Parametric modeling aids the design for a complex paving pattern at a corporate campus.

By Jessica Canfield, ASLA

When stepping off the city sidewalk and into the site of the Cummins headquarters building in Indianapolis, there’s an immediate sense of arrival into a distinct landscape. David A. Rubin, FASLA, the principal and lead designer at DAVID RUBIN Land Collective, says that the site is an expression of choice, with amenities for collaboration and contemplation, “allowing people the capacity to choose where to be most creative.” This could be in a cluster in the amphitheater, in movable seating, at an isolated bench, or around the long, Wi-Fi-enabled community table, dubbed the “High-Tech Harvest Table” by the design team.

Located in downtown Indianapolis, the Cummins DBU (shorthand for Cummins Inc.’s Distribution Business Unit headquarters) site spans a full city block. Along the site’s western edge is the Indianapolis Cultural Trail, a citywide bike and pedestrian path. Sweeps of vegetation planted atop elongated berms extend inward from here to guide circulation and, as Rubin describes it, to create a sense that the landscape was intentionally carved back to reveal the underlying hardscape. The main path, which zigzags east–west to connect the building entry and parking garage, widens at the heart of the site to become the central plaza. This multifunctional gathering space is framed by amphitheater seating and can accommodate performances and special events. Just adjacent is a communal work space, the Social Hub, where employees can bring their laptops and connect to power and Wi-Fi. A more secluded area, the Dell, offers benches for quieter work. These distinct subspaces are threaded together by a continuous two-toned paving pattern, creating a unified surface and visual identity for the site.

The eye-catching paving pattern, comprising alternating bands of light and dark concrete pavers, echoes the calibrated facade of the new Deborah Berke Partners building and is emblematic of a checkered flag, in reference to the Cummins diesel engine enterprise. The design team first explored concepts for the paving pattern through sketches and a 3-D model. According to Land Collective’s project manager, Henry Moll III, Affiliate ASLA, “Early studies included larger concepts of fading patterns and pixilation,” but ultimately they went with the more geometric and focused pattern. After selecting the two-toned scheme, the team turned to Grasshopper to further explore and refine the pattern’s scale and color distribution.

Grasshopper, which is now included in Rhino 6, is a visual scripting tool used for parametric modeling. In parametric modeling, design outcomes are created through the application of scripts, which establish and define relationships between components within given constraints. In a design workflow, a script can be used for ideation or for accomplishing a specific task. Moll describes Grasshopper as ideal for working with repetitive elements, because you can automate complex goals, which lends itself well to developing patterns.

The paving script for Cummins included the boundary of the hardscape area and multiple sizes of unit pavers and joint orientations. Once those components were created and their relationships established, the team could experiment with a variety of pattern and color options. They tested variables such as different levels of contrast between the bands, different amounts of a specific paver color, and how it would look if one paver color were shared between geometries. When adjustments to the bands’ geometries were made in the model, the pattern would regenerate to fit within the new constraints. These studies, which were accomplished in minutes with the parametric model, would have taken hours to produce manually. Since the patterning possibilities were essentially limitless, Moll says establishing a clear design goal was important.

A mix of colored pavers achieved the paving pattern’s dappled effect. Each of the light and dark bands combined three colors in quantities of 15 percent, 42.5 percent, and 42.5 percent, proportions that the parametric model helped establish. Precast concrete pavers were chosen for purposes of durability and maintenance and because they offer more opportunities for nuanced pattern making, as compared to cast-in-place concrete. The design team initially explored using pavers in warmer colors, but ultimately chose a palette of cool grays after examining an on-site mock-up. Land Collective worked with Harold Henry, a technical sales director at Hanover Architectural Products, to customize the palette colors.

For the Cummins DBU project, Hanover produced six custom blended colors of its Traditional Prest Brick paver in two thicknesses. Vehicular-grade pavers were used for the main east-west path and central plaza to support special event infrastructure. Secondary pathways use pedestrian-grade pavers, tightly placed onto a bituminous setting bed with sand-swept joints. Both paving systems employ four-by-eight-inch solid concrete interlocking pavers, which are supported with reinforced concrete slabs and a compacted aggregate base. Color and texture in concrete pavers come from the parent materials used and finishing techniques, which Henry explains can be achieved in different ways. Variation is created through the types of aggregate, sand, cement, and pigment additives. To create lighter color pavers, like those at Cummins DBU, white cement is used instead of gray. Hanover produces its pavers in south central Pennsylvania, “in the heart of very good aggregate country,” Henry says.

Though permeable paver systems are available, the design team chose to manage stormwater more visibly with a system of infiltration gardens that capture runoff and direct overflow into an underground cistern for irrigation use. Rubin says that all stormwater runoff is captured on site, with additional infiltration occurring in the green roof atop the parking garage and in the meadow that blankets the southern half of the site. The meadow’s generously wide mowed paths double as usable green space and suggest the location for an expanded plaza in a future development phase.

The paving script developed for the Cummins DBU project evolved during the design process, starting first as a tool to generate graphic patterns, then as a means to calculate paver color percentages for the manufacturer and contractor. Though the script could have assigned a color to every paver in the pattern, that level of specificity would have been overly prescriptive. Instead, information about the pattern was communicated through enlargement plans, illustrative details, and implementable rules. For example, to ensure even color distribution within a given band, the design team developed a rule that no more than three pavers of the same color should align. Another rule stated that only whole or half pavers should be used along transitions to ensure a crisp delineation between bands. With this guidance, contractors were given leeway when installing the pavers, which arrived on site in palettes that the manufacturer prepackaged in the specified color proportions, as defined by the design team. During construction, Land Collective was on site regularly to examine the pattern and adjust individual blocks as needed. “We were expecting some pushback on the complexity of the design, but we worked closely with the contractors, and it actually went a lot smoother than we anticipated,” Moll says.

“One of the biggest takeaways from this [project], and something I am trying to continue to improve upon, is to first think about the options for constructability,” Moll says. “How are you going to be able to communicate this to the installers? The question becomes a big influence for how you might approach the assembly of the script; otherwise you might create outcomes that look intriguing but are ultimately not executable in built form.”

Despite its mainstream use within architecture, parametric modeling remains fairly uncommon within landscape architecture, according to the 2019 design software survey, authored by ASLA’s Digital Technology Professional Practice Network and the professors Benjamin George, ASLA, and Peter Summerlin, ASLA (see “Get with the Program,” LAM, November 2019). When ASLA members were surveyed about which technology applications they use to operate effectively and efficiently, 20 percent of 482 responders indicated they used parametric modeling, and only 18 percent of those not currently using parametric modeling indicated interest in future adoption.

At Land Collective, parametric modeling is regularly used for discovering and testing ideas and as a time-saving tool for tedious tasks. Rubin says the studio works equally in writing code and in hand-drawing: “We endorse both.” Together with his colleagues, Moll has developed an in-house library of scripts that can be tailored to analyze slopes, generate topography, customize handrails, and more. Moll, who learned parametric modeling in practice, says constructing an initial script can sometimes be time-consuming and frustrating, but with a solid understanding of a project’s real-world constraints, it is a worthwhile investment.

Not all scripts start from scratch. Many are open source and easily found online for download. The parametric modeling community is very supportive, Moll says. Outside Grasshopper’s own resources, there are numerous online tutorials and forums for learning, including Generative Landscapes, a blog created by Joseph Claghorn, a University of Sheffield lecturer in landscape architecture. LinkedIn Learning also provides a range of resources for self-paced exploration of Grasshopper and Rhino.

Before developing a script, Moll suggests diagramming it by hand, establishing what it should accomplish and what all the possible variables could be, since a script typically grows in complexity as it is developed and applied. The beauty, he says, “is that you can go back and change parameters. It’s not a linear process,” and through experience you learn what to script and what not to script. Being comfortable with ambiguity is also important for parametric modeling. “A lot of times I’m given a problem to solve, not knowing exactly what to do or how to solve it, but through testing and script building, hopefully I find the solution.”

Jessica Canfield, ASLA, is an associate professor of landscape architecture at Kansas State University.