Design Space Exploration with Dynamo

One of the most challenging aspects of the architectural design process is determining how to organize form to fit an overall parti. Facing endless possible geometric configurations, making sequential alterations towards a fitting result can be difficult without a means to measure suitability. During the initial phases of design research, an architect gathers essential information such as program requirements to meet a clients needs, zoning and code information for a provided site, environmental and material influences, and aesthetic preferences. These assets serve as the foundation for a constraints based design approach where parameters can be assigned in an effort to influence and control form.

Constraints in design are rules or vocabularies that influence form through the design process. An inherent feature of the architectural process is that design must be performed within a set of given parameters. Parameters help to focus the scope of an architect by narrowing the range forms and formal relationships may take within a design solution... Constraint based design takes the parameters associated with a design problem and links them to the attributes of the formal components and relationships of a solution. (Dustin Eggink, )

Dynamo is an ideal platform for constraints based design because the visual programming environment allows you build a parametric model that can be quickly adjusted with changes to input values.

Once you have a functioning Dynamo definition, all of the nodes can be consolidated into one Custom Node by dragging a selection window over everything and going to Edit > Create Node From Selection. This will transition everything to the custom node editing mode -- you can always tell when you are in this mode because the background is yellow.

To create a custom node, the first step is to give it a name, description of what it does, and category (where it will be saved in the Library). All of the input number blocks (far left side) must be swapped out for Input nodes, generally named for the variable they represent. Output nodes also need to be added after the final nodes in the definition (far right side) that are providing the finalized geometry. When these steps are complete, save the node. Back in the Dynamo node space -- also known as the canvas -- number sliders can be added to the newly-created custom node. It is helpful to click the down arrow on the left side of the node to set the minimum, maximum, and step interval because large numbers can take awhile to process or crash Dynamo while zeros will often create null values and turn the majority of your definition yellow with warnings. Now you have a fully parametric custom node that allows you to explore a range of formal configurations with the simple adjustment of number sliders.

Developing custom nodes for form making allows for use with the Dynamo Customizer -- a web-based viewer currently in beta for viewing and interacting with Dynamo models real-time. This platform has a lot of potential for sharing designs in the future and allowing colleagues or clients to experiment with their own manipulations of the design.

Check out this example for the twisting tower here: Dynamo Customizer - Twisting Tower.
DISCLAIMER: you will have to request Beta access and sign in with your Autodesk ID to view this. For step-by-step instructions, visit:

After guiding parameters have been established, a design space can be generated for testing all possible variations of a few select variables of a design. Design space exploration is a concept involving a virtual -- or non physical -- space of possible design outcomes. This allows the designer to simultaneously see a wide range of options and extract only those that satisfy pre-determined criteria of fitness.

The core essence of this workflow is the use of Cartesian product which facilitates comparison of all possible pairings of variables. This mathematic operation can be understood as an array of combinations between x, y, z and 1, 2, 3 (below left) or as a slope graph of all possible correlations between the two lists of variables (below right).

Using the List.CartesianProduct node calculates all possible combinations of the number range values however all of the geometry is instantiated in Dynamo at the origin point, making it appear that only one object was created even though the count shows 132 (below left). Thanks to Zach Kron and the Design Options Layout node from the Buildz package, the nested list clusters of geometric objects are arrayed according to the Grid Size spacing value (below right). Using list management logic -- such as List.Transpose, List.Map with List.Transpose, etc. -- before the Design Options Layout node will re-arrange the list structure and result in different compositions of objects.

To set up a design space in Dynamo, the inputs to the custom node are fixed values. Whichever variables that you want to test must be left empty on the custom node and number ranges are connected to the List.CartesianProduct node. The number of list inputs in List.CartesianProduct must match the number of inputs left open on the custom node. It is also important to note that the list order for the number ranges in the List.CartesianProduct node must correspond to the same order of inputs in the custom node. The total number of values from each number range will not only determine the scale and form of the resultant geometry but count of values in each list will determine the overall size and shape of arrayed objects -- this is critical to remember because an excessive number of input values may take several minutes to process or potentially crash Dynamo. After the ranges of values have been set up, the List.CartesianProduct node is connected to the Design Options Layout node which arrays all possible combinations in 3D space. Depending on the geometry being tested, the Grid Size input determines the spacing between objects. When everything is connected correctly, Dynamo will display an array of forms which can be altered by changing the number range inputs and re-running the definition. If Dynamo crashes, geometry disappears, or there is an insufficient amount of variation in the forms, continue to calibrate the number ranges and explore the limitations of the parameters in your custom node.

A successful design space arrays all possible options along two or more axis utilizing the concept of dimensionality. Design space is theoretically unlimited, however the visualization of the virtual design space is limited to the constraints of graphic representation. Color can be added to provide visual differentiation of a third dimensions such as the analysis of generated outcomes, or could represent any of the associated properties of variables. Criteria for evaluation of fitness refers to the means by which the best solution is determined.

For example, a calculation of height of twisting tower forms can be colorized on minimum to maximum gradient (below left). Another representational technique is selective omission or hierarchical modifications to the representation (below right).

Ultimately design space and its subsequent representation is nothing more than a tool, designers still have to make decisions. Design space should function as a method of exploration to make informed, confident, substaintiated decisions.

Portions of this blog post were developed in collaboration with Jamie Farrell for our course Advanced Revit and Computational Workflows taught at the Boston Architectural College.

AEC Technology Symposium 2015


AEC Technology Symposium 2015
Hosted by Thornton Tomasetti (NYC)
Baruch College
September 25th, 2015

Measurement Moxie
Christopher Connock - Kieran Timberlake

Christopher Connock emphasized the importance of approaching each architecture project as an experiment — an opportunity to test new technology and ideas — a philosophy that Kieran Timberlake incorporates into all of its projects. They are particularly exploring the frontier of data capture using a wireless sensor network to gather building performance analytics. The comparison of plant diversity and placement against soil moisture and temperature sensors in a green roof can help assess drainage, plant health, and solar gain over time. Temperate and relative humidity sensors can be used to investigate an entire space or focus on a particular application such as the performance of materials in a building envelope. Hundreds of different sensors placed throughout a building can track and transmit environmental changes across a given day or even across seasons. Kieran Timberlake implemented a wireless sensor network to help inform their renovation of their new offices in the Ortlieb Bottling Plant in Philadelphia and then developed an in-house app to capture post-occupancy feedback from their own employees about the overall comfort of the space and to identify abnormal conditions. All of this research contributed to the development of tally — a life-cycle assessment (LCA) app and add-in for Revit that evaluates the environmental impact of building materials among design options and promotes a much more eco-conscious approach to design.

Grow Up, Grasshopper!
Andrew Heumann - NBBJ

Andrew Heumann believes in the need to change the perception of design technology in the AEC industry and integrate it more into practice. He showcased an extensive portfolio of projects that have used Grasshopper for Rhino and custom written apps to simulate inner office traffic patterns and the importance of sight lines, the use of human location and city data for urban planning, and the tracking of digital tools in the office to identify focus areas for development and support. All scripts and tools developed in-house at NBBJ are documented and packaged into products for use by project teams. In addition, custom dashboards and user interfaces help reduce intimidation and increase universal adoption — for example, reducing a complex Grasshopper script to a series of slider bars that control the inputs of a parametric design. Andrew also advocated for the use of hackathons and similar hands-on user meeting formats to promote design technology as a facet of culture and process. He shared an example of how a brief hackathon with senior partners at NBBJ led to the funding of a proposal for further development of an innovative tool for optimizing healthcare patient room configurations.

Evolving Modes of R+D in Practice
Stephen Van Dyck and Scott Crawford - LMN Architects / LMN Tech Studio

The Tech Studio was founded to support the prominent role of research and development at LMN Architects in Seattle, which led to an expanded use of analytic and generative tools to drive design. They have not only embraced the use of custom digital tools for the creation and visualization of complex forms but regularly construct scale models for material testing and to explore modular strategies as part of their iterative design process. Working with fabrication in mind facilitates improved precision for collaboration with engineers and consultants. In addition, they have found that a thorough digital process and physical models help better communicate design ideas thus resulting in increased positive community feedback. The development of a ray-tracing tool for exterior acoustics studies, custom panel creation for balance in musical acoustics and aesthetics, and a highly parametric pedestrian bridge spanning a major Seattle highway are a few examples of projects that demonstrate how research is guiding principal for design at LMN.

Collaboration and Open Source - How the Software Industry’s Approach to Open Sourcing Non-Core Technology Has Created Innovation
Gareth Price - Ready Set Rocket

This presentation provided insight to the current state of technological innovation through the lens of a digital advertising agency. Gareth Price emphasized that individuals should not be hesitant to share ideas out of fear that another company will benefit from them. Particularly in the AEC industry, companies do not have the overhead to for pay tool creation and requisite support, nor can they cover the cost to pay an outside software consultant. The reality is that other people are busy with their own work and do not have the time nor resources to steal your ideas and commodify them. More importantly, it is advantageous to share ideas for a project because they may elicit constructive criticism, or inspire others to contribute to those ideas and improve them. Also, do not get too entrenched on one idea and know when to pivot — the next great idea may come as an unexpected derivative of the original intention.

Key quotes:
"purpose is the DNA of innovation"
"failure is the new R&D"

How Opens Source Enables Innovation
Mostapha Roudsari and Ana Garcia Puyol - CORE studio / Thornton Tomasetti

Mostapha Roudsari and Ana Garcia Puyol exhibited many examples of digital tools that have emerged out of CORE Studio - the research and development arm of Thornton Tomasetti. The majority of examples presented originated during previous CORE AEC Technology Hackathons and were then further developed into more robust products. Nearly every tool required collaboration from multiple individuals, with expertise in a diverse mix of software platforms, and oftentimes representing different companies. The takeaway from this presentation was the value of open source and hackathons as a means for getting a group of talented people into one room to create new tools for AEC design and representation. Mostapha wanted to make it clear that more important than software tools, code, and machining is the strength and power of the user community. If you want to be at the forefront of the movement, be a developer, however the community is just as important make an effort to share ideas and spread adoption.

Here are some of the many tools presented:

  • vAC3: open source, browser-based 3D model viewer. This project led TT to further develop Spectacles.
  • Spectacles: allows you to export BIM models to a web interface that allows you to orbit in 3D, select layers, and access embedded BIM information (demo HERE)
  • VRX (Virtual Reality eXchange): a method for exporting BIM models for virtual reality viewing via Google Cardboard
  • DynamoSAP: a parametric interface that enables interoperability between SAP2000 (structural analysis and design), Dynamo, and Revit
  • Design Explorer: "an open-source web interface for exploring multi-dimensional design spaces"
  • Pollination: "an open source energy simulation batch generator for quickly searching the parameter space in building design"

For more information, check out TT CORE Studio's GitHub, Projects, and Apps pages

Open Source: Talk 3
Matt Jezyk - Autodesk

Matt Jezyk provided an introduction to Dynamo including its history and the most recent developments. Dynamo may have started as a visual programming add-in for Revit but it is quickly transforming into a powerful tool for migrating data and geometry across numerous software platforms. The talk highlighted the role open source has played in the empowering independent developers to create custom content that expands capabilities and makes interoperability possible. By keeping Dynamo open source, it has benefited from contributions by individuals with a wide range of expertise looking to satisfy specific requirements. As part of a larger lesson taken from the growth of Dynamo, Matt emphasized that the key to the emerging role of technology in practice and the AEC industry as a whole is less about learning specific tools but about codifying a way of thinking — tools are only the implementation of a greater plan.

Beyond Exchanging Data: Scaling the Design Process
Owen Derby -

Flux has been a frequent topic of conversation lately. The company initially marketed a product called Flux Metro which boasted the potential for collecting the construction limitations of any property based off zoning, code, municipal restrictions, and property records — an ideal tool for developers and architects to assess feasibility or use as a starting point for the design process.

The company has since pivoted to focus on creating a pipeline for migrating and hosting large quantities of data for many software formats. Their new product line features an array of plugins for transferring data between Excel, Grasshopper, and Dynamo, with plans to release additional tools to connect to AutoCAD, SketchUp, Revit, 3DS Max and more in the near future. Data exported from these software programs is hosted to a repository in the cloud where it can be archived and organized for design iterations and option investigation. Flux has great potential for achieving seamless interoperability of data and geometry between software platforms, and significantly improving the efficiency of AEC design and production process.

Holly Whyte Meets Big Data: The Quantified Community as Computational Urban Design
Constantine Kontokosta - NYU Center for Urban Science + Progress (CUSP)

The NYU Center for Urban Science + Progress (CUSP) is using research to learn more about the way that cities function. Buildings, parks, and urban plans are all experiments built on assumptions in which the true results don’t emerge until years and decades later. How do you measure the “pulse” of a city? How do macro observables arise from micro behavior? Constantine and CUSP have set out to test these questions by collecting and analyzing: NYC public internet wireless access points, Citi bike share, 311 complaint reporting, biometric fitness devices, and social media. They use these urban data sources to make better decisions and form initiatives for future community improvement projects. The results also have positive implications for city planning, city operations, and resilience preparation.

Data-Driven Design and the Mainstream
Nathan Miller - Proving Ground

Nathan Miller is the founder of the Proving Ground, a technology consultancy for Architecture, Engineering, Construction, and Ownership companies. In his experiences providing training and technological solutions he professes the importance of equipping staff with the right tools and knowledge to adequately approach projects. There is an intersection between managers and leaders responsible for projects and staffing, and those who are actually doing the work. It is imperative to focus on outcomes and not get deterred by the process.

The Biggest IoT Opportunity In Buildings Is Closer Than You Think
Josh Wentz - Lucid

Energy consumption, mechanical systems data, thermal retention, and other metrics are not recorded for the majority of buildings worldwide. These are incredible missed opportunities for evaluating the overall performance of a building and collecting real-time research that can inform better construction techniques. Lucid has developed a product called BuildingOS that offers 170 hardware integration options to collect robust building data. This data has the potential for helping facilities management departments better track efficiency and maintenance of their systems, in addition to contributing to the international pool of data to help us better understand how materials and systems perform over time.

Capturing Building Data - From 3D Scanning to Performance Prediction
Dan Reynolds and Justin Nardone - CORE studio / Thornton Tomasetti

This presentation highlighted CORE Studio's use of various technologies for capturing existing conditions data and testing architectural responses through computation. They have utilized drones for capturing the condition of damaged buildings and structures by assembling fly-by photos into a point cloud.The development of in-house GPS sensor technology accurate to 1 centimeter anywhere in the world has enabled measuring the built environment and construction assemblies to a high level of precision. CORE Studio has also investigated the use of machine learning for exploring all possible combinations of building design parameters and calculating embodied energy predictions. All of these design technology advancements are helping Thornton Tomasetti design more accurate better-informed systems.

Data-Driven Design
Luc Wilson - Kohn Pedersen Fox Associates PC

Luc Wilson thinks of data as an “Urban MRI” - a diagnostic tool for measuring the existing configuration of cities and predicting future growth. Multiple FAR and urban density studies were presented that exhibited how comparison to precedents and benchmarks helps to conceptualize the data and make visual sense of the analysis. The key to prediction is the ability to test thousands of designs quickly, which Luc has perfected by developing digital tools for quickly computing all possible combinations of input parameters and producing measurable outcomes for comparison. One of the most exciting portions of the presentation was the mention of a 3D urban analysis tool called Urbane, which Kohn Pederson Fox is working with NYU to develop — could this be the new replacement for Flux Metro?

Cellular Fabrication of Building Scale Assemblies Using Freeform Additive Manufacturing
Platt Boyd - Branch Technology

Platt Boyd founded Branch Technology after realizing the potential for 3D printing at a large scale by imitating structures found in nature. Branch has dubbed their technique “cellular fabrication” where economical material is extruded with geometric complexity to construct wall panels that are lightweight and easy to transport. Their process seeks to reduce the thickness required by traditional 3D printing technologies and the intricate geometric structure provides equivalent strength to that of a printed solid. The wall panels are printed free-form with a robotic arm on a linear track and then are installed onsite where insulation, sheathing, and finish material are added to reflect the same condition as traditional wood or metal stud construction. It will be really interesting to see Branch continue to refine their methods and start to tackle complex wall conditions for use in real-life building projects in the near future.

Watch videos of all the presentations HERE.