LMN’s transformation of the San Antonio Public Auditorium into the Tobin Center for the Performing Arts required the old theater and stage house to be removed and replaced with a contemporary theater and back-of-house areas. The concept for cladding this addition was to wrap all of the new construction in what appeared to be a simple, delicate surface whose appearance would constantly change in different lighting conditions. The design team referred to this piece of the building as the Veil. While the concept was simple, the actual system was fairly complex. There are over 18,000 panels, with 8 unique panel types occurring in various patterns. Luckily, even though the system was complex it was not complicated.
Parametric modeling was not used during the design of the Veil, but the modeling work-flow was like a mental parametric model where the critical parameters and logic of the system were held in the minds of a few individuals instead of a Grasshopper definition. The designer developing the study models for the Veil was very rigorous in forming the logic of the system and was capable of producing a large number of studies by continuously modifying or remodeling the Veil using linear arrays. LMNts was asked to help develop a Building Information Model that could be used by the contractor to do more detailed quantity takeoffs as well as address a few issues that proved difficult to solve through the direct modeling process.
There are a number of similarities between the layout of this system of panels and the way that we developed the base geometry for the Cleveland Medical Mart. Both approaches begin with a shell model that is broken down into a grid according to the width and height module of the base unit panel. An issue that often arises with modular systems is what happens when the dimensions of a face aren’t a multiple of the module. For example, the north and south elevations of the Medical Mart were 195′ wide and we were using a 10′ module. This left us with a difference of 5′ to the closest module multiples of 190′ and 200′. The difference was taken up in uniquely sized corner panels, a piece of the facade that was going to be unique no matter what.
The Veil is not a water tight enclosure so the layout flexibility of the system was taken up in the gap between panels which could range from 0.25″ to 0.50″. This was originally causing problems in the direct modeling approach because even slight changes to the base geometry required the panel spacing to be recalculated for each face. Once we had the Veil modeled in Grasshopper it was possible to calculate the number panels that needed to occur on each course and then adjust the gap to be within the acceptable range. Adjusting the gap size also afforded us more flexibility when trying to align the panels with openings and structure.
The design team had established a variety of panel types and organization patterns for how the panels were arrayed across the building. Here again there was a series of rules that the patterns were following, making it fairly straightforward to replicate the design. The challenge with parametric modeling process is incorporating exceptions to rule. For every exception there needs to be a system of logic defined to always pull out the exception and handle it accordingly. These exceptions end up carrying over into the fabrication and construction process as well. The more exceptions, the more room for potential errors to occur.
In the above image, there are a series of trapezoidal panels that occur along the top edge of the Veil. Each of these panels are unique and there are over 1300 of them. The current fabrication plan for the project is to field cut full-size panels. We’ve digitally modeled every one of those panels so there’s the potential that shop drawings could be provided for all of them. To demonstrate this potential we put together a work-flow that writes to a spreadsheet the left and right edge height for each panel and these values are then used as a Design Table for an Autodesk Inventor sheet metal file. We don’t often use Inventor but fabricators do so we thought the fabricator building the Veil might be interested in this information.
Turns out we were wrong for a few reasons. They were intrigued that we could provide them with that information, but at that time the fabricator was only responsible for making the panels and another contractor was responsible for installation. Therefore, the time investment in machine code setup, tagging and crating the 1300+ panels had no payoff for them. From our point of view, shop cutting the panels seemed like it would offer more control and likely a better end product. The increased setup time and effort involved with tracking the unique panels could potentially be offset through material and labor savings.
In the end, means and methods are not within our scope of responsibilities and we’re not responsible for selecting who builds what on the project so there are limits to how much we can influence the decision this time. However, more and more we are developing models that contain a significant amount of information that could be useful in the fabrication process and we’re looking for opportunities to share that information rather than simply handing over 2D drawings and asking someone to take their best shot. We’re now trying to make our consultants and collaborators more aware of the control we have over our models and the information we can extract from them with the hope that someone will take full advantage of this information.