Beyond Building (But) Inside Architecture
Beyond Building (But) Inside Architecture
Beyond Building (But) Inside Architecture
Dana K. Gulling
ABSTRACT: In recent years, mass customization and computer aided manufacturing (CAM)
technologies have transformed off-site building component fabrication. At the same time,
traditional repetitive manufacturing still dominates building component production. Bricks,
bathroom fixtures, window mullions, and door hardware are all repetitively manufactured.
Ironically, CAM’s computer numerical controlled (CNC) machines have made the fabrication of
custom molds for repetitive manufacturing easier. This allows architects to customize
repetitively manufactured architecture components. I am using the term customized repetitive
manufacturing (CRM) to refer to this type of work. Through my ongoing research, I have
identified a wide range of historic and contemporary examples of CRM for architecture
components.
INTRODUCTION
In recent years, mass customization and computer aided manufacturing (CAM) technologies
have transformed off-site building component fabrication. At the same time, traditional
repetitive manufacturing still dominates of building component production. Bricks, bathroom
fixtures, window mullions, and door hardware are all repetitively manufactured. Ironically,
CAM’s computer numerical controlled (CNC) machines have made the fabrication of custom
tools for repetitive manufacturing easier and thus have reduced costs. CNC milling machines,
electrical discharge machines (EDM), and hot-wire foam cutters are used to creating tools for
repetitive manufacturing. Reduced tooling costs therefore require smaller production runs to
offset those costs. Architecture benefits from smaller production runs, as architects can now
consider customizing repetitively manufactured products for architecture components. I will use
the term customized repetitive manufacturing (CRM) to refer to this type of work.
Through my ongoing research, I have identified a wide range of historic and contemporary
examples of CRM in producing architecture components and have presented in-depth
investigations into selected case studies. The case studies of CRM in architecture are located
around the world and use a variety of manufacturing processes. Using those examples, I have
identified three categories of customized architecture components. The categories are custom
components, architecture prototypes, and architecture products and they are based on the
architects’ design, intention, and application of their components. Custom components are
designed by the architect to be custom manufactured for a particular building. Examples
include Renzo Piano’s sand-casted, ductile iron truss members for the Menil Collection (1987)
and Tom Phifer’s contact-molded, fiberglass reinforced plastic (FRP) ceiling coffers for the
North Carolina Museum of Art (2010). Architecture prototypes are full-scale, mock-ups used to
test and architectural idea. Examples include R. Buckminster Fuller’s Prefabricated Bathroom
for the Dymaxion House (1940) and houminn practice’s Drape Wall (2008). Finally,
architecture products are architecture components available for mass consumption. Examples
23
include Zaha Hadid’s ZH Duemilacinque doorknobs for Valli&Valli and Robert A.M. Stern’s
Rhythm light fixtures for Lightolier.
Although building component design is outside of standard architecture practice, architecture
components are simultaneously part of architecture. Components are attached to the building
and it is the aggregation of components that physically makes a building. In order to use CRM
for architecture component design, the architect must decide that the available building
components products do not meet the needs of the design. Additionally, because the CRM
component is repetitively manufactured, the architect must believe that a new component’s
design is strong enough to warrant multiples. Designs for CRM components require
coordination between the architect and the repetitive manufacturer. Both must balance capital
costs with production runs, mold complexity with manufacturing parameters, and size
restrictions with manufacturer’s abilities. How the architect and the manufacturer balance these
parameters depends on the component’s application.
This paper highlights architects’ design work beyond the profession’s standard definition of
architecture design. For this research, I use contemporary and historic case studies to define
the three categories, study the lessons learned by the case studies, and draw conclusions.
The three categories—custom components, architecture prototypes, and architecture
products—provides particular constraints on CRM that the architect should consider. By
grouping the examples together, I am able to draw out common themes, challenges, and
constraints that should be considered in each category. For example, by examining the custom
components together, I discovered the challenges for architects in educating themselves about
potentially unknown manufacturing processes. This paper is part of my investigation to
understand the overall conclusions that my collection of CRM case studies provides.
CRM balances the value of repetitive manufacturing with the ability of the designer to
customize a building component. There are forms, materials, and finishes available in CRM
that are not available in CAM. Processes such as precision slumping glass and clay, blowing
glass, and contact molded fiber-reinforced plastic (FRP) are available in CRM but not CAM. In
comparison to subtractive CAM processes, repetitive manufacturing typically uses only as
1
much materials as the mold, pattern, or jig needs . By reusing tools and reducing raw material
requirements, customized repetitive manufacturing can have little to no production waste.
Manufacturing tolerances for most of these processes are high and have the potential to rival
the tolerances of CNC equipment.
In comparison to CAM, architect need to consider more variables in CRM. In CRM, one must
consider desired materials, shapes, allowable production runs, capital costs, and finishes in
order to select a manufacturing process. For example, if a production run is small and costs
are required to be low, then a designer may want to consider thermoforming for plastic rather
than injection molding. Additionally, the mold media affects the manufacturing process. In
thermoforming plastic, changing the mold from wood to aluminum (with imbedded cooling
lines) increases the cycle time, tolerances, mold costs and production run lengths. Conversely,
there are fewer variables in CAM than CRM. A CNC router is consistent in its operation
regardless of the media, shapes, production runs, and finishes. Media and finishes may affect
production speed, but the operation of the machine, tolerances, and production run lengths
remain the same.
ARCC/EAAE 2014 | Beyond Architecture: New Intersections & Connections
Re-Disciplining: The Rise, Fall and Reformation of the Disciplines. History, Theory, Historiography, and Future Studies
24
The three categories of architectural components—Custom Components, Architecture
Prototypes, and Architecture Products—affect the variables of CRM production. Therefore, in
order to design for manufacturability, the designer should consider the category in which their
design is. For example, if an architect-designed component is a product to be made of metal
and available for mass consumption, lost wax casting with injection-molded patterns could be
considered. If a component is custom for a particular building application, then it may have a
smaller production run and could be sand casted. The capital costs and finishes with the lost
wax casting with injection-molded patterns are substantially higher than those of sand casting
metal are.
Table 1: Table listing selective case studies of CRM custom components. The listed case
studies have been limited to examples from the past 10 years.
Repetitive Process,
Year Project Name Location Designer
Component
Herzog and
2003 Slumped Glass, Windows Prada Store Tokyo, Japan
de Meuron
Mexico
Wood Molded, Blown
2003 Hesiodo Hierve Diseneria City,
Glass, Screen
Mexico
Renzo
2003 Cast Metal, Skylights Nasher Sculpture Museum Dallas, Texas
Piano
Rubber Molds, Concrete Utrecht, Wiel Arets
2004 Utrecht University Library
Panels Netherlands Architects
Kraaijvange
2005 Explosive Forming, Panels Theater Castellum Alphen, Holland
r Urbis
Minneapolis, Herzog +
2005 Stamped Metal, Panels Walker Art Center Addition
Minnesota deMeuron
Carlos
Fiberglass-molded Precast Rice University Data
2007 Houston, Texas Jimenez
Concrete, Walls Center
Studio
Extruding Clay, Column Francisco
2008 Spanish Expo-Pavilion Zaragoza, Spain
Cladding Mangado
Extruding Aluminum, Dee and Charles Wyly
2008 Dallas, TX REX
Screen Theater
2010 Slumped Glass, Windows VAKKO Fashion Center Istanbul, Turkey REX
Mack
Yale University Health New Haven, Scogin
2010 Extruding Stiff Mud, Bricks
Services Building Connecticut Merrill Elam
Architects
Contact Molding FRP, Foster and
2010 Walbrook Office Building London, England
Exterior Louvers Partners
Contact Molding FRP, North Carolina Museum of Raleigh, North Thomas
2010
Ceiling Coffers Art Carolina Phifer
Rubber Molded Precast Cleveland Medical Mart LMN &
2013 Cleveland, Ohio
Concrete, Panels and Convention Center URS
Nieto
Contact Molding GFRC,
2013 Contemporary Art Center Cordoba, Spain Sobejano
Panels
Arquitectos
25
Figure 1: Selected Images of custom components. From left to right: Prada Store by Herzog
and de Meuron, flickr: (InfoMofo, 2006); 290 Mulberry St. by SHoP, flickr: (joevare, 2008);
Walbrook Office Building by Foster and Partners, flickr: (Myxi, 2009).
Custom manufactured custom components allow for customization from the designer while
balancing the need for repetition in order to remain cost effective. Since customization is done
on a per-project basis, the architect has to pay particular attention to the production run length
of the custom component. Required production runs for custom components can vary between
designs, applications, and production methods. Prior to the use of CNC equipment to fabricate
tooling, tools were fabricated by hand and were thus labor intensive and costly. Therefore,
historically an architect would be required to consider tooling costs for the designs of their
custom components.
Historic examples of custom components include Frank Lloyd Wright’s cast textile blocks for
his California concrete block houses (c. 1920) and the stamped aluminum panels for the
Harrison & Abramovitz’s Alcoa Building (1953) in Pittsburgh, PA. Both examples used different
methods to offset the additional tooling costs. Wright’s textile blocks were manufactured on
site, by hand, in a multipart mold. The mold was similar to a springform pan—its sides
unhinged so that they could be separated from the mold’s face. Tooling costs were reduced by
breaking the mold into parts. The mold’s sides were used for every block and two different
faces were used to create the flat and textile block faces. The mold was made from aluminum,
and by reusing the sides for all block production, less fabrication work was required to make
the tools. The Alcoa building used a different approach for offsetting tooling costs. A thirty-
story, high-rise building, the Alcoa has a custom, stamped aluminum panel under each
window. Because the Alcoa Building is a high-rise, additional costs for the custom component
2
is offset by the number of panels required to clad the high-rise.
Today, because of the availability of CNC equipment to make tools, custom components are
easier to produce than they were historically. CNC equipment fabricated the tools for the
Utrecht Library, Walbrook Office Building, NC Museum of Art, 290 Mulberry St. and Cleveland
Medical Mart. The case studies listed in Table 1 demonstrate the advantages of CRM over
CAM. Sometimes CRM is less costly than CAM. For example, Foster and Partners had
investigated using CNC equipment to make the louvers on the Walbrook Building, but then
learned that repetitively manufacturing them with a mold was more cost effective. Oftentimes
CRM produces less waste than CAM; such was the case for the Cleveland Medical Mart. Here
the design team used a CNC-milled pattern multiple times to produce rubber molds and then
used the rubber molds multiple times to create the precast panels. If the tools, were not used
repeatedly for production, more manufacturing waste would have been created to make the
panels
26
Analyzing the custom component case studies offers lessons for designer considering CRM for
production. First, in order to reduce production costs and waste, the designer should consider
creative uses for the production tools. Similar to Wright’s multipart, textile-block molds, some of
the case studies have considered molds that can be broken down or subdivided. At 290
Mulberry St., SHoP designed the building’s precast concrete and brick composite panels so
that they could be manufactured using one large rubber mold that was sub-divided. At the
Cleveland Medical Mart, the toolmaker used dams to make different rubber molds from a
single CNC-milled pattern. Both examples allowed for multiple variations using a single tool,
thus reducing cost and waste. Second, a designer must consider both the production run and
the manufacturing processes. For some manufacturing processes, high tooling costs cannot
be reduced and the only way of balancing those additional costs is through high production
runs. These processes typically use steel tools and include extruding clays, metals, and
plastics. There are over 27,000 extruded clay pieces on the Spanish Expo-Pavilion and over
300,000 custom bricks for the Yale Health Services Building. Alternatively, some
manufacturing processes have low tooling costs and therefore can support small production
runs. This includes the explosive formed panels for the Theater Castellum, the precision
slumped glass for the VAKKO Center, and the contact molded FRP coffers for the NC Museum
of Art. Finally, since custom components are done on a building-by-building basis, architects
may not have prior experience with their selected repetitive manufacturing processes. This
often requires architects to find manufactures who are willing to collaborate for a particular
design. Such was the case with FiberTech who worked with Phifer’s office for the NC Museum
of Art, and Ceramica Cumella and Ceramica Decorative who worked with Mangado for the
Spanish Expo-Pavilion.
Innovative architectural practices, architecture researchers, and studio courses tend to explore
design ideas through architectural prototypes. Today, CNC equipment and robots fabricate
most prototypes. This may be because of the accessibility of CNC equipment in architecture
academia, because CNC equipment can be programmed directly by the designer, or because
CAM equipment has little-to-no capital costs. Despite those advantages, there are a handful of
architecture prototypes that have made use of CRM for their component production. See
Table 2. In these examples, CRM was selected because it offered something the CAM did not.
For example, thermoforming metal was a cost-effective method to get 3-dimensionally formed
tiles out of metal for the Busta Line project, and for the Dragon Pavilion CRM was more
beneficial than CAM because the students could fabricate their own molds rather than gain
access to CNC roller equipment.
27
Table 2: Table listing current, selective case studies of CRM in architecture prototypes
For some architecture prototypes, the design team manufactures the CRM architecture
components, directly. This was the case for Drape Wall, EcoCeramic Envelope System, and
the Dragon Skin Pavilion. By working directly with the manufacturing process to build the
architecture prototype, designers had freedom to experiment with the process. For example,
the Drape Wall design and fabrication team purchased vacuum-forming equipment to make
the prototype. According to an interview with Mark Swackhamer, the team researched the
thermoforming industry’s manufacturing parameters before experimenting with the process
themselves to make the prototype. Although the manufacturing parameters were stated as a
given, the team discovered that some parameters could be altered. Their investigations with
Drape Wall have led the team in two directions. First, they continued to develop their
4
architectural ideas into subsequent prototypes. Second, lessons learned about vacuum
forming plastic have led to research into manufacturing and they have been investigating the
5
possibility of using a dynamic mold for thermoforming plastic.
If architects intend to use CRM for their architecture prototypes, then there are specific issues
that they should consider. First, in order to reduce costs, the architects themselves or other,
less-skilled laborers often make the architecture prototype. For examples, Emmi Kerskisarja et
ARCC/EAAE 2014 | Beyond Architecture: New Intersections & Connections
Re-Disciplining: The Rise, Fall and Reformation of the Disciplines. History, Theory, Historiography, and Future Studies
28
al made the plywood scales of the Dragon-Skin Pavilion, houminn practice’s team
manufactured all Drape Wall’s thermoformed plastic bricks, and Jason Vollen (with CASE) ram
pressed the structural tiles for EcoCeramic. Second, CRM processes that use little or no
complicated equipment are most often selected. For example, Drape Wall’s plastic bricks and
the scales of the Dragon-Skin Pavilion were both made in university fabrication shops. Third, to
reduce capital costs, architects may fabricate the CRM tooling themselves. This happened in
all of the case studies listed in Table 2. Finally, for prototypes, the architect will want to
consider manufacturing processes with low capital costs, and thus will allow the small
production runs associated with prototypes. All of the CRM processes listed with the case
studies have low capital costs and thus small production runs.
Table 3: Table listing current, selective case studies of repetitively manufactured architecture
products. For brevity, the table offers only one sample product for each component type.
Figure 3: Selected Images of Architecture Products. From left to right: Agape Nivis washbasin
by Shiro Studio, Valli&Valli Fusital door levers by Zaha Hadid (top) and Jean Nouvel (bottom),
Duravit Starck 1 Pedestal Ceramic by Phillipe Starck.
Unlike custom components or architecture prototypes, which are developed by the architect for
a particular application, the architecture product’s success depends on the consumer.
Customers must desire the product, and therefore companies must promote the products.
Companies promote products through advertising and are facilitated by the companies’ brand,
the quality of the design, or the designer’s name recognition. Such is the case with the
Valli&Valli Fusital Series door levers, which include designs by Zaha Hadid, Robert A.M. Stern,
Frank Gehry, Jean Nouvel, and Richard Meier.
Beyond buildings [but] inside architecture
Dana K. Gulling
29
The value of architecture products to the architectural community is perhaps a difficult one to
access. On one hand, it is beneficial to the architectural profession to specify building
components designed by other architects. Since architects have not been involved with the
design of the majority of our building products, architecture products allow architects to
promote the value of our profession throughout all aspects of the building. This in turn offers
architects more design opportunities beyond that of building designer. On the other hand,
since architecture products are available to the public, products may be seen either as a
commodification of design or as a perverse extension of a designer’s fame (Deamer). For
example, the Valli&Valli door levers demonstrate the importance of the designer’s fame to the
7
product—as each image of the door lever includes a head shot of the designer.
Architectural products affect the parameters for repetitive manufacturing. First, depending on
the anticipated popularity of an architecture product, architecture products could use
manufacturing processes with high capital costs that support large production runs. With high
production runs, an architecture product may support injection molding instead of
thermoforming. This would in turn affect the component’s finish, design, detailing, and
materials used. Second, unless specifically marketed to have a handcrafted feel, consumers
purchasing architect-designed products have high expectations of quality. This may require
manufacturing processes that can produce items with high tolerances and quality finishes—
such as the smooth, high-gloss finishes are available in lost-wax casting but not sand casting
metals. Third, architecture products may be required to be certified by third party agencies,
such as the Underwriter’s Laboratory (UL) certification for light fixtures and American Society
of Testing and Materials (ASTM) certification for hardware. The product’s designs must be
compliant with the certifying agency.
CONCLUSION
As the case studies demonstrate, CRM is rooted in both history and contemporary practice.
There are many recent and global examples of architects using CRM for the design and
production of architecture components. In comparison to CAM, CRM has more variables that
an architect should consider for architecture component design and production. These
variables include materials, shapes, required production runs, capital costs and finishes. From
my collected case studies of CRM in architecture, I have identified three categories of
architectural component applications. They are custom components, architecture prototypes,
and architecture products.
This paper organizes these case studies into a larger discussion. My future goal for this
research is to create a guide CRM for architects. The guide will include an introduction to
different repetitive manufacturing processes, parameters for possible customization, and
architectural case studies. By presenting this paper as an overview, my goal was to establish
commonalities between the case studies within the category. Each of three categories
provides particular constraints on the architect for the design and production of an architectural
component using CRM. Custom components require the architect to consider production run
lengths, creative use of molds to distribute costs, and collaborate with manufacturers.
Architecture prototypes often use lower-skilled manufacturing processes, manufacturing
processes that are accessible by designers, and allow for greater experimentation by the
architect. Architecture products often use manufacturing processes with tight tolerances and
high quality finishes, they are certified, and may allow for less collaboration between
30
manufacturer and designer. Understanding the constraints on the designer at the beginning of
the design process is important, as they influence the component’s design.
REFERENCES
Deamer, Peggy. “Branding the Architectural Author.” 2005. Perspecta 37: Famous.
Cambridge: MIT Press.
Gulling, Dana. 2010. “Mobile Home Breakdown: A Study of Building Product Design Lineage”
Association for the Collegiate Schools of Architecture (ACSA) NE Regional Conference:
Urban/ Suburban Identity. Conference Proceedings
Joevare. “290 Mulberry Street (SHoP Architects).” Taken October 4, 2008. Flickr.
Piller, Frank T. et al. 2004. “Does Mass Customization Pay? An Economic Approach to
Evaluate Customer Integration” Production Planning & Control: The Management of
Operations. (2004) London: Taylor & Francis Ltd.
Myxi. “New London Rhythms II.” Taken September 26, 2009. Flickr.
United States Patent Office. Fuller, Richard Buckminster. “Prefabricated Bathroom”. No.
2,220,482, 5 November 1940.
ENDNOTES
1
Subtractive CNC machines are those machines that take material away in order to produce
the unit. Subtractive equipment includes drills, lathes, millers, routers, surface grinders, EDM’s,
plasma cutters, water-jet cutters, laser cutters, knife cutters, hot-wire foam cutters, punch
presses, and oxyfuel cutters.
2
Additionally, the building was a showcase for architectural uses of aluminum and so the client
had a financial interest to offset the added cost of a custom component.
3
The Cellophane House was designed and fabricated for a Museum of Modern Art exhibit in
New York.
4
Subsequent prototype iterations by houminn practice include Cloak Wall and OSWall.
5
In a 2010 interview with Mark Swackhamer, we discussed the practice’s research into using
dynamic molds for thermoforming. According to the practice’s website, they recently presented
their variable vacuum forming research at ACADIA 2013.
6
For architecture products, I am including only 3-dimensional architecture products, but not 2-
dimensional components such as carpets and fabrics. This keeps the types of components
listed in architecture products similar to those listed in the categories of custom components
and architecture prototypes.
7
Headshots of each designer or architect are presented with all of the door levers. For an
example of Zaha Hadid’s door lever and head shot, visit
http://www.vallievalli.com/en/site/vallievallicom/ValliValli-
USA/Products11/Fusital/Handles/H356/
Beyond buildings [but] inside architecture
Dana K. Gulling
31
8
The distancing of an architect from the manufacturer may be even worse than this exampled
provides. Based on a recent story broadcasted on National Public Radio (NPR) a companies
are now licensing their brands to products that they may not manufacturer. This means that
Black & Decker may not have made the toaster oven that bears its name. Bobkoff, Dan. “How
Much is NPR’s Brand Worth? $400 Million”. Aired November 1, 2013.
www.npr.org/blogs/money/2013/11/01/240285576/how-much-is-nprs-brand-worth-400-million.
Accessed November 2, 2013.
32