[AnaE]

Hello,

Recently someone mentioned SOLIDWORKS Flow Simulation - I have never heard of it, so I am curious about its capacities and there is not much information available.

Is it possible to perform compressible flow analysis with this software?

Thanks you!

[fluid23]

It is fairly limited, but good for conceptual or quick-look analysis. You don't have a lot of control over the mesh nor any options when it comes to turbulence modeling, but it can get you decent answers for a lot of problems. For the money, its fairly good. Although, I tend to opt for power/efficiency/user friendliness.

As for compressible flow... yes.

[aeriform]

http://www.solidworks.com/sw/product...simulation.htm

I had a chance to try this out in a 1 hour session a couple months ago where we were altering a design for car exhaust.

The focus is on simplicity. If you want to nit pick about details of meshing then this isn't the option for you. It does provide relatively quick setup and you can pump out many fast (<1 minute to compute) steady state simulations for optimizing changes to a 3D model. From my perspective, there is a higher focus on the 3D modeling optimization side rather than a focus on the CFD itself. In other words how does the average pressure of the exhaust change as I change my internal model rather than detailed characteristics of the airflow in or exiting the exhaust.

Once you have the boundaries setup initially, it's very quick to make changes to an internal SolidWorks design and reuse the setup boundaries. If you already have SolidWorks Premium then this could be an ideal program depending on what you're focused on.

Primary uses are for electrical cooling and pipe flow:
https://www.youtube.com/watch?v=d6T-dvi2znk

For pipe flow you can take a look at:
https://www.youtube.com/watch?v=X3ecq8MNfk0 https://www.youtube.com/watch?v=9HhjrZWDsbs
Fan flow:
https://www.youtube.com/watch?v=NauURWekZmY

[AnaE]

Thank you for replies.

Ok, so I guess that the software is fine for simple simulations,and to have a first impression of the physics of the problem.
But for example, to simulate a complex case (geometry with a lot of details, mesh with very different sizes), which for example in Fluent has 1M volume elements, is not feasible in that software. Am I right?

[aeriform]

It's possible to create complex meshing by meshing parts of an assembly separately. If you only have one part you can slice it up with surfaces then mesh each slice separately.
http://help.solidworks.com/2015/engl...al_meshing.htm


the biggest issues I've found are with sharp geometry, small details in large parts, and multiple parts or features that are just touching but not overlapping or have small separations.
http://help.solidworks.com/2015/engl...il_to_Mesh.htm

[Boris_M]

I'm sorry but this is all completely wrong.
Let's start with the first post here:

@ AnaE:
FloEFD or SolidWorks Flow Simulation is very well capable of calculating compressible flow and not at all a rough but fast results tool. This tool is used by the aerospace industry as well as Automotive industry and not in some "rough results" analysis but for simulations where the results have to stand against other CFD tools and measurements.
Sure it doesn't have all the capabilities such as the other tools for example sloshing of a tank with free surface flow or spray modeling and Fluid-Solid-Interaction etc. but the capabilities it has it performs extremely well.
If you let me know what application you are looking at I can post some references here for you.

@ MBdonCFD:
As mentioned above it is not quick-look CFD. And you can influence the mesh a lot but you don't need to do surface fitted meshes and consider y+ etc. It applies solver approaches that do not require such meshing methods as well as the turbulence model k-e is used in 90% of all actual industry applications so unless you work in the rest 10% you don't need any other turbulence model and especially since it doesn't use the standard k-e but a modified model which has extended capabilities compared to the usual ones.

@ aeriform:
In a 1 hour session and a few videos you won't grasp the full potential of the software.
Most simulations with it are also not done in less than 10 minutes. Probably the model you did was a simple ball valve because this is something shown real quick to see how it is done and calculates in less than a minute. Most cases take hours to calculate and have more complex geometries than a ball valve and require a bigger mesh and therefore longer solver time.
You are right, it is ideal for very early simulations in the design stage and fast in doing multiple clones of the projects with slight changes in the design to see how the improve the results but that's the setup, the solver takes as long as it takes depending on the model and physics you consider.
The focus is not on simplicity but on ease of use which doesn't mean only simple models can be calculated. It means the user interface is intuitive and the meshing automatic and fast etc.
As for the links in your last post, those are not from Flow Simulation but from the Stress Simulation tool in SW and they are not the same at all. The meshing in FloEFD has no problems with complex geometry and it doesn't consider surfaces and slicing up geometries. You can use geometric objects you create and only use as reference regions for mesh refinement definitions or use the geometries of your assembly itself for local meshes. It has absolutely no problems with overlapping geometry or small features. This is not Flow Simulation you are talking about.

@ all:
There is no problem if someone who actually really used the software comments about its quality but it is never a help if someone who only used it once for an hour or just saw videos about it or from heard something about gives any comments about it.
Please let those comment that actually use the software and know its capabilities.
A good way of finding out if someone actually knows the product would for example be to look at the profile and see where that person post the most comments and if it isn't in this forum than it CAN be a hint that he is probably not using the software. In any case, always ask how long someone has used the software rather than listen to everyone posting.

In my case, I've been using FloEFD for 8,5 years now and know the software inside out.

Regards,
Boris

[aeriform]

Originally Posted by Boris_M

I'm sorry but this is all completely wrong.

@ aeriform:
In a 1 hour session and a few videos you won't grasp the full potential of the software.
Most simulations with it are also not done in less than 10 minutes. Probably the model you did was a simple ball valve because this is something shown real quick to see how it is done and calculates in less than a minute. Most cases take hours to calculate and have more complex geometries than a ball valve and require a bigger mesh and therefore longer solver time.
You are right, it is ideal for very early simulations in the design stage and fast in doing multiple clones of the projects with slight changes in the design to see how the improve the results but that's the setup, the solver takes as long as it takes depending on the model and physics you consider.
The focus is not on simplicity but on ease of use which doesn't mean only simple models can be calculated. It means the user interface is intuitive and the meshing automatic and fast etc.
As for the links in your last post, those are not from Flow Simulation but from the Stress Simulation tool in SW and they are not the same at all. The meshing in FloEFD has no problems with complex geometry and it doesn't consider surfaces and slicing up geometries. You can use geometric objects you create and only use as reference regions for mesh refinement definitions or use the geometries of your assembly itself for local meshes. It has absolutely no problems with overlapping geometry or small features. This is not Flow Simulation you are talking about.

@ all:
There is no problem if someone who actually really used the software comments about its quality but it is never a help if someone who only used it once for an hour or just saw videos about it or from heard something about gives any comments about it.
Please let those comment that actually use the software and know its capabilities.
A good way of finding out if someone actually knows the product would for example be to look at the profile and see where that person post the most comments and if it isn't in this forum than it CAN be a hint that he is probably not using the software. In any case, always ask how long someone has used the software rather than listen to everyone posting.

In my case, I've been using FloEFD for 8,5 years now and know the software inside out.

Regards,
Boris

All completely wrong? You must have had a bad day. I'm pretty sure I mentioned that I had little experience with SW flow simulation specifically. To clarify, I only had 1 hour personal hands on training where it's me on a computer with SW Flow Simulation implementing my own ideas in order to get a feel of optimizing a model for pressure, temperature, speed of pipe flow that has nothing to do with ball valves. I'm aware that the product, like every product, has limitations in what it's capable of and these capabilities change over time.

I would love to know more about the flow simulation from SW which was what the original question was about rather than Mentor Graphics FloEFD.

Personally I'm biased in favor of SW because I already use the mechanical, drafting, and modeling aspects on a regular basis. Adding the ability to optimize flow in and around models I already have in an interface I'm already familiar with is appealing. Personally, having programmed meshers and simulations from scratch, I think the FloEFD is a cleaner way of meshing. The SW products I use are not the most stable software when the user is interacting with the software, but I've never had any must reboot the program issues during a stress analysis regardless of run time. I'm assuming the same with the flow simulation analysis. It's a small annoyance, just make sure you save your work often.

The understanding I got from SW personnel when I specifically asked about meshing was that there are only 2 meshing libraries used in SW products. One is for their paid products, the other is a crippled version that is implemented in trial software. So the meshing library used behind a SW mechanical analysis or SW flow analysis is the same. There could, admittedly, be different parts of the SW mesher used or accessible by some SW products and not others...

One of the best things to do would be to contact a reseller and present them with a case you'd like to solve. In most normal circumstances they will have seen similar cases and either have or know someone who runs SW with that expertise. You can get a quote in the meantime which is probably a larger deciding factor if all you're concerned about is if SW can work with compressible fluids.

Another main deciding factor would be what are you currently using to create models and do you intend to continue using that software? You can get more out of SW flow simulation if your model uses a SW model with SW parametric features.

Are you intending to do basic research or optimization of components or networks? You might get more out of SW flow simulation if you're doing optimization.

If you're doing longer runs, do you have the capability of splitting a flow simulation across a network? I'm pretty sure SW is still not able to split a job across a network, some other CFD packages can. I think you're still limited to a single computer.

Do you intend to have porous media, heat transfer, moving parts controlling flow, particles controlled by flow, transient, transonic, or supersonic flow? This is probably doable and in some specific cases made very easy/simple to implement.

Do you intend to model multiphase flow, heavy particles interacting with flow, phase transitions other than steam, combinations of fluid flow with plastic deformations, or chemical reactions? You may not be able to do so with SW flow simulation depending on what your specific needs are.

[Boris_M]

Hi aeriform,

As I mentioned "I'm sorry" because I read the thread and it sounded mostly like all negative comments from people who haven't used it enough to get the experience of the capabilities. That kind of made me mad as I try not to say anything about a product that I don't know except I don't like it because ... and not it cannot do this and that although it might be able to do it and I just don't know it because I barely used it.
I'm sorry if you felt attacked and I don't want to say rather buy FloEFD than SWFS. It is as I said a large portion the same product with some less capabilities in SWFS but still capable of most of the tasks you might need for pipe flow.
I was also referring to the valve because that is the first tutorial model that comes with the software and of course not the most suitable one to show the more complex and sophisticated things that can be done with the software.

As for your SW environment, Yes, SWFS is the right tool just like FloEFD standalone as it uses the SW modeler as you will find when you look at the user interface. This is the reason why FloEFD is integrated also in Catia, ProE/Creo and NX - to have the same environment the user is used to.
The stability of the software I noticed is a lot related to the installation of the users computer. I know of cases where it crashed constantly and on my computer it worked perfectly. There are just too many variables that can cause a crash for different computer configurations. In such cases it is important to pester the support to help you fix the issue. At the end SWFS code issues go back to Mentor Graphics development as it is the code that is developed by Mentor. Any stability issues with the stress simulation code in SW would need to be handled by SW themselves as it is not a product from Mentor Graphics.

No, SW Simulation (the FEM stress tool) and Flow Simulation "SWFS" (coming from Mentor Graphics) are two entirely different tools just using the same CAD integration. The SWFS has its own mesher which is based on a finite volume method Cartesian mesh with octree refinement. It doesn't have anything to do with the stress simulation tool except that you can map the thermal and pressure load from the CFD tool onto the FEM tool as boundary conditions.
In FloEFD and SWFS there is just one meshing type and that doesn't change due to installations. I think there might still be a teaser version in SW, it was once called Flow Simulation Express or so. The limitation is more to the range of boundary conditions and fluids you can use etc. It should just give a small peak in what you could do and how it works but not the full spectrum of capabilities. But uses the same mesh.

Yes, SW reseller should be able to give you an idea if there are other customers of theirs or in the SW world that use it for such an application already. Worst case, they can contact their SW contacts and ask them if they know any customer as a reference. Not every reseller knows all the users as they might be customers of another reseller. You can also have a look at the success stories on the Mentor Website for FloEFD or in the Engineering Edge Magazine on the website. There might be also a similar application in one of the stories there.

Yes, SWFS and FloEFD are currently limited to shared memory (single computer for one calculation), but usually you don't need a cluster in most cases. I know you always want the results ideally yesterday :-)
You also have to consider that most time with traditional methods is spent on meshing which can occupy your own time (not the CPU) for days whereas with FloEFD/SWFS the meshing is automatic but the solver might take longer than on a cluster. But CPU time is cheap compared to man-hours.

SWFS and FloEFD have the capability of using porous media, heat transfer, particle study (pure post processing but controlled by the flow, just not influencing the flow, that would be more DEM), transient, transonic and supersonic (even Hypersonic with the Advanced Module in FloEFD, not SWFS). Moving parts controlling flow depends on what type of movement.

Multiphase also depends on what type as there are different methods considered as multiphase but free surface flow such as a sloshing in a tank is not possible. There is phase transition for any condensation/ice film and also in the advanced module of FloEFD (not SWFS, doesn' have that module) you can have gas combustion, so some chemical reaction type.
But yes, some of these physics are currently considered as highly advanced physics and in itself need often a deeper understanding of the physics to be able to judge the results or interpret them etc.

Sorry again for coming over too aggressive.

Boris

[freierfall]

I am a newbe in SWFS, so I am not able to compare things. And I do only flow without heat, moving walls or without transient.

no Boris, i was also shocked reading the statements before. I was wondering, because SWFS is powerfull and sure on some points really easy to use. The first question should be what is not possible in SWFS. You both aeriform and Boris said it already.

I would be happy if my cell definition and calculation will be done in 10 min. But right now takes all in all several hours, per study. But this is actually not a reference point for a good CFD software.

[Boris_M]

Hi Sasha,

Yes, the best way of asking is by describing your task in detail (size of the model smallest detail that is important what the task of the product ist etc.) and what "physics" (meaning transient, with heat conduction, porous media or any rotating parts etc.) you need to consider in your task and then ask if the software is capable of doing that. Otherwise there are just a hundred different answers that don't really help you as you have a completely different need.

Well not all calculations are setup in 10 minutes and the mesh is done in 2 minutes of course. This always depends on the complexity of the model and the amount of boundary conditions you require to setup the project. If it is just a pipe flow it is only about inlet and outlet conditions and that's about it for the boundary condition setup. If you have a electronics box you usually have many different materials, heat sources and fans etc. so it will take some more time to specify all of them on the specific components.
The next step is then the meshing. I've seen really complex geometries being meshed with 5 million cells in 15 minutes and other models took 8 hours. But this time was pure CPU time in which the user can work on other things. You might run 2-3 meshing iterations until you get to the desired mesh but changing the mesh settings can be done in just a few seconds to a few minutes before you hit the mesh button again and let the CPU do the rest.

Boris