02-19-2013
A Parametric Primer
When it comes to parametric modeling, everything starts with a point. Two points make a line; two lines make a surface. With this basic set of components, the possibilities are endless. In much the same way that languages comprise a unique combination of nouns and verbs, parametrics synthesize innumerable permutations of points and lines. As with learning a foreign language, the parametrics learning curve begins steeply, but with a basic skill set, the possibilities are truly infinite.
New Frontiers
Understandably, approaching a new technique can be intimidating. The easiest way to grasp the concept of parametrics is to take an iterative or algorithmic approach. An algorithm is commonly a step-by-step procedure of calculations that are usually done by a computer; that said, algorithms can be created manually as well. Architecturally speaking, if something can be manually modeled, then it is highly probable that the same design can be generated from a script, or built as a parametric model. It is important to start small. I recommend a baby-step approach. Take time to analyze the steps that would be employed to model something manually, and then write these steps down sequentially. This creates a sort of "pseudo code." These pseudo code steps can then be used as a roadmap to generate that very same design - only this time the design would be generated using a parametric modeler.
The act of creating a parametric model is an inherently front-loaded process. That is to say, it may seem more practical to manually model a design because it is initially faster. The problem with a manual model, however, is that once the geometry has been created, it is static. That means that the geometry would require rebuilding, in the case of new inputs, and/or when the end goals are changed. With a parametric model, though, there is never a definitive object – instead, the model illustrates a series of relationships. For example, if point A moves in one direction, then point B will move the same distance in another direction. A well-designed parametric system allows for instantaneous results whenever inputs are updated.
Parametrics in Practice
I am currently working on a school in Kenya that will require partitions to separate men and women. In addition to divvying up the space, the panels will allow for passive ventilation to cool rooms as necessary. My first step has been to take an existing, culturally significant pattern and arbitrarily determine how easily it can be scaled in order to test its opacity.
Using Grasshopper, I am able to test how easily the pattern can be scaled. However, in this case the tool is only being used to create an arbitrary global condition across the entire panel. The next step would be to take into account some contextual inputs. For example, a secondary input is the fact that the screens could be more transparent above typical eye heights. With this in mind, I have used a linear gradient to test the results of having dense patterns at the base of the panel and a more open condition at the top. Once the gradient benchmarks have been determined, I can start testing the patterns on different materials and fabrication techniques. In the case of this exercise I will be using GFRC that has been milled using a CNC router. Taking into account the width of a router bit at the base and the minimum thickness that GFRC would allow for at the top of the panel, I am able to bracket the gradient based on the fabrication constraints.
With the basic inputs in place, I can test the screens in a practical application. In this situation I will use the screen to span 4.2m floor to floor slabs. Using a new Grasshopper definition I can use a curve to tune the size of the screen openings. This curve can be adjusted to accommodate privacy constraints, air flow, and slab adjacency, while also considering material and fabrication limits. This is accomplished by moving the curve control points away from the Y-Axis in different increments and then testing the results.
A Powerful Tool
Even in this basic example of parametric modeling, the flexibility, strength and ease of this system/approach are undeniable. I took a primitive modeling technique and applied it in a complex way. Simple offsets of a common pattern accommodate aesthetic, programmatic, ecological, material, fabrication, and functional constraints.
Ultimately, the greatest success of the system is not based on the technique that was used to create it, but how well it functions as an architectural element. The best projects are not defined by their parametric process; instead, the process falls into the background, and seamlessly supports the project's overall concept. Parametric modeling is a tool, similar to rendering and BIM, to be employed as necessary on a project to project basis. While this is an example of parametric modeling in an iterative way with an outcome that was somewhat predictable; greater knowledge of parametric models can be used to venture into the truly unknown, and therein lies the promise of mathematical computation.
by Mark Bearak
New Frontiers
Understandably, approaching a new technique can be intimidating. The easiest way to grasp the concept of parametrics is to take an iterative or algorithmic approach. An algorithm is commonly a step-by-step procedure of calculations that are usually done by a computer; that said, algorithms can be created manually as well. Architecturally speaking, if something can be manually modeled, then it is highly probable that the same design can be generated from a script, or built as a parametric model. It is important to start small. I recommend a baby-step approach. Take time to analyze the steps that would be employed to model something manually, and then write these steps down sequentially. This creates a sort of "pseudo code." These pseudo code steps can then be used as a roadmap to generate that very same design - only this time the design would be generated using a parametric modeler.
The act of creating a parametric model is an inherently front-loaded process. That is to say, it may seem more practical to manually model a design because it is initially faster. The problem with a manual model, however, is that once the geometry has been created, it is static. That means that the geometry would require rebuilding, in the case of new inputs, and/or when the end goals are changed. With a parametric model, though, there is never a definitive object – instead, the model illustrates a series of relationships. For example, if point A moves in one direction, then point B will move the same distance in another direction. A well-designed parametric system allows for instantaneous results whenever inputs are updated.
Parametrics in Practice
I am currently working on a school in Kenya that will require partitions to separate men and women. In addition to divvying up the space, the panels will allow for passive ventilation to cool rooms as necessary. My first step has been to take an existing, culturally significant pattern and arbitrarily determine how easily it can be scaled in order to test its opacity.
Using Grasshopper, I am able to test how easily the pattern can be scaled. However, in this case the tool is only being used to create an arbitrary global condition across the entire panel. The next step would be to take into account some contextual inputs. For example, a secondary input is the fact that the screens could be more transparent above typical eye heights. With this in mind, I have used a linear gradient to test the results of having dense patterns at the base of the panel and a more open condition at the top. Once the gradient benchmarks have been determined, I can start testing the patterns on different materials and fabrication techniques. In the case of this exercise I will be using GFRC that has been milled using a CNC router. Taking into account the width of a router bit at the base and the minimum thickness that GFRC would allow for at the top of the panel, I am able to bracket the gradient based on the fabrication constraints.
With the basic inputs in place, I can test the screens in a practical application. In this situation I will use the screen to span 4.2m floor to floor slabs. Using a new Grasshopper definition I can use a curve to tune the size of the screen openings. This curve can be adjusted to accommodate privacy constraints, air flow, and slab adjacency, while also considering material and fabrication limits. This is accomplished by moving the curve control points away from the Y-Axis in different increments and then testing the results.
A Powerful Tool
Even in this basic example of parametric modeling, the flexibility, strength and ease of this system/approach are undeniable. I took a primitive modeling technique and applied it in a complex way. Simple offsets of a common pattern accommodate aesthetic, programmatic, ecological, material, fabrication, and functional constraints.
Ultimately, the greatest success of the system is not based on the technique that was used to create it, but how well it functions as an architectural element. The best projects are not defined by their parametric process; instead, the process falls into the background, and seamlessly supports the project's overall concept. Parametric modeling is a tool, similar to rendering and BIM, to be employed as necessary on a project to project basis. While this is an example of parametric modeling in an iterative way with an outcome that was somewhat predictable; greater knowledge of parametric models can be used to venture into the truly unknown, and therein lies the promise of mathematical computation.
by Mark Bearak
