Is there any CFD software available in ProE?

(1). I have several books on Pro/Engineer. It is a CAD program. I hope your ProE is the same as my Pro/Engineer.

Although CAD-Systems like ProE or CATIA already have 3D meshing and FEM capabilities I dought that there are any efforts to implement a CFD module.

ICEM-CFD has an interface for Pro/E, which in turn can interface with most commercial CFD codes.

However you can set the meshing parameters i Pro/E, but have to perform grid generation in ICEM.

For further info:

http://www.icemcfd.com/

Best regards,

Paal

No. Pro/Mechanica offers FE stress capability (with some level of automatic optimisation). There's no prospect of native CFD capability IMHO in Pro/E for some time. General puprose CFD is too much hassle for companies like PTC based on the potential revenue. A viable option for PTC would be to do it in conjunction with Fluent or CD but for the amount of work involved for a native capbaility again, I doubt it would be cost effective.

It looks like geometry interfacing technology utilising native file formats or interfaces built into Pro/E using Pro/Develop will be the norm for quite some time to come.

JohnC, Pro/E is used by about 1/4 million people. Big potential market for CFD vendors regardless of whether you think CFD is 'ready' or not.

Fred.

Thank you for your answer. Is it possible to write a programme (or script)in Pro/Develop as a CFD code to solve FD equation (like N-S equation). And how can I?

Lev

Possible, don't know I'm afraid. I think Pro/Develop is NOT suitable for heavy number crunching at all, it could be used to link another CFD program in though?

Fred.

I suppose that Blue Ridge Numerics support Pro/E (and IDEAS too) in CFD area, but I never use this package. You can see it on the http://www.brni.com/Products/ProEngineer.asp

Juro

CFDesign can read the native Pro/E data (can it write it? don't know), e.g. asm and prt files. The CFD is still conducted 'outside' of Pro/E though.

There are many levels of integration. Even 'soft' integration is often good enough as far as productivity goes.

Fred.

Fred,

Have you used CFDesign with Pro-E? can you comment on it's user interphase and level of training required for non-CFD people?

Is there no cfd code that has been fully integrated with Pro-E?

Prashant

(1). "Is there no cfd code that has been fully integrated with Pro-E? ", I have already said that Pro/E (not Pro-E) is a CAD program. (2). It is widely used and you can easily visit their website. I have already done so, and I don't think that there is a cfd code integrated in it. (3).ICEM has put some mesh interface in it. But even ICEM is not a CFD solver. Most CFD solvers have their geometry and mesh generation codes. I have used ICEM to build geometry and meshes for CFX-TASCflow. To go all the way back to CAD is just not the most efficient way.

CFdesign from Blue Ridge Numerics has an excellent interface with Pro/E. Once your flow volume is created in Pro/E, then CFdesign is launched from Pro/E. If any changes are made to the Pro/E assembly, those changes are automatically updated in CFdesign. It really is easy to use and still quite powerfull.

Adding to Scott's comment above, CFdesign 5.0 for Pro/ENGINEER was debuted at the Pro/User show in Reno june 2001, winning an order over ICEPAK immediately.

CFdesign runs under Pro/MECHANICA using Pro/Mesh, and has full associativity with the exact Pro/E geometry. Changes in the Pro/E model such as a pipe radius are automatically reflected in CFdesign making parametric studies of, for example, exhaust-manifold balancing very rapid.

For complex Pro/E assemblies with numerous parts this saves a lot of time in going the old Pro/E to IGES to CFD geometry translator route.

(1). If I have to solve a turbomachinery stator/rotor stage problem, how many codes do I need to solve the problem? I guess I will need to buy Pro/E, Pro/MECHANICA, Pro/Mesh,and CFdesign? all together? (2). Can this code do it? or it can only solve flow in exhaust manifold or pipes ? (3). So, looks like you still need two codes to solve a CFD problem? and the user must buy both codes?

(1). I have visited the website of CFdesign, but there was no validation results of standard test cases. (2). I would be interested in this "finite element code"'s ability to produce accurate CFD solution.

Hi John,

I would not suggest you go out and buy Pro/E to solve CFD problems. What we are offering is a route into CFD that existing Pro/E design sites can now take advantage of, instead of having to redo their turbomachine geometry in a CFD package. The added advantage being that design changes can be done in the Pro/E environment and automatically passed into the CFD model, thus getting CFD into the design cycle earlier than before.

Anyone doing analysis in a Pro/E environment will use MECHANICA, which comes with Pro/Mesh. Simply add CFdesign as a menu option in MECHANICA!

We also support all the other major CAD formats of course.

We do have several standard test cases and a technical description in our product documentation. This is not available on the website at present.

Finite-element methods have been used for many CFD applications previously to great accuracy and success (e.g. FIDAP), and have the great advantage of easily fitting with the CAD/FEA world, a problem finite-volume has had. Even unstructured f.v. codes like STAR-CD still prefer a largely regular structured mesh for the best (or any) solution, from personal experience.

One thing we do in CFdesign is provide mesh-enhancement near wall surfaces. This provides layer(s) of prismatic elements linking with the bulk tet or hex mesh, enabling good resolution of wall effects.

Blue Ridge / CFdesign has been around since 1992 and our extensive customer list attests to the good results and benefits we are providing our largely commercial and government clients.

Thank you for your interest John, and I am glad to introduce you to CFdesign.

(1). Since you seem to be an honest person, I am going to give you some negative side of integrating solver into CAD, especially for existing design. (2). The reason is very simple: the CAD designer think differently in creating the geometry. And even if the engineer knows how to use CAD, it is still doing the geometry in the CAD way. (3). So, at one time when I was doing some turbomachinery components analysis, I had to come up with a new design approach and then implement the method in ICEM to create the geometry and mesh.(for CFX-TASCflow analysis) This is because, the available Pro/E CAD model is not suitable for analysis and design. (4). Once the CFD solutions were obtained through various CFD geometry redesign, the final geometry was then shipped over to Pro/E through IGES files so that CAD designer can re-modeled it in the Pro/E way, for CAD purposes. (5).In other words, design (geometry and mesh changes)using CFD approach requires different kind of geometry. The geometry used in CFD must be consistent with the solution and the design iteration philosophy. (6). It is possible to use existing CAD geometry for CFD analysis, but then subsequent design will be limited and constrainted by the CAD geometry. (that is, how the CAD geometry was created) (7). So, the geometry for CFD analysis should be different from that for CAD application, even though both are describing the same object. In other words, the way the CAD geometry is created is not suitable for the CFD analysis in most cases.

Hey, John, a concise and turly accurate answer! Whatever next

Many people are talking about 'CAD interation', 'shared data models', 'design integration' etc. but these people (who have indeed identifed a real market need) have not appreciated the effort involved in the geometric mapping/transform/preparation/simplification (take your pick) necessary when taking production quality CAD definitions into the analysis realm. Far more than feature supression is required. CFDesign and DESIGN/Star seem to both have direct CAD native data sharing but neither offers any geometric preparation capability (apparently, according to marketing blurb, Steve Rawnsley like to comment?).

In my field of electronics thermal analysis other codes include ESC (integrated with IDEAS but again with no geometry preparation), ICEPAK (with an interface within ProE but requires manual 'abstraction' of ProE parts into Icepak objects) and FLOTHERM that has an interface module called FLO/MCAD that does offer preparation capabilities in terms of the removal of small holes/bumps, levelling of near level faces, removal of small blends etc. all in an automatic way. CADFIX from FEGS (or an advanced version of CADFIX) also appears to offer some geometric preparation capability but seems to be focused on data translation and healing.

So, BEWARE! If it demoes well and looks like it'll solve of all your problem....it probably won't

(1). It's 3:28 in the morning. And I have been studying texture mapping and Bezier surface earlier last night. (2). I have been thinking about this CFD geometry earlier also. And I have to get up in the middle of the night to say something. (3). This is serious thinking. So, if your are for business, you can bypass it. It will only make the business harder for you. On the other hand, sometimes, we have to look beyond the horizon. (4). If we say the flat plate is just a flat plate, then we can simply draw a very flat rectangular and the job is done for the CAD guy. (5). Now, for flow over a flat plate, what are we going to do? How many CAD designers understand laminar boundary layer theory? and how many understand the turbulent boundary layer flow over a flat plate? I am sure that you are not going to find a single undergraduate student in the ME or AE department who can derive the boundary layer equation and also solve the equation. (6). Why is this so? Because, I learned how to solve the boundary layer equation when I was working on my PhD in graduate school. (7). Sometimes ago, someone was also looking at the drag over a flat plate here. And I think, I saw a rather complete report about the effect of the mesh distribution on the solution and the drag. (8). My feeling is, if you put the mesh generation in the CAD, then the designer is likely to create a nice and uniform mesh from the CAD geometry. Then everything after that would be a nightmare for the CFD solver. The guy is going to look for the boundary layer solution for the rest of his life, with this nice and smooth uniform mesh. (9). So, he needs to understand the boundary layer solution in the first place, before he touches the geometry and the mesh. To generate the mesh, he needs the geometry (the analytical one) so that he can distribute the mesh consistent with the solution. And this is the reason why it took the master like Prandtl to discover the boundary layer theory. (10). So, by making the mesh closer to the CAD geometry (or using the CAD geometry directly), it will move the true solution futher away from the CFD solver. And this is only one example. A rather trivial one. (11). It becomes clear that creating a mesh consistent with the CFD analysis and the modeling, is utmost important in getting the right solution. The requirement for the proper selection of Y+ near the wall is another example.(and I guess everyone has been faking the geometry because of the rather rigid implementation of the wall function) (12). I thought I had mentioned this before, that is, the geometry for CFD analysis should be created from the solution side, not from the geometry side.(this is nothing new, and has been studied by many researchers in eraly days. if you can't find any, try NASA/Ames first) Then how do you create the geometry and the mesh, if you don't know the solution? It is one of the biggest difficulty and it is also the wide open opportunity in CFD. There are already many methods available to move the mesh and modify the geometry closer to the solution. (further away from the CAD geometry) (13). The discussion here is just personal experience and opinion. So, it will not affect anyone's business practice at all. And I know that if I use uniform mesh for a 3-D problem, then it is going to take forever to get the right answer. (14). So, I guess, everyone would agree that non-uniform mesh is essential. I got the feeling that it's better to create the mesh closer to the solution, because in this way, one can optimize the mesh size and have the opportunity to get a good solution. (15). Now we are looking for the unknown mesh based on the unknown solution, so that the proper geometry can be created. Isn't that very interesting? That's the power and charm of CFD. In mathematical terms, the nature is coupled and non-linear. And it is easier to say "what you see is not what you get". (which is the opposite of the graphic world)