Work With Us

Our Product Engineers’ Best FEA Simulation Tips

Finite element analysis (FEA) simulations are one of many tools in our product development toolbox. As product engineers, we use FEA simulations to develop, test, and refine designs. We stick to a few best practices that allow us to move quickly while keeping simulation results accurate, honest, and affordable.

To get the most out of FEA simulations for engineering design, we balance them with first principles and empirical testing. We find this helpful regardless of the type of analysis we’re conducting — fluid, thermal, electromechanical, or structural.

In this post, we’ll discuss some of the tips and tricks we’ve incorporated into our process.

Use First Principles Early in the Design Process

Hand calculations, or first principles, are a great place to start, no matter what you’re building. These ground your analysis in reality and force you to think through the problem, including the critical parameters that will influence the outcome. They can help identify boundary conditions (fixtures), loads, and opportunities for symmetry.

No matter what you’re building, hand calculations can help identify boundary conditions (fixtures), loads, and opportunities for symmetry.

You may be surprised by how far a simple textbook analysis can go. Simple hand or spreadsheet calculations may be adequate for simple geometries, such as cantilever snaps.

Roark’s Formulas for Stress and Strain is a useful handbook with tables and equations for solving more complicated loading conditions and geometries. It covers a broad range of cases that can be used to add reality to simple hand calculations.

The more complex the geometry or loading conditions, the more FEA simulations come in.

Use hand calculations, or first principles early in the design process

Let Hand Calculations Inform Your Simulations

First principles can often inform your FEA simulations.

When we designed a door for a commercial toilet roll dispenser, we started with hand calculations to define the loads we wanted to apply in the simulation. For the initial analysis, we decided to apply a single load to the hinge, rather than multiple parts with distributed loading. This simplification allowed us to move much faster and iterate before we got deeper into the design.

Once you have a simulation, you can verify your results against hand calculations.

Use Simulation When Empirical Testing Isn’t Feasible

In product development, the best advice is to build early and often, but this isn’t always possible. From a cost point of view, part size and the ability to prototype (fabrication costs, time required to source parts, etc.) might not make it feasible to do empirical testing early on. In those cases, you can put FEA simulations to use sooner.

In product development, the best advice is to build early and often, but this isn’t always possible.

When we designed the toilet roll dispenser, we didn’t have a representative model of the second-generation device we were developing, but we did have a physical sample of the original design.

We chose to conduct empirical testing on the original design, got real world data on how that model performed, and used that data to adjust the parameters of the simulation, so that our new design was grounded in reality. This provided a robust, relatively accurate simulation to be used for subsequent design refinements.

Use Empirical Testing When Simulation Is Overly Complex

On the flip side, the complexity of a simulation might lead you to conduct empirical testing earlier in the process. The material you’re working with may be very complex, for example hyperelastic materials or complex geometries. The loading may be difficult to model, or the cost to prototype may be relatively low. In those instances, empirical testing may be a better option.

When Bresslergroup set out to redesign a very large ceiling fan, we wanted to know how design variations would impact airflow, but because of the size and scale of the design, it would have been very expensive and time-consuming to test physical models.

Instead, we used benchmark data from flow testing of the existing fan and applied CFD (computational fluid dynamics) analysis to three different iterations. By analyzing the benchmark model, we were able to get a relatively good idea of how design changes would affect the airflow. This was enough to help us in the early stages of design.

If your goal is to optimize a design, simulation lets you progress through multiple iterations quickly to get more rapid results.

In product engineering, valid analysis depends on the accuracy of your assumptions

Valid Analysis Depends on the Accuracy of Your Assumptions

Keep in mind that valid analysis depends on the accuracy of your assumptions.

What assumptions have been made about the material you’re using? What are the parameters at which the data you’re using was published? What are the assumptions you’re making about part service, and how do they reflect reality?

We use SOLIDWORKS Simulation, which is effective at lowering the difficulty of creating an accurate FEA (or CFD) Simulation, but it does the same for creating an inaccurate FEA. If results are to be trusted, your assumptions must be solid and grounded in hand calculations.

As they say, “garbage in, garbage out.”

A Few Tips for Plastics Simulations

Linear FEA tends to be conservative, particularly when you’re designing with a plastic that has a nonlinear stress rate curve. If your design intent is to be more aggressive, to optimize, or to reduce costs, you might want to use the actual stress strain curve of the plastic.

Ask yourself how much you actually know about the material. Is it a filled plastic? Fiber orientation can result in different material properties. Do you have the stress strain data or just tabulated data from a spreadsheet?

How much do you know about the environment where the parts will be used? Moisture and humidity can impact material properties.

Determining When A Simulation Is Good Enough

How good a ‘model’ you need depends on your goals and where you are in the process.

In early phases, when you’re working with more concepts and want to move quickly and evaluate feasibility, you can get away with a lower-accuracy model. Here, relative performance might be more important than accuracy. You may not know the exact material you’ll use. The load case may be estimated, and you might be working with simplified CAD models.

You can get away with a lower-accuracy model in early phases, when you’re working with more concepts and want to move quickly and evaluate feasibility.

There are a few ways to simplify your model so you’re able to iterate more quickly, especially in the early analysis phase. You might avoid using small faces and knife edges. You might choose to de-feature parts, or you might simplify assemblies into single parts.

As you advance in the design process, your model should become more complex. If you’re looking for verification, you’ll need a medium level of accuracy, and if your goal is optimization, you’ll need a highly accurate model.

Cost Versus Complexity Tradeoffs

There’s a point of diminishing returns when higher density mesh might needlessly increase your solve time without providing additional useful data. At this point, the accuracy of your data compared to your empirical results is probably going to be good enough, and the cost of refining the model further won’t be worth it.

Another way to evaluate the quality of your model is to consider the Factor of Safety (FoS), also known as safety factor (SF). This is the structural capacity of a system beyond the expected or actual loads. Essentially, the SF is how many times stronger the system is than it usually needs to be.

Another way to evaluate the quality of your model is to consider the Factor of Safety (FoS), also known as safety factor (SF).

Factor of Safety can be based on things like: the probability and severity of a product’s failure, how confident you are in the validity of the model and assumptions, established standards, how much it will cost to increase your FoS, and the feasibility of verification testing.

In product engineering it's crucial to validate your 'model' with empirical testing

Validate Your ‘Model’ with Empirical Testing

It’s crucial to verify your simulation results with real-world (lab) data. If properly set up, your simulation data should be within 10 to 15 percent of the lab results.

One type of empirical testing you might choose is drop testing. This tests a highly dynamic impact scenario on an assembly that is comprised of complex injection-molded and diecast components.

Verify, Verify, Verify

Choosing when to use first principles, simulation, and empirical testing is a bit of a balancing act.

Absolutely accurate results are not always necessary when evaluating relative performance. There are cost versus complexity tradeoffs to weigh. FoS can vary based on cost, confidence, and acceptable risk, and material properties of plastics can be quite complex.

Determining which to use will depend on your goal and resources. Regardless of the path you take, remember to verify, verify, verify.