It feels like we’ve been talking about simulation driven design for a really really long time. Doesn’t it? The whole idea is to setup and run simulations of a product’s performance early in the design cycle and base design decisions off those results. This approach is in contrast to using simulation towards the end of the design cycle, prior to prototype and testing or even just prior to design release, to validate and verify product performance. It might seem like a nuanced difference to some, but theoretically it can make a huge difference.
While this approach seems to hold a lot of promise, it’s been hard to find many engineering organizations that have put it to use successfully. Why? We can come up with a variety of potential reasons such as cultural or process change push back or potentially a lack of software usability or automation. But I think there’s another more fundamental problem. In my mind, I think it’s critical to have four types of skills to perform a simulation driven design approach.
- Background in Engineering Science: To really use simulated performance to drive design decisions, you first have to know what the the results of any calculation, by hand or by computer, first mean. This includes the understanding of various engineering sciences such as statics, dynamics, thermodynamics and many others that span other engineering disciplines. Without this background, there’s no context in which to understand what the calculated values actually mean. For example, making a design decision is risky when you don’t know the difference between Von Mises and shear stresses.
- Understanding of the Computational Methods: Next, to make sound decisions based on simulations, you need to understand the underlying method used to calculate resulting values. For example, the Finite Element Method (FEM) and Finite Difference Methods (FDM) can provide very different results based on how calculation models are setup. Furthermore, each method uses some basic assumptions and simplifications. If you don’t know the fundamentals of each of them, then you can interpret the results incorrectly.
- Familiarity with CAD Software: The entire purpose of a simulation driven design effort is to actually use the results to help make design decisions as opposed to just verifying and validating performance before design release. That means results should drive changes in the design and ideally the design model. However, the design model is often built up using specific methods used to capture design intent. In some parts, the thickness of specific walls must be maintained. In other parts, a specific dimension must be based on another dimension. That’s why the design needs to be changed through the definitions created in the design model. But more fundamentally, CAD models are often simplified to enable faster simulations. For examples, some rounds that have no impact on the results might be removed.
- Knowledge of Simulation Software: While simulation calculations of very simple designs could be performed by hand, anything close to a real part or product leads in to mathematics far too complicated for manual calculations. Simulation software automates much of the work to setup, run and then review calculation results. This would include activities such as defining loads, constraints, complex material properties and the like. This requires software skills above and beyond CAD software knowledge.
So there’s four sets of skills that you need to really perform simulation driven design. Where’s the problem? The issues lies in the fact that very few, perhaps no one, in the engineering organization has this combined set of skills. Let’s take a look at some common roles and see.
- Drafters – For the sake of today’s conversation, let’s say drafters spend more than 75% of their time creating, modifying and standardizing drawings. Given that, drafters may have skills in area #3. That is they might know how to use CAD software to manipulate the 3D model in the boundaries of original design intent. But they often do not have the educational background to address areas #1, engineering science, and #2, computational methods. Furthermore, they often don’t have the experience for area #4 in how to use simulation software.
- Designers – Again, to set a baseline, let’s assume designers spend more than 75% of their time creating and modifying 3D models. But let’s also assume they do not have a bachelors degree in engineering. Given those assumptions, individuals in this role would readily have skills in area #3, CAD software, and area #4, Simulation software that would be embedded in CAD software. However they would be lacking in areas #1, engineering science, and #2, computational theory. They would be able to generate a color fringe plot of stresses on a component. However they wouldn’t have the confidence they made the right fundamental assumptions around setting up the simulation model nor in interpreting the results.
- Engineers – As we have for the past two roles, let’s define a baseline here. Engineers have a bachelors degree in engineering. Furthermore, let’s assume they have lifecycle responsibilities for products. That means they’re shepherding a product through the entire development cycle, not just defining it in the design phase. They would have the knowledge in areas #1, engineering science, and #2, computational methods. However because they are often addressing ‘firedrills’ everyday, they are very infrequent users of CAD. So their skills in area #3, CAD software, may or may not be enough for a simulation driven design approach. Also, I’d expect they are don’t have time to setup and run simulations themselves. I would imagine they wouldn’t have the skill set in area #4, Simulation software, either.
- Analysts – As for a baseline for this role, let’s assume that an analyst dedicates more than 75% of their time to performing simulations. Furthermore, let’s assume they have at least a bachelors degree in engineering, if not a more advanced degree. This role has the educational background in area #1, engineering science, and area #2, computational methods. They also have deep knowledge in area #4, simulation software. Area #3, CAD software, would most likely be their one weak area. However the integrations between CAD and Simulation software could at least in part address that shortfall. By and large, this role has almost all the knowledge and skills to drive simulation driven design. The only remaining issue would be bandwidth. There are often far fewer analysts than engineers. And they often are dedicated to the big, nasty, hairy simulation problems around verification and validation for design release or change orders.
Now there may truly be mitigating circumstances. There are training programs to address the fundamentals of engineering science and computational methods for non-engineering degreed roles. There are easier to use CAD software for engineers like Spaceclaim and the upcoming Creo from PTC. There are integrated CAD and Simulation environments for engineers and analysts such as NX from Siemens PLM and CATIA/SIMULIA from Dassault Systemes. But have do these efforts really mitigate the concerns around these four knowledge and skill sets?
What are your thoughts? Is a background in engineering science and computational methods really necessary for simulation driven design? Do you think the CAD and Simulation software offered today enable engineers and analysts to truly perform simulation driven design? Sound off and let us know what you’re seeing in your job.
Take care. Talk soon. And thanks for reading.