Compare CFD tools the right way (Part 2)

In my previous post, we discussed why you shouldn’t simply read what you see online without a grain of salt. Instead you should see and evaluate CFD tools by yourself. In this part, I’ll cover some other things you should avoid.
So let’s have a look at some more “how not to do it” cases:

Only compare CFD tool against CFD tool

I have seen someone comparing CFD tool A against CFD tool B countless times. Often the tool that gets trusted more is often their known tool which has given correct results countless times. This is like saying your car will never break down. There are several reasons why comparing CFD Tool A with CFD Tool B may not be the right approach. Here are just a few:

• Different users will get different results on the same tool. This is due to the set-up (including the mesh and other settings) being done differently from user to user.
  
• Different tools might use different approaches to solve the task. In Simcenter™ FLOEFD™ for example there is a completely different meshing technology (SmartCells™) which cannot be compared simply by the number of cells used. This would be like comparing the velocity of a speed boat with that of a cruise ship except you’d still force the speed boat to carry as many passengers.
  
• Everyone can make a mistake unintentionally. If that is done in one tool, then the comparison with the other tool is useless.
  
• How do you know whether CFD tool A is + or – 5% from the reality?
  What if there’s an 8% deviation with Tool B? It might either be just +3% from experiments; therefore closer to reality than the -5% of tool A. Or 8% in the same direction of tool A and therefore further away from reality (-13% from experiments).

The only good comparison is the one between simulation results and physical experiments. With that said, bear in mind that even in experiments there might be errors. Compare it with your existing products where you can actually have the confidence that the experiments are correct. If you want to reduce the probability of any error, then test the CFD tool on 2-3 different applications or products of yours.

Never ever contact the vendor for support

The worst thing you can do is not seeking help from the vendor when something is not working during the test. Or when you are unsure about how to use it correctly. Even worse, is not contacting supporting after buying the product and you have issues.

While evaluating products, it is your right to get support. I also recommend that you invest in a support contract after purchasing the software. Here are two key reasons why this can be important:

• You might be a beginner with the product, especially during an evaluation phase. Make sure you are using it correctly to get accurate results with the help of the vendor’s support team. Contact the vendor if something is not right or when you are not sure how to use it.
  
• By reaching out to them before purchase, you can experience their service firsthand. See how open, friendly and most importantly knowledgeable as well as fast-reacting they are.

But do bear in mine that everyone has bad days. They may be having an off day or may be new to the support function. When I first started, I was doing my best to help. If I didn’t know something, I checked with my colleagues and got back to the customer, perhaps a little bit wiser every time. Everyone has to start somewhere.

If you are not happy with your direct contact or the sales partner, try to reach someone else. We are all human and we know that sometimes some people don’t get along. That’s why we switch car mechanics when we think the service is bad. It doesn’t mean that the product is bad.

From my personal experiences

Here are a few examples I have experienced personally to show how things simply can go wrong:

• A customer had serious issues with the stability of the software. It was constantly crashing and he was unable to work with it. He never contacted support to get help.
  As you can imagine, there are a gazillion things that can go wrong. Anything from operating systems, hardware settings, other software, firewalls, permission rights, some strange IT settings etc. The support and the development teams can usually figure out what is causing the issue. After all, if the software runs on a thousand other computers, then usually there is no reason why it shouldn’t work on your computer. They will be able to find the issue. If your car breaks down, you don’t let it sit there and forget about it. You call the car dealer to get it fixed. The same goes for CFD software.
  
• In another case, Simcenter FLOEFD was compared against another CFD tool during the evaluation stage and the results were off. The user said that our tool does not give accurate results and they trusted another CFD tool (which they had been using for a long time). We knew that our software would provide accurate results. We have customers doing the same thing every day with good results. It turns out that the user had made a mistake in the settings which led to wrong results with the other CFD software. After they corrected it, the results matched.
  Things go wrong and everyone can make mistakes. So make sure to never blindly ignore potential errors. Ideally compare results with something you really know is accurate from multiple verifications over time.
  
• I also had cases where users of other CFD tools were used to very fine meshes and resolving the boundary layer in detail etc. They had a rough feeling of how many million cells they used to have in their CFD tool. So when they tried to set up a similar model in Simcenter FLOEFD, they tried to create an equal number of cells. This of course led to massive CPU times that were not efficient. Simcenter FLOEFD uses intelligent technology which gives accurate results with up to an order of magnitude fewer cells. I know, for an experienced CFD user this sounds like witchcraft. But if I told you that my car doesn’t require gasoline because is it an electric vehicle, it is not witchcraft either. There are technologies that you might not be familiar with, but you should be open to new smart and innovative ideas of tackling well-known problems – just like Uber, SpaceX and Netflix .

“Any sufficiently advanced technology is indistinguishable from magic.”

Arthur C. Clarke

Conclusion

I know this is a long post. But I hope I was able to shed some light on some common things that can go wrong. So make sure you get your information from valid sources and take everything with a grain of salt especially if it is from a forum or a source with an unknown level of experience and even questionable intent. Ideally, ask for references from similar customers in your industry or application or even better your colleagues.

Also, remember not to compare apples with oranges. Compare results and whether the tool is suitable for your processes, users, application and brings you the benefits you are looking for.

Oh, and remember to always contact the vendor for support if you have issues. Especially during the evaluation phase when you are still learning and need to make a decision for what and where your company needs to spend its money at the end.

Now, of course, I would be happy if you decide to try Simcenter FLOEFD or any other Simcenter products, but in the end, the decision is yours.