How Algorithmic Design Improves Collaboration in Building Design

Algorithmic design

Design, like everything else in a construction project, is a collaborative effort. Even with digital tools, collaboration across design disciplines is not yet optimal. An experimental project thus set out to test whether algorithmic design could help streamline the interaction between architects and structural engineers.

Design data originating from an architect is used in several engineering tools for visualization, analysis, and calculation. Ideally, changes in the architect’s design would propagate automatically across all the software. Unfortunately, the process is in fact mostly manual. Hence, the design data is seldom, if ever, in perfect sync on all systems.

Two companies, A-Insinöörit and Geometria Architecture, joined forces to test algorithmic design collaboration to see if it could solve many of today’s data exchange problems. They found their solution could speed up the process, eliminate many errors, and alleviate the frustration of doing updates by hand. It could also have a positive effect on the final outcome of the project.

Karjalainen and Wikar
Petteri Karjalainen and Markus Wikar

The Emerging Algorithmic Design

Petteri Karjalainen is a structural engineer at A-Insinöörit. He has been working over the last two years in the company on international projects, especially with industrial clients. He’s also involved in developing an algorithmic design process, the theme of his recently completed master’s thesis.

“We have been speeding up our efforts in this field. Our company leadership sees potential in the practice and has encouraged me and our computational design team to apply more and more of these advanced methods to company’s daily routines,” says Karjalainen.

Algorithmic or computational design uses sets of instructions to perform certain tasks, for example, to generate a digital model of a structure. The instructions have parameters that generate variations of the same code. Algorithmic design is especially suited for architectural forms that are non-conventional and that can be constructed from repeatable elements.

A design algorithm example
Design algorithm, example

Experimental Architecture

Demanding, non-traditional forms are the bread and butter for Geometria Architecture, the brainchild of two architects, Markus Wikar and Toni Österlund. The company is both a design practice and a consultant for architects, engineers, and contractors. They cover parametric modeling, algorithm-aided design, and digital fabrication.

“At Lahdelma & Mahlamäki architects, my previous employer, I was in charge of the warped geometries of POLIN, the Museum of the History of Polish Jews in Warsaw,” Wikar explains. “Today, the company is our client.”

Algorithmic design opens new perspectives for architectural expression. It makes the use of non-conventional forms feasible and cost-efficient to build. In addition, it gives designers freedom to test and present dozens of alternative solutions, which in a traditional process would be very time-consuming or simply impossible to do.

A FEM model
A FEM model

Experimenting with Algorithms

In February 2018, A-Insinöörit and Geometria Architects started experimenting with how to use algorithmic design for collaborations between architects and structural engineers. The project got funding from the national KIRA-digi digitalization program.

They chose to experiment with an imaginary building, specifically a swimming pool covered with a steel-structured curved roof. The aim was to test typical design tasks and data exchange between the designers. The platform that the experimenters used was Grasshopper, an extension of the Rhinoceros software.

The design process
The roof design process

The architect designed several variations of the roof. The final design was formed with parable arcs, generated through so-called dynamic relaxation. This resulted in an optimized structural system.

After the architect had algorithmically created the curved roof and generated the line geometry, the structural engineer took over the data. They used Grasshopper-RFEM Link; an extension developed by A-Insinöörit. This allowed them to analyze and dimension the structural steel framework created in Grasshopper. Furthermore, they used Trimble’s Grasshopper-Tekla Live Link to build a BIM model of the structure.

Making Data Flow Between Systems and Processes

The experimenters needed a way to exchange algorithmic data between the architect and the structural engineer over the design life-cycle. They chose Speckle, a cloud-based platform. It links the data intelligently between design parties and models.

Speckle allows designers to share geometric data across several design models and to create an aggregated model from discrete parts. This means that when, for example, the architect changes the original geometry, the data is automatically updated wherever it has been referenced. After the engineer has designed the structural model, they, in turn, can provide the architect with the updated data.

In its purest form, all design data resides in the algorithms and the traditional BIM model is generated only if and when it’s needed. However, at the moment, a combination of algorithms and models seem to work best.

Data transfer
Data transfer

The Automated Future

“Imagine we modeled a double-curved shell exactly as it will be built. Say, someone then comes up with a change that implies a one-degree change to every rod of the structure. With traditional methods, the change would imply a huge job, but with an algorithm, it would be a breeze,” Karjainen proclaims.

Both Wikar and Karjalainen believe that the use of algorithms and artificial intelligence represents a huge potential for the whole industry. They will not only enhance collaboration, but also free designers from routine tasks that in many cases constitute the larger part of their work. Furthermore, algorithmic design makes digital fabrication a reality. Building parts can be robotically manufactured in factories and installed at the construction site.

After the KIRA-digi experiment, Geometria Architecture has continued to collaborate with A-Insinöörit around the Helsinki–Tallinn tunnel project. It will be the world’s longest undersea railway tunnel.

“We have been thinking through the data interfaces with the around 10 disciplines involved. We’re figuring out the smallest common design denominators or parameters that need to be exchanged between parties,” says Wikar. “Our goal is to allow the experts to focus on their core competence areas and not to have to struggle with software issues,” he concludes.

An exterior view

The project illustrations are courtesy of A-Insinöörit and Geometria Architecture

Share This