sweep or loft comand for bladed part?

[eng1234]

I'm trying to create a bladed part through three specific cross sections and having some difficulty. I have three closed cross sections at various location in space and want to create a solid wing shape through them. The difficulty I'm having is the front edge of the blade is to be straight while the rear edge curved. I have these line geometries created as well.

Using the sweep command, I can't seem to chose more than one line for the sections to follow? Using loft, it generally draws a traight line between the sections, but than not able to have the desired rear curved edge? I have also tried to open up the sections and use connecting rails between the section, but no success yet? Any other suggestions would be appreciated.

Thanks!!

 

[RCDLtd]

The short answer if you wish to accurately define this "blade", which sounds like it is a freeform/organic shape rather than a geometric shape, is to create it using a surfacing system and import it into I-DEAS.

Probably not a helpful answer!

If you're happy with a more approximate shape for your blade then how about:-

Assuming both edges a planar create this geometry in wireframe on a sketch plane and convert to Reference Curves.

Sketch on path on one of these edges, probably the straight one would be easier to control. Select your primary plane to be coincident with one of your cross-sections.

Create your cross-section on this plane Intersecting the other edge (curved) so that you get a blue point to constrain your wireframe to.

Select Variable Sweep and let the system sweep that cross-section along the path, while it is controlled by the curved edge.

Modify the Variable Sweep introducing new sketch planes coincident with your other original cross-sections.

Modify the geometry at the new cross-sections to match the originals.

Check your result to see if it's anything like what you expected.

....That's just an idea, there's a few different ways it could be done, it all depends on the shape you're trying to define.

Good Luck.

 

[eng1234]

Thanks RCDltd,

The geometry is a airfoil shape that changes geometry from the top cross section to the bottom cross section. The front edge of the airfoil is to be straight from the top cross section to the bottom cross section. While the rear edge of the foil to be curved.

I have specific geometry created for each cross section and each is placed relative to each other for the correct position. I would prefer to keep the surfacing inside the IDEAS program.

If you were interested I could send a low res. jpg of the sections I'm trying to create? Probably asking too much, but this is hard to explain. I'm also reading about variational sweep to see if that is a possibility.

Thanks,

 

[RCDLtd]

Now you say an Aerofoil I understand the problem. As the blade has a twist through it, it really is freeform! You will be struggling in I-DEAS.

How about breaking it into two sections, ie Cross-section 1 to 2 and cross-section 2 to 3, then using Loft 2 Rail between each, with the second Loft constrained with Tangeny to the first. You'll have to break the two edges at the centre cross-section.

You'd also have to split the blade through it's thickness, so that you actually build the part in four sections.

If you require tangency and curvature matching (or even just tagency matching) at all surface boundaries then you're just not going to get their with I-DEAS alone.

 

[christoph]

eng1234: Sounds like you're on the right track with Loft, except you just need at least two more cross sections, to obtain the desired curvature definition along the rear (curved) edge. Try Loft again using at least five cross sections, instead of three, and let us know if it doesn't create the desired result. I'd probably recommend sketching each cross section with 2D, planar, fully-constrained wireframe.

If I recall, the rails don't matter much; only the cross sections create definitive Loft constraint requirements. Good luck.

 

[eng1234]

Christoph,

I will try and add at least two more cross sections (fully constrained) to see if that will give me the desired rear curve between sections. I was fearful of adding to many sections and not having a smooth surface between sections.

Adding the rails seem to be a pain since they want open sections and the rail (lines) need to be at the point of tangency of each section.

Here goes

 

[christoph]

eng1234: Now that you mention it, I think that fear is well-founded. Too many cross sections (if they're the least bit off) could make a surface rough. Good point.

I'd like to retract my comment that "rails don't matter much." As I vaguely recall, Loft allows you to have only two rails, the two outer rails. If I recall correctly, Loft is required to pass through each cross section, and secondly will try to follow your two outer rails if at all possible. So now I'm slightly confused on why your loft didn't follow your two outer rails when you tried it with your original three cross sections (?). It doesn't sound like it would have been overconstrained.

BTW, are your cross sections oriented laterally across the blade? Or longitudinally along the blade?

 

[ConX]

I've done something similar with the variational sweep command, but it's a bit*h to set up. You need to do as RCDLtd stated, but you have to set up your constraints very carefully. You need to keep at least one edge completely constrained perpindicular to the sweep reference line. You then have to lax your constraints in the areas you want the curve to vary. You may need to also add another reference lines as a second sweep path.

I did it a couple of years ago in MS7 using variational sweep. I created a compound sine sweep in both the X and Y directions. Too sum it up, here are the steps.

1. Create first sweep line (straight).
2. Create second sweep line (connecting same point on different cross sections).
3. Create reference lines of the two curves.
4. Create a reference plane on the path of the reference curve (the straight one).
5. Sketch your cross section making sure to connect the two intersection points of yor reference curves to your cross section.
6. Add your constraints carefully, as they will decide how the model is created.
7. You will need at least one line fully constrained for the others to follow suit. Ideas will lengthen/shrink/twist the other lines but keeping one line fully constrained will allow the other to vary in a manner you want.
8. Use Variational sweep, use the help file as a reference.
9. If it doesn't work, try changing your constraints.

I can't stress enough that your constraints will make the feature work right or not. It took me a while to get my sweep done correctly, but it was all in the choice of constraints.

Good luck.

 

[christoph]

RCDLtd and ConX wrote:
>> Convert the two sweep path curves to reference lines. <<

(1) What advantage does using a reference line here give you over using just a fully constrained, regular line?

>> You will need at least one line fully constrained for the others to follow suit. <<

(2) When you say "line" in your above sentence, you're referring to one of the lines within a cross section, not the sweep (reference) line. Right? (3) And should that fully constrained line be constrained to either a horizontal or vertical ground? Or can it, as well, be constrained by two pin grounds?

(4) Shouldn't all lines within a cross section be fully constrained, but only one should be constrained to, e.g., horizontal or vertical ground?

(5) What's the main advantage (or difference) of using Variational Sweep here instead of Loft? When should a person choose one over the other? Thanks.

 

[RCDLtd]

1. The advantage of refernce lines over wireframe curves is that the Refernce Lines don't get absorbed into the operation and can therefore be used in further operations saving the time to recreate them.
My choice to use them in this instance was one of habit rather than necessity.

3. Which ever type of constraint captures the Design Intent and Editability that you are after.

4. A Variable Sweep is used when the cross-section is required to change as it is swept along a sweep path. The constraint system needs to accomodate that change and can't therefore be "fully constrained"
Regarding Horizontal & Vertical Grounds, I use these as little as possible, they are not very editable. A focused CSYS axis is my preference.

5. Loft doesn't follow a path, so the cross-section orientation is not controlled. Loft 2 Rail comes close but only allows two cross-sections. Loft is a good for blends where tagency is required to the features adjacent to the Lofted Surface. Variable Sweep is good for Flanges that keep the same "concept" but vary their dimensions.

 

[RCDLtd]

Just thought about point 4.
The Cross-section could be "Fully Constrained" and still accomodate the changes along the Sweep Path. As long as the wireframe is constrained to the item that is varying. ie. in this case and intersect point on the curved edge.

 

[ConX]

(1) What advantage does using a reference line here give you over using just a fully constrained, regular line?

There are two advantaces. First, you can reuse the line as RTD stated. Second, it instantly constrains the line when you convert it to a reference line, this is an added bonus when dealing with complex geometry.

(2) When you say "line" in your above sentence, you're referring to one of the lines within a cross section, not the sweep (reference) line. Right? (3) And should that fully constrained line be constrained to either a horizontal or vertical ground? Or can it, as well, be constrained by two pin grounds?

I was referring to a single line which intersects each cross section at the same point. I then make this line my reference line and sweep along it. By making the sweep lines referrence lines, you don't need to constrain them. This will allow the cross section to follow both curves and vary depending on how you set up your constraints.

(4) Shouldn't all lines within a cross section be fully constrained, but only one should be constrained to, e.g., horizontal or vertical ground?

You will need to have some lines of your cross section not constrained to get variational sweep to work. You should constrain each line differently depending on the area you want to change. For example, say you want to change a box to a rectangle along a sweep path, you would choose to fully constrain the height, but not constrain the width, this would allow the width to vary it's shape along the v-sweep path.

I hardly ever "ground" my sweep path. The system can crash if you do this, instead, I would fully constrain one line, then choose to constrain the other edges to vary in it's form. It is possible though to not constrain anything and still get the sweep to work, but you really have to know what you are doing and it will only work for surface geometry I believe.

(5) What's the main advantage (or difference) of using Variational Sweep here instead of Loft? When should a person choose one over the other? Thanks.

Again, I'll refer to RTD's response. He's right on. With a loft, you can't controll the path, just the cross sections. You can manipulate the path by changing the tangency influence to approximate the path, but you can never get the exact path you're looking for.

Good luck!!!

 

[RCDLtd]

ConX, I've got to be a little "picky" with your statements about constrained geometry.

I personally don't feel that generating reference curves constrains an unconstrained wireframe profile. Agreed, you can't directly modify the Reference Curves without rolling back to the underlying wireframe geometry, but if you do roll back it will still be unconstrained.

I would fully constrain all geometry before generating Reference curves from it.

And while I'm being picky, I'd generally fully constrain variable sweep cross-sections to ensure the "variable" is controlled in the way I want it.

This may be a reflection of the environment I tend to work in where the client's rule book for modelling Best Practices is very long and very precise.

 

[christoph]

eng1234: I apologize for misguiding you. You were right. Three cross sections should work better for your application than five, to give you a smoother surface, as you said. However, I had forgotten that Loft allows you to add rails (right click and select Add Rails) only if you have two cross sections! But you need three cross sections in your application.

Two-cross-section, two-rail lofts will exactly follow the rail paths. For lofts with three or more cross sections, the Loft does not follow rails and goes off course from your "rail" curves.

So I concede to RCDLtd and ConX that you must use Variational Sweep. By the way, Variational Sweep works only if you create each cross section using Sketch on Path! (FYI, what RCDLtd and ConX describe herein for setting up a Variational Sweep is identical to what one would do to also perform a two-cross-section, two-rail Loft. So everything said herein equally applies to the two-cross-section, two-rail Loft, except the two-cross-section, two-rail Loft doesn't require you to use Sketch on Path.)

>> I'd generally fully constrain Variable Sweep [and two-cross-section, two-rail Loft] cross-sections. Cross sections can be fully constrained and still accommodate the desired changes along the sweep [or loft] path. <<

I agree with the above statement by RCDLtd. You want all wireframe in each cross section to be blue (fully constrained). Being blue doesn't mean certain aspects of the cross section can't change along the sweep path. E.g., RCDLtd's rectangle example is fully constrained. Its height is a hard-coded (unchanging) constraint, whereas its width is fully constrained by the initial distance between the rail curves, which changes along the path.

I think I would tend to agree that we should avoid pin grounds, which are unreliable and uneditable. But I vaguely thought horizontal or vertical ground is safe and generic (?). So I'm slightly confused on what is a good, safe, fundamental, primary constraint for a cross section?

 

[eng1234]

A big thanks to everyone since I originally posted this question. I have since been able to break up the three desired cross sections, add a front straight "rail" and a rear curved "rail" and loft the part. I split each cross section of the desired blade along the major axis. After lofting, I now have a thin surface for the right and left side that followed my exact desired geometry. I now need to figure out how to make these surface into a solid. I think I can use stitch?

Thanks again to everyone for their suggestions.

 

[christoph]

But a Loft having more than two cross sections won't allow you to add rails. When you right click after selecting three or more Loft cross sections, notice there is then no longer an "Add Rails" command in the right-click menu. It might look like your lofts followed "rails," but they didn't (and therefore you might find stitching the surfaces together prone to failure). Someone please correct me and explain if I'm wrong about the rails.

I currently can't answer your question regarding how to create a solid model from a closed, hollow, surface model. I hope someone can tell us how.

eng1234, are your loft cross sections oriented laterally across the blade (loft direction longitudinal along the blade)? Or are your cross sections oriented longitudinally along the blade (loft direction lateral across the blade)?

 

[RCDLtd]

Loft 2 Rail will only work with 2 cross-sections so perhaps eng1234 can confirm if by "been able to break up the three desired cross sections" means that he's split each side of the blade into 2 sections. ie x-sect 1 to x-sect 2 and x-sect 2 to x-sect 3.

For creating the solid, end1234 has the answer - stitch.

 

[eng1234]

RCDLtd and christoph,

You are both correc that I ended up only using two cross sections the extreme top and bottom ones. I did however need to make it an open section so I split the cross section laterally across the blade. The rails were then at the front edge (straight line) and rear edge (curved). This turned out to be much simplier than I originally thought. I initiall tried to break it up into section 1-2 and section 2-3, but wasn't getting consistent surfaces that I could stitch together.

Once the two halves of the blade were created (leaving me thin surfaces), I than created a top and bottom surface using boundry by curves. The tops of the blade worked great to chose the boundaries. This allowed me to than make a top and bottom surface that created the desired solid (with the stitch toggle switch on).

Thanks again.

 

[ConX]

RTD, just got a chance to respond. . .

ConX, I've got to be a little "picky" with your statements about constrained geometry.

I personally don't feel that generating reference curves constrains an unconstrained wireframe profile. Agreed, you can't directly modify the Reference Curves without rolling back to the underlying wireframe geometry, but if you do roll back it will still be unconstrained.

You are correct, but in function curves, it is impossible to constrain the curves without grounding them. In my example, I was constraining a sine curve defined in the function command from ideas, so I still feel the best option is using reference curves.

I was also under the impression eng1234 was using similar a similar type profilem to sweep/loft around, so I figured reference curves/sketch on path was the best option for him.

I would fully constrain all geometry before generating Reference curves from it.

If at all possible, I would agree, but sometimes it's is difficult if not impossible to fully constrain curves if you are using functions. In this instance, I think Reference curves are the best option.

And while I'm being picky, I'd generally fully constrain variable sweep cross-sections to ensure the "variable" is controlled in the way I want it.

This may be a reflection of the environment I tend to work in where the client's rule book for modelling Best Practices is very long and very precise.

You may run into problems with a fully constrained wireframe when you try to run v-sweep depending on how the constraints are set up. You should try to constrain the wireframe as much as possible, but sometimes if the wireframe is fully constrained (all blue), Ideas will fail to generate a solution, espically if your dealing with sweeping along a function curve.

I understand your best modeling practices rules, and I think that's a real good thing. Unfortunately, at my company, results are the expected outcome with correct modeling being a distant thought. Hence, I am stuck using a lot of shortcuts to get the results in a short period of time.

I am in no way trying to teach bad modeling practices, I am just trying to get the desired outcome, a product of my work enviornment as well.