Compound Blend

[VitaminM]

I am modeling a compressor blade in NX4. The blade calls for a "compound" or "eliptical" blend between the airfoil and platform. This means a small radius blend grows into a large radius blend between the two surfaces.

This is different from a variable blend where the user can select points along the edge and specify a radius.

Does anyone know how to do this?

 

[Xwheelguy]

Take a look at Face Blend and see if that will do what you want. Using Face Blend, you can specify the cross sectional shape (circular, elliptical or conical), as well as make it variable.

The documentation should help you with all the options on the dialog.

Tim Flater
Senior Designer
Enkei America, Inc.

http://www.enkei.com

Some people are like slinkies....they don't really have a purpose, but they still bring a smile to your face when you push them down the stairs.

 

[hudson888]

You can create a law conical face blend which will in effect have an elliptical section, and drive in along an edge as the spine curve. To create a variable blend use the law curve option as either conical or linear. The system allows you to specify a setback rather than a radius size. Depending on what you mean by the radius you may have to calculate it in terms of setback for yourself. To me the idea of setback in this application makes more sense.

Cheers

Hudson

 

[jerry1423]

Also look at soft blend.
It will do the job but when using that you will need to specify the tangent edges of the blend - sometimes they are easily put in with curves (and sometimes not so easy).

 

[Dave K]

Check out: Face Blend - Isoparameter Type.

If you look in the help files, it was designed primarily for turbine blades.

-Dave
Everything should be designed as simple as possible, but not simpler.

 

[EDCSRPM]

See this thread -

http://www.eng-tips.com/viewthread.cfm?qid=221778&page=12

This topic was discussed there.

 

[VitaminM]

Thanks all.
It looks like the conic face blend is the way to go, but it can't wrap around the airfoil's leading edge. The blend fails here.

Gunman: I see FACE-BLEND... ISOPARAMETER in the help files, but I don't see it in my menus/toolbars. Is this in MODELING or another application?

Soft blend is an option, but I haven't given up on face blend yet. The advantage to face blend is that the model will be defined by the actual dimensions on the print.

TurbBLD: thanks for showing me the other thread. I swear I searched before posting. I never found this one.

 

[hudson888]

I suspect that the radius of the leading edge is too small, or the conical blend too large. Either way it means the same thing. If the tangents are setback further that those of the leading edge blend then the surface may tend to self intersect as it attempts to follow the corner. Either that or the leading edge not being a true radius approaches being a smaller conic than the one you're applying to the tip.

I'll try to explain.

When you apply a true radius blend using edge blend with say a smaller blend on the leading edge and a larger one on the tip then the software will create a three sided face on the corner. It is as if the larger blend were applied first and the smaller created later on. Try it with the edges of a cube for example and you'll see what I mean.

In face blend the true radius geometric equivalent of the edge blend is termed spherical. That relates to the application of the blend as if it were a rolling ball contact swept between the two faces. The three sided faces you get from edge blends are in fact trimmed spheres, so that is a unique case.

This was historically a problem that required careful ordering of blends in earlier versions of UG. The software has become a little smarter but the range of possible geometry still falls within the same practical limits.

Conical blends pretty much always want to be able to produce four sided faces. In all but a few geometric types, spheres, planes, cylinders and cones, surfaces are described as basically U and V meshes. This can mean that the absence of a fourth side or any condition approaching a focal point where the side becomes too short will be at risk of self intersection. You will almost certainly be facing such a dilemma.

I doubt that you'll be able to achieve the exact result you expected in that case, but I haven't seen your part, so in order to know you'll need to look at the faces with some face analysis tools. In this case using face analysis>radius analysis should tell you what you need to know. I would use minimum radius method, but if the surface is complex try analysing just the U or V direction depending on which way the mesh runs. Select just the face on the leading edge where it won't wrap around to analyse and adjust the range to get a sense of how tight the curvature is at its minimum radius.

You may want to create the conical face blend without trim and attach so that it builds what it can of the surface. Then you can also use that in analysis to compare with your leading edge observations.

You may succeed if you vary either the size of the conic or the amount of blend setback to the tangents. Otherwise it is up to you whether you choose to vary the leading edge geometry or simply build a mesh surface that nearly enough approximates your desired result, and attempt to apply the result as a sew and patch.

Good luck with it.

Hudson

 

[VitaminM]

Good info Hudson, thanks.
After monkeying around with this, I stumbled on a "solution".
The platform at the base of the airfoil is cylindrical, so I had modeled it as a cylinder, to be trimmed later.

For some reason, the face blend could not work with this cylinder. (I think it has to do with the Normal Direction arrow appearing at the bottom of that cylinder face - 180° from where the airfoil is). However, if I only used a partial cylinder there was no problem at all. The attached picture should clarify.

I guess this is just one of those NX quirks.

Thanks again to all who helped. I will refer to this thread in the future.

 

[hudson888]

Who'd have thunk it.

Just a tip to mention that a smaller Jpeg image would have done nicely.

Cheers

Hudson