Parametric spline modification (rotation/translation)

[giulianoj.zennaro]

Hello,

After multiple searches on the forum, it appears to me that there is no way to parametrically edit a spline. The scale tool is the only one that is usable, whereas there is no way to move the spline in a rigid manner. Constraining one of the points of the spline to something which is parametrically defined in space also does not work as the spline is not updated properly when the parameter is changed. 

This issue seems to affect few people, however it has been the case for years with already several people mentioning it and no viable workaround. Here is an example of a post clearly describing the issue with images : https://forums.autodesk.com/t5/fusion-design-validate-document/how-to-fix-the-relative-shape-of-a-sp...

In my specific case, I need to design a wing parametrically since I want to export multiple step files with slight geometric variations for CFD simulations. One of the parameters I need to modify is twist and so the profile (spline) at one end of the wing has to be rotated parametrically. Another parameter I have involves the translation of a profile. Obviously, being an airfoil, the splines are made from far too many points to manually constrain them all (I use airfoil DAT to spline).

Have there been any updates on this necessary feature, new viable workarounds or a separate possible workflow for my specific use case ? 

Thank you in advance for your time and suggestions !

PS : for the twist variation, I have a possible workflow which I still haven’t tested. I could draw one profile at each end of the wing and multiple construction lines representing the chord along the wingspan. The construction lines can easily have their angle changed from parameters and I can connect them all in 3D with lines or splines which then can be used as rails for a loft. For the second profile, I can then manually rotate it each time, export the model and then rotate back to zero before choosing the new desired twist. My issue here however is I’m not sure I can select a curve which is composed of multiple splines or other curves, even if they’re coincident and tangent (another major issue I have run into), as a rail.

[giulianoj.zennaro]

Quick update on the situation.

I have had issues mirroring the winglet and so I ended up remodeling the mirror side of it and lofting again. All winglet parameters seem to update fine.

Now the issue I have left is with the twist parameters. Most of the time, when changing them the winglet becomes broken (up to right before making the winglet, everything is acceptable). Now I think the reason for my issues related to the winglet is the way I sweep the wing. Since I am using a non-vertical path, the surface I work with is not horizontal. Maybe this is not the only issue (or the issue in the first place), nonetheless, I want to solve it because the current geometry has lots of minor issues related to this.

Would there be a way to sweep with a vertical path and then use 2 guide rails ? What I would like geometrically speaking is to have a naca0012 profile, remaining flat and parallel to the ground, move vertically and scale+rotate to have both ends of it follow the construction line ends (rib ends). I would prefer using straight lines (non tangent) between each rib, but I can accept a spline like it is now.

I heard it is not possible to select a path + 2 guide rails. Is there a workaround ?

[davebYYPCU]

Loft will accept many Rails.

Loft requires at least 2 profiles.

Not sure about the new criteria will check file bit later on.

Might help…..

[giulianoj.zennaro]

That's the issue with loft is the two profile requirement. I have been trying to think of ways to achieve that, but it brings us back to the initial problem of having no way to update the spline composing the profile. Also not sure the profile would be kept flat (symmetry requirement along a vertical plane passing by the chord line).

Feels a bit strange to no be able to select two rails for the sweep. That would solve the problem. I think I'd have everything sorted after that.

Thanks again for your time !

[laughingcreek]

---

@giulianoj.zennaro wrote:... Now I think the reason for my issues related to the winglet is the way I sweep the wing. Since I am using a non-vertical path, the surface I work with is not horizontal. ...

---

as I mentioned before, if the start and stop of your path is constrained to be vertical, you will end up with a profile that is parallel to the origin plane.

[davebYYPCU]

Are you saying the current LE and TE spline curves, for a single sweep, (fast improvement over this request)

should be segmented for 24 straight lines, and then 12 Sweeps, for the faceted construction output?

(If so delete the curves, and draw in the segmented straight lines.  A Sweep from the preceding rib (selecting the body face is available) as a join, (12 times) will avoid the 2nd profile a Loft would require. Almost the same as your original file)

Can you explain how and why the winglet mirror was unsatisfactory or not working?

I note that the rib twist angles are nominated and not from a formula, which is what I thought you were after.

[davebYYPCU]

Both the LE and TE curves are 3d and so it does lean the profile over a tad.

Might help....

[laughingcreek]

---

@giulianoj.zennaro wrote:...That's the issue with loft is the two profile requirement. I have been trying to think of ways to achieve that, but it brings us back to the initial problem of having no way to update the spline composing the profile.......

---

there are a number of ways to parametrically place the second upper spline profile. 2 possibilities come to mind (which I have used before).  probably other ways.-

1-create a surface patch of the spline profile and parametrically move or align it. (in a previous post you were using the first move type in the move dialog.  it's not parametric and won't keep a parameter.  the other ones, how ever, are parametric for bodies, but not for components.)  surface bodies can be used as profile inputs for sweep and loft

2-place the sketch of the spline profile in a component.  joint the component in place.  easiest would be to joint it to a sketch article (i.e. the cord you've already created).  the whole component, along with the sketch, would then move parametrically when the cord is moved.

I think to use a loft of this type, you would need to use 4 rails to help the loft keep it's form and not collapse a bit along it's length, which would defeat the purpose. 

[laughingcreek]

---

@davebYYPCU wrote:

Both the LE and TE curves are 3d and so it does lean the profile over a tad...

---

only when the ends are unconstrained.  constrain the beginning and end  of the sweep path to be vertical by applying a vertical constraint to the tangent handles, and you can make the profile end up parallel to the origin plane, even at extreme twist- 

[davebYYPCU]

Can't be vertical, pivot point is ahead of the Path.

[giulianoj.zennaro]

@laughingcreek, the issue is not having the ends flat, but the whole sweep process being flat. One way to think about it is sweep with a vertical path and two guide rails. Another way to think about it, but I did not find a way to achieve this, is to do 12  individual sweeps with a vertical path, a guide rail (nose of the profile) and a twist angle. It seems I cannot use 2 guide rails and I cannot apply twist if there is a guide rail.

A faceted shape is fine for me and I would be more comfortable dealing with straight lines rather than splines (not a requirement). I have tried doing multiple sweeps, but the issue is obviously the paths are not tangent to the surface... Anyway, this wouldn't solve the issue of keeping the sweep profile flat.

The issue with mirroring is that since the top surface is not flat, it is also not symmetrical along a vertical plane passing by the chord line at the winglet base. Since the winglet is meant to be symetrical, the base obviously has to be as well.

The twist angles are indeed nominated. I have a python code that calculates those angles since I cannot directly insert my formulas as one smooth non-linear twist (btw, the formulas would be really long anyway). Having discrete intervals which are linear is fine by me.

[giulianoj.zennaro]

Would a loft with two guide rails and a straight centerline (for example the hinge line) work ?

I have to admit, I don't know exactly how the loft command is geometrically defined, but if the centerline imposes the profile to be flat and the guide rails serve only to guide rotation, translation and scaling, then we should be good right ?

Then for the second profile I can use the same technique with a joint as for the winglet. This then leaves the question of tangency of the rails...

[davebYYPCU]

Would a loft with two guide rails and a straight centerline (for example the hinge line) work ?

No.  Loft requires all rails to connect to all profiles. Segmented rails must be tangent connections.

The stand alone pivot centre line, is now ahead of the profile so it cannot be used.  

Sweep - can be Path and Guide rail with scaling, needs just one profile.  This profile will roll with a curved Path orientation, and therefore does not stay planar to the Offset planes.

Loft requires both (can be horizontal and twisted) profiles scaled to size.  No rails will give point to point straight segments.   Curved rails will work.

I believe we are full circle, Loft with scaled profiles -is the workflow to get what you are requesting.

Might help...

[giulianoj.zennaro]

Thank you very much ! I'll try that as soon as I get access to my computer again. 

I'll have to use joints onto the current chord lines. Not sure how to insure that the rib joints will be colinear to the the chord and not the origin. I'll figure it out like for the winglet....

Quick question about the center line. I didn't get to try it out yet, but I heard it does not need to connect to the geometry. In that case would the hinge line not work ? Or actually it may not be needed if all I need is straight segments ? 

[davebYYPCU]

Centre Line will not be needed if you align the profiles correctly.

You just need a sketch centre line in the profile (Paste New) components to use and make colinear with the chord lines, then scaled to the same size.

Might help...

[giulianoj.zennaro]

So I finally got the time to get back to the model... I did everything as you suggested with the loft and it seems good !  For some reason, two artifacts remain.

1) I seem to have the same issue as in the past to combine all of the bodies together after mirroring even though there is clear overlap. I suspect it is due to edges that are very close to overlap but not exactly. I see that from the small faces that are created. Actually, that's something I have to sort out because small faces and edges cannot be meshed properly and the geometry must first be repaired.

2) A strange error is given when I change the twist parameter. In the initial loft of the winglet, it says "rails cannot be tangent to profiles". I assume that has to do with the tangency constraint I give to the profile, as when I use "connected" instead, it works. Not quite sure what the difference between the two is and what the best practice is.

I finally feel very close to the end of the model ! Once these two slight issues solved, I'll try to post something summarizing how we tackled the initial problem and click accept solution on that post and your relevant posts for the next users. What do you think of the current workflow ? (I included the file)

Thanks again !

[giulianoj.zennaro]

Glad to be able to say, I figured out the two remaining artifacts in one go myself... The tangency constraint in the loft is impossible to follow since rail 1 is not tangent to the curvature of the trailing edge. Solution is either to make it tangent or remove it. After making it tangent (project curvature of the arc and a spline matching the tangency requirements), for some reason the issue remained. I didn't investigate any further and just decided to remove the rail as it seemed to give good results nonetheless.

For whatever reason, that also solved the combine issue ! Also, I cannot figure out why the tangency issue was only raised after changing the twist parameter and never before since twist shouldn't change the underlying issue. Who knows...

Now everything seems fine other than there's a useless edge on the trailing edge after the mirroring. Might just remove that in ansys. I'll try to find time to mess about with all the parameters and see what works and what doesn't. I know for example that a twist parameter of 0 breaks everything. I can live with that though. Thanks again for everyone's help ! And special thank you to @davebYYPCU for teaching me new workflows and tools !