Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

Register Log in

Surfacing Challenge: 3 Sided Surfaces

J

jimshaw

New member
Hi folks,


What is the best practice for creating a HIGH QUALITY 3-sided boundary surface? In this case, my boundaries are 3 splines which have been drawn on 3 orthogonal planes.


Below is a snapshot of the results using two methods that I know of. There's a boundary blend on the leftand a filled surface on the right. The filled surface is constrained (curvature) to 3 extruded surfaces, which are hidden from view.


View attachment 5899


The curvature on the left is obviously terrible, but I'm not happy with the one on the right either... Any ideas on how to get a suh-weet surface out of these 3 splines?


Thanks in advance everybody!!
smiley4.gif



Jim Shaw
http://www.linkedin.com/in/shawengineering
 
In the past Iused a sweep to generate a 3 sided boundary. We were creating a rugby ball type shape and it used to turn out pretty well although we didn't do a huge amount of analysis on the surface at the time as it was purely for concepts.


I'll try and have a look at it later
 
The models I'm talking about weren't symmetrical....sorry, shudve gone into more detail, we modelled up an 1/8th of a rugby ball shape.......it was basically what you have shown.


I'll break open the SW later tonight and show you what I mean.
smiley4.gif
 
Hi Jim,


Having viewed the image, I have'nt made a sample model like this butdoes adding surface points on the same plane able to control the curvature? I assume this surface is having non-uniform points. is it?


Ivan who


SW2010
 
Someone brought up a good suggestion with using a sweep.

Draw a straight line going from one end to the other to act as a stable direction vector, in the sweep that'll be your path. Pick your profile and then your two splines as guide curves. This will sweep to a point, now it will make a degenerate surface so be aware of that as it may have difficulty shelling but it should calculate very smoothly.

Sweep takes your profile and rebuilds it every "n-th" times along your path, you may need to constrain your profile spline to stop it wobbling, definitely consider turning on the "proportional spline" option in the spline PropertyManager.


Edited by: Kevin De Smet
 
What Kevin has described is what I was talking about. I still haven't had a chance to look at it yet Jim. Make sure the sweep profileis contrained to the guide curves using the 'pierce' relation.
 
Thanks for the responses!


@jacek: Your tutorials are great. I had already downloaded a few for reference before, actually. I tried to patch up the bad area in the bottom right with a 4-part boundary, but it seemed there were issues, as there was local high curvature at the edges of that boundary.The method was working, but the geometry was not.


@Ivan: I've added as few control points to the splines as possible, so that the surface is not over constrained. I'm confident I can describe the geometry I want with 2 and 3-point splines.


@kevin: I will try the sweep method and report back. I always think that the sweep requires the beginning and ended profiles to be the same, but I guess if the guide curves intersect at the end, the profile essentiall goes to zero, and then that areacan be patched.


@ Michael: Pierce relation. Roger that.


I was able to tweak the curves to get the fill to work. I realized that it's important to select the EDGES of those surfaces, and not the underlying sketches, so curvature can be calculated properly. The issue with this is, extremely small changes in the splines can yield drastic changes in the surface, so this model can't be "pulled and pushed" as robustly as I want it to be. So, right now I have compromised the spline geometry to get a decent surface. It's not ideal, but it's working. Here's the current surface with the same curvature scale as the pics above. You can see the overall curvature has been lowered.


View attachment 5904


You can see the bottom right corner is still weird, but ultimately the model will be cut there, so this will work.


Jim Shaw
http://www.linkedin.com/in/shawengineering
 
Alright,


I could not get the sweep to work, so maybe somebody can show me? I tried adding a straight line from one end to the other, but it did not seem to help out at all. I also don't like how the sweep doesn't give you that many control options.


The loft, on the other hand, showed some potential. Here are three lofts, with each curve taking it's turn being the guide curve, and the other two used as start/finish profiles.


View attachment 5905View attachment 5906View attachment 5907


These 3 surfaces are using the original curves, but ultimately I updated the splines to be normal to each other at the corners, then chose "normal to profile" as tangency constraints, but ideally I don't want the splines to be normal to each other. I've tried using "direction vector" but that always yields some crazy curvature at the corners... also, there are cases where I want to use two direction vectors - one at each end of the spline, you know what I mean?


Jim Shaw
http://www.linkedin.com/in/shawengineering


BTW, Here is that first one with the bottom left corner patched up with a 4-part boundary (highlighted in orange). This the nicest one yet using those original curves!


View attachment 5908
Edited by: jimshaw
 
You must log in or register to reply here.

Sponsor

Articles From 3DCAD World

Back
Top