Lofting from intersecting edges

[GeneralFault]

I am using the latest version of Fusion 360 to try to create a wingtip that has a fairly organic shape. I have been attempting various lofting methods with no success. As the wingtip has multiple complex curves, The profiles are at right angles, and there are two intersecting edges.  Honestly as a beginner, I'm not certain that I am even taking the correct approach at this shape, and I could really use some guidance.

[TrippyLighting]

I believe your approach is correct, but perhaps the finicky loft tool has a limitation. I've not gotten this to work either. Lets see what the lofting experts say:

@cekuhnen @michallach81

[cekuhnen · Accepted solution]

@GeneralFault @TrippyLighting

thats a tricky shape - rounded tops are hard

2 approaches:

1. many cross sections at end

2. one rail at tip only and I build it as half and then mirrored it (check timeline for steps)

but the quality of the surface transitions at the tip are not the best if you consider technical surfacing or Class A quality.

[GeneralFault]

Thank you @cekuhnen & @TrippyLighting I greatly appreciate the help! 

@cekuhnen, that looks very close to what I am attempting to accomplish. However there are a couple of critical differences. The trailing edge should extend out and down. Accordingly the top and bottom half are not symmetrical (though I think I can see how to do the asymmetrical profile from your first example). This was a defining feature on this model of airplane, so I am keen to replicate it as closely as possible. 

I've read from this thread that I can in fact loft from an edge (is that what you did here). However if you see in my file that the edge "wingtip root" can only be selected as a single piece. I see in your example that it is split into a top and bottom half. Please forgive this very beginner question... how did you split it into two edges? I suspect this is a core part of my problem. That and perhaps the existence of the trailing edge.

-Cody

[GeneralFault]

@cekuhnen Using your excellent example, I was able to accomplish exactly what I was looking for. The problem it seems was that I needed to bisect the airfoil to create two edges from which I could loft. It is at least good enough for my 3d printer. I've included the wing in case you are curious as to what I was going for and if you have some advice for doing it better (the trailing edge is not straight for example, but has an interesting and acceptable curve). 

[davebYYPCU]

I had a look at your wing tip lofting now that I am home again, Good job really.

Couple of heads up, as my 2 cents.

the tip profile has caused a small amount of undercutting nearer the outside front, that can be filled no worries, but there is also a small amount of concaving on the bottom surface loft - generally near the back of the leading edge stock intersection, call me picky but that area would need some fill as well in my model.

To see this better I used the section analysis on YZ plane and pulled it forward to about 161mm,

May I suggest another rail profile up near the leading edge stock position,

However if you're happy, go with it.

Browny

[cekuhnen]

@GeneralFault Glad this helped. Your file had some feature issues so I quickly build a demo scene

[GeneralFault]

I would be surprised if it didn't have a few issues. I am grateful for you assistance.

There was no attachment on your post. Am I missing something?

[cekuhnen]

@GeneralFault I was refering to the first files I uploaded for you. The process is the same.

Fusion creates a rail surface with the loft - which coming from Alias with technical surfacing background is not ideal because it is not considered mathematically very smooth.

However I not sure if this really matters in your case as we are not talking about a car body.

Such rounded end shapes I normaly would build this way and you can see how labor intensive this is.

So I am very happy that such shapes Fusion loft can do actually quite well.

https://www.youtube.com/watch?v=PRbf_eGHWtc

the first example is similar to what Fusion builds (note the poles with many CVS!)

Go to min 8 to see how I would have to do it in Alias ...

[davebYYPCU]

Thanks Claas,

That video explained a lot if my troubles with early loft results, I was looking for section 3, results from Section 1, inputs.

Now the light is on.

Browny

[GeneralFault]

@davebYYPCU

Thanks for the pointers. I'm overwhelmed by the positive support I've seen among the Autodesk community - even for us beginners. 

I see your point about the undercut at the tip. I assume this is what you are talking about? It is well within the tolerances for my purposes. But if I wanted to fix it, what would I do?

 

The concave at the back of the airfoil is part of the JM-3511ST airfoil. I was a little surprised by it as well when I noticed it, but I had to do quite a bit of fiddling to try to get it right. You can see the profile in the "Airfoil" canvas. Unfortunately at scale, there is a bit of blur in the canvas image making the exact profile hard to nail. But I think I'm close enough for a model airplane.

 Thanks again for your help.

[GeneralFault]

Thanks for that interesting video. I need a few minutes to sit down and really take it in. I have to make some changes to the nose of my model (to make it less razor sharp deadly) and the information from this will certainly come in handy.

Similarly, I think I need to spend a little time going over the examples you sent to understand the differences to your approach. Fortunately I have a trip for work that will involve many hours in an airplane and in a hotel room in which to dig into this properly.

What little I've learned so far makes it clear that setup and structuring the relationship between constructions, sketches and features is critical. Mistakes made early on in the project can have large impacts further on.  But I'm finally learning to be able to go back and correct some of those mistakes without backing out (ctrl-z) all the way through a lot of work. That timeline feature is useful!

-Cody

[davebYYPCU]

Gooday Cody,

Your screenshot was not clear to me if you found what I found on the outside front of the wingtip. 

(If looking from the model nose front on but towards the tip,)

I edited your model in that area, and got a respectable result, by adding two more rails right up the front, one at 155, and the other at 162mm

There is still pinching going on right at the front, the actual alignments used don't help with a sweptback design of the tip, but now it looks better to me.

Open the file, and you will see the two new rails, delete/add them from the loft selection, to see the difference between the two.

I didn't see any other problems in the rear area of the airfoil, that looks all ok. I was going toi screencast it, but I learn more by rolling the timeline up and down, opening the dialogue boxes in "Edit Feature" to see the selections, sizes and options used, etc.

Hope that makes it a little more clearer, and will save a bit more bog.

Browny

[michallach81 · Accepted solution]

Hi Cody, I think that after all these advices you know much more, about surfacing in Fusion. Nevertheless, I would like to show you, how to deal with such a shape without changing your initial data. There could be several approaches, each have its flaws and pitfalls, but usually, quality of surface depends on quality of curves and topology (by topology I mean which curves your able to define. As you notice, all guys did change your initial curves, add some, remove some. These different curves will create different surface flow).

Before I'll start reviewing some of possible ways you can take, I have to say that with out whole wing design, my help is just a guide.

First step is to recognize which part of our design is most problematic. In your designs there are to places I've marked:

Point and a line creates problem for loft to succeed. There are two most reasonable approaches to consider.
One is to loft "cross section" (dot to curve; curve to curve; curve to line)

Second, to loft "guids" (red to green; green to blue)

For both examples "line as a profile" is the biggest problem (unwanted resaul or failure), and "point as a profile" creates bad geometry.
In first example loft tool is trying to match endpoints of selected profiles, which creates turn near "line":

In second example, loft can't use as guide in one direction, and opposite in the same time:

Problem with point is less urgent, and in worst case we can live with it.
What's next? To avoid "line as a profile" problem, we have to loft top and bottom separately:


That method have one main flaw, a cross outline edge we don't have curvature continuity maintained. Because our profiles (curves in sketch 2 and 3) are smooth surface is smooth only near them:

The best method would be to create our wingtip out of even more patches.
First we need to build surface in the middle:

This way we've build primary surface (surface that uses only user defined data; no data based on approximation; undegraded surface).
Next step is to build the "nose". There are two way to do it:

I don't like this method, but it's not bad.
Second one:

This one is better because we could match curvature if we would have rest of a wing.
Last is a tail, but we need to make it in two steps, top and bottom (because of "line as a profile" problem). In order to do that we have to split your primary surface (we need to profiles from one edge):

There was some trick behind the scene, but I'll will record screencast also.
Now finally two lofts for tail:

At the end, we could have ask two questions. Why I've picked Tangent not Smooth for surface continuity? Briefly smooth, applies more extreme values and as results we get more extreme shapes on smaller area (very briefly, wobbly surface).
Next question, why we've started from middle part, instead of "nose", for example?
When we create surface just on curves it's a primary (clean) surface. When we build surfaces using data from already existing surface engine need to approximate certain data and results are slightly degraded. These data are more degraded if more data need to be counted (like curvature continuity). This is why we call those surfaces secondary (less clean). We have also tertiary surface (dirty), those are build no data that is already just an approximation. In our case "nose" is secondary surface, if we would build it as first, we would have to build middle part as secondary (degraded) surface. Problem is that, "nose" even as primary surface is bad, because surface converges at one point. That "quality" will be "reflected" in the middle one.
In the end we would end up with both of them being not very clean.

Certain thing I didn't explain, you can have some troubles with handling curves, but please watch video below, and if you will have some question, ask.

[cekuhnen]

@michallach81

I was considering this approach as well - but Fusion tools cannot go this well because the loft command cannot use G1 or G2 along rail edges.

In your model a lot of the edges are G2 and some only G0

In the showed process I am not even sure what surface continuity you would have where the tip flows into the wing body.

I wish the loft (with rails) and patch tool would support the same edge surface continuity options I have in Alias.

Then such quite complex shapes would be technically a lot cleaner.

[cekuhnen]

@GeneralFault Why is it ok for the end of the wing top to be sharp as a line? 

Besides this making it hard for the modeling software you also will end up with very thin geometries.

Could that end line be a very small U shape so that the loft will be happier? We can still later patch stitch and nicely fillet the hard edges or even use the patch with g2 to fill blend it.

[davebYYPCU]

HUGE Thank you Michael,

You have been able to show me why loft does what it does, the theory, beautiful.

Let's say I could drive Loft, self taught, but did not understand some results.  

Certainly, start in sections, work towards bad areas is excellent advise and is why I had so many problems in the past.

I did not change Cody's original data, just added 2 rails.

You have missed,

that the design at rear trailing edge, should be a curve, similar to the original lofts from Cody, he didn't have a profile there but your straight edge line is minor particular.

in the file I reviewed, the main wing panel was not visible, but was there.  My Rails were constrained horizontal and vertical to help the transistion.

I will go back and review why you drew and then deleted the short lines from the data points, were you checking for point project?

Big thank you I have learned a lot from your responses.  Sorry if I have been unhelpful.

Browny.

[cekuhnen]

One can also build this without breaking it apart into multiple segments.

In this try I used a U profile for the loft end instead of a flat line and it seems Fusion with the way how the loft code works has some issues

because the surface somewhat self intersects sadly.

But I also roughly rebuild the profiles so this could be due because of my curve network.

[michallach81]

Hi guys,

In my post I didn't explain how to split curves. In a screencast you can see how I have left parametric environment and I keep working in direct environment. Some of you also notice that I've draw some lines in order to split curves. 

First, I had to stop record design history ("leaving parametric environment"; Do not capure Desing History tool ) because in paramertic, features are ordered. If I would try to edit "Wingtip Root", "Sketch2" and "Sketch3" would disappear, as they were created after "Wingtip Root":

Only way to be able to edit all sketches while all others are visible is to have them all in direct modeling environment, either by turning off history, or by creating Base Feature, and then developing them insiede Base Feature. Of course, nothing comes for free, while we can see other sketches, any reference will not update when we'll make changes.

Good practise would be to develop sketch network in parametric environment, then turn off history, split curves and make lofts. When we know which method will suite best our model topology, then recreate all sketches in parametric, but this time adjusting sketches/features order, to the knowledge we just gain. A bit laborious workflow, and not suitable for Bottom-Up designs, but should be quite good for Top-Down approach.

In my screencast I didn't care about workflow, I was just looking for quickest way to split curves.

Now it's time to anwser your @davebYYPCU question, why I draw those short lines? In Fusion we have Breake tool in sketches. In order to split line, we need some sketch entities to cross our line, split will happen exactly at crossing point. Take a look (note red X icon):

There is one more thing you should know about Break tool, especially useful in this Wingtip model. To split closed profile you need to provide at least 2 crossing points:

There are many other gems and traps in Fusion sketches, but I don't have time to cover them, also it would require separate thread/tutorial to make (and if anyone is curious my firstborn son just born, so I'm occupied in 100%. All answers, I do post while I'm at work  )

@cekuhnen

I would like to point one more thing, you've post an image with some curve combs.I don't know if you aware of it, but break in curve comb for that middle surface does not indicate G0 it's in fact G2. I'm not saying that you don't know that already, but I would like to explain it for those who are confused when they see that.

When we split surface or curve they remain smooth. When we apply curvature comb, even if we pick both splitted parts, calculations are made to them separately. Exaggeration of scale depends on proportion between radius on a curve and curve length. Because these proportion usually are different for each side of split, we can see it immediately. Sometimes proportion are similar (like on a fairy flat curve) and curve comb looks good until we scale it up. Here is a simple example:

When proportion are the same everything look ok. One more example with big mismatch of these proportion:

Of course I don't claim that after split everything is perfect, it is perfect up to Fusion kernel tolerance. Because every build on splines and brep will be only approximation, and that could cause problems (this is usually main reason why solids in different CAD could come as set of surfaces).

@GeneralFault

At last I have to tell, that even if I build wingtip without changing your curves, you should consider such a change. I don't know how whole model looks like, but when we have tool that has some limitations, we should consider some adjustments to our initial plan.

[cekuhnen]

@michallach81 Duh - I forgot that Fusions curvature comb is useless when you scale too much because it breaks the individual segments scales.

Thanks for reminding me about this. We should put this up as an idea.