CATIA is probably the most preferred 3D modeling tool in the automotive and aerospace industries. This specific surface (it can be handled well by the experienced users) is well suited to use parametric design principles and hence result in great products. One of the issues we have in the automotive industry is to work with large surfaces for body exterior (main body, gate, hood.) or interior trim (door, headliner.) panels. CATIA enable us to design surface perfectly through the lines. Also, assembly features for intricate designs (gearbox, exhaust, engine, suspension.) are more than adequate. I did not find any inefficiency while designing products with CATIA. Only the interface and handling of the menu and mouse is not the best - or at least you need time to get used to it.
It appears to be evolving more towards large users with the 3D EXPERIENCE, while becoming less focused on small users like me, becoming more expensive and limiting the number of cores while most PC's now can easily run 8 cores. Of course it is great for non-linear and highly non-linear scenarios, and especially good at combing a huge variety of element types. I guess it is not best suited for linear analysis due to its high price, but even in this case you have to put in the balance the ability to make the best choice of finite elements and being so straightforward about actually using them.
Create and modify designs from within the CAE environment. Although it has very basic capabilities, it is quite capable of generating 2D and 3D parametric designs. I have even generated some "fancy" designs which ended up being a challenge for draftsmen in specialized solid modeling CAD. I also like the fact that being somewhat limited, it forces you to stick to simple and effective design.
Addressing structural instabilities such as snap-through or buckling. This was such a challenge when I started using Abaqus, I had to take a course on "achieving convergence". Coming from there, I can see how SIMULIA has become increasingly able to give you the numerical tools or tricks you need to achieve convergence consistently. In the past I often feared running into different convergence issues as I moved across different produce sizes, leading to changes in the analysis approach, which would make it more difficult to compare them. I just went through a 12-size family in 5 dimensional scenarios each, without a single issue after ensuring convergence with the first couple of sizes.
Beautiful pictures. The post-processing of results enables me to generate highly illustrative, fairly easy to understand and elegant presentations, by controlling transparency and results shown independently on different groups, which I can select by material, location, etc.
Exporting sketches. For example, you can import dxf for sketches, but you can't export dxf. This is a major drawback for me, because I often communicate with customers through dwg or dxf sketches. If I can't export dxf, my sketches in CAE are "dead". I have to redo them in the dwg sketch. It is so inconvenient, I often end up making the sketches in DraftSight, so I can export them to CAE but I still have the originals in dwg format.
CAE doesn't remember element type assignment by sets. Sometimes when I reconfigure or modify a design, I regenerate sets and surfaces, and this in turn updates material assignments, interactions and loads almost automatically. This doesn't happen with element types, so I often end up submitting a job with the wrong element types.
Mesh regeneration and re-mapping in 2D within the job. I need this a lot to model axisymmetric assemblies with sharp indentation the destroys the original mesh. I had it in MSC/MARC and it worked great. Abaqus has it only for 3D. I have tried alternatives like lagrangian/eulerian and eulerian domains but it's still more complicated.
Extremely well organized and friendly, reflecting the latest approaches in solid modeling and adding the FEA part so seamlessly for the user. It's a joy to use. I'm aware some people would complain about the meshing capabilities and thus prefer using HYPERMESH. Fortunately not my case. I did wish I had HYPERMESH a couple of times, but for most of my projects CAE meshing is fine.
I generally contact with IT or internal CATIA support team, so I cannot comment in detail. I remember we have struggled to move to v6 update, and it required a long and detailed work to update all the systems. There were some times we had limited support from Dassault, but it is normal for a company of our size. So I cannot criticize much.
I keep getting the feeling that it is drifting away from small users. I don't feel so comfortable with the SWYM community approach. I felt much better when you could just email or call HKS and you would feel like you were talking to a real expert who understood your situation. When I was reassigned to reseller in Argentina, which is like the other side of the world for me, I appreciated the cultural closeness, but I had to say goodbye to the level of support I had from HKS and learned to "support myself."
Once major advantage I realized when using CATIA is surface modelling, and modifications or changes in part model later. SOLIDWORKS provide a really good and quick way to part modelling. However, part modifications and assembly modifications take time. In CATIA, complex geometries can be easily generated, when compared to SOLIDWORKS. Managing and updating designs of large scale models and complex geometries made me realize an advantage of CATIA over other CAD software.
MARC can do the job from an FEA point of view. It even appears better at a couple of things such as 2D remeshing and surviving with highly distorted meshes. So I regard it as a very competitive alternative. I prefer SIMULIA because the GUI is so much better, especially the ability to create actual parametric designs in CAE. ANSYS doesn't cut it for me. I tried it for 1 month and I would see how much I would struggle with convergence in large deformation and large displacement situations with hyperelastic materials with which I have to deal all the time. Also the GUI is not as well built an integrated as in SIMULIA.
The ease of modeling helps a lot to save time, in both modeling new parts and also assemblies, which in the end makes a great positive impact on the development.
I don't even bother doing an ROI. Without SIMULIA I would be pretty much out of business. Except for the fact that I could still have an alternative with MSC/MARC.
Price is becoming a more difficult hurdle for small users. It appears I have survived because there was a 50% discount on my small 6-token package. Without this special price I would have to close my business. I'm currently exploring alternatives.
A more flexible scheme enabling to add tokens over short periods of time or paying per use could be an interesting possibility.
