Inventor & non-manifold topology?

[Hopinc]

Hi Guys,

 

I might stir up a hornets nest with this discussion, but what the heck, here goes.

 

I have only been using Inventor now for 2.5 years, compared to over 22 years using S.D.R.C's I-Deas Master Series. Because of my previous experience it has become obvious to me that Inventor is seriously lacking by not using non-manifold topology.

 

What is non-manifold topology I hear you say? Basically it means that the software only updates parts (and those directly associated) that have changed. It does not rebuild and recalculate all the parts and constraints of the model/assembly when it is unnecessary.

 

It became fairly obvious to me that Inventor seems to be lacking in this department. I have been modifying some quite complex assemblies recently and have been amazed at the length of time taken for updates. Escpecially when the modifications were carried out on parts at the end of the history tree. As such it should not have necessitated the dreary updating of the whole assembly piece-by-piece every time I made a minor change.

 

Now don't misunderstand me, I am not saying that non-manifold topology is the answer to everything. For instance if a modifications takes place on a step early in a long history tree it can still take a while to update, but most of the time it really does speed things up.

 

If I am wrong about Inventor not employing some form of this technology please shoot me down in flames. However, if it does I would humbly suggest that it needs some serious overhauling.

 

Anyone agree/disagree with me?

 

Regards.

 

Dave

[cuneyt ustun]

Hey how do we reply to that

[cuneyt ustun]

Ok seems I found that

 

Dear Dave ,

 

I am a new user of inventor since 2 months. So I can hardky tell agree or disagree...

 

While attemting to work on a transferred part from NX It gave the error :

"The attempted shell operation had Non Manifold inputs . Try with manifold inputs."

Do you know any way to workaround that?

 

Thanks

Cuneyt

[Hopinc]

Hi,

 

I believe that this error message means that the material thickness is not uniform - it varies. This is sometimes a problem when transferring some sheet metal parts from the I-Deas/NX systems. It is NOT a problem with Inventor.

 

I ran into this myself and after interrogating the offending part discovered that thickness errors were the cause. On the plus side some parts will work OK, others won't. I don't know why this is so, it appears to be quite random.

 

The trouble is that it is so long since I came across this I cannot remember exactly how I resolved it.

 

On reflection I think that I imported it as surfaces, not a solid, deleted the inner and edge faces and thickened the part. You should then be able to convert it to sheet metal and create a flat pattern without problem. This is not an elegant fix but it might work for you.

 

Please let me know how you get on?

 

Regards.

 

Dave

[JD Mather]

"The attempted shell operation had Non Manifold inputs .

Try with manifold inputs."

Do you know any way to workaround that?

 

What Options were set on import?

Sign up at http://au.autodesk.com and check this class material

http://au.autodesk.com/?nd=class&session_id=3056

 

or...

attach your original file here (not Inventor file).

[Hopinc]

Yes, JD is quite right I should have suggested that also.

 

When you load your IGES file select Options at the bottom of the menu. Here you can set parameters to import the part as a single item (if required), select the entities to be imported -solid, surfaces, wires or points and you can also check the parts as they load and enable advanced healing.

 

Dave

[Hopinc]

Hi Guys,

 

I revisited one of I-Deas IGES parts to see if I could make a translation into Inventor. Unfortunately I was not entirely successful.

 

Picture 1 shows the part I brought in using the options I mentioned above. I then stitched the composite surfaces together to form a solid. Turned it into a sheet metal part, setting the thickness parameter to 1.5mm, and made a flat pattern (picture 2).

 

You can see that Inventor has not recognised every bend.

 

Has JD any thoughts as to why this might be so?

 

Regards.

 

Dave

[JD Mather]

I would have to see the actual part - my guess is that there is an area that is not constant thickness (no matter how small) or not planar, cylindrical or conic (Inventor can unfold native sheet metal loft, but not imported geometry that is not planar, cylindrical or conic).

There are Inventor add-ins that can, but from the picture it looks like that part should be easy to fix (or simply remodel as native).

[Hopinc]

Hi JD,

 

Quote: "There are Inventor add-ins that can, but from the picture it looks like that part should be easy to fix (or simply remodel as native)."

 

First of all, where are these add-ins available from?

 

Secondly, remodelling as native parts is not the answer.

 

I except that legacy data can be difficult to handle but if you have a lot of it to cope with having to remodel each part is a definite "no no." If there are add-ins available that can overcome these issues why doesn't Inventor handle legacy data better than it does?

 

Dave

[JD Mather]

Sheet metal includes quite a bit of variation in flat pattern due to material, bend radius, bend angle, thickness and process machine. The other tools like Autopol caution you to verify with trial parts. More real world experience is expected as it starts to get into an art. I suspect Autodesk is geared more towards the "least common denominator" designer who barely knows what they are doing, so to avoid helping them make a mess of things (and then cussing Autodesk for the money lost - thereby giving Autodesk a poor reputation) Autodesk sticks to tools that are fairly basic and suggests 3rd party apps for stuff like this.

 

I think once you identify the particular feature that is causing the unfold problem you can quickly fix on other similar parts from the same system/designer. But if you get files from all over the place......

.....heck, most of the stuff I see done native in Inventor I can go through the history tree and see it is garbage.... .....then compound the problem by not having a sketch/feature history tree to see where the designer went wrong...

[Hopinc]

Quote: "Sheet metal includes quite a bit of variation in flat pattern due to material, bend radius, bend angle, thickness and process machine." - absolutely 100% agree with you JD.

 

Quote: "More real world experience is expected as it starts to get into an art." - most definitely agree.

 

Quote: "Autodesk sticks to tools that are fairly basic and suggests 3rd party apps for stuff like this." - this is a wrong philosophy. Autodesk is shifting responsibility onto 3rd parties instead of providing better translation facilities themselves.

 

Legacy data is not an issue from me now since I set up my own company, but had I still been at my old place of employment it most certainly would have been. We had 22 years of I-Deas usage. Even if only 10% was required to be reused we would have certainly been amongst those "cussing Autodesk."

 

I think you and Autodesk both underrate the importance of legacy data.

 

Quote: ".....heck, most of the stuff I see done native in Inventor I can go through the history tree and see it is garbage" - surely some of this is down to poor training courses?

 

I well remember when I did my Inventor Pro training and that I was appalled at the way the course was laid out. Even I, with over two decades of solid modelling experience ended the course shy of some of the basics. It was only afterwards that I filled these gaps.

 

There were only three of us on this course and one left without the slightest idea of how to use Inventor. He was totally confused by the chopping and changing from one file to another.

 

I had some sympathy for him. About 1/3rd of the way through the course he started to really struggle. I heard later they had to bring him back in and give him some one-to-one tuition in order to save face.

 

The jist of this is that I don't think you can always lay blame for poor technique on the end user. Sometimes it's the quality of the training.

 

Regards.

 

Dave

[JD Mather]

I think I have a pretty good number that only around 10% of all users (including trainers) really know how to use the program.

When you panalyze that statement it gets interesting....

[Hopinc]

Yes, that is food for thought!

 

I don't think though that the instructors lacked training in this instance. It was because they were obliged to stick to the training course as layed down by Autodesk. I hope that it has been revised.

 

Certainly the tutor that took me for a one-to-one on advanced surfacing new his stuff. Together we finished it in record time and covered a lot of other stuff to my satisfaction as well.

 

Have a good weekend.

 

Dave