Computational Design for Assembly: Automating Design Workflows for 3D Concrete Printed Freeform Staircases
Interview with Philip Schneider + Timo Zollner
In this interview ahead of their presentation at CDFAM, Philip Schneider and Timo Zollner discuss how computational design and additive manufacturing are opeming new opportunities for concrete construction.
Building on Scawo3D’s development of the world’s first commercial inkjet print head for cement paste, enabling their Selective Paste Intrusion (SPI) process.
They outline a shift from traditional CAD–CNC workflows to fully automated design-to-production methods for freeform concrete staircases.
Their talk at CDFAM will demonstrate how their 3D StairDesigner plugin, developed in Rhino and Grasshopper, dramatically accelerates design iteration, integrates voxel-based and implicit modeling, and bridges digital automation with physical fabrication in concrete 3D printing.
Could you start by describing Scawo3D’s work and give an overview of what you will be presenting at CDFAM?
Philip: Scawo3D is the inventor of the world’s first inkjet print head for cement paste.
It serves a novel 3D concrete printing process called Selective Paste Intrusion (SPI) that is based on a large particle bed. We both develop the print head and machinery required for SPI production, while also providing 3D concrete printing as a service.
Due to the company’s history in staircase construction, we have always used the stair as a benchmark due to its complexity, both geometrical and structural. Building stairs digitally all the way from 2D planning through to 3D modelling and ultimately digital fabrication has been one of the main objectives in developing a particle-bed-based 3D printing method for concrete.
Once that the SPI method started to produce sound results in 2022, we shifted the focus to developing a plugin based on Rhino and Grasshopper that enables us to design any kind of geometrically complex stair with only very little restrictions and in a matter of an hour instead of multiple days.
At CDFAM we’ll talk about the plugin, 3dStairDesigner, and the actual results it produced to date – ultimately, it’s a linear story from computational design through additive manufacturing and assembly.
Your presentation outlines a shift from a fairly manual CAD-CNC workflow, to an automated design process for 3D concrete printing. Can you explain how this workflow is structured and what this approach enables?
Philip: The very initial workflow was based on AutoCAD. This was owed to the fact that it had to produce G-code for CNC-milling EPS blocks – our old method of producing formwork elements for stairs.
The workflow was perfectly suited for CNC-milling but had too many geometrical limitations for exhausting the geometrical freedom of the SPI method. We could no longer use it to design the first SPI stair ordered by a client.
Thus we resorted back to modelling by hand using… What a mess it was, nervous breakdowns on the human’s as well as the computer’s side included. On the upside, it made us realize that we need an automated solution and brought in Timo who went on to save us a lot of time and nerves!
Timo: With the old workflows, the modeler would spend many hours creating the complex shape of the formwork. The boolean operations involved caused many challenges, such as channels for rebar and added stiffeners. Changing the geometry was practically impossible.
We then devised a workflow where everything happens in 2D, except for modeling the boundary conditions (walls and similar). You can draw the stair path, the stair profile, the step shape, and input values for thicknesses and other properties. We then autogenerate a stair constrained to the boundaries of its space. This makes precise modeling much easier and allows for quick adjustments to the geometry.
To address the issues with boolean operations, we convert the model into a voxel-based representation at a resolution slightly lower than that of the printer, which makes boolean operations robust. Finally, we automatically split the stair into individual modules ready for print. The estimated time reduction in design is around 95%.
Which software tools and modeling techniques form the core of your process, particularly regarding SubDs, meshes, volumetric, and implicit modeling and how are they integrated?
Timo: Rhino and Grasshopper with a UI on top. For SubDs and meshes, we use Rhino’s native elements, and for voxels, we use Dendro, which is an open-source implementation of OpenVDB. For implicit modeling, we use the Isopod plugin by Daniel Piker.
How does design data move from initial 2D geometry through automation to the final 3D print-ready model, and what checks or adjustments are typically required along the way?
Timo: We allow the user to input all information in 2D, either in plan or rolled-out elevations. This approach covers nearly all stair types. No adjustments in 3D are typically required. However, many checks are needed, and we have built several tools to assist with these. These checks include ensuring sufficient concrete thickness (after casting) and confirming that the division of elements works with both printer and construction constraints.
Another important point is that we do not just press GO, pray, and hope for the best. We first build a simple model for quick review, then a more complex one.
Based on that, we adjust the positive and negative shapes added to the main stair model. We then export this model (as a mesh) to Rhino and run another tool that prepares it for printing and performs the boolean operations.
Digital: Boundary conditions, structural checks and details for assembly and reinforcement.
Physical: SPI printing and unpacking, assembly and completed result.
In what ways has the Selective Paste Intrusion method influenced the parameters or constraints you use in your computational design workflows?
Timo: The main influence is on the resolution of voxel geometries, which depends on printer tolerance. When dividing the geometry for printing, we also consider the strength of the elements and printer bed size to avoid making them too large, and we ensure that all loose sand can be removed from voids. These considerations affect the user’s choices within the software rather than the software itself.
What do you hope to share with, and learn from, other attendees at CDFAM regarding computational design for large-scale additive manufacturing in concrete?
I want to share the idea that no single geometry definition is better than others. By using the strengths of each representation, you can combine the best of all approaches.
Timo Zollner
I want to show how it is possible to build truly responsive software with Rhino that is fully tailored to specific fabrication needs. I also hope to meet many interesting people in the industry and learn from their experiences.
Philip: With our presentation, I’d like to spread the word about SPI and what effort it took to start exploiting its possibilities. In the end though, stairs merely scratch the surface of what SPI is capable of producing in the context of our built environment, and I’m mostly looking forward to engaging with everybody at CDFAM to hear their take on where SPI could go in the future.
Thanks for letting us join CDFAM!
Philip and Timo’s work reflects the broader shift toward automation and computational design that is reshaping how complex structures are conceived and built.
Their approach, bridging algorithmic design, voxel modeling, and additive manufacturing illustrates how digital precision and material intelligence can converge in practice.
At the CDFAM Computational Design Symposium in NYC, you’ll have the opportunity to connect with Philip, Timo, and other designers, engineers, and researchers developing similar methods across disciplines.
Join the conversation on how computation is transforming the design-to-production process, from materials and products to large-scale architectural systems.
