What is Algorithms-Aided Design

ARCHITECTS DO NOT CREATE BUILDINGS, THEY DEFINE WHAT A BUILDING SHOULD BE.

-Ondrej Slunecko, All I Ever Had Was an Idea-

This is the quote that started the catalogue for the exhibition “All I Ever Had Was an Idea”. The exhibition took place in September 2019 at Godsbanen, Aarhus, Denmark and I worked on it alongside my dear friend Alina Croitoru.

So let’s step into a time machine and go back to 2019 and discover what is Algorithms-Aided Design and how you can use it to make great pieces of architecture.

What is an algorithm?

An algorithm is a set of operations to be performed in order to produce a solution to given conditions. Or to perform a specific task by defining precise set of instructions. Algorithms divide complex problems into smaller steps, that can be calculated. For an algorithm to properly operate we need to fulfil several criteria; the instructions must be unambiguous, the input and output must be clearly defined and instruction steps must be in logical sequence.

Although often associated with computers, something as simple as instructions how to assemble a furniture is a basic version of an algorithm. We are given the pieces needed for assembly and led through the process step by step. Therefore if we follow each step carefully, we will end with a finished piece of furniture, exactly same as all other people who bought the same product and followed the instructions.

If we talk about computers, we can define our own algorithms using a standalone editor, such as programming languages (Python, C++, Java etc.) or we can use special embedded software such as Grasshopper for Rhino or Dynamo for Revit. The advantage of the latter category is that they are easier to use and are meant to operate within realm of geometry. Their development allowed architects to experiment with generating of form as well as automating labour-intensive tasks and enabled manufacturing of complex structures.

Use of Algorithms

The initial impulse for the exhibition was the exploration of one of the pieces of software, Grasshopper for Rhino, which led us to seek which architects use these tools and how. We were fascinated by the complex forms. That is usually the first thing you see if you search for algorithm-aided design; complicated shapes, abstract geometry and organic structures. So we studied among others works of Patrik Schumacher and Zaha Hadid. However the deeper we went, the more we realized, that the form that emerges by using AAD is form for form’s sake.

As a reaction to that realization we aimed our search in a slightly different direction. We looked for architects, who use AAD more pragmatically. Based on our second wave of research, we developed 7 examples how AAD can be used in variety of design tasks pragmatically. These examples are not meant to be definitive versions or ultimate tools. They should serve as a guidance and showcase of possibilities.

Through our process we developed a procedure to follow while designing with a help of AAD tools. Our procedure can be seen on the diagram below. Simply put, it is not enough to develop an algorithm and expect it to operate perfectly and serve universally. As with traditional design, AAD should be seen as iterative process. After setting up the inputs and definition of operations, we analyse the outcome and execute changes in inputs, as well as the definition itself. And only after certain amount of iterations, are we close to what could be seen as optimal solution.

This procedure allowed us to explore different possibilities for AAD, which can be divided into several categories. Among the most obvious are analysis and designs based on the environmental conditions such as light, wind or thermal properties. The form of the building can dynamically respond to those and the outcome would be different based on the geographical location. The second category includes using AAD in relation to the structure, where load bearing parts can change and be optimized.

The third category deals with socioeconomic variables. And the final category helps with automation of processes and manufacture.

Now that we understand the basics, we are ready to look at the scenarios we developed and descriptions of driving forces which led to the final form.

So next time we will look at individual case studies and think how can we use the tools we have at our disposal to make cities better, one idea at a time.

Cheers