Parametric Vaulting Definition

Comment

Comment by Benjamin Golder on December 7, 2009 at 6:00pm

Parametric Vault 4.ghx

Here is the definition in its current state. Feel free to modify/edit. I would be really excited to see anything you make with it.

Comment by Karthik Dondeti on December 2, 2009 at 3:07am

Awesome! I have a presentation this Thursday (in 2 days). I would not mind the "messy" version either, and would be most excited to explore this! I am planning to eventually develop a definition based on Philippe's method, though I guess it will be a while before I can invest time. Would love to discuss further with you regarding this! Thanks so much!!

Comment by Benjamin Golder on December 2, 2009 at 12:18am

Cool Karthik! Absolutely! My definition does not incorporate Phillipe Block's work or thrust network analysis, which I am sure you already know. Though I wish that it did. I would love to see what you make. I am happy to upload it, but as I said before, I should clean it up and I am really busy at the moment. If you really can't wait, send me a message and I'll either clean it sooner or send you a messy version.

Comment by Karthik Dondeti on November 28, 2009 at 1:18am

I am currently studying the "Thrust network analysis" by Phillipe Block and I am hoping to build a grasshopper definition in the near future. Is it possible to share the one you developed? I could, if you permit, present this for a study am compiling on TNA in the design school at Harvard.

Comment by Benjamin Golder on November 23, 2009 at 1:00am

I will definitely post a grasshopper definition, but I need to clean it up a bit and will be pretty preoccupied with other tasks for the next month or so.
The physical model has been scrapped at the moment. It was a group project and this model got voted down.

Comment by corneel cannaerts on November 7, 2009 at 3:21pm

thanks for the info. I'll have a look the medial axis definition you mention.

I was aware of axel kilians work and found Mos's applet work while making this modeller. Although i assume they were made with a different intent, they all use the same principle: vertexes represent a particle with a certain mass, while edges represent a spring with a certain force, a particle-spring physics integration is used to calculate the forces. In my case i used the traer physics library for processing using Runge-Kutta 4 integrator.

My interest in this lies more with the possibilities opensource software has in developping project specific, dynamic and physically grounded models, rather than structurally optimized vaulting. I'm also developing these into physical models, and currently working on some extensions of these (topological changes, surface populated CA's, and more ;) )

keep us posted on the physical modelling development.

any chance of uploading a grasshopper def of this?

regards,

corneel

Comment by Benjamin Golder on November 5, 2009 at 3:32pm

Thanks Corneel! Your catenary modeller is fantastic! Great interface.

These are definitely made with a different logic, and they are far from being any sort of optimized vaulting system. I am mostly just excited about modeling a vaguely plausible vault with a variable shape and number of connections to the ground. I am also pretty excited about medial axes and finding more potential for using them to create structural optimizations of irregular shapes.

I would love to learn more about how you've created your catenary fabric modeler. I have seen the catenary applets by MOS and Axel Kilian and have wanted to understand them for a while.

In this definition, I used a graph tool to change a varying number of points on 0 to 1 interval, which I then use to evaluate a circle. I then found the mid points between these points and pulled them towards the center and up to create the greenish degree=2 curves for the sides (please correct me if I'm wrong, but I think that would make the edge curves parabolic). I also use these edge curves in plan to create a polygon that I plug in to Daniel Piker's medial axis definition, and with a bunch of filtering methods, I extract a number of points along that medial axis (in purple). Then I raise the axis points with a catenary function, [a*cosh(x/a)] in which I can adjust the "a" value to raise or lower the axis points, and in which x is their distance to the center of the circle. Then I find the closest point on the two closest edge curves to each raised axis point, and use the resulting sets of three points as control points for curves, which you can see in a sea-green color. In some of the screenshots I have rebuilt these "surfacing curves" to get different quantities of control points and curves of varying degree.

I made this for a structures class that I am in here at UC Berkeley, so I expect to develop this into a physical model.

Any advice that you have is greatly appreciated, especially with regards to ways of creating a better approximation of a structurally optimised vault.

On a side note, have you checked out Phillipe Block's dissertation on Thrust Network Analysis? They have a great blog on a vaulting project over there.

Comment by corneel cannaerts on November 5, 2009 at 2:54pm

nice!
although i assume these are generated trough a different logic, the result looks similar to catenary modeller i made in processing: http://www.introspector.be/index.php?/research/dook/

can you tell us some more of th process involved?