systems engineering for all of the United States’ nuclear weapons, advanced modeling is critical to ensuring joint reliability.

“It’s an extremely large, nebulous problem that we cannot solve using traditional methods,” shares Matthew Brake, a principal research scientist at Sandia, where he’s organized an international research institute to tackle the issue. What was once considered well-established physics has been found lacking, beset by overcomplexity and uncertainty in the models.

It’s a balancing act in more ways than one. Brake, who is also the American Society of Mechanical Engineers’ (ASME) research committee on the mechanics of jointed structures, points out that almost everything that is manufactured has joints and even something as simple as a chair is often over-engineered to guard against failure.

The research will be especially impactful in “high-consequence areas” where the cost of failure is steepest. Examples include automobiles, airplanes, satellites, civil engineering structures and weapons.

As a key partner in the National Nuclear Security Administration’s (NNSA’s) stockpile stewardship program, Sandia has a compelling reason to study joint reliability in missiles and nuclear weapons. Since international rules prohibit the flight-testing of nuclear weapons, the scientists have to ensure that their modeling and simulation tools are as precise as possible.

In one report, the engineers concluded that stiffness measurements of manufactured-to-spec bolted joint hardware can vary by as much as 25 percent, and energy dissipation measurements by up to 300 percent. It’s these kind of inconsistencies that are leading the investigators to question the physics of their usual models.

They’ve identified friction as a particularly thorny culprit for being difficult to accurately model. Brake is among a group of experts who cited friction as a “grand challenge” for experimental and theoretical mechanics in the 21st century.

Limitations with fine-mesh finite element analysis led Brake and his group to explore an alternative technique, called Reduced Order Modeling (ROM) that uses equivalents to reduce the number of equations that must be solved.

Although these approaches are costly from a computational and cost standpoint, they reduce the need for even more costly experiments.

“Building prototypes is even more expensive and time consuming than building models,” says Brake. “If a test fails and you need to redesign something, it’s a year and a half until you can test the next iteration. By developing a predictive and efficient analysis tool, we can circumvent this lengthy process.”