EvolvArk | Computational Design Practice, Copenhagen

Design through computation

Parametric massing options generated by the feasibility tool

Architectural Feasibility Tool

Computational tooling · Feasibility modelingSeattle, United States

Daylight analysis of tunnel portal variants

Tunnel Portal Daylighting Optimization

Environmental simulation · Infrastructure lightingOntario, Canada

A single wire module aggregated into a sculptural wall

Aperiodic Wire Polyhedra

Computational fabrication · Form findingSeattle, United States

Automated 3D site context generated from Danish open geodata

Automated Site Context

Computational tooling · Site modelingCopenhagen, Denmark

Evolutionary Algorithms in Urban Design

Research · Evolutionary design

Services

Parametric Design & Automation: associative models that keep geometry, data, and documentation in step as the design moves.

Computational Scripting: custom tools in Python and Grasshopper that remove repetitive modeling work from the practice.

Environmental Simulation: daylight, solar exposure, and energy studies used to steer design decisions rather than confirm them afterward.

Multi-Objective Optimization: evolutionary and generative searches that resolve competing targets: area, cost, daylight, structure.

CAD & 3D Modeling: Rhino, Revit, and AutoCAD production models built to survive coordination and revision.

Architectural Visualization: diagrams, renders, and analysis graphics that make a technical argument readable.

The brief

Feasibility decisions are made early, often before a team has explored more than a few options. The aim was to widen that first look without slowing it down.

Approach

We made the site and its planning rules the inputs of one parametric model. Every option it generates is compliant from the start, so none needs a separate compliance check.

Area and unit figures come directly from the geometry. When the massing changes, the figures change with it, and every option is compared on the same basis.

Delivery

A ready-to-use tool with a simple interface for non-specialists, handed over with documentation and a training session. The team can run its own feasibility work and extend the tool as planning guidance changes.

The brief

Drivers entering a tunnel by day meet a drop in brightness faster than the eye can adapt. The portal geometry and the artificial lighting both affect that transition, and a change to one changes what is needed from the other. The work had to make that relationship clear while the portal design was still developing.

Approach

We parameterized the portal and automated the cycle of generating variants, simulating their daylight performance, and ranking them against the target transition.

The analysis puts built form and installed lighting side by side across the options.

Delivery

A ranked set of design options with the simulation data behind each, and the workflow that produced them. When the design or site conditions change, the workflow can run again without being rebuilt.

The brief

The wall is built from welded wire polyhedra and reads as a collection of three-dimensional stones. It had to look organic, fit a given volume exactly, and stay simple enough to build by hand, without a visible grid or repeated pattern.

Approach

We designed the system that grows the wall. The module braces itself, the connection rules control how units join, and the field guides how the wall fills the volume.

Each placement is evaluated against the field, so the wall does not resolve into a regular grid. Every version can be measured, compared, and regenerated with different rules.

Delivery

A generative design tool with the module, the rules, and the field exposed as controls, so the artist can take the composition in new directions without rebuilding it. It produces candidate compositions together with the schedules needed to fabricate them.

The question

Architecture and urban design increasingly have to answer environmental, social, and economic questions at the same time. Evolutionary algorithms are well established in performance-driven and sustainable design as optimization tools. The paper asks what they can offer earlier, while the design question is still open.

Approach

We treat the evolutionary search as a means of exploration. Set up for this purpose, it generates a broad and varied range of design options for the designer to work with.

We argue that this exploratory use supports innovation, adaptability, and diversity in early design, and that it suits the complexity of urban environments, where many factors have to be weighed together.