Introduction to Optimization and Design Decision-making
Computational Design III (Level 2) advances data-driven design methodologies by integrating optimization, data analysis, and machine learning into architectural decision-making processes. Building on parametric workflows, the seminar explores how environmental simulation tools can inform iterative design through measurable performance metrics. Students engage with both single-objective and multi-objective optimization techniques to generate and navigate complex solution spaces. Expanding on this, the course introduces clustering methods and regression analysis to identify patterns and relationships within large datasets, supporting deeper insights into design performance. Through the application of multi-attribute decision-making techniques, participants develop transparent and reproducible strategies for evaluating and ranking design alternatives within complex architectural contexts.
Building on the previous terms, this course will focus on translating acquired knowledge into structured and informed design decision-making processes. Moving beyond purely generative approaches, this course explores how computational methods can support the evaluation, comparison, and selection of design alternatives. Participants will explore processes that bridge conceptual ideation and analytical reasoning for a more rigorous and transparent approach to design.
Throughout the course, we will explore how performance evaluation informs design decisions. One aspect of this is environmental simulation, through the plugin Ladybug, which will allow students to assess key microclimatic factors to generate measurable inputs that guide iterative design development.
Building on this, this course introduces optimization methodologies as tools for navigating complex design problems. Students will engage with single-objective optimization to refine specific performance criteria, and multi-objective optimization to generate diverse sets of solutions that highlight trade-offs between competing objectives.
To support the interpretation of these solutions spaces, this course incorporates data analysis techniques.Students will apply clustering methods such as k-means to identify patterns and group similar design alternatives, alongside regression modelling to understand relationships between design variables and performance outcomes. These approaches enable a more informed evaluation of large datasets and support evidence-based design reasoning.
Finally, the course introduces multi-attribute decision-making techniques that allow students to evaluate and rank design alternatives based on multiple criteria and varied performance priorities. Students will learn how to develop, interpret, and adapt clear and transparent methodologies for design decision-making within complex and multi-dimensional architectural contexts.
At course completion the student will:
The objective of this project was to compare the effectiveness of different machine learning strategies applied into the exploration of a recursive model using the exercise worked on in the assembled architecture course as a proof of concept. .Recursive Processes in Architecture But this approach has left us with a question: is it possible to … Read more
This project turns wind and snow into design tools. Instead of starting with a fixed form, environmental forces drive the geometry, generating adaptive architectures that evolve through performance. Project Description The site spans 39 acres and consists of two adjacent properties: Trumbull Park Homes, established in 1938 and owned and operated by the Chicago Housing Authority (CHA), occupies 21 acres and contains 454 housing units; … Read more
From the catalogue of designs for Final Assignment 2, we selected one of the iterations we had explored, as it was the most suitable candidate for optimisation. Following the circulation analysis of the terrace, the base points of the pavilion were located. However, key performance criteria — including solar radiation levels for user comfort, sky … Read more
A TerraFormed production CONTEXT TerraFormed is an ongoing project developed within Digital Matter Studio 2026 that implements osteomorphic blocks as a standardizable and industry-compatible approach for producing a novel geopolymer material within a sustainable construction framework. DESIGN OBJECTIVES Osteomorphic blocks are primarily governed by the parameters of the sine curve that defines their geometry. This … Read more
Abstract This project uses parametric design and multi-objective optimization to develop an interlocking masonry brick. A Grasshopper workflow generated brick geometries by varying sinusoidal width, depth, and cavity parameters. The designs were evaluated using structural and geometric performance criteria, including load transfer, web thickness, horizontal area, volume, and interlocking capacity. An evolutionary optimization process identified … Read more
Reciprocal Frames Reciprocal frames are self-supporting 3D structures. They consist of three or more sloping beams arranged in a closed circuit. Each beam rests on the preceding one and supports the next, creating a completely interdependent grid. This design eliminates the need for a central pillar or column. Traditionally these frames were made of wood, … Read more
Abstract Morphologies of Shade explores the design of a computationally optimized pavilion for the rooftop of the Institute for Advanced Architecture of Catalonia (IAAC) in Barcelona. The project responds to the environmental challenges of excessive solar exposure and strong wind conditions that limit the usability of the rooftop throughout the day. Inspired by natural canopy … Read more
This project investigates the use of construction and demolition waste (CDW) as a primary material resource for the development of prefabricated concrete slabs. It combines computational design, material optimization, evolutionary algorithms, and physical simulation to explore how reclaimed concrete and ceramic aggregates can be systematically incorporated into new construction products. A parametric workflow was developed … Read more
Using climate clustering, vernacular precedent research, and evolutionary optimisation to create climate-responsive design typologies when no established local precedent exists Overview Lovingly named after the very first test location, Project Michigan is a tool that provides climate-responsive design guidelines when no local vernacular precedent exists. Through scraping climate data and global vernacular information, it helps … Read more
