Other Applications "Reading" Surfaces?

[REDesigner09]

Hi,

I have a turbine blade and the airfoil concave surfaces or face has a very noticeable questionable vertical surfaces or faces in the middle of it. This is most noticeable when Shading with Edges is on & of course, when shading without edges is off, this area looks perfect.

My question is that is CAD model is considered a "validated" model, which will be handed over to ANSYS operators & I think CNC operators. If they inspect or analyze this model, will they also see these questionable surfaces or faces?

More importantly, how are these questionable surfaces / faces affect end ANSYS analysis & CNC programming or any other CAD model analysis?

If this is an issue & not just an appearance issues then I need to justify the "risks" or "ramifications" to the Design Engineer & Management that not fixing this area will cause ....???

Thanks

 

[REDesigner09]

Hi,

I attached a picture for reference. What is your opinion if this "validated" CAD model is used for ANSYS analysis, CNC Programming or any other application?

Will it affect the quality of the product if these various applications can "read' these questionable surfaces, faces or edges? Or is it just a visual thing that will have minimal to no impact towards quality?

Thanks

 

[Hobbles]

Hello REDesigner09,

Could not view the image attached.

CNC machines or ANSYS will read any model you give them. However, they are not designed to predict what your intent is. A CNC machine will operate based on the coordinates of the model, and will machine, cut, drill, etc based on what you tell it to do.

The basic rule: "Garbage in, Garbage out" applies.

I assume a "validated" model is one that meets your specifications; you might as well not analyze it otherwise.

Hobbles

 

[REDesigner09]

Hmm,

Not sure about the picture, but here's the same pic with a different file name.

I also agree with the GIGO rule of thumb, but wanted to get some other perspectives, particularly when this CAD model goes into other applications, such as ANSYS or CNC / CAM programming & how these questionable surfaces will affect these or other applications.

For the drawing, it would appear & be a pain to manually edit out or just wouldn't look good.

Thanks

 

[Jabberwocky]

What is "questionable" about this surface to you? Do you think it lacks tangency? Curvature continuity? Is it inverted? Are there discontinuities? Holes? Downstream packages will pick up on those issues and it can result in errors or at the very least increased processing/machine time.

If the model made it through "validation" then assumably it is sufficient for those downstream tasks, or your validation processes are insufficient, we can't answer that stuff for you here.

 

[Enginerd9]

It looks to me like the view resolution and accuracy of your set up is pretty low. What you may be viewing is a numerical aliasing in the mystical magical math engine driving the surfaces, etc. If you increase the accuracy of your session, and increase the acuracy / resolution of the model, you may see an improvement.

Point is, it may be an optical dellusion. I suspect your "finished product" would come out looking smooth from the mill.

Of course, I've also learned that ProE doesn't like to do surfacing with only 3 edge / spetch references. It wants a completely bound, even number boundry when creating blends and such. I don't know how your model tree looks, but if you've got surfacing and blending you may also have some weird ProE fails going on with tangency at surface joints.

I did an exerisze where I created a elliptical pill shaped surface model. What I found was I had to trick ProE into creating perfect tangency closer to the ends. Some ProE guys call it a "toupe method"... and the process is quite convoluted. You start with a trick, create a 4 boundry blend. Cut out the potentially faulty sections with a trim. Then re-insert another boundry blend using the newly created, finite 4 boundry condition. (I made the boundry blend assumption because with things like fans or turbine blades, it seems like the mose common modeling method in ProE.)