Computational Design In Aerostructures: Topology Optimization For Conceptual Design And Trade Studies
Interview with Brandon DeMille – General Atomics Aeronautical Systems
Brandon DeMille works in the Advanced Manufacturing Technology group within Airframe Engineering at General Atomics Aeronautical Systems, where he assesses emerging manufacturing methods and the design and analysis tools needed to make them viable for airframes.
At CDFAM in DC he will present a workflow that applies topology optimization during the early stages of airframe design. Rather than aiming for a finished structure in one pass, the approach generates structural mass estimates across a wide range of materials and manufacturing methods, from machined titanium to laser powder bed fusion, so that weaker options can be eliminated quickly, informing trade studies and decision-making before detailed design begins.
First up, from golf clubs to aerospace structures, that must have been quite a transition, so, can you give us a brief overview of your role, what your group at General Atomics Aeronautical Systems, Inc. (GA-ASI) is responsible for, and what you will be presenting at CDFAM?
Moving from golf equipment design to aerospace structures was a little tricky at first. It took me some time to get familiar with all the aerospace rules and regulations. There’s also a bunch of aerospace lingo I had to learn … there are so many acronyms! In a lot of ways, I’m still very much a novice when it comes to aircraft design, but I’m surrounded by smart people at GA-ASI and I’ve learned a lot over the last few years.
On the other hand, when you get down to the engineering fundamentals, it doesn’t matter whether it’s golf clubs or aerostructures. I’ve been fortunate to have had opportunities to explore a range of topics during my career and many of my past experiences have come in handy in my current role.
Composite materials design and manufacturing methods, finite element analysis, additive manufacturing, and structural design optimization are all skills I brought with me to GA-ASI.
I’m part of the Advanced Manufacturing Technology group in the Airframe Engineering department at General Atomics Aeronautical Systems. In my role I’m looking into new and emerging manufacturing methods to assess their viability for airframes.
Typically, these new manufacturing methods require accompanying design and analysis tools to reach their full potential. Sometimes the tools don’t quite mature at the same rate as the manufacturing method. Part of my job is to figure out how to fill those gaps.
One of the ways I contribute to new aircraft development is through trade studies during the early stages of airframe design. I’m going to present a workflow we’ve been using lately. It uses topology optimization to help estimate some airframe metrics and inform decision-making early in the design process.
Topology optimization has been discussed in aerospace for some time, but adoption at the conceptual design stage has been slower than the technology might suggest. What has changed that makes it more practical to apply these methods and into what manufacturing processes?
The most important thing that allows these trade studies to work is the support from my GA-ASI leadership and colleagues. To do these studies well, we need input from a wide range of technical subject matter experts. The GA-ASI team has bought in and supports trade studies with contributions from a variety of groups and departments.
Beyond that, I think this approach relies on the shared understanding that we’re not trying to design the final, detailed airframe in one shot. We’re really starting with a very wide range of materials and processes and trying to eliminate the bad ideas as quickly as possible.
When we approach the problem this way we can use these design optimization methods to generate structural mass estimates and quickly trade off the pros and cons of different approaches.
We’ve used this method to consider a wide range of materials and manufacturing methods. Materials we’ve considered so far include steel, titanium, aluminum, and carbon fiber composites. Manufacturing methods in recent projects have covered CNC machining, prepreg hand layup, compression molding, investment casting, sheet metal stamping, laser powder bed fusion and directed energy deposition.
Your case study involves simultaneous optimization of skin thickness, substructure geometry, and overall shaping. How do you manage the interaction between those three materials and processes when they may have conflicting constraints and optimization potential?
The optimization solver we’re using calculates all those sensitivities for a given combination of materials and manufacturing methods. To capture all the combinations of interest, we run variants of the baseline finite element model (FEM) and then compare the results.
If we want to evaluate 15 or 20 different combinations, we’ll typically need to run 15-20 different FEMs. Fortunately, the jobs can be run in parallel, which can help save some time and effort, but this is an area that is ready for automation. I’d love to discuss this while I’m at CDFAM and see what I can learn about how to make this part of the process more efficient.
How does your workflow keep manufacturability in the loop given there is often a lack of real constraints in engineering software, and what does that mean for how optimization results are actually used?
We’ve devised ways to approximate the limitations of each manufacturing process, but it’s not always easy or straightforward.
As you noted, some manufacturing methods are difficult to capture directly using the typical out-of-the box manufacturing constraints so we have had to be a little bit creative. Most of the manufacturing constraints are best captured using a two-step optimization process where load paths and a refined design volume are identified in the first step. Then the design volume is remeshed and subsequent optimization problem is defined with the refined design space in the second step. That allows us to approximate the design-for-manufacturing rules for a wide range of manufacturing options without things getting too computationally expensive.
The two-step versions of the optimization process take more time and effort, but we’ve done them enough times now that we know roughly what kind of mass reduction we can expect in the second step. In many cases, that’s enough to get the job done for this stage of the design process.
Speaking of software, what tools are involved in this workflow, and how does data move between the conceptual design environment and the more detailed analysis and manufacturing stages?
Right now, we’re using Optistruct for most of the design optimization, but we don’t expect to use it exclusively going forward. We use Siemens products for design and PLM and then there are a few other tools we’ve explored and not fully implemented yet like nTop, SimCenter and others. NASTRAN remains the standard for any verification of designs. Within NX, the realize shape toolbox has been an excellent option when converting topology results to surfaces and solid bodies.
The way we’re currently moving data between the various tools isn’t great. Optistruct and NASTRAN use very similar syntax and formatting so that data transfers fairly well, but the other data transfers can be cumbersome. When moving from topology optimization result to our CAD systems we are still relying mostly on STLs, which isn’t ideal. This is another area I’d love to discuss with the CDFAM attendees during the conference.
What are you hoping to share and take away from the conversation at CDFAM in DC this year?
I’m really just presenting a way of combining several well-established design optimization methods and organizing it all into a workflow framework. I’m hoping it’s useful for people to see these methods used in a way they maybe hadn’t considered before.
The more I learn about this topic the more I realize there is so much more to learn. The possibilities for improving this kind of workflow are endless.
I want to figure out how to add automation, combine this with implicit modeling techniques and multidisciplinary design optimization methods, improve the level of detail in the manufacturing constraints, and enable tighter integration with cost models.
At CDFAM, I’ll be looking to make connections and hopefully initiate some new partnerships to help expand on this work.
CDFAM brings together engineers and researchers applying computational design across materials, structures, and systems. Brandon and others working on these problems will be at CDFAM in DC, July 15-16 to compare methods and find new collaborators. If you are working on similar questions, it is a place to connect with the people solving them alongside you.
