Free-Forms’ Intelligent Structural Design Process by DATA Manipulation 110/45

Intro

According to the description represented in the Intro, every step from gathering data, manipulation, and generating models would be defined in GH3D. Therefore, the idea of using karamba3D is found in designing and analyzing in an integrated mode towards optimization processes. This is the mainstream of this workshop comprises the path to understand how design works accentuated via material impact, different load cases, how form follows forces, diversity of supports’ locations, and, more importunate, whether design iterations are qualified enough to be considered reliable to discuss as an optimized iteration. The joint section would be set by using different ML approaches to surrogate optimization results in a feature-based clustering setting. We will use *.csv files and logs as our reports and dedicate the clustering to studying which design has what characteristic. Accordingly, we will use the Jupyter Notebook via Anaconda, so every aspect of such a procedure can be achieved conveniently. The final back-and-forth data collection, model generation, and optimization would be converged to Machine Learning based analysis. It is assumed that every group could demonstrate the whole open-source surrogate optimization that is suggested by this workshop as a procedure of how data manipulation, besides high qualified algorithms, can make a difference within conceptual structural analysis toward reliable yet flexible  decision-making. You can follow Mahdi on Medium to have a better perspective on such titles as Generative X: Bio-Intelligent DATA-Flow on Free-Form Structures or Generative Design and How to implement it in Architecture.

The workshop perquisites

Keywords: Structural Analysis, Optimization, Strength through Geometry, Free-Forms, Machine Learning
Required Skills: Grasshopper3D; knowing Python is a plus but not mandatory
Required Software: Rhinoceros 6/7, Karamba3D, Anaconda
Required Hardware: Laptop/PC
Maximum number of participating students: N/A

The Instructor

He is known for his researches on Integrated Design in Architecture. Mahdi also knows how to deal with DATA flows from different disciplines, especially regarding Engineering Design. There is a diverse range of Algorithmic Engineering consultancy he has set in high-qualified projects in Iran. He is the lead developer of Caddisfly 01.01, an exclusive Grasshopper3D add-on for detail designing free-from structures, which brought up Ardaena.com as a network for professionals. Since his researches are now dedicated to AI and dealing with a series of data flows algorithmically integrated, he is now working on Generative Design based on Functional Paradigms through Geometry.

Course Sections

There are 5 sections in this course:

Stage 1: we are going to define a model generator so we can have different generations of our design

Stage 2: we will record our optimization processes data to have our iterations in a CSV file

Stage 3: Python will come up with Jupiter notebook and anaconda, and we will use it to cluster our recorded data by unsupervised machine learning algorithms according to the features we want.

Stage 4: We take our result back to the Grasshopper3D again by hacking number sliders to manipulate their parameters and recall our selected iterations

Stage 5: We will use Microsoft Power BI to demonstrate the results in an elaborated yet readable format