- Computers and software, now commonplace in construction design, shocked the system when introduced to the industry in the late 1980s. Computational design is positioned to make a similar shift in the way we design by replacing codebooks with algorithms and human analysis with artificial intelligence.
- Computational design is an emerging design method set to change the landscape of our industry as we know it. At the 2022 Digital Agility Summit, Anthony Zuefeldt, a leader in technological transformation within the AEC industry, proclaimed, “Every facet of the AEC industry will eventually be affected by [computational design], and some have called it the ‘defining moment’ of this decade.”
- So what exactly is computational design? And how will it affect the design and construction processes widely used now?
What Is Computational Design??
Computational design is a design method that uses a combination of algorithms and parameters to solve design problems with advanced computer processing. Every step of a designer’s process is translated into coded computer language. The software program uses this information alongside project-specific parameters to create algorithms that generate design models or complete design analyses. Once the initial programming is completed, design becomes a dynamic and repeatable process.
Traditionally, design is passive—a designer uses their knowledge and intuition to create designs with a computer-aided design (CAD) program. This method of manual drafting limits the number of design options that can be considered and is restricted by available time and resources. Once implemented, computational design is an effective and useful tool for increasing productivity and creating more robust designs.
Designers must break down their design process into measurable steps to implement computational design. These steps create a set of instructions, complete with recognizable patterns and trends, that lay the framework for algorithms to solve design problems.
Computational Design Tools
Computational design gives designers the power of programming without the need to learn code. That is because most computational design tools use visual programming as opposed to lines of text-based code. With visual programming, users connect outputs from one node to inputs of another, creating a program that travels from node to node by connectors. The end result is a graphic representation, or essentially a flowchart, of the design process.
These visual programming tools are typically plug-ins that pair with design modeling software like Tekla Structures, Autodesk Revit, Trimble Quadri, and Bentley MicroStation. Two of the leading computational design plug-ins are Dynamo, compatible with Revit, and Grasshopper, compatible with Tekla, Quadri, and Rhino.
Dynamo is a visual programming tool from Autodesk. Users import and export data from their 3D model, Excel, or even image files to populate the scripting interface. The program displays complex geometries allowing designers to analyze their projects and make visual modifications.
Grasshopper predates Dynamo and is likely the most popular computational design plug-in. With this algorithmic modeling tool, users create design rules with the node-based interface. Designers can also take advantage of the extensive node library and third-party design tools.
Types of Computational Design
Computational design is very much evolving and changing as standards are established within the field. As it currently stands, there are three subsets of computational design: parametric design, generative design, and algorithmic design.
Parametric design is an interactive design process that uses a set of rules and inputted parameters to control a design model. The rules establish the relationship between different design elements. The parameters are project-specific values that define the design model, like dimensions, angles, and weights. When a parameter is modified, algorithms automatically update all associated design elements based on the set dependencies.
Parametric design is a step above traditional 3D modeling, where a designer is responsible for updating each design element individually. Instead, a designer would simply update a parameter, and the parametric algorithms would make all of the associated updates. Parametric design is ideal when designing complex and uncommon architectural geometrics.
A parametric model is easily modified and can be adjusted in real-time. This allows a designer to explore many possible design options. Rather than drafting hundreds of columns all with independent offsets and dimensions, designers input symbolic parameters that define how the columns relate to each other and to the building itself. If the columns need to be shifted in the future based on new design information, the parameter is updated, and the entire model adjusts based on the stored algorithm.
The term parametric stems from parametricism, coined in 2009 by Patrick Schumacher, principal of Zaha Hadid Architects. He argued that a new design style had emerged that was “rooted in digital design techniques and takes full advantage of the computational revolution that drives contemporary civilization.” Parametricism can be used to describe a particular contemporary, avant-garde style of free-form structures that are typically designed with parametric design tools.
How Parametric Design Is Used
Parametric design is already infused in many designers’ workflows. With visualization tools like Grasshopper coming pre-installed on Rhinoceros 6, parametric design is accessible, and the visual programming is intuitive. Designers simply input parameters, like dimensions, angles, or offsets along with the applicable design requirements to receive an output. The parametric design process is done in a live building information modeling (BIM) program and will update in real-time if attributes are changed.
Certain computational design plug-ins, like Tekla Structures, come pre-programmed with a library of nodes that produce algorithms based on current industry codes and standards, eliminating the need to input design rules on the frontend. Tekla Structures enables structural engineers to create complex curved structures by visually inputting data into an algorithm-based editor.
Generative design is an iterative design process that uses user-defined inputs to produce multiple design concepts that meet specific goals. The inputs are rules and parameters that define design requirements, similar to parametric design. With generative design, the user also inputs success metrics that will evaluate the results. Artificial intelligence (AI) and cloud computing generate tens or even hundreds of design options, ranked by these metrics.
Success metrics are criteria used to optimize the design, such as building positioning, spatial planning, life safety analysis, structural loading capacity, number of building units, or cost data. The program will produce a wide range of design options and the designer will refine the optimization criteria accordingly. Generative design combines the power of AI to create hundreds of possible designs with the intuition of human choice to narrow down the results.
Designers left to their own devices tend to create predictable results. While some level of trial and error is inherent to the design process, it is unfeasible for a human to produce and vet all possible design options. This causes architects to lean on tried-and-true designs or those used on past projects, sometimes rather than the ideal option.
Generative design leads designers to solutions they never even imagined that are beyond their normal thought process. These solutions are referred to as “happy accidents” in the generative design world. Designers use the process of optioneering— the in-depth consideration of multiple design options—to evaluate all results, refine their criteria, and land on the best design outcome.
How Generative Design Is Used
Generative design is a process for design optimization. Designers use these tools to maximize the number of locations served by a road, minimize the number of structural members needed to achieve a specific design load, or achieve a specified thermal capacity dependent on multiple construction surfaces or materials.
Algorithmic design is a design method guided by algorithms. The term is often used interchangeably with computational design and could be considered a type of generative design. Algorithmic design uses algorithms—a set of instructions that determine the solution to a problem—to produce architectural models. In other words, a set of rules is used to define a system rather than defining each element individually.
While the goal of generative design is to produce as many design options as possible for analysis, algorithmic design is the opposite. A higher level of detail and scrutiny is placed on the input rules and parameters in an effort to produce just one or a few desired results. Often, algorithmic design looks like a separate line of code or connectors between nodes that can be traced back to each individual building element generated.
Relationship Between Parametric, Generative, and Algorithmic Design
How do these different subsets of computational design relate to each other? It is up to interpretation as the field of computational design develops and standardizes.
Let’s first simplify the definition and intent of each design method:
- Parametric design uses parameters and rules to create a design solution that is easily modified.
- Generative design uses algorithms to generate a batch of design options for evaluation.
- Algorithmic design uses algorithms to produce a design model.
It is easy to see how there is some overlap between the terms, especially the broad definition of algorithmic design. Algorithmic design is a type of generative design because it uses algorithms to produce a design result. It can also be considered a type of parametric design if those algorithms rely upon a set of parameters.
Parameters and rules are key components of both parametric and generative design. Both design methods also rely on strong input data to produce reliable results.
Engineers use visual scripting to unlock the full power of parametric design - building customized workflows, automating repetitive design tasks and handling complex shapes. Learn more about parametric design using Tekla and Grasshopper here.
Parametric design is an interactive process—elements of the design model are associated with each other, allowing real-time design modifications that update across the entire design. The process uses software plug-ins that rely on accurate input parameters and element relationships.
Generative design is an iterative process—the software produces many results, ranked by the user-provided constraints or success metrics. The process uses advanced algorithms and artificial intelligence, but still requires human intuition to finalize the design choice.
Researchers at the Frontiers of Architectural Research further break down the difference between these subsets of computational design in their report Computational design in architecture: Defining parametric, generative, and algorithmic design.
What Are the Benefits of Computational Design?
Implementing computational design methods requires a cultural shift and extensive programming on the frontend, but once a design firm overcomes the initial learning curve, they will be able to:
Design Better Solutions – Designers can explore hundreds of design options rather than just the few they would produce with manual drafting. They can also take advantage of the unique design solutions generated that veer away from conventional thinking. Design algorithms can be refined to continuously improve outcomes.
Automate Repetitive Tasks – Updating a dimension or renaming a surface is simple when it just applies to one element, but it becomes tedious and profit shrinking when necessary across hundreds. With computational design tools connected to modeling software, a designer can create an algorithm that modifies the entire model in real-time.
Improve Productivity – Once firm-specific design processes are programmed into a computational tool, designers can essentially outsource design tasks to these programs. With computational design, architects can design faster with fewer iterations—improving productivity and accomplishing more with fewer resources.
Mitigate Design Risks – Iterative design processes and easy-to-use visual programming tools enable a designer to improve design quality above human capabilities. Artificial intelligence can be leveraged to test a design under multiple scenarios. Error-free designs reduce risk and liability for all parties involved.
Reduce Project Costs – Shifting tedious design tasks and design thinking over to computational design tools reduces the level of staff needed for a project. Additionally, algorithmically-produced designs will have fewer errors, reducing the likelihood of field design changes. With fewer resources and changes, projects costs will shrink.
Designing structures with interesting shapes such as those seen on the Twickenham Riverside Development Project, are made possible with computational design. Read more about the award-winning project here.
How Computational Design Is Used Now
Although computational design is an emerging concept to many in the industry, it has been used in practice across numerous building and infrastructure projects.
Parametric Design at the New Orleans International Airport
The design of the Louis Armstrong New Orleans International Airport terminal project began in 2011. It would be the first major airport to replace a terminal in the previous ten years. To meet a fast-paced schedule and achieve the crescent-inspired look, the design team, a joint venture between Atkins North America, eStudio Architecture, and Leo A Daly, knew innovative solutions would be necessary for the design approach.
One of the solutions used was a parametric model to assist with the design process. The spherical roof and radial grids were parametrically controlled with the visual programming plug-in Grasshopper enabled with Rhino modeling software. With the parametric model, designers were able to effortlessly adjust the geometry of the structure as the design was fine-tuned, allowing the design process to finish on schedule even after numerous design changes.
Generative Design Used By Japan’s Daiwa House Group
There is an inflated demand for urban housing in Japan. With 9 in 10 Japanese citizens residing in densely-packed cities, Daiwa House Group, Japan’s largest homebuilder, is tasked with the challenge of maximizing housing opportunities across the scarce available land. To do this, they utilize generative design practices.
Daiwa House Group uses generative design tools to speed up their workflow and present customers with unique house plans that maximize their building lot rather than rely on conventional methods. “Generative design … [offers] possibilities that deviate from conventions in positive ways. I think that is the technology’s greatest appeal,” explains Daiwa’s Project Director Masaya Harita.
How Computational Design Will Be Used in the Future
Computational design is positioned to alter the landscape of the AEC industry in the same way CAD and project management software have. Once initial barriers to entry are overcome, computational design will be a revolutionary addition to architectural design, engineering, and construction.
Computational Design in Construction
Computational design methods are beginning to transcend the scope of design and enter into construction. Contactors will input parameters about their unique construction sites and receive optimized data about procedures to enhance efficiency and reduce costs, including site improvements such as:
Optimized Equipment Positioning – For some construction projects, the entire schedule is based on the availability and movements of integral construction equipment. With generative design, contactors can receive project-specific optimized data on the ideal number and location of tower cranes and other heavy equipment.
Reduced Material Waste – Zero waste, or at least minimizing waste, is always the goal of a project, but proves difficult in practice. With computational tools, a project design could be optimized to reduce waste by using raw material data and generated waste evaluations.
Improved Order of Operations – Stacking and scheduling trades is a complex task that can make or break a project’s schedule. Computational design tools enable a project team to create logic that sequences the installation of building components. The sequence could then be optimized and improved upon for efficiency.
Implementing computational design across the AEC industry will require a cultural shift.
Although computational design tools are relatively easy to use thanks to visual programming, implementing the design method across a company or an entire industry is no small feat. Therefore, designers, engineers, and contractors intrigued by the efficiencies and optimizations computational design offers should advocate for its use, one project at a time.