Big Ideas Lab - DYNA3D

Episode Date: November 12, 2024

What does designing hurricane-proof buildings have to do with heart surgery and light beer? Surprisingly, the answer lies in a groundbreaking computer code developed more than 50 years ago.A code so powerful that automakers use it for car crash simulations, beer manufacturers rely on it to design cans, and surgeons turn to it to understand how fluid flows through the heart.This is the untold story of DYNA3D—a revolutionary code that transformed industries by simulating real-world physics and reshaped innovation as we know it.-- Big Ideas Lab is a Mission.org original series. Executive Produced and Written by Lacey Peace. Sound Design, Music Edit and Mix by Daniel Brunelle. Story Editing by Daniel Brunelle. Audio Engineering and Editing by Matthew Powell. Narrated by Matthew Powell. Video Production by Levi Hanusch. Guests featured in this episode (in order of appearance): John Hallquist, Inventor of DYNA3DRoger Werne, LLNL’s Senior Advisor for Innovation and PartnershipsKim Budil, Director of LLNLBrought to you in partnership with Lawrence Livermore National Laboratory.

Hurricane Ian officially made landfall as a category 4 hurricane with maximum sustained winds at 150 miles per hour. Question. What does designing buildings that stand up to hurricanes have to do with heart surgery? And light beer? Thousands die every year in car accidents because they don't buckle up. Vince, we're dummies. We don't wear safety belts. And a couple of crash test dummies from the 90s. Well, the answer is a small section of code written more than 50 years ago.

Code that automakers use in car crash simulations. Beer manufacturers have used to design cans, surgeons used to understand how fluid flows through the heart, and meteorologists used to understand how structures will hold up to weather events. Code that was shared collaboratively with experts and hobbyists,

and set the model for open source movements like Wikipedia, GitHub, and even AI development. This is the story of Dyna3D. You are listening to the Big Ideas Lab, your weekly exploration inside Lawrence Livermore National Laboratory. Hear untold stories, meet boundary-pushing pioneers, and get unparalleled access inside the gates.

From national security challenges to computing revolutions, discover the innovations that are shaping tomorrow, today. Picture an egg in your hand. What do you think would happen if you threw that egg into a wall? You could safely assume that the eggshell will shatter into hundreds of pieces and the yolk will splatter all over the wall. Sure, but what if you had to predict where every crack would form in the shell, or where

every piece of it would form in the shell, or where every piece of it would land on the ground. You would have to make calculations on the fragility of the shell, the force of impact with the wall, and model the fluid movement of the yoke. That would be pretty tough. But this was the same problem that engineers were facing at Lawrence Livermore National Laboratory. Except on a scale that was hundreds of millions of times larger and more complex.

They were designing a bomb that would be dropped at low altitude at high speed. Would the warhead survive the crash landing intact without damage? Where would it and its debris land? They had no way to affordably test this and needed to test it through a simulation code that would give accurate predictions for a lot of possible crash landings. My start date at the Lord Silverware lab was more than 50 years ago. I think it was March 17, 1973. The problem? It was the 1970s. The 70s were rich in disco, bold colors, and rebellious fashion. But what wasn't nearly as abundant? Advanced simulation systems. The technology at the time was far too limited to provide accurate solutions to the questions

Lawrence Livermore needed to be answered. After all, they had to build a computer model of a highly complex bomb when we were still two years away from Pong being played on your home television. Meet John Hawquist, the inventor of Dyna3D. When I first joined the lab, it was the Nuclear Test Engineering Division, where they were looking at vulnerability of protective structures such as missile silos and so on.

And in that role, I was an analyst running some of the eight finite element codes. And then after I found out that none of those codes, they never gave correct answers. They were developing a new one megaton bomb that was going to be dropped from about 120 feet above the target at a very high speed. They had absolutely no way to analyze the impact. When modeling the behaviors of objects in physical space,

computer scientists and mathematicians break up the space into sections using the finite element method. This might sound complicated, but do you remember in art class when you'd lay a grid over a drawing and then recreate the drawing square by square? It's kind of like that. All those grids, in this case millions of them, recombine to give you a full picture.

Each of those grids is called an element and just like a drawing, when John joined them all up, the models were stuck in two dimensions. They were using two dimensional codes, which really can't model a 3D impact. So they wanted to develop a new code to look at the impact of this weapon on a hard target. That's how Dyna3D got started. At that point, computers could only process finite elements in two dimensions.

As the name suggests, Dyna3D operates in three dimensions, allowing supercomputers like those at Lawrence Livermore to simulate the effects on structures in a way that mirrors the real world. John was by far the most productive and creative software engineer that I've ever seen. That's the lab's Senior Advisor for Innovation and Partnerships, Roger Werne, who has spent more than 30 years of his career at Lawrence Livermore. He has seen firsthand the evolution of Dyna. The unique feature of Dyna 3D was that it was able in three dimensions to model the folding collapsing of metal

structures on themselves. Until he did this, a finite element code could not model a structure that collapsed on itself. They would just pass through one another in a non-real way. Well, John's algorithm was able to do this to metal structures. John was working late nights and weekends dedicated to his mission to complete the code. And in 1976, the hard work paid off. The very first version of Dyna3D was released, a breakthrough in the modeling capabilities for the time.

By 1976, the first version was completed. But John wasn't satisfied. It was implemented on a CDC 7600, which had 65,000 words of fast memory and 256,000 words of slow memory. We could only run 2,000 elements, maybe 5 to 10,000 elements. We could perform basic simulations but nothing all that complicated. But in 1978 we received a Cray-1 computer which was revolutionary. So at that time I completely recoded the code so it was almost 100% vectorized

and we could start looking at real impact problems. But then, the original project that John had been working on was canceled. But the bomb that we were developing was canceled at that time. We received additional funding to pursue the development of Dyna 3D for other types of structures used in defense. And they funded it then for the next few years. And the software started to be used quite a bit by the engineers within the lab

for virtually all applications. While the weapons program that prompted the need for Dyna3D was canceled, John Holquist realized the potential for his code, which he believed could handle complex 3D simulations for industries beyond just the military. In 1978, I applied to have the code released into the public domain. Suddenly, technology born in a nuclear lab would be accessible to everyone.

Software engineers around the world were quick to get their hands on Dyna3D. These early adopters played a vital role in helping not only evolve this massive and still growing software package, but explore never before tested use cases for the code. It was a foundational moment in the open source code movement. In those days, software was available for the taking. We would share the software widely with universities. And the requirement we had was if you make an improvement to whatever you find a bug in the code you send it back to us and we'll fix it or do an improvement thereof. So we had a group of users in the outside world you know 50, 60, 70 of

them who gave us feedback on this thing on a regular basis. By allowing Dyna3D to be open source the code was free to use, modify, and apply across any use case by any company or organization. That kind of sharing and transparency are core principles of the digital revolution, and Lawrence Livermore National Laboratory established them from the very beginning. Of course, we had users inside who were using the code as well. Let's say I was running a problem and I would find a bug in the code, something wouldn't work. So I take the code to John and show him the problem,

and the next day it would be fixed. That happened throughout the lab community and throughout the outside community. When he found a bug in the code, he would work all night fixing that code so that the next day you could go back and do your calculations. And I think John took that ability to rapidly fix problems and make the user more productive by not having to wait weeks to fix the bug. I mean, it was done in days.

That was a huge benefit to the outside world because, you know, to the outside world, commercial world, time is money. Thanks to the flexible and scalable nature of the code, Dyna3D quickly evolved with the help of a dedicated and loyal following. The code base grew to 10 times its original size and began popping up across a diverse set of industries. Dyna3D became the workhorse for crash simulation in the automobile industry. It saved the automobile industry billions of dollars per year. They no longer had to do many, many real full-time crashes. They could model the crash on a computer, do the changes they need in order to strengthen the structure, the auto structure, and then do a final test crash

to validate the models that they had developed. In the early 1980s, Hallquist recognized the code's importance to these industries. So it's a huge success from the standpoint of our impact on the commercial world. It may very well be one of the biggest successes we've had. So John decided to take his work to the private sector, forming the Livermore Software Technology Corporation. I thought that if things really didn't work out, I would move to Michigan and work for one of the tier one suppliers

to the automotive companies. At that time, I knew a lot of people. So I figured the worst thing that could happen is we wouldn't survive and I would move out of California. But it seemed to me that there was a huge market that wasn't being addressed. But John would still face stiff competition in the private sector. So as part of an effort to update the original code, John enhanced the software and reintroduced it as LS Dyna.

ESI and HKS were the two big competitors. And there's also a company called Mecalog. And Mecalog and ESI both worked with Dyna3D from 1981. I wasn't so sure that we couldlog and ESI both worked with Finite 3D from 1981. I wasn't so sure that we could feed out ESI because they had almost 100% of the automotive crash market at the time. But eventually we took over about 90% of the crash market. If you ever watched a car safety test,

you know, with the crash test dummies, you'll usually see a computer screen in the background with a simulation of the crash on screen. More likely than not, that simulation is LS Dyna. John's journey from government researcher to entrepreneur expanded the code's application from impact simulations to aerodynamics. Kim Badil, the director of the Lawrence Livermore National Laboratory, tells us more. We've used those tools to do things like model the aerodynamics of trucks. So we're trying to make big semi-tractor trailers more fuel efficient on the road by designing the cowling and the

shielding that goes on the truck so that the air flows more smoothly. So we're saving emissions by making our trucking fleet more efficient. After widespread success in the automotive industry, the technology really took flight. Our aerospace is the big one for us. At LSTC we had usually two meetings a year with the jet engine manufacturers Boeing and the FAA because the engine manufacturers were certifying modifications to the engines without doing testing and they wanted to use the simulations to justify the changes because the testing cost millions and millions of dollars to change and the FAA was interested because they wanted to understand how analysis could be used to make design changes without doing the

physical test afterwards. So I think we had almost all the world's manufacturers, the jet engines, using Alistina for that type of work. And usually it's a bird impacting the outer fan blades and then being adjusted into the turbine blades. And they'd model the entire engine and do the simulation. I got it, I got it! In the 1990s, even the Coors Brewery turned to the Dyna code for the company's high-speed production lines, using the code to determine exactly how much material they needed to use for their cans.

much material they needed to use for their cannons. Explosives, automobiles, airplanes, even light beer. The use cases for Dyna reach far and wide, and new applications for the code are being discovered every day. One of the features that I think LS Dyna has done is they've modeled a combination of structural performance and heat transfer characteristics within the model. You can do both simultaneously. You can actually do chemical kinetics as well. And I know that battery technology, in which you look at the failure of a battery due to the heating of the battery components and the eventual destruction of the

battery, I know that there are companies who are using LS-Dyna in battery modeling, and it turns out to be a very valuable thing. The coupling of electromagnetics, heat transfer, and structural response via these codes is turning out to be a very valuable new feature. Perhaps the most important potential application of Dyna to date has been the promising work done

in the healthcare industry. Today, bioengineers employ the code to model impacts during various surgical procedures and bodily injuries, and even to design medical equipment. DINA 3D and various other codes have been used to model the impact on human bodies. For example, you make a skeletal structure, you make the tissue structure of the human body, and then if it's in an automobile and hits a steering wheel or hits an airbag, you can study the impact of that structure.

I was really aware of this work where they were looking at their pacemakers, which then requires a complete model of the heart and then the fluid flowing through the heart, as well as a very complex constitutive model to model the behavior of the heart. You know, you have to have muscles modeled that contract and pump just like the heart does. And then the electromagnetic capability is needed

to model the pacemaker and the electrical current hits the muscle and the muscle has to contract. Because Dyna3D is open source and available to all, it would be a daunting task to calculate how many companies or individuals are using it. In recent years, it was discovered that Dyna-based code was prominently used by 18 aerospace companies, 9 atomic energy firms, 13 automakers, 37 research labs, and 25 engineering corporations. Every one of them using this code to simulate crucial tests in their research.

DINA, once designed for a singular purpose, has evolved into an unsung hero, safeguarding us in the skies, on the roads, and in our bodies. Exceeding the confines of early computing, it was able to simulate the intricacies of the real world, a feat that laid the foundation for advanced 3D simulations we rely on today.

From its origins in public funding to its current success in the private sector, Dyna has transcended its initial purpose, becoming an essential tool for industry, academia, and research alike. It is intertwined with the very fabric of modern engineering. As one engineer interviewed put it, Dyna is to finite element codes what Hershey is to chocolate bars and Kleenex is to tissues. Whether it's modeling the splatter of an egg, understanding the impact of a nuclear device, or digitally replicating the human body, Dyna is a staple in the engineers toolkit. As we look to the future,

Dyna is poised to enable breakthroughs we can scarcely imagine today. From designing safer vehicles to unraveling the mysteries of protein folding, its relentless optimization and incredible versatility ensure Dyna will continue to expand the boundaries of the possible for decades to come. Thank you for tuning in to Big Ideas Lab.

If you loved what you heard, please let us know by leaving a rating and review. And if you haven't already, don't forget to hit the follow or subscribe button in your podcast app to keep up with our latest episode. Thanks for listening.