Maximizing Simulation's Value:
Aerospace/Defense Firms Target Key Constraints
Ora Research Letter on Digital Prototyping, Simulation & Analysis August 4, 2009
By Bruce Jenkins, CEO
Last time – state of simulation practice in aerospace/defense, and key next challenges targeted by this practice-leading industry. Today, top obstacles A/D firms are working to overcome, to maximize simulation's value.
What constrains aerospace and defense companies from achieving their goals for maximizing simulation's business value?
CAD-CAE gaps A perennial constraint has been the technological gaps that exist between product definition geometry on one hand, and simulation models on the other – and the resulting penalties in time and, sometimes, accuracy exacted by the need to prepare geometric and functional models for input to analysis:
“Typically we receive the product definition geometry, then we massage that to create the geometry on which we build our analytical models. That is, we start with a model, then we use all kinds of high-powered math to optimize that model for analysis – which can be quite costly. Then once you have it optimized, you have to ask how the new geometry relates back to the product you have to build. Because you have optimized the model, but not the product. This is significant, because transferring the optimization of the geometry back into the product is not that simple. What would be great would be to optimize the product in its operational environment, rather than taking an abstract model and optimizing that. That would be a very great breakthrough that would alleviate a lot of the cost currently incurred in setting up our models.” – Aerospace/defense company F
“Among the constraints on [‘bringing fidelity forward’] – doing more detailed analysis earlier, the biggest one is figuring out how to be able to construct the models quickly enough to be able to do the analysis quickly. So it isn’t the threat of a 6-hour solver run that’s holding you up; it’s how you are going to produce the geometric and finite element models fast enough that the 6-hour run does become the gating factor. If it takes you 6 weeks to prepare the input, the 6-hour run time is irrelevant. But if you can get the models ready in 15 minutes, the run time becomes significant. We would like to get to the point where the run time of the solver code is what’s holding us back.” – Aerospace/defense company A
Cross-discipline analysis gaps Similar barriers impede sharing of results between analysis tools in different disciplines:
“Because computers have gotten faster, there are analyses we can do today that we couldn’t five years ago. But doing a full-blown aeroelasticity problem, for example, still involves compromises. We still have not reached the point were we can take a CFD code and a structures code, and run them together efficiently. What will fix that? Increasing automation, especially in automatic mesh generation. The show-stopper has always been that when you try to couple together the very best codes you have, the primary model you’re using has to be the computational grid for one or the other of those codes – and that prevents the other code from being an equal partner in the process. If the geometry representation can be made code-neutral, so that both codes are sharing equally in the modeling, that will improve the situation.” – Aerospace/defense company A
Advanced-materials characterization needs At the same time, today’s growing use of composites and other advanced materials is making the need to characterize materials properties a constraint on the value simulation can deliver:
“The biggest thing holding us back is the time it takes to integrate new and advanced materials – to qualify and analyze new materials, basically structures technology. If there’s a weak link in the chain, that’s it. It has as much to do with the difficulty of performing advanced analysis of composite materials as anything else.” – Aerospace/defense company A
Need for better simulation data/process management Another chief constraint is the need, seen by many as yet unmet, for robust, capable simulation-specific data management and process management tools:
“We’ve done a very good job of automating geometry data management, as well as change and configuration management. My focus now is simulation data management. The challenge is to ensure that design studies, and the models and simulations used in them, are kept coordinated. We’ve developed processes and methods to do that, but it’s less efficient than we would like because today it’s all done essentially manually. The consequences are lost productivity, missed opportunities for innovation, reduced quality, greater difficulty obtaining certification, and inefficient knowledge management.” – Mark E. Miller, Senior Technologist, GE Aviation
“PDM for CAE really isn’t solved yet. There’s lots of work to do. The commercial codes are simultaneously helpful and show-stoppers in making real progress – both solvers and data management tools. Commercial data management codes have tremendous capability, but also impose on you a situation where the engineering data and product data are held captive by a proprietary system. And when you’re putting together a system that integrates many different codes, having your data held hostage by any vendor’s PDM system is a problem. SOA approaches and ProSTEP help a lot, but I think the real solution has probably not been conceived of yet. The solution is going to come in finding the right mix of dealing with commercial software. There has tended to be this all-or-nothing approach – we either roll our own, or buy a whole system from some vendor. I think the final answer will be a mix of internal software and external solvers, capabilities and packages.” – Aerospace/defense company A
Need for more highly automated model preparation/problem setup However, some feel that until the lack of fast, efficient model creation and problem setup capabilities is solved, this constraint will limit the benefits that could be had from better data/process management:
“It’s not my perception that we’re being held back by a lack of formalized and capable PDM tools; it isn’t our lack of ability to keep track of the data that is holding us back today. It gets back to the 6-hour solver run time. If it were the case that the 6-hour solver run were the thing holding us back, then keeping track of the results of that run would be a big deal. But if it takes 6 weeks to set up the problem, that makes the analysis results disposable, because you’re only 6 hours away from reproducing the results, in a 6-week process. But if we were dealing with a 6-hour cycle time, and it took a full 6 hours to retrieve your results, then storing that data better would be important. But that is just not what is holding up the process at this point.” – Aerospace/defense company A
This research is excerpted from our white paper Strengthening Simulation's Business Impact: New Strategies in Aerospace & Defense. In this project we interviewed discipline leads and methods experts at major aerospace and defense OEMs and subcontractors around the world. Our investigation focused on business drivers for sustaining or increasing simulation investments, current state of industry practice, constraints on maximizing simulation’s value, and new strategies for overcoming these constraints. Request your copy of the white paper | 
