2018 WhiteheadR Portentous and Predictable
2018 WhiteheadR Portentous and Predictable
Abstract
This paper will examine how the problems of Kresge Auditorium stemmed from a structurally
inappropriate building form, a set of under-designed and documented drawings/specs, and a building
process that lacked oversight and engagement on-site from the design team—all deviations from
established thin shell design and construction standards. This paper will argue that many problems
were predictable and avoidable, but were exacerbated by contractual separation of responsibilities and
lack of integration between the designers: Eero Saarinen & Associates, Ammann & Whitney, and the
builder: George A. Fuller Company.
Keywords: Kresge Auditorium, Saarinen, Ammann & Whitney, George A. Fuller Company, Shell Construction, Shell Failures
The conference also featured two central participants in Kresge’s development: the structural
engineering partner, Charles Whitney and the construction foreman, Douglas Bates. At the time of the
conference, both Whitney and Bates knew of the difficult and costly design and construction
conditions that proceeded this moment, and the current plans to support the sagging arches with steel
columns. Following Saarinen’s lead, neither mentioned this in their presentations. Whitney only
mentioned Kresge once—conspicuously saying shells like it, “must be judged as special architectural
and structural problems and there are no general rules as to their justification.” (Whitney, [4]).
Figure 1: Exterior and Interior view of glass wall with supplementary steel columns shown.
2
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
shell projects in Europe in the previous decades, Kresge’s design and construction reflected more of a
traditional mid-century American architect/engineer/builder roles.
Like most architectural projects commissioned in the mid-20 century in the U.S., Kresge was
th
contractually set-up by MIT to follow a “design, bid, build” process. As the architect, Saarinen was the
contract holder with MIT and his firm would hire consultants (including structural engineers) who
would design the project and produce construction drawings and specifications that would be bid upon
by several General Contractors and built by the lowest qualified bidder.
As a consequence, during the design process there would be no opportunity for Saarinen and his
consultants to collaborate with a builder on critical issues—all the technical expertise had to come
from the project team and be contained and communicated to the eventual contractor through
construction documents. This arrangement stood in contrast to the predominant “designer / builder”
roles practiced by the most successful shell builders: Candela, Nervi, Tedesko, etc. The separation
between design and construction, as mandated by a bidding process was highly consequential.
Figures 2A-2B (left to right): Fig. 2A, Sketches for shell geometry and support, 1952, Fig. 2B, Final model
showing three points of support and edge beams, 1953.
In a letter to his future wife, Saarinen described how he became fond of one model, which was initially
rejected by his design team because of the lack of structural logic (and dubbed the Vulgar Freak). He
explained that a three-pointed dome would match a fan-shaped auditorium plan and would be able to
defy the “earthbound” aesthetics of other domes—and boasted of the ability to demonstrate the form
with a grapefruit peel (Saarinen [10]). Saarinen readily admitted that the form wasn’t derived from
engineering-logic—a central tenant of his practice philosophy held that forms didn’t need to—although
he later admitted that this approach was “dangerous, but true” (Saarinen [11]). He had chosen a
paradoxical relationship between aesthetics, acoustics, and spanning, explaining that, “the spanning of
3
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
a great space has more authority (than acoustic constraints) in terms of design and the determination of
basic form.” (Saarinen [8]). This scheme became the basis for the final design—a design which would
become the longest free-spanning thin shell concrete structure in the U.S. (Figs. 1A-1C).
2.2 Separation of Activities and Responsibilities
Saarinen hired Ammann & Whitney, in part due to the expertise of two of the firm’s founding partners:
Whitney and Boyd Anderson in designing concrete shells. Whitney had developed a new analytical
method for the plastic theory and ultimate strength methods for reinforced concrete design and had
overseen the writing of the recently completed Design of the Cylindrical Concrete Shell manual for the
American Society of Civil Engineers (ASCE). Anderson was a national award-winning engineer known
for his ability to design suspended steel construction and concrete shells (Whitney [12]).
According to Saarinen’s design partner, Kevin Roche, Saarinen did not involve Whitney or Anderson
during the firm’s extensive deliberations on the shell’s form because the structural constraints were just
one of many issues that needed to be balanced by the project’s design. Whitney was “largely absent from
any design interactions,” assigning Anderson as the project engineer (Roche [13]). It was also perhaps
not surprising for Saarinen as two years earlier at a conference he called for a “greater level of
integration” among all the design professionals working on a project but strangely didn’t list structural
engineers as part of the list of potential collaborators [5].
Having the architect define the form and asking the engineer to “make it work” is understandable in the
historical context of a traditional project development and contractual responsibilities, but the exclusion
of structural consultation in the development of Kresge’s form was ill-advised and it became the
project’s central liability for all ensuing problems.
And yet separating design from technology was nothing new for Saarinen. He had also institutionalized
this separation of project tasks in his own firm as well. Among his employees he had three groups:
Design (headed by Roche), Technical (headed by John Dinkeloo), and Construction (headed by Bob
Lacey). Technical issues that could adversely affect the design were expected to be brought to Saarinen’s
attention for consultation, but in practice this rarely happened (due in part to Saarinen’s busy schedule).
Like the engineers, the Technical team of architects was tasked with “making it work” as designed.
4
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
of Cambridge would guarantee large build-ups of snow and ice at the foundations. Eventually, during
inspections conducted by Ammann & Whitney from 1973- 1979, they found serious long-term
deterioration of the edge beams at the abutments including large areas of corroding reinforcing caused
by freeze-thaw conditions. The building was closed as the lowest portions of the abutments were
repaired (Boothby, et al. [15]).
If the project would have been a negotiated project, or one in which the contractor and engineer could
have collaborated together to resolve certain complicated construction and detailing issues like this,
perhaps the arches would not have been designed in this way and these issues could have been avoided
or mitigated. As it was, the design team had to try and anticipate architectural, structural, and
construction issues through the documentation put forth in the drawings.
3.1 Documenting Double-Curvature
One of the challenges in preparing bid documents is the need to define the scope of the project
accurately so that someone else can build it without being prescriptive to a single means of
construction (so bidders can establish a potential advantage based on their construction plan and cost).
As a consequence of this separation between drawing and building, certain drawings are created that
may not be very useful and certain details are developed that become unexpectedly difficult to build.
This drawing approach is in stark contrast with the shells created by designers/ builders such as
Candela and Nervi in which the construction documents often included drawings that specifically
address the means/methods of forming and casting the shell (Candela, [4]).
Because Saarinen’s team had never designed or documented a shell before, and perhaps didn’t want to
over-define its means of construction, they created a series of construction documents that were mostly
concerned about defining the building’s geometric form than its means of assembly. Although the
building’s form could be easily described geometrically, but it was exceedingly difficult to document
in the construction drawing phase. One of the primary strategies used by Saarinen’s Technical team
was to establish a 10’ x 10’ grid reference system on their plan and section drawings along with
coordinated control point system of reference (like a surveyor) in which a series of radii would define
the curves (e.g., Point A Radius: 210’ etc.). This makes sense when one is drawing these geometries
using orthographic projections in the x and y axes, but a radius point of reference that is shown several
stories below the ground isn’t very useful for builders. Further complicating this approach was the
changing reference points of the basement and ground floors which both had fan-shaped auditoria that
stepped down from grade. Finding a specific reference point from within the building using the
drawings would be an incredibly complicated surveying problem.
3.2 Abdication Or Autonomy?
“Truly, I can think of no type of construction where closer cooperation between the architect,
engineer, and builder is more important (than shells). -M.J. Roach, Turner Construction Company, at
MIT Conference on Thin Concrete Shells, 1954 [4]
The specifications allowed the contractor to determine the order by which the building would be
built—either casting the auditoria first and then building the dome over or building the dome and then
the auditoria [6]. This was somewhat understandable because it was a project that would be awarded to
the lowest bidder and certain contractors might have a better approach in mind.
There were other more serious problems and omissions left the construction documents that could
have been better documented, particularly considering Ammann & Whitney’s experience with
concrete shells. The Concrete section used broad language related to the testing, casting, and placing
the concrete even though each of these conditions is markedly more difficult for shell construction.
Placing concrete on double-curved surface is tricky, particularly the challenge of workability and
slump—if the concrete isn’t stiff, it will flow down the forms and consolidate. If it is too stiff, it
wouldn’t be very workable. A thin shell need concrete to flow easily between the reinforcing so it can
be appropriately vibrated to avoid pockets of air. Adjustments to the concrete slump were made during
construction (Specifications [3]).
5
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
Most surprisingly, the specifications explicated granted the contractor the exclusive ability to
determine the “design, fabrication, and erection of the shell falsework and forms, the pouring
procedure (one pour or in sections with joints), and the de-centering procedure.” Any of these
conditions would normally be thoroughly coordinated between the design and construction parties. By
abdicating responsibility of these decisions to the contractor, Saarinen and Ammann & Whitney
essentially removed themselves from a critical collaborative role in the bidding and construction
process.
4. Construction Complications
In May 1953, George A. Fuller Company had won the bid to build the project. Because it was based on
a standard contractual arrangement, the contractor would only hold a contract with MIT and not
Saarinen; the construction documents and specification serving as the only bind between those two
parties. These documents granted them broad, and in many cases, the sole authority to define the
construction procedure and sequencing so little pre-construction coordination occurred between
Saarinen’s office and the contractor.
Because the auditoria seating and basement foundations needed to be cast in a way that made them
integral to the building foundations and buttresses, they chose to build these portions of the building
first—saving the shell construction for later. As a consequence, the roof construction would be delayed
until the upcoming winter months.
In his speech to the Conference on Thin Concrete Shells, the job’s foreman, Bates, described the first
portion of the construction project as uneventful particularly when compared to the “special
conditions” of developing a geometrically accurate and stable formwork for the shell. The process
Bates described for how the roof was formed was practical and accurate—and completely different
than the methodology shown in the construction drawings. Instead of relying on a grid and large radius
points, they started with the most difficult element—the arches (Bates [4]).
Figure 3: Edge Beam formwork, latitudinal rings, and view of scaffolding under dome, 1954.
Because all three arches were the same and symmetrical, they built a full-sized model for half of an
edge beam in a hangar on MIT’s campus to use as an accurate template of the formwork. They built all
the formwork off-site and then placed the formwork around the building perimeter from one support to
the next. Once they established the proper height at the center of the arch and the outward angle of
inclination, they supported it with scaffolding from below. Because the arch formwork accurately
defined the boundaries of the shell, they used it as reference points for the spherical curvature of the
roof that filled between the arches. A system of curved ledger beams were built atop the highest level,
serving as latitude lines around the dome (Fig. 3). The formwork was very labor intensive and
expensive but it was relatively easy compared the difficulty the contractor faced when pouring the
shell.
6
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
separate segments because a single pour wasn’t possible on a project this large. The specifications
didn’t even address whether or not the arches should be poured separately or simultaneously from the
shell. From a technical and structural perspective, these options aren’t equivalent. A shell relies upon
the monolithic behavior of the membrane; by allowing the shell (and arches) to be poured at separate
times with potentially different concrete mixes and adjoined with “joints” simply invited the types of
problems that eventually occurred. Ammann & Whitney didn’t have the authority to intercede for
Kresge, but on ensuing shell projects with Saarinen (TWA Terminal and Dulles) they recognized the
importance of continuous pours and incorporated these requirements into their documents (Whitehead
[16]).
Not surprisingly, the builders started separate pours simultaneously at the base supports and continued
moving upwards in a series of pours that ultimately took more than three months to complete. Making
matters worse, the roof was poured during the coldest months (December 1953-February 1954). Cold
weather creates many difficulties for casting concrete at it requires additional attention and preparation
for how the concrete is mixed, placed, and cured to make sure the concrete achieves its target yield
strength.
Having a double-curved surface is always hard to pour (and screed), particularly for a shell with
relatively large rebar and minimal thickness (3.5” uniform thickness), but it was made more difficult by
the integrated arch beams—especially at the supports. Because of the steepness of the slope at the base,
the first pours needed counter-forms which unfortunately hid the visibility of concrete placement and
limited the ability to vibrate the concrete. This was the location where inspections in the 1970s showed
decay and deterioration in the concrete and rebar.
Construction photos of the building’s construction show the simultaneous difficulties of these pours. In
some cases, concrete is being transported by shovel and no particular order of operations is apparent—
certain portions of the shell are being cast at the same time as the integral beam but at different depths,
vibrators aren’t visible, and cover over the rebar in the shell surface seems to occur at a variable thickness
(Fig. 4).
Figure 4: Difficulties pouring the shell at edge beam (4A), curved roof (4B), and abutment (4C), 1954.
5. Consequences
Kresge Auditorium was an expensive and poorly performing structure that was difficult to build. As
TIME magazine had predicted, it did indeed establish a precedent for future architecture projects.
Unfortunately, instead of serving as an example of the inherent efficiency and expressiveness of true
shell structures, Kresge served as a precedent for other free-form expressive projects claiming to be
“shells” including the TWA Terminal (by Saarinen and Ammann & Whitney) and the Sydney Opera
House.
Unlike other successful shell projects, these poor examples nearly always separated the design,
engineering, and construction considerations and participants from each-other—relying on the architect
for design authority of the form. Certainly most of Kresge’s problems were consequences of the form,
but there is little doubt that the lack of correspondence between the designers, detailers, and builders
contributed to a design that was more likely to allow certain potentially avoidable problems to occur.
7
Proceedings of the IASS Symposium 2018
Creativity in Structural Design
References
[1] Scott, N., M.I.T. Auditorium: An English View, Journal of the Royal Institute of British
Architects, Feb. 1955.
[2] Cohen, E., Dobbs, N., and Combs, W., Inspection, Analysis, and Restoration of MIT Kresge
Auditorium, Rehabilitation, Renovation, and Preservation of Concrete and Masonry Structures,
ACI Special Publication, SP-85, Detroit: American Concrete Institute, 1985, 92-125.
[3] Eero Saarinen Collection (MS 593), Series IV, Project Records, Job 5007, Manuscripts and
Archives, Yale University Library.
[4] Proceedings, Conference on Thin Concrete Shells, M.I.T., June 21-23, 1954.
[5] Saarinen, E., Our Epoch of Architecture (speech transcript), AIA Gulf States Regional
Convention, October 24, 1952.
[6] -, Tripod Built on Tricky Formwork, Engineering News-Record, May 27, 1954, 30-32.
[7] Pennsylvania State University’s Architectural Engineering Lab, Case Study #3, Historic
Preservation of Thin-shell Concrete Structures, Retrieved from,
www.engr.psu.edu/ae/thinshells/module%20III/case_study_3.htm
[8] Saarinen, E., The Trend of Affairs, The Technology Review, M.I.T., vol. 57, no. 8, June 1955,
387-388.
[9] J. Merkel, Eero Saarinen, Phaidon, p. 114, 2014.
[10] Saarinen, E., Letter to Aline Louchhein, 1953, Aline and Eero Saarinen Papers, 1906-1977,
Archives of American Art, Smithsonian Institution.
[11] Saarinen, E., General Statement about the sculptural, curved shapes…, undated (between 1958-
61), Eero Saarinen Collection (MS 593). Manuscripts and Archives, Yale University Library
[12] Whitney, C., The Committee on Masonry and Reinforced Concrete, Subcommittee on Thin Shell
Design, Design of Cylindrical Concrete Shell Roofs, American Society of Civil Engineers
(ASCE) in the Manual of Engineering Practice, No. 31, 1952
[13] Roche, K., Personal Communication/Interview, 27 May 2014.
[14] R. Whitehead, “Saarinen’s Shells: The Evolution of Engineering Influence,” Proceedings
International Association for Shell and Spatial Structures, 2014.
[15] Boothby, T., Parfitt, M., and Roise, C., Case Studies in Diagnosis and Repair of Historic Thin-
Shell Concrete Structures, Assoc. for Preservation Technology International, Vol. 36, No. 2/3,
2005, 3-11.
[16] R. Whitehead, “Formative Experiences: Saarinen’s Shells and the Evolutionary Impact of
Construction Challenges,” Proceedings 5 International Congress on Construction History
th