Complex Geometry on a 3-Axis CNC Mill

The 3-axis CNC machine is most commonly used for one sided milling (profile or contour cuts). In order to create more complex geometry or surfaces with two finished faces, jigs will have to be used.

This example will look at how jigs were used to facilitate two sided milling for fabrication of the table legs of a QuaDror table. 

Setting up the part:

This tutorial begins with you having the object fully modeled in Rhino and oriented flat on the xy plane. We're going to duplicate the object and rotate it so there are two objects with each side facing up. It is helpful to know the distance that these two objects are displaced from each other.

Milling notes - consider tabbing the part if it wont be resting on the table after the final cut

Setting up the jig:

The process of designing the jig is relatively straightforward. The primary concern is alignment. Making sure you can place the jig on the spoil board in the same place with respect to the 0,0 point in the rhino model is extremely important as well as ensuring that you can align the stock to the jig.

The first step is to use the reference points defined in the previous step. An easy way to do this by doing a 'Save As' on the file you created the geometry in and removing the extraneous geometry. The jig in this example was created by rough cutting and stacking mdf on top of each other. The rough geometry was then brought into rhino, and the CNC was used to cut the part to the precise dimensions we needed (tip - you can create an irregularly shaped box stock by using the 'Stock from Selection' command.

The tool-paths we ran on the jig were; once along the raised face so we had a straight face to press the board to, a surfacing path along the bottom of the jig so we knew the elevation of the piece precisely, and guide holes for the jig to spoil board and the jig to the part.

Milling the object:

Because of all the prep work we've done, the actual milling of the piece is pretty straightforward. We first run the jig locator tool path and mark on the table where we should place the jig. We place the jig on the table and secure the stock to it. We mill the first side and make sure that included in this tool path is a way to locate the part on the jig when we flip the stock. Depending upon how the part is set up, you might need to include tabs in the part so it stays attached during the mill process. We flip the part, and locate it using the guide holes and run the tool path for the second side. If you've tabbed the part, you'll need to take the piece into the wood shop to cut the piece out.

And the final result is...

This tutorial was originally written for the University of MN's Digital Design website.

Designing With Rules

We had our first review of our project with outside critics. We presented to Renee Cheng, Blaine Brownell, Andrea Johnson, Gayla Lindt and John Comazzi (our advisor). Our [100 word] synopsis of the project is below.


The housing demand that came as a result of the massive growth of Seoul during the mid-20th century was primarily met through the construction of high density superblock housing complexes that were designed with no regard to context, safety or identity. Plans for redevelopment of these complexes repeat these same mistakes. The ubiquitous nature of the problem necessitates a design intervention that can be rapidly developed and deployed, which can only be delivered through a parametric process. We aim to create a toolkit that architects can use to design locally sensitive interventions that re-purpose and re-imagine these buildings and their role within their context."

Our process for the project has been to tackle the problem from a variety of different fronts; study the urban condition of Seoul (and also superblock housing projects around the world), research pedagogically important collective housing projects around the world, define values from those projects, and develop a series of tools that evaluates and incrementally modifies the geometry of the site with respect to those values (the above image being one of the twelve different parametric tools that were developed).

The feedback from the review was generally positive. The reviewers all seemed to think that the problem was a worthwhile one to tackle. One question that was brought up had to do with answering the question, "how do we know that we're creating something better?" It's easy to hide behind the idea that we're creating a series of 'agnostic' tools that will do what is asked of them and won't make a value judgement. I think our next step in the process is to start tying together these different tools that have been developed in isolation and use them to start creating an architecture that we can evaluate (both qualitatively and quantitatively).

Seoul and Superblock Housing

The chosen site for our final project is the Dunchon-dong superblock housing complex that was built in four phases during the early 80s in the southeast section of Seoul.

These buildings currently exist as single use high rise housing complexes. We are critically examining that current configuration with respect to value, resiliency, and resident experience.

Seoul Maps by Sangyong Hahn

Seoul Maps by Sangyong Hahn

Because of the ubiquitous nature of our site, we are positioning our project as a critical response to the sort of development that is primarily concerned with delivering as many units to an area without regarding the context, safety and identity of the architecture. The scale that this project exists in the world necessitates a design intervention that can be rapidly developed and deployed, which is why we are looking to parametric design as the tool to realize our solution.

We're approaching this project in a multifaceted research process that looks at identifying the economic factors that created this housing in the first place, issues related to dwelling and identity in high rise housing and also through the perception of urban modern living through film.

We've been challenged to start proposing ideas while the research is ongoing, and the below image is an example of one of those. All of our studies so far have been primarily additive with a more surgical hand when it comes to demolition. Most of the redevelopment and parametric urbanism examples we've seen have been the tabula rasa type. This is an interesting distinction we've noticed between what's been done before and what we're proposing. It will be interesting to see how this develops over the semester.

Currently reading/watching...

"The Neighbourhood Unit" by Clarence Perry, "Complexity Economics: A Different Framework for Economic Thought" by W. Brian Arthur, "Matter Matters" by Manuel DeLanda,  Mon Oncle, The Pruitt-Igoe Myth


Master's Final Project - Departure Point

Something that's been present in a lot of the work that I've done (both professionally and academically) has been this interest in automation of the design process. Specifically looking at what sort of processes can be automated and critically evaluating the effect it has on the eventual product. A common attitude that I've encountered in both conversations with design critics as well as writings about automation (and I might start using the word parametrics interchangeable for automation even though I think there are subtle differences) is that the aesthetic effects of it should be fought against. I've always felt that this wasn't the most productive attitude because I've always thought of automation as a tool (and I'm paraphrasing Chris Erwin, a coworker I had at Thornton Tomasetti here) to, "automate the tasks that I don't like doing so I can focus my creative energies on problems I enjoy." This attitude I've also found mirrored in the work of Sergio Albiac, an artist whose process moves back and forth from digital and analog processes. Here's how he describes his process: Generative sketching or writing computer programs togenerate images that will be used as painting input, allows me to modify this traditional cycle of idea-sketch-work and transform the route and the goals with a degree of freedom that greatly stimulates my creativity. Now, the “sketching” begins with a very abstract idea in mind. Once you get used to transform ideas into computer code, generative sketching allows the exploration of artistic alternatives and it produces inspirational feedback in a way that is radically different as the traditional sketching process. Then, the generative sketch can become the final work or it can “demand” the rendering of a picture using traditional media.


Generative Sketch by Sergio Albiac

 Being reflective on the role that automation has in your process is one thing... I'm also interested in investigating sorts of problems that can only be solved through a rigorous process of defining the rules of action and acting on those rules in an uncompromising way. An architectural example that come's to mind is Christopher Alexander's "Houses Generated by Patterns." And finally, this project is also providing a testing ground for what a truly collaborative design project could look like as I've decided to team up with two of my extremely talented classmates Sangyong Hahn and Hwan Kim (hence the use of "we" and "I").

We've been asked by our critic to create a number of theses statements with a different number of word counts (100, 50, 25, 12, and 6). Our fifty word one is probably the closest to the spirit of the project as I'm currently envisioning it:

"Architecture is fighting a losing battle against obsolescence. The technology that surrounds us is constantly updating and improving, and any belief that the buildings we make can be timeless is wrong. Our project is to develop a strategy to adaptively reuse a ubiquitous building stock everywhere it exists in the world."

Design and Tool Making

The role of tool making in architecture and product design is often an ignored part of the process. For this seven week design exercise, I partnered up with fellow classmate Hank Butitta to develop a series of analog and digital tools to assist in the fabrication of the QuaDror geometry as designed by Dror Benshetrit of Studio Dror.

We isolated the 3-axis CNC routing machine as the ideal CAD/CAM machinery to design our manufacturing process around because it represents the most accessible type of CNC equipment. In theory, this process could be replicated at any of the thousands of cabinet shops in the country that have access to a 3-axis CNC machine.

We developed two different types of processes for two different types of products; a high-end hardwood version and low-end plywood version. The hardwood system was optimized for speed and efficiency. We built off of a Grasshopper definition that was provided by Dror that is used by his office to define the articulating QuaDror geometry by inputting the desired width and height. Our modifications would automatically create and bake the Rhino curves into layers that would create a toolpath that could be used for the three different routers that we were using (dovetail, chamfer and 1/2" endmill). The hardwood version couldn't be simplified to a degree that we could just lay our stock material on the bed of the CNC machine and mill the desired shape. Instead we had to create a jig that would place the part in the correct location and allow us to flip the stock after the first pass and place the piece in the correct location for milling on the reverse side.