Dear all,

I am inquiring on the commericial CFD code for the application in Internal Combustion Engine (ICE).

I want to know which one is more popular, and more powerful for ICE modelling, scuh as KIVA3, Star-CD, Fluent, CFX, FIRE, WAVE, etc. I think academic and industrial applications should be different. so which one is academic favourite, and which one is industrial favourite.

Anybody would like to give me some suggestion on it. I would like to discussion it with you.

many thanks

Hi, I think to make yr life much more easy, i advised u go for AVL fire coz avl is design purposely for ICE whereby others are not only focus only ICE. Thus, a lots of features had build inside the avl fire and u no need to manual do it by yr own.

Regards, Ray

The answer depends on what you aim to simulate.

1) In-cylinder flow.

2) Combustion.

3) Performance.

4) Emissions.

5) Acoustics (exhaust/intake noise).

If you are looking at overal engine performance and acoustics, choose a 1D code (from your list that would be WAVE). If you want detailed in-cylinder flow prediction, any of the 3D codes will do (and also consider VECTIS).

according to my knowledge, KIVA is also developed purposely for ICE simulation. Star-CD established its reputation originally in ICE field. So, what do you think? FIRE has any other advantages?

many thanks

Hi, steve

Thanks for your reply. I konw that any of the 3D codes can do that (such as the incylinder two-phase flow). I want to know which one is more popular and suitable. In other words, which one is the favourite of the professionals in ICE field.

For the real (not simplified) engine simulation, if you look around all car companies, STAR-CD is the code.

so Why is not KIVA, I guess KIVA is generally used in academic application. Is that right?

In order to perform any sensible research on numerical simulation an academic has to understand and have access to source code. In your list this means only KIVA and, I believe, access to this is now limited. All good academics will use code developed in their own group or a group with which they have a strong working relationship. Unless you wish to establish a working relationship with an academic group I would suggest what they use is not particularly relevant.

Industry now predominantly uses commercial codes. The choice of code is often historical. That is, not necessarily the best choice today for someone starting but the one they have gained experience with over the years and, therefore, the best for them.

A proper study of what is best for your situation will depend on the information you want from the CFD studies and who you interact with.

My answer will be very close to the others.

Based on literature, academic research is done with KIVA (see Wisconsin University in NA or Chalmers in Europe. Chalmers tends to use FOAM now it is open source as KIVA but it is rather new).

Companies use Star-CD (more than FIRE). Few months ago, Renault SA (European Car Man.), was in a hire process for developing a CFD-based development team to have new research tools. They were looking for a Star-CD expert.

It does not give an absolute answer but it may help you in your own opinion, based on facts.

The idea that academics HAVE to have full source code is baloney. It depends what you are studying and what the code you use lets you do. Some commercial codes (I'm familiar with STAR in particular) let you add in a wide range of physics via user subroutines. Why should an academic have to write or need to look at an efficient, scalable, parallelized solver when all they want to do is tweak a combustion model or a turbulence model? They ought to spend their time doing something new and original instead of re-inventing wheels. Not all fluid dynamicists need or want to be expert computer scientists.

My opinion is: With KIVA you have the apsolute flexibility (since you have the source code) but it is not easy to start to use if you do not have somebody to ask about it, because it is user unfrendly, and the documentation is pure. Fire has an excellent user interface, documentation,support,mesh creation flexibility (accepts unstructured mesh) and it is much easier to use. For what concerns the results, my supervisors prefer KIVA. Another big adventage of Kiva is the price (if you don't have any of the softwarws mentioned). So Kiva is better for scientists and academics, Fire for industry. I did not try the other options (Star CD, Fluent..)

So, dear friend,

Where can you get KIVA source code?

Cocerning about the FIRE, I think it is mainly for ICE simulation. While, other CFD code, like Star-CD can be used with much wider applications. i am not sure if it is right? I am just wondering why it is Star-CD but not Fire that is more popular in car company.

You get it in the package when you order it, together with few examples and the poore documentation. In fact it is all you get. Try on http://www.osti.gov/estsc/

Some infomation from my experience of both STAR and Kiva.

(1) price STAR-ICE is really costly as you should pay maintenance fee every year for pre/post, solver and mesher (Prostar, Star, ICE modules). For kiva just one tenth if you are in uni.

(2) difficulty of meshing STAR handles both unstructured and structured mesh. Meaning, by STAR-ICE, you can generate a relatively good quality mesh even though you have a very complex cylinder head geometry.

Kiva allows only the structured mesh. if your engine geometry is complex, say, 4valves, pentroof, or bowled piston, you have to spend 90% of your time to generate the engine mesh. With the default prep.f, it is nearly impossible. Even with a commercial mesh tool, say ICEM CFD, you will enter a painful learn process. If your engine is flat head diesel engine, prep.f will do.

(3) Touching fortran program If you don't need to touch the main algorithm, your effort for implementing fortran program into Kiva or into STAR would be the same. Say, if you need a new wall film model or a new spray model, you can access subroutines for both.

If you specify me what particular topic you are interested in and what sort of area you are in, I may give you more useful information.

Thanks a lot, Yong.

According to your information, I want to inquire two more questions,

1. what is ICE modules? Are they individual modules that will be needed in ICE applications? Are they included in Star-CD code, or ordered separately.

2. I am interested in the fuel injection, spray and atomization, combustion processes in ICE. Which code is better, you think. Do you know about FIRE, I think it is pretty suitable in ICE calculation as well.

My experience from ICE simulations is a purely industiral one, and I have mainly used STAR-CD for ICE applications. However, I have seen presentations of the other main commercial packages (not KIVA) recently and some of my comments are:

STAR-CD/es-ICE: This requires not only STAR-CD licences but also an es-ICE license. This is probably the costliest code, but it is highly popular with industry as it has been well adopted for industrial needs for a long time. STAR-CD is an excellent code to run if you want to do (manual) geometry optimisation, i.e. you want to run a simulation 15 times when you keep changing some small geometric detail. This is hardly an academic task, but it is core business in industry. Star-CD can be used in all fields of industry, allowing for the same code to be run both for powertrain applications and for say external aerodynamics. CD-adapco employs several highly skilled engine experts, and consequently STAR-CD features most of the important physical models needed/wanted for engine simulations. Furthermore, the code is robust and many people know how to use it well. On the other hand, this is probably the least modern code on the market, and the development prospect is not looking that good at the moment.

FIRE: These comments concern only Fire 8. AVL is a company of engine experts, and this is reflected in the presence of a large number of unique physical models. Their special tool for diesel simulations appears to be a highly efficient software, well adapted for automatic optimisation. FIRE 8 gives a modern impression, but rumour has it that it is quite CPU demanding. On drawback of FIRE for a car company is that AVL's general CFD tools do not make a fantastic impression on external aerodynamicists, requiring a car company to run different CFD softwares for engine and other applications.

Fluent: The largest vendor of CFD software in the world is working hard to develop a competetive ICE tool. By now, most of the required models are in place, but do they fit together? I am not convinced that any major developper of four-stroke gasoline engines use Fluent, and somehow I have the impression that if you are the first one, things may not run 100% smoothly... Fluent has a highly efficient solver, and generally speaking good physical models, and thus I am convinced that they will succeed in entering the ICE market eventually. One drawback of Fluent though, is the comparatively poor quality of the TUI which makes scripting more difficult than in STAR-CD.

CFX: Two years ago I not sure that CFX were even considering the ICE market, now they have a tool which is up there among the very best it appears, at least for gasoline applications. When looking at their solutions for ICE applications it is apperant that CFX is being used for four-stroke gasoline engine simulations today. The turbulence modelling as well as the GUI/TUI's appear to be top class. Maybe this is the leading ICE tool of the near future. To attain this status they must improve their methods for simulating diesel engines though, because in that area they are well distanced by STAR-CD and FIRE is at a completely different level.

For your inquireis,

(1) STAR-CD is selling ezy-ICE as a separate module. Major function of ICE is to generate mesh and set up boundaries and event setup(moving piston, moving valves and valve opening and closing)

You need a separate pre/post and solver module. CD-Adapcdo has a number of products for pre/post module depending on your degree of CFD skills and budget.

Don't forget the total cost involvement if you decide to go with the commercial software.

(2) Reknown engine CFD software.

(a) Developed and used widely in research area. Most popular one is Kiva group. Economical and accessibility to the source code. It got good models for diesel engine model Not attractive in terms of meshing and pre/post.

(b) Developed from automotive/engine industry. FIRE from AVL(engine and vehicle development company) VECTIS from Ricardo(engine development company)

Specialized in Engine CFD. Excellent in in-cylinder motion but limited access for additional modelling. Relatively hard to use.

(c) Developed from commercial CFD software company. Say, STAR-CD ICE, Fluent. Very versatile and easy relatively easy to use. Good for meshing/ pre/post processing. Ability to plug in 1D CFD code, say WAVE or GT-Power if you need to have transient boundary conditions at the ports.

For the purpose of spray model, droplet propagation evaporation and in-cylinder combustion, STAR and Kiva is bit ahead. In prinple, dealing turbulent model in the cylinder is not much different. (k-epsilon or DES)

Therefore if you need an engine simulation from spray to combustion in one set, I would recommend those two as they are relatively well organised.

For your information. Most of the automotive and engine developers are using more than two softwares simultaneously as well as their own in house codes.

Let me know if you need more help

please help me if u can