2005 Pamphlet Architecture Entry
Lateral Thinking, Both Ways, for Both Objectives
While there are significant precedents for structural expression as an architectural theme with
regard to the way a building or other structure resists the vertical loads of gravity, we have not
yet fully explored the potential for expressing the way lateral forces of earthquakes are resisted.
Structuralism as an Architectural Theme
A classic case of elegance in architecture combined with elegance in engineering terms would be
a bridge by Robert Maillart. From standard architectural texts (Gideon, 1941) to books written
by engineers (Billington, 1979), Maillart’s arch bridges have been well presented to a large
But as with most other examples of structuralism in architecture, the well-known example of
Maillart’s works only have to do with vertical gravity loads, not lateral seismic loads.
The Design Challenge: Design a Building With “Maillart Bridges On Their Sides”
So, long story short, I set for myself the challenge of imagining a Maillart bridge turned on its
side and then explored both the architectural and engineering aspects of what resulted when you
try to make that into a building.
This is an example of what could be called “lateral thinking, both ways, for both objectives.” It
is lateral thinking both in the sense meant by the originator of that phrase, Edward De Bono (De
Bono, 1967), and in the seismic sense of resistance to horizontal forces. De Bono has explored
innovative visual and mathematical ways of opening the mind to the discovery of creative
solutions by thinking around a problem rather than jumping to the most obvious conclusion. One
of his sayings concisely explains his approach: “You cannot dig a hole in a different place by
digging the same hole deeper.”
The other mode of lateral thinking is the literal one, thinking about and visualizing how a
building or other structure will perform when loaded laterally by an earthquake.
“Both objectives” refers to the two reasons why one would want to incorporate lateral thinking
into a building’s design:
1. To produce new forms of interesting aesthetic effect;
2. To achieve efficiency in seismic design, which, for sites where approximately one-
third of the world lives (GSHAP, 1999), is a much more difficult structural problem
than dealing with gravity loads.
Differences in Lateral and Vertical Loads
In simplified form, the differences between lateral and vertical loads can be reduced to a few key
factors that must be considered and which enter into the design concept illustrated here.
The motion of the ground in an earthquake inertially generates seismic loads in the structure, and
because that ground motion is complex, so are the effects on the structure. The key aspect
relevant to the design issue here is that the ground moves in various directions (including
vertically but those accelerations are significantly less severe than the horizontal), thus the
structure feels lateral forces in various directions. Gravity forces on a structure are only exerted
in one direction, down; an earthquake generates forces that must be resisted in any horizontal
We can typically analyze a structure along one of its primary axes, with the seismic loads acting
first one way and then back the opposite way; then do the same for the other axis. These are the
typical governing cases, as compared to the various off-axis forces resulting from other
directions of ground motion impulses. This is diagrammed in the accompanying illustration.
The implication of this is that the structure’s horizontal resistance must be balanced in plan: You
can’t take a structure designed to resist one-way (gravity, downward) loading, turn it on its side,
and obtain a stable structure. The “Maginot Line” approach of providing resistance only in one
direction doesn’t work for earthquakes.
There are of course other differences between gravity forces and seismic forces. Along with the
fact mentioned immediately above, two others are listed in the table below.
Table 1: Comparison of Gravity and Earthquake Effects
Architectural Implications of Seismic Design
Resistance to lateral forces must be balanced in plan
For large earthquakes, it typically is not feasible to provide
enough strength to resist the forces elastically; some
inelastic behavior—cracking, permanent bending, etc.—is
Ground motions and resulting forces vary by the split-
second; frequency and damping characteristics of the way
the structure “gives” affect the magnitude of the forces
In the design explored here, these three basic seismic factors have been integrated as follows.
The typical building has walls that only resist in-plane lateral forces; for forces in the out-of-
plane direction, the walls merely add to the problem (their mass adds to the load), and they must
hang on to floors and the roof (structural diaphragms in that horizontal role) for support.
Meanwhile the walls oriented the other direction do all the work. The building must be designed
for the worst case on both axes, “throwing away” half the total resistance (not being able to
mobilize the strength of the out-of plane walls).
The design produced here takes a “Maillart bridge on its side” as its basic structural unit and then
designs it as a lateral-force-resisting element that can resist loads either along its own axis or for
loads imposed perpendicular to it. This in itself is a great achievement in engineering elegance,
accomplishing more with less.
The second consideration in Table 1 has to do with the fact that “ductility” is a central concern in
earthquake engineering and is explicitly recognized as a requirement in every building code in
the world that contains seismic regulations. (IAEE, 2000) Seismic forces in a large earthquake
are so large and difficult to resist that it is not economical to provide so much strength via
“bunker” construction that a building will come through the event without a crack, that is, remain
elastic. Ductile behavior—cracking, permanent deformation of steel frame members or
reinforcing steel, etc.—is explicitly designed into buildings by structural engineers. But ductility
is also damage: We need to control where the ductile behavior, i.e., damage, occurs.
The design here provides for the ultimate receiving point of lateral loads to be the four corner
cores. The “Maillart bridges on their sides” convey all loads to those points. We now have the
opportunity to pack tremendous amounts of strength into these windowless pylons, but if the
earthquake pushes them past their elastic limits (if the reinforced concrete cracks and spalls, the
reinforcing is stretched permanently), this damage could be easily repaired while the building is
Put a ton of weight on a bridge or a roof and there will be exactly a ton of load conveyed down
through the structure, unchanging from one moment or one day to the next. In an earthquake,
however, the building feels accelerations—it feels the effect of moving in one direction at one
instant, then another direction at another instant. Its velocity changes abruptly (that’s how
acceleration is defined in physics) as if the building were alternately stepping on the gas pedal
and the brake, lurching around erratically.
How does that affect the design? We can “tune” the overall structure to have an inherent
tendency to vibrate back and forth at a particular rate (fundamental period of vibration).
Actually, we want to make it “out of tune,” that is, to avoid vibrating in sync with the motions of
the ground. If the building tends to resonate with the earthquake, the forces can be several times
greater than if it does not.
Again, the corner cores provide a strategic location for what are called “advanced technology
devices” in the earthquake engineering field. Entire buildings and bridges in California and
Japan rest on special bearings that can filter out the frequencies of damaging motion of the
structure, and can add damping or “squishiness” to reduce the forces further.
The building volume that is contained within the four corner cores moves as one rigid unit,
nicely braced by its “Maillart bridges on their sides,” while the seismic isolation devices at the
bearing points at the cores absorb the impact of the earthquake and provide a cushioned ride. To
my knowledge, there is no building in the world that uses such a “sideways” isolation scheme:
The usual approach is to place isolation devices in the basement under the structure.
We can also introduce seismic isolators under the four cores in the conventional manner. This
gives us a doubly-isolated building. The side structures of the building span vertically off the
ground as well as laterally spanning to carry horizontal forces, and hence the building stands on
the ground in only four places, at the cores. In practice, shake table testing and analytical
simulations would be conducted, for a given site (and thus a given “suite” of earthquake ground
motions expected at that site in the future), to efficiently design the two sets of isolators work
efficiently as a team.
Aiming Higher: Earthquake-Proof, not Earthquake-Resistant
“Earthquake-proof” is a forbidden term in earthquake engineering today, because as explained
above, the typical design solution is to explicitly anticipate some ductile behavior (damage) and
design it to occur in particular places so that the structure remains safe. In this design, however,
we aim higher, and we use the term “earthquake-proof” by intent and unashamedly. The seismic
goal is to provide such a high quantity of strength, via the “Maillart bridges on their sides” and
corner buttresses, and with a double isolation scheme (sideways isolation of sides to cores,
vertical isolation of cores to ground), that our building will undergo its maximum design
earthquake with no damage at all. We’ve been settling for much less for too long.
The schematic design illustrated here (Figure 1) visually speaks for itself, so I will only add two
I initially found the “Maillart bridge on its side” to be a fascinating object as I rotated it in my
mind, then sketched it. The sweeping curve of the arches that is beautiful when oriented
vertically as in a bridge is also, I think, a beautiful thing arranged horizontally. The arch ribs are
narrow bands, for obvious fenestration reasons to allow light and view from the inside looking
out, and also to provide more visual interest when the building is viewed from the exterior.
The shape that results from this plan form, when roofed, results in the shell structure shown in
exploded view. In practice, such roofs of significant span, while sometimes theoretically
possible to design so that they produce no lateral thrust on their supports, in practice exert some
outward lateral forces, if only because of asymmetric wind loads. Such horizontal forces,
however, are easily resisted by the top-most “Maillart bridge on its side” and conveyed laterally
to the corner structures. Not shown for clarity is the possibility of dividing into separate stories
one or more of the lower levels of the interior space, but by intent I want to achieve a complete
open-plan volume at least in the upper portion of the structure, if not to use the entire interior as
one grand public space.
Billington, David, 1979. Robert Maillart’s Bridges: The Art of Engineering. Princeton
De Bono, Edward, 1967. The Use of Lateral Thinking. London: Jonathan Cape.
Gideon, Sigfried, 1941. Space, Time, and Architecture. Current edition 1967, Oxford University
GSHAP, Global Seismic Hazard Assessment Program, 1999. “The Global Seismic Hazard
Map.” International Council of Scientific Unions, published on-line by the Swiss Federal
Institute of Technology: http://www.seismo.ethz.ch/GSHAP/
IAEE, International Association for Earthquake Engineering, 2000. Regulations for Seismic
Design: A World List. 1996 edition with Supplement published in 2000. Tokyo, Japan:
International Association of Earthquake Engineering.