Computational design has become common parlance in the field of architecture, and its associated technologies are now broadly incorporated into front-end design processes. These computerized means help us to rationalize and optimize complex geometries; forms that we otherwise could not model or even fathom building without the help of computer-aided design tools such as Rhino, Grasshopper, CATIA, Dynamo, Revit, and the like. The softwares are typically exploited early on in the design process to iterate multiple design options, but we rarely have the opportunity to use such tools during construction administration to advance execution of a particular element in the field.
This entry delves into the collaborative processes undertaken to construct three unique, compound curved, acoustic plaster soffits dubbed “the eyelids”. These eyelids are the underside of three north facing skylights soaring over two 5,500 SF learning studios in the University of Kansas Medical Center Health Education Building. The learning studios are half sunken into the earth, and the skylights carve into a manufactured green-roof landscape of hills, reminiscent of the surrounding Kansas countryside. These soffits negotiate between the geometry of the curved hillsides on the roof and a flat, interior ceiling.
Once construction commences and the design team hands off the Building Information Model (BIM) to the general contractor, they begin to dissect these models into the bits and pieces of information relevant to each trade and subcontractor. These subcontractors then produce “shop drawings” for the work they are going to execute. The drawings are typically planimetric representations in plan, section, and elevation, used for fabrication in the shop or for a construction worker to implement with the tools at his disposal in the field.
This process starts to fall apart when the construction geometry does not lend itself to planimetric representation and when computer-aided shop fabrication is not utilized to prefabricate unique geometry. Take for example, the compound curve geometry of the soffits with an infinite number of unique plans and sections. These soffits are framed with straight sections of standard 2’x2’ ceiling grid, so shop fabrication cannot assist with construction of the geometry. You might imagine this leading to construction workers in the field, scratching their heads, holding tabloid PDF printout of a shape that doesn’t translate onto the piece of paper in their hand. This urged us to work closely with the GC and subs for the KUMC project to develop a process that translated the three 75’ long soffits into 837 measurable work-points in space; data that they could use in the field.
This process began with measurements the general contractor provided from the primary structural steel of the eyelids. We developed a set of live parametric surfaces based upon these heights in order to start coordinating the eyelids’ finished surface. We were focused both on making sure there were no collisions with the steel that had been erected and ensuring that the eyelid design transitioned seamlessly with the final ceiling surface.
We took the opportunity to explore and compare parametric tools which would provide us with the final ceiling layout locations. The first was developing a node-based script in Grasshopper for Rhino. While facilitating rapid development, the Grasshopper script required creating quite a few nodes just to define the leading edge of the eyelid surface. The portion highlighted in green shows what was needed to define that curve.
For comparison, we setup the same curve as a parametric sketch in both CATIA and OnShape. Both support a more intuitive and visual approach to laying out and manipulating the data-driven geometry. The network of connected nodes defining the curve in Grasshopper contrasts with the image below of a constrained sketch in OnShape. With OnShape, any manipulation could be done easily and without exhaustive instructions to consultants and subcontractors.
After the eyelid surface models were set up and could be adjusted automatically, we went through a series of modifications to find the optimal shape to meet all the design constraints and coordination conditions. For the final deliverable, we provided the work-points for the ceiling grid to the GC as a point list in Excel. The automation allowed us to meet all our design criteria quickly while providing the contractors with data they could easily execute.
In the field, the sophistication of the digital tools used to develop design and geometry are only as good as the translation of the design into readily usable information on the construction site. Through a collaborative, team-based approach, we learned what data the construction team required to execute this complex design feature, and worked with them to determine how best to derive it from the design.