CAD/CAM discussion forum > Other CAD/CAM Technology > Options for surface tangency?

Options for surface tangency?

    
  Subscribe Topic

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 1 of 27

 Options for surface tangency?
25-05-2006 01:16 . pm   |   View his/her posts only
I'm wondering whether there are options besides the standard loft command that allow you to build surfaces to other surfaces while maintaining the tangency and continuity of all faces.

The N-sided and fill gap command give the option for edge tangency, but what I find is that the continuity is not very clean. I would like to be able to repair one surface and be able to put a chained radii over top of the ridgeline that is selected. I find that the loft command usually is my only hope, and the down side of this is that I have to rebuild a surface where the 3rd and possibly 4th open edge would be.

A better solution for placing these surfaces if it would only allow tangencys (especially if an entire edge line is not sewn) would be using the bi-rail loft, or drive curve loft.

What I find is that it is much less likely to make a radii if the edges are not contiguous.

Any thoughts?

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 2 of 27

25-05-2006 06:11 . pm   |   View his/her posts only
Hi Paul

First thought is, can you provide an example file?

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 3 of 27

26-05-2006 12:22 . pm   |   View his/her posts only
Hi Chris,

I've attached the file that I struggled with prior to this post. It's not real complicated nevertheless I still can't apply an overedge radii to this without cleaning up the green surfaces in this file.

I need to apply a .1875 overedge radii. As you can see I was able to get a larger overedge radii around the impression in blue.

I had to tweak the 2 far green radii to a larger size.

The closer green surfaces I ended up removing the long surface and replacing it with a drive curve loft that is more tangent than the existing one. I erased the remaining 2 green surfaces and did some fudging from that point on.

I would of liked to just place a N-sided surface tangent on all 4 sides, but this had poor continuity on all edges and wouldn't allow for an overedge.

My 2nd choice would be to use a birail surface, but that was not tangent and does not have good continuity.

If you can't decipher what I am talking about from this file I can try building another example.

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 4 of 27

27-05-2006 12:20 . am   |   View his/her posts only
Hi Paul



The main problem is not to do with the ability of the VX surface functions to hold tangency as much as it is to do with the disposition of the original geometry. You cannot have a tangent patch that also gives curvature continuity (needed for a successful tangent fillet blend) when each of the original surrounding faces meet abruptly or do not even meet cleanly. There is therefore more to replace/repair than just the faces you have highlighted in green. Image1 shows that an original fillet, as you have highlighted, is way-off, such that the adjoining face must have the curve continuity of it's edge fixed first (use a blend curve, then replace face boundary). We can then define a new face for the region that is tangent with it's neighbours - Image2. However, another original fillet also has tangency that is way off (comb analysis of edge continuety, Image3) - it is this that prevents the fillet size that you require, not the long green face above it. Image4 shows that the edge at that green face will fillet OK providing the fillet does not reach that lower non-tangent edge (R0.12). Certainly, the 0.1875 fillet required is very large in relation to the nature of the edges involved, increasing the difficulties.

There are other tools that can help - Use Edit/Face/Modify Nurbs/Elevate to increase the DOF of a face, and from the Free-Form tab the match tangency command is good for correcting the non-tangent fillets.

I think you have made your task a little more difficult by the way the file is managed. What I suggest you do next time is to keep each of the original part models in their own seperate (root) Part Objects. Insert them into your die design as assembly components and merge them. With this method, when you find a problem such as the ones found on this project, you can go back to the original part and modify that. A regen of the die model will refresh it with the modified geom of the original parts. This lends two advantages. Firstly, it is easier to work on the original part seperately. Secondly, the design intent of the part needs to be considered. In this case, a fair bit of change is required in order to accomodate a fillet of any size on the tool. There could be a time when you would have to go back to your customer and request approval of the changes. Looking at the problem fillets on the part model, I would, if I could, go back to them and ask for a better model (or perhaps a non-filleted model if they are unable to improve their tangency -looks like it may have been modelled in Catia v4?). I hope this helps you at least a little. Good news is that the Florida team are working on a new, more powerful fillet engine for VX.

Rank: 1

OldForumPost

Newbie

posts: 0

Registered: 2012-1-14

Message 5 of 27

27-05-2006 03:42 . am   |   View his/her posts only
Hello Paul,

Here is my solution, please step true the history, there are still some real small tangent problems if you check it. But I think good enough.

In general if you work with a tolerance 0.01 MM than it is allowed for VX to have max. tangent errors with 0.3 degree. (Sorry I am located in Germany and my "world" is metric). The final thing here was the FEM Patch, with the option TANGENT on the edges and for the U DIRECTION of the surface was the Z DIRECTION used. BUT before I used the FEM PATCH I made on the edge where I do not like to have a tangent area a CURVE.

During the picking of the 3 curves for the FEM patch I used the pick filter, I picked one time the CURVE, that I made before and than the two other edges.

The FEM patch makes a trimmed tangent surface, that is the most time better than the N-Blend

I hope it helps ... if not please ask

Best Regards

Thomas

Rank: 1

OldForumPost

Newbie

posts: 0

Registered: 2012-1-14

Message 6 of 27

29-05-2006 02:42 . am   |   View his/her posts only
...there was a problem with the file upload...so I try it again

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 7 of 27

30-05-2006 10:17 . am   |   View his/her posts only
Tom TE and Chris,

Thank you for looking at and applying what you would do to fix a problem such as what I have.

There appear to be many different ways to attack the same problem. The reason that I asked is because even though I can get past the problem it still is quite a nuisance. I use certain tools with every job that I do, but there are many tools in the tool box that I probably haven't used.

I have read through your suggestions but unfortunately have not been able to view your results. I can not view the pictures that Chris submitted and can not read the file the Tom linked. (I get a file size upwards of 33,000 kb but no objects or part appear.)

If you have any suggestions in how I could view these images and file I'd appreciate it. I'd like to see exactly how you did this.

Chris: normally I don't encapsulate but I just wanted to remove the insert dependencies for demonstration purposes.

Rank: 1

OldForumPost

Newbie

posts: 0

Registered: 2012-1-14

Message 8 of 27

30-05-2006 12:33 . pm   |   View his/her posts only
...which VX version do you using. I made it with VX 12.41. If you work with a VX 11.X version I will do it again in that "old" version.

Regards

Thomas

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 9 of 27

30-05-2006 01:22 . pm   |   View his/her posts only
Tom,

I just downloaded a newer version of a RAR extracter and was able to read the file in.

From what I see however, even though the file name is different the contents appear to be unchanged.

I'm not sure if this is something else on my end, if this was from an earlier file or if it was some kind of translation error.

Rank: 1

OldForumPost

Newbie

posts: 0

Registered: 2012-1-14

Message 10 of 27

30-05-2006 01:40 . pm   |   View his/her posts only
Hello Paul,

I saved the new VX file with the name "3453dienew.VX" that should make no problems if you open the file and if you step true the history.

Regards

Thomas

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 11 of 27

31-05-2006 09:45 . am   |   View his/her posts only
Tom:

At 1st look it appeared that the model I submitted and the model you modified were identical. I went through the history as you suggested and could see what you actually had did.

This is quite interesting. You followed some of the same procedure I would take yet you used a couple functions that I have not really tried out.
1) Line at angle to face; it was interesting how in picking the edge with the face you were able to match the drafted sidewall.
2) FEM Patch; The tangency looks great but the the continuity is off considerably. However this does not stop the creation of the overedge that I was after. (To my surprize)

Thanks a lot for your suggestions, this gives me 2 more tools that I will without a doubt be using and I think will save me time over the methods I used.

Sometimes you have to use geometry that is far from perfect, in these situations it is important to find the most painless way to get past the flaws.

I am looking forward to the filleting improvements that VX is working on, perhaps it will be able to overcome geometry like this.

Rank: 1

OldForumPost

Newbie

posts: 0

Registered: 2012-1-14

Message 12 of 27

31-05-2006 10:58 . am   |   View his/her posts only
Hello Paul,

I am happy that is helps, it is every time the most important thing ...at has to be done ... of cource it can be never quick enough.

Some of the new filleting is based on the "Create Face" that can be found under the "Free Form" on the left side.

Please see my example for the "Create Face".

So you must still wait for the new filleting but there is already a strong function with "Create Face" that can help if the solid or surface fillet does not give the right result.

Regards

Thomas

Rank: 7Rank: 7Rank: 7

Paul

Moderator

posts: 344

Registered: 2011-9-17

Message 13 of 27

31-05-2006 06:53 . pm   |   View his/her posts only
Hi Paul,
Just curiosity here.
I looked at the geometry you are working with. Do you know what Frankenstien created it?
It just looks plain ugly, to use a kind term, and it amazes me that VX manages to work with at all.
I am feeling kinda lucky that I originate the parts - and don't have to deal with others efforts.
All good learning tho - as a spectator.

Cheers,

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 14 of 27

01-06-2006 09:55 . am   |   View his/her posts only
Hi Paul

Good to hear that you have a reasonable way to move forward. Sorry you could not see my image - not sure why that is. If anyone else is reading this, please reply to the thread, just to let us know if you can or cannot see the image. I have now also attached it to this message.

I spent some time in the motor vehicle industry (having originally learnt my trade in the aircraft industry). and I was shocked by the poor modelling quality that I was hit with. It is caused by two main things - sadly, the number one cause is the User - all too many people out there that do not pay enough attention to what they are modelling, from a CAD model perspective. I have this week seen some huge models of parts intended for casting - without any draft angle applied anywhere! The second cause though is the CAD application itself, especially one that only has an "implicit" modelling mode, no parametrics or history. Such models, over time, become hugely complex in their geometry description, with non-tangent patches here there and everywhere. As the User nears his deadline, more compromise in model quality is made, and you end up with things like bosses not intersecting their parent face completely and really unwise shortcuts such as mods to the 2D drawing without the model being updated. On the model data that you have had to deal with Paul, the problems are brought about by your customer's poor quality data - the lack of tangency where fillets are applied and odd "nips and tucks" where they have bodged a modification or two. Still, for those that like grey hair, all this stuff is wonderfull

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 15 of 27

01-06-2006 11:40 . am   |   View his/her posts only
The quality of the customer model is fairly poor yet we regularly work with worse data than this.

In our Think3 days it was impossible to seal the models up. Instead of being able to stitch things tight it would all unravel.

Customer models which are seperate from their parametrics can be bad.

Unfortunately however we cause much of our own geometric mess. Regularly a customer will need the forging to be slightly modified to be more condusive to machining or perhaps real world draft needs to be implimented into the model. A new inclusion to this mess is where customers try to simulate trimmed flashland at the partingline (many times for their FEA). In trying to patch things together we end up with a convoluted mess. We do not have the time luxury to start from scratch. Rarely will the customer modify their base model.

In our business it is quite ironic that modeling the base forging shapes is much more difficult than the nice, neat and precise end product.

Chris: I had someone double check your original JPEG and they also can not read it. I can see what you are saying from your last attachment however. I do wish there was an easy button with the bad tangencys.

Tom: It is interesting that some of the new filleting modeling will be based around the create face command. Sometimes we have good luck with it's intelligent blend. Sometimes things just don't work with the geometry however. (and this is a flat parting line example)

Rank: 7Rank: 7Rank: 7

Paul

Moderator

posts: 344

Registered: 2011-9-17

Message 16 of 27

01-06-2006 10:20 . pm   |   View his/her posts only
Hi Chris, while you were sleeping (GMT +12) (I am only assuming Chris sleeps - no evidence whatso ever) I looked at the jpg. It's just beautiful.
Cheers

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 17 of 27

05-06-2006 02:51 . pm   |   View his/her posts only
What would really be nice is if you could get a surface tangency or curvature continuity with a bi-rail loft that would allow for a radii to fall.

This would make it easier to take 3 open edges and seal a problem area in fewer operations.

The loft command gives the ability to have 2 of the opposing edges hold tangency but many times this falls in place quite a lot different than what the 3rd edge is (even if the along edge option is used).

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 18 of 27

05-06-2006 08:32 . pm   |   View his/her posts only
Hi Paulverisor

Could you provide a couple of small example models where tangent bi-rail lofting would be a better choice than current options (fill, mesh and patch)?

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 19 of 27

06-06-2006 11:04 . am   |   View his/her posts only
I will come up with something simple.

Typically however the geometry that I work with that would exhibit what I'd like to show isn't that simple, there is lots of curvature.

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 20 of 27

06-06-2006 05:56 . pm   |   View his/her posts only
Hi again

Representative is what we need - if that means complex, that is OK. It just wants to be small - i.e. we do not need to see an entire model, just a region, a collection of faces, the "nitty gritty"

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 21 of 27

08-06-2006 11:11 . am   |   View his/her posts only
This would be a representative example of where a birail loft with tangency on the edges would be helpful.

Our customer model includes a simulated flashland and overedge radii. In order to actually sink this in a die with the true overedge radii we have to remove the entire simulated flashland.

What we end up with is a drafted gap that needs to be filled. The way I do this is with a lot of untrimming and extending. In some situations we can not untrim the existing geometry because it veers away from the other surfaces that are extended. (Whether this is the case with this example I don't know, but I want to just give you an idea of the application that gives us problems). When this occurs we delete the surface which leaves 3 open edges. Downstream this has to be subtracted in a die with an overedge radii so the replacement surface has to have tangency or continuity along it's edges. There are other ways around this type of problem but it would be so much simpler if the replacement surface could be tangent to the 3 open edges.

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 22 of 27

09-06-2006 06:21 . am   |   View his/her posts only
Hi Paulverisor

That is a very nice example of your problem. It might be the case that a VxDal program would be a help - combining the operations you use to trim-extend the faces and trim back to the parting face. I can see that a variational sweep would deliver a perfect join, both sides of the parting line at once, replacing the existing faces altogether (new faces start tangent to existing fillet edge [fillet edge = overedge radii]), but because the edges are facetted, there would be some preparation work to do to make the sweep work - a similar amount of time to your current solution.

I am assuming that the faces representing the flashland plus the fillets to those would be removed all around the entire object. There is a factor that makes things extra tricky - your parting face does not intersect the center of the flashland all the way around the part - is that a typical scenario?

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 23 of 27

09-06-2006 09:33 . am   |   View his/her posts only
Hi Chris,

Yes the entire flashland/radii combination does get removed around the entire model. To save time I patched in our existing model in the areas that were not changed. Regardless, just to get to the point where the model is sealed up at the parting line took me 11.5 hours. (Normally to seal up a customer model only takes a few minutes)(Of course this time includes doing the model twice as the customer changed their model a couple times after I started working on it).

I'm not sure how the variational sweep command would work in this application, and I'm also not familiar with VxDal.

Do you see why I need tangency with the 3 open faces in my example?

I do fairly well with the VX software in getting my work out, but there are many tools or techniques that I'm sure I don't understand. I like being able to post here because it allows me to challenge the way I do things.

To answer another question of yours. This model is indicative of only a small percentage of how we receive of our customer data. It may be true that the parting line does not intersect the exact center of the partingline all the way around the part. I borrowed the partingline sketch from a couple of iterations ago. When I untrimmed and trimmed back the new model I did notice that some geometry intersected the exact center, but other geometry intersected somewhere else.





Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 24 of 27

09-06-2006 11:44 . am   |   View his/her posts only
Hi Paulverisor

It is often a good idea to ask the customer if their CAD system is history based like VX. If they are changing the model anyway, it would take them very little time to suppress the flashland feature and they could then deliver a superior model to you.

I can see your need for tangency all the edges. However, part of this is down to how the problem is tackled. You are working on one face at a time. I would be looking at VX methods that work with multiple faces/edges at the same time as this will normally retain edge relationships between adjoining faces and of course would be quicker. The difficulty in this example is the facetting of the face edges - i.e., where it really should be one continueous edge of one face, the customers software has produced multiple edges. We see this a lot from ProE and PowerShape in particular - these programs use a facetted method to describe trimmed surfaces (faces), unlike VX which uses a superior analytical description (Class "A"). You can usually clean these edges up using the commands "delete tiny edges", followed by "concatenate edges".

I am going to request a bi-rail loft tool that holds tangency. When it arrives, it probably will not solve your problems in many cases and that is because of the relationship between the edge profile to be swept and the edges to be used as the rails - if there is a lack of tangency at the start point, the result will not be as you require. This is true of your example case. None the less, it would be better to have this extra power than not have it. Clearly, it will be a real challenge to code so please be patient with us.

Alternative individual face approach would be to insert a blend curve to close the edge gap. Blend curve will match the curvature of the edges it joins. Thus, you have a four-sided gap and more options on how to produce the face - both the mesh patch and fem patch methods have tangency controls to play with.

Rank: 1

Paul

Newbie

posts: 0

Registered: 2011-7-12

Message 25 of 27

12-06-2006 01:27 . pm   |   View his/her posts only
Chris:

I did ask my boss above my boss to check if we could get a model without the partingline crap. He may or may not have checked with the customer, but regardless he asked me to compile the extra time that it took so hopefully the customer knows how much of a detour this causes.

I would love to be able to solve the whole problem all at once, but as you have said things just don't come together nicely.

I think my aversion towards using blend curves is that typically they want to reattach themselves to a completely different location if you go through the history tree again.

Thankyou for pitching for a 3 sided option for me (as in the Bi-rail). I can see where that might be a bit complicated. Especially if you consider the solid birail options, which would be wonderful to have as well.

Rank: 7Rank: 7Rank: 7

Paul

Moderator

posts: 344

Registered: 2011-9-17

Message 26 of 27

14-06-2006 06:42 . am   |   View his/her posts only
Hi Chris,
I'm outside the loop of the issues you are dealing with, BUT.
When using tangency controls on the VX dialoge form there is usually a slider . Some controls, multiple sliders. Is there any reason why there isn't a tangency value associated with this? I have had a number of occassions where being able to reproduce the same tangency was required. With out a value it aint parametric! e.g I cannot reproduce it reliably.
I dont need a value controller, although a connection to the Step size might be nice. Just an indication in suitable increments.

Cheers

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 27 of 27

14-06-2006 11:36 . am   |   View his/her posts only
Hi mudcrab

With the cursor on the slider control, the tan value is reported in the tool tip. I agree that the slider controls are not helpful when you want to quickly match one of several values in some of the options forms, though some do of course offer the chance to save all values to a bundles file.
See also
X