How to Tackle an CAE Problem — From Static Cases to Complex Nonlinear Simulations

Fresh out of graduate school, I started working as an FEA engineer at a well-known consultancy firm in Stockholm. My very first assignment was to take over a simulation that had failed at a previous consultancy. The task: analyze a radio filter housing for the telecom industry—a rectangular box (20×5×5 cm³) containing 50 components mounted on a printed circuit board (PCB).

The client had performed a design of experiments (DoE) and wanted to evaluate material combinations under thermal cycling. In simpler terms, it was a complex multiphysics problem involving assembly sequences, material interfaces, and thermal fatigue behavior, all while accounting for the laminated PCB’s nonlinear characteristics.

This project ended up being the most advanced simulation I’ve conducted in eight years as an FEA engineer. But the technical details are not what I want to focus on. Instead, my biggest takeaway was about how to approach problems of this complexity.

As a young mechanical engineer, I had only performed some simple simulations during university and I had never seen anything this complex before. While I was ambitious and full of energy to tackle the problem, I was in a rush to solve it without seeing the whole picture. I was lucky to have senior colleagues who guided me on how to approach it, and I still use this approach for every engineering problem I get, whether it is FEA-related or not.

It makes sense when you hear it, and you might say that we always do this, but to really stick to it and do it step by step—especially with tight deadlines—is not an easy task. The approach is well known to many: split the simulation into simpler problems and add complexity step by step.

For this specific project, we took a 4 cm vertical cut—like a piece of cake—out of the whole case, since the geometry was repetitive. We started very simply by connecting (tie contact) the structure together and added the complexity step by step. The steps we followed were:

1. Took out 4 cm from the whole geometry, meshed it, and ran a linear-only structural simulation using tie contacts to glue everything together.
2. Added temperature and ran a simple thermal simulation.
3. One-way coupling of structural and thermal simulation.
4. Returned to structural and added material nonlinearity and contact.
5. Added the soldering step during assembly.
6. Added the thermal cycling step.
7. Tried two-way coupling—do we really need it?
8. Added PCB laminated characteristics.
9. Applied different boundary conditions.
10. Ran the design of experiment: 16 simulations on a smaller model.
11. Meshed the whole geometry and ran the 16 experiments on the full model.

To make a long story short, we conducted a detailed simulation for a big telecom company that had a large budget for R&D. In the end, we delivered around 60 pages of a detailed report and presented the results—and our client seemed to be happy with it.

Perhaps the main takeaway of this blog, as you might think, is to split the engineering problem into smaller tasks. However, before jumping into the problem, I believe that there are several steps that every engineer and manager with a good technical team should be aware of—and that is: Do I really need to run a simulation?

During my career, it has often happened that a simple hand calculation or writing some lines of code could avoid running weeks of simulations. Running CAE simulations requires not only significant theoretical and applied expertise but also HPC infrastructure, license costs, and many other costs that small- to mid-size companies usually outsource. However, many times—especially when you’re tight on a budget—you can do hand calculations and iterate through prototyping steps. This depends on the soundness of the technical team and how well they can refine their assumptions, which I explain in another blog about assumption testing.

Another step before dividing a CAE problem into smaller tasks is to understand the problem by heart. This step often requires communication skills with the client or the design team—which mechanical engineers are not always the best at. Most often, with a little digging and curiosity, you can get more information about previous communication and dialogue and understand the whole picture before jumping into the problem. Try to gather as much information as possible before even opening a CAE software.