© Nathan William Smith

Columbia University GSAPP, 2011, c. David Benjamin

a tall tower proposal for masdar *See complete project progress at the GSAPP website

Structural Filigree

Using parametric modeling tools, fused with evolutionary computation, my proposed tower focuses on two concurrent architectural interests: structure and program. The model is set to run in two separate tests: A 4-floor structural module which uses programmatic rules to obtain particular objectives and a whole-building programmatic module based on L-System growth which is created with customized rules based on specific site constraints. These tests are then fed back into a new model which uses evolutionary computation to find the highest performing combinations of structure and program as part of a full-building assemblage.

1 Structural Module

Using a multitude of thin-tubed members (3” diameter) as a new form of structural diagrid, the skin acts much like curtains. The diagrid can be made more or less dense in particular areas based on internal forces.

1.2 Model Set-up (Parameters)

A 4-floor rectangular module is created in Catia, with each side of each floor being broken up by structural points. Each point searches for the next closest point on the floor below it and creates a structural member in-between. This works via a top-down method with the assumption that there will be more structural members needed toward the bottom of the tower because of gravitational weight and building mass.

2 Multiple Paretos

A typical procedure with automated evolutionary computation is to develop a complex enough problem which has multiple high-performing results (towards exploration) versus a single Pareto design (towards exploitation). The most current test takes this fact into consideration and is actually creating multiple Pareto sets of Pareto designs based on programmatic necessity. Using Multi-Criteria Decision Making [MCDM], a post-testing facet within modeFrontier, the final test results are weighted for each programmatic objective to provide structural modules with imbedded program efficiency.

3 Program

At the same time, a whole-building programmatic model is tested based on specific site constraints which are used as objectives. This model is based on L-System growth, yet with a customized parametric logic which pertains to the particular programs existing within the building. What results are multiple programmatic solutions which are evaluated according to relative locations to each other and to site parameters.

3.2 Model Set-up (L-System Parameters}

L-System logic was used as a base-code for the creation on initial programmatic logic. This code was then edited based on programmatic desires [i.e. basic programmatic locations, minimum sizes, maximum sizes, etc.] After these conditions were set, they were also used as the parameters inside the EKL script which ModeFrontier could access in order to create high-performing results.

4 Combining Test Results

Using the structural designs as base conditions for programmatically based structural sections, these modules are then embedded in the Pareto result from the whole-building programmatic test. What results is a full-building simulation which is based on programmatic desires with a structural system that reacts to the necessity of that particular program.