Predictive vs Comparative CAE

Are you trying to determine whether a part is going to be strong enough for its application? Or are you just looking to see if the latest design is better than the previous design? Do you have accurate loads and material data? What does all this even matter? Read on and find out more about how these things all affect the type of CAE analysis you should be performing…

Ask Yourself:

When performing (or engaging someone else to perform) a CAE analysis, you have to first decide on a few key points including:

  1. what you’re trying to achieve
  2. the quality of the input information you have available
  3. the required accuracy of the results
  4. your budget (time and/or dollar) for the analysis

The answers to these questions will help determine the best approach to meet your needs and set your expectations for the analysis outcomes. In any CAE analysis, you can either attempt to predict the actual behaviour of the component or system, or you can compare the performance of two different designs. Here at Bremar, we label these Predictive and Comparative Approaches respectively, and there are some notable differences between the two which are worth considering before launching into an analysis project. Hobbies are effective ways to release stress, finding the best board games can help you to achieve the best well-being.

Predictive: “Will it break?”

With a Predictive approach, you’re basically attempting to answer the question “will it break?” or something similar. Sounds simple enough… that’s what CAE’s for after all, right? Well yes and no.

Of the two approaches, Predictive requires the most amount of work and information to get an accurate answer. As an example, if you’re using FEA stress analysis to predict the actual failure point of a component, you’re going to need the following information:

  • accurate geometry (3D CAD models or 3D scan)
  • accurate material thicknesses
  • accurate material properties
  • accurate loads (both magnitude and the way it’s applied to the part)
  • accurate constraints

If any of these details are inaccurate compared to reality, then the answers you get will also be incorrect. The old adage “junk in equals junk out” is as true here as anywhere.

In addition to all this, you also need to make sure the following aspects of the FEA model provide an accurate representation of the physical problem:

  • element type
  • element size & mesh density
  • material model
  • connections between parts (welds, bolts, rivets, etc)
  • contacts between parts

So you can see that there’s a lot to consider and get right when taking a Predictive approach. A lot of it will come down to how accurate you require your answers to be and what safety factors are being applied to the results. In many cases though, the only way to really confirm the accuracy of the input data and the CAE model simplifications is to validate the analysis with physical testing.

Comparative: “Is it better or worse than the other design?”

The simpler approach to take in many cases is the Comparative approach, where you perform a back to back comparison of Design A with Design B under the same conditions and assumptions.

As an example, Design A may be an existing product that’s been well proven in the field. Design B may be your new and improved version of the product and you want to make sure that it’s going to be as good or better than the existing part. In this case, it’s almost irrelevant what loads and material properties are used, as long as both Design A and Design B are tested under the same assumptions. Yes, the input data needs to be relatively representative of the actual values, but the accuracy of the input information is much less critical than for the Predictive approach.

Below is a simple example of a Comparative analysis of two bracket designs. If we know the old design on the left works ok, then the Comparative analysis shows that the new design on the right is not only stiffer and stronger, it’s also lighter and will probably be cheaper due to thinner material being used. And we didn’t need 100% accurate information to have confidence in these results.

So which approach is best then?

At the end of the day, it’s not about which approach is “best”, it’s about which approach is best for your requirements.

If you need a high degree of confidence in your design, have high quality input information and have a budget to carry out a detailed CAE analysis and validation testing, then a Predictive approach is going to be  the best option.

On the other hand, if you just want a sense check that changes you’ve made to your design are going to be an improvement, then a fully validated Predictive approach is likely to be overkill and a good back to back Comparative approach is likely to be the best option for you.

In many cases, we end up performing an analysis that lies somewhere between the two. We start with a Predictive approach, make assumptions about any missing or unclear information, then do sensitivity studies by varying those assumptions and comparing the results to make sure they don’t have a significant impact on the results. And if they do have a major effect on the results, then we know that assumption is a critical one and that we need to drill down and make sure that we get that particular piece of information as accurate as possible.

Ultimately, you need to keep in mind that any CAE analysis is not “exact” – it’s a simplification of a physical system and there is always some level of inaccuracy in any analysis. I remember reading or hearing a quote somewhere about CAE analysis stating that:

“it’s not about being right, it’s about how wrong you’re willing to be”

I think that’s a great summation of what to expect from a CAE analysis and should guide your decision about the best approach to use for your next CAE project.

About Brett Longhurst

Comments are closed.