Swept blends cause slow-downs?

[patricks]

Has anyone tried using swept blends for anything? Have you experienced any system slow-downs with them?

I had the idea of creating a model of the Gateway Arch in St. Louis, Missouri using a swept blend. The cross-section of the arch is an equilateral triangle that goes from 54 feet per side at the base to 17 feet per side at the top. It is 630 feet wide at the base and 630 feet tall at the center.

The shape of the arch is a catenary shape, which can be expressed by a mathematical equation. I found this equation on Wikipedia, then plotted out a few height points at 50-foot horizontal intervals. I then created a mass and used the Swept Blend tool to create one-half of the arch. For the path, I decided to use a spline curve in order to get a curve shape to touch all the points I had plotted in elevation.

After mirroring the shape and finishing the mass (which BTW it would not let me join the two halves even though they did touch at the top), I then applied generic walls to the faces, and it took quite awhile to apply those walls. When I try to open any view cutting through the model (plan or section) it takes 30 seconds or more to generate and open the view. However in a 3D view, it spins around and pans just fine with no problems.

I copied the mass and the walls to a blank project (no template), saved, and the file is still over 8MB, so I don't think I can post it here. Does this spline curve make for the rather large file size and long view generation times? Or could it be the sheer size of the mass object? The only things in the file are the mass, 6 walls (3 walls for each side of each half of the arch), a few reference planes, and a plan and elevation view. Yet still the file size if over 8 MB. Any ideas?

[Scott D Davis]

We use the St. Louis Arch as a demo to show the new swept blend tool. Our file with only the Mass is about 2.5 megs. And we can join geometry between the halves before finishing the mass family. I will post it here. I'm trying to skin it right now to see what happens to file size.

Creating more complex geometries will increase file size, processing time, etc. That should be no surprise!

[patricks]

Yeah, but I didn't expect it to slow down THAT much on my quad-core machine w/ 4 GB RAM.

I'd like to see how you created the path for the swept blend for the arch.

[Scott D Davis]

Here's the RVT file Zipped.....

The arch is a partial ellipse. All the ref planes are here for the top and bottom profiles, and for the arch shape in the south elevation.

[patricks]

Ah, I see how you did it. I was trying to make something more like the actual arch, which is a catenary shape. The bases of the Gateway Arch are not vertical, like they are if you make it using an ellipse.

Here's a screen shot of what I made. The only way I could see to make a more "true" approximation of the catenary curve was to use a spline for the sweep path.

*edit* WOW when I deleted the six wall instances and left the mass only, the file size dropped from over 8MB to under 3 MB! :shock:

[aaronrumple]

Here's the RVT file Zipped.....

The arch is a partial ellipse. All the ref planes are here for the top and bottom profiles, and for the arch shape in the south elevation.

Being from St. Louis, I'm insulted that Autoesk is demo'ing the swept blend using an ellipse rather than the proper catenary curve..... ;-)

[Scott D Davis]

Ah, I see how you did it. I was trying to make something more like the actual arch, which is a catenary shape. The bases of the Gateway Arch are not vertical, like they are if you make it using an ellipse.

Here's a screen shot of what I made. The only way I could see to make a more "true" approximation of the catenary curve was to use a spline for the sweep path.

*edit* WOW when I deleted the six wall instances and left the mass only, the file size dropped from over 8MB to under 3 MB! :shock:

I didn't build our version, but recognized that the ellipse wasn't a "perfect" match for the catenary curve, but close enough for demostration purposes. :) Spline would make a better approximation, as you could "tweak" it to just the right curve.

There's alot going on in the 6 walls that make up the arch. 4 of them are double compound curves. The outside faces just curve in one direction. But these walls all clean up where they hit each other. Lots of calculations to describe that geometry in code....

[patricks]

Yeah I can see that.

It would be nice to have a tool that would make a curving line that passed through the control points, instead of curving around with the control points off to each side. I remember FormZ having a tool like that.

[aaronrumple]

Yeah I can see that.

It would be nice to have a tool that would make a curving line that passed through the control points, instead of curving around with the control points off to each side. I remember FormZ having a tool like that.

You mean like AutoCAD, CorelDraw, Illustrator, Photoshop and every other program out there? That would be novel. Maybe even a trim and split?

Doing casino's without a good spline tool is a real trick.

[iru69]

Since it came up again, +++++

You mean like AutoCAD, CorelDraw, Illustrator, Photoshop and every other program out there? That would be novel. Maybe even a trim and split?

[tomnewsom]

Splines are much more complex curves that elipses, which is why you're getting the slowdown. However, given the weird behaviour of splines in general, I'm sure this performance can be improved.

[Tyveka]

Technically, the way that both of these are constructed is still not quite correct. The arch is a catenary curve but it is segmented. Each set of metal panels is a particular height as the curve goes up, starting with, if I remember correctly, 18 feet high. And actually, the set of metal panels is comprised of a subset of metal panels to make that initial metal panel. So there is no "curve" on the arch, just a series of segmented lines that approximate a catenary curve.
I should know because I've actually had the arch built in Revit since Revit 2.11, based on the actual drawings housed at the JNEM. I've even got the visitor's center and museum modeled out, not to mention the grid system and the ramps leading to the trams.
My next project is to get specs on the trams (if possible) and model them, then drop them into the project. I'm slowly working on getting a full animation of this building done, with photos taken from the top. Yes, I did say since Revit 2.11. It's been a labour of love...

[troberts]

Here's a screen shot of what I made. The only way I could see to make a more "true" approximation of the catenary curve was to use a spline for the sweep path.

Patrick,
I am trying to duplicate what you have done in your Gateway Arch 2 model, but I am lost.
I see the three faces of the Arch, I see your reference planes, but nothing else. Can you help a beginner? Where do you define the spline sweep path? How do you define the taper?
Thanks, Tim