Hi All !

We plan to introduce CFD-simulation in our company this year. We are just at the beginning of the process.

It would be nice to have some impressions from those who made this kind of implementation.

Which are the main points/criteria for this ? Most common errors to avoid ?

Thanx JFS

Hello JFS,
I think, that the following points are important:
- What kind of CFD tool do you want to use? The answer will depend on the kind problems you like to cover.
- How can you implement the new "department" into your company's production processes?
- How big should a CFD department be? (From my point of view: At least one person the whole day. CFD is not something what you can do besides your main work.)
Don't hesitate answering by mail, if you need more information.
Greetings, Marc

Hello,

I totally agree that CFD is not a part time job. Another issue is that you should avoid ending up with a single CFD analyst sitting alone for years trying to prove his worth. It is for sure not very tempting to start off directly with an entire department if you haven't been into CFD at all before, and a short low-key break-in period where you try to see in what applications CFD can make a difference in your company may well be justified. On the other hand, I also know for a fact that CFD is not going to pay off unless you are willing to boost the volume of the activities up past a certain critical level. There is simply no way that a single person simply can efficiently develop the CFD applications of your company, support your business (firefighting and troubleshooting), and keep up with the development of CFD itself, all on his own. So my advice to you is to spend no more than a year on the break-in. After the first year, you should employ at least one more CFD analyst.

Regards

Lars Ola

Hello Marc, hello Lars Ole,

At the beginning, it will be a one-man division. It is not very clear now what will happen in the future. It will depends of the success of this new department ans the quantity of job to be made. But we are not a big structure (about 1000 people in the company), so i think we don't have the capacity and the need to have a huge simulation department.

First of all, i plan in the next weeks to consult all the Business Units to analyse our needs.

One of the big question is : who is going to do the job ? It is planed i will do the calculation at the beginning. I am responsible for the implementation of the software, but should i do the simulation job ? I'm not familiar with our products and maybe a product-developer is the best man ? What do you think ?

Thanks !

JFS

Try to get some help from a external cfd company at the beginning. This helps you to identify your needs and you can also collect a lot of know how for your own work.

The problem i would like to have an independant point of view. If i ask CFX- of Fluent-poeople, they won't have an independant and impartial approach...

There are also Ingenieurbueros (www.beilke-cfd.de) around, which are not the code vendors itself.

Choosing the right software is probably the 3rd step on your way. At first you have to find out what problems in your company can/should be solved using cfd. Without experiences in this field you don't know what is possible, what is easy and what is really hard to calculate.

Then you have to create good connections to the CAD people if you want to get models, which are suitable for a cfd calculation. Depending on your cad system this might take you more time than you think.

Hello again,

I never meant to say that you need a huge CFD department. Quite as you say, that depends on the size of your unit. What I wanted to say is that with a single analyst (and a thousand man company), it will be very difficult to reach a break-even situation where the benefits of doing CFD outwheighs the costs.

For one thing, he has to build a position entirely from scratch. That not only means to identify and define potential CFD applications, try to simulate them and develop suitable methodologies to deal with them. It also means that he has to create a space for himself in the conciousness of his colleagues - make himself and his activities known to the rest of the organisation. And the relation he builds to the others such that they are aware of what he can do for them (and actually comes to him with work) whereas he can rely on them to provide him with geometries, boundary conditions etc.

All this takes a lot of time and energy. It takes even more energy when he has to do other things besides CFD, like participate in pointless three-hour meetings about things that has nothing to do with fluid flow, prepare colourful slides for commercial ads bragging about your company's advanced simulation procedures and capabilities, prepare budget proposals, etc. And it takes yet more time and energy if he doesn't have the clear, active, and outspoken support of his superiors.

And time is something that just isn't available today. All too often, he'll find that the design has changed before he is even done with the meshing, and before he has come up with an answer to an urgent problem, the running warranty costs have forced the company to apply ad hoc solutions that might have saved that particular situation but that aren't going to save you from similar ones later on.

Which brings me back to where I started: A single person working with CFD in an environment where nobody understands diddly-squat of what he is doing is simply a lot less efficient than two people doing the same together. Remember, it's a 100% increase in manpower, and the synergy effect of that is most significant.

As to the qualifications of the analyst, it certainly doesn't harm if he knows the products. However, it is much more important that he is really good in fluid dynamics because he will be seing things that nobody in your company has sen before and must be able to tell them what it means in practice.

Regards,

Lars Ola

Hello,

CFD or any other FEM system implementation should consider the strategical needs of the product development process inside your company.

Some points are:

- you should use a 3D-CAD system like SolidWorks or Pro/ENGINEER

- from these models you can derive the fluid flow volume

- if you want also calculate heat transfer problems you have to use also the solid parts (example: electronic cooling problem or water cooling of an engine block)

- these kind of problems mainly have quite unregular geometries, so you need a 'free' mesher

- if you use a 'free' mesher you need also the capability of an automated mesh enhancement near the walls for a precise calculation of the boundary layers with respect to the shear forces and the heat transfer

- an additional very helpful tool is the monitoring of the solving process (mostly iterative solvers) by showing an 'convergence' plot on velocities, temperatures, vorticity, etc.

- an additional necessity is the possibility of an automated 'conditioning' of the solver settings (e.g. 'automatic turbulence start-up algorithm, 'adjustment' ot the boundary layer sizes, automatic change of relaxation factors,...)

- if these tools are implemented in the CFD system a trained design engineer is capable of using such a system in standard situations !

These points are - mostly - the reasons for me to use - and also sell - the system 'CFdesign' from Blue Ridge Numerics ( http://www.cfdesign.com )

If you want to contact me, please see our home page: http://www.pretech.de

With best regards

Stefan Reul