Está en la página 1de 5

14

TheStructuralEngineer
May 2012

Project Focus
US Air Force Memorial

The United States Air


Force Memorial
Use of impact
dampers to control wind
induced vibrations
Synopsis

The United States Air Force


Memorial (USAFM), overlooking
the Pentagon in Washington DC,
comprises three stainless steel
spires which evoke an image of
aircraft in a bomb-burst manoeuvre.
The elegance and simplicity of
their architectural form belies the
complexity of their engineering
design. Structurally they consist of
a stiffened stainless steel shell with
the lower two thirds of each lled
with concrete. A second component,
essential to the integrity of the
structure, is also hidden by the steel
skin; a series of large, steel-coated
spheres, free to roll in oversized
padded boxes, are located inside
each spire. The purpose of these
impact dampers is to stabilise the
motions of the spires in high winds.
This article provides an overview of
the spire structures and focuses on
the challenges encountered during
the design, development and test of
the damping devices.

TSE5_14-18_USairforce.indd 14

Project history
Situated at the summit of a natural ridge
overlooking the US Pentagon, and adjacent
to Arlington National Cemetery in Washington
DC, the Air Force Memorial evokes the spirit
of ight through its stainless steel spires (Fig.
1). The memorial, conceived by the late James
Ingo Freed of architects Pei, Cobb, Freed &
Partners (PCF&P), was inspired by the US Air
Force Thunderbird F-16s bomb burst ight
manoeuvre (Fig. 2).
The road to the construction of a
memorial to the United States Air Force was
a long one. Arup began work with PCF&P
on the design of the memorial in 1997. This
original design was for a site just north of
Arlington National Cemetery near the Marine
Corps Iwo Jima Memorial.
In 2001, Congress relocated the site to land
adjacent to the Navy Annex buildings. The
Air Force Memorial Foundation, a private
organisation formed to lead the design and
construction of the memorial, launched a
second international competition; Arup again
entered the competition with PCF&P and won
the design commission for a second time.
From the earliest design meetings with Mr
Freed, it was clear that the spires were to be
as sleek and as slender as possible in order
to maximise the impact on the observer. The

Air Force Memorial Foundation was led by


retired Air Force Major General Edward F.
Grillo Jr.

Overview of the structure


The spires range in height from 61 to 82m
with the tallest spire only 4.2m wide at the
base. The geometry of each spire was
derived from the intersection of a horizontally
oriented cylinder which forms the spires
back surface and two opposing, horizontally
oriented and intersecting cones. The resulting
shape is a curved spire whose triangular
section tapers towards its tip. The tapering
triangular shape of the spires was formed
by bending and welding 19mm stainless
steel plates. As two faces of each spire
are sculpted along conical surfaces, these
surfaces are in fact warped out-of-plane,
creating an added degree of complexity to
the form. The prole of the tallest spire was
devised such that the combination of its
slenderness, curvature, overhang, and the
extent of internal reinforced concrete, serves
to balance the structure about its base.
This approach ensures that the foundation
beneath the tallest spire is uniformly
compressed under the structures selfweight, minimising the potential for differential
settlement with time and, ultimately, reducing

23/4/12 15:21:46

www.thestructuralengineer.org

15

Figure 1
The United States Air Force
Memorial, Washington DC

TSE5_14-18_USairforce.indd 15

the The United States Air Force memorial


(USAFM), overlooking the Pentagon in
Washington DC, comprises three stainless
steel spires which evoke an image of aircraft
in a bomb-burst manoeuvre. The elegance
and simplicity of their architectural form
belies the complexity of their engineering
design. Structurally they consist of a stiffened
stainless steel shell with the lower two
thirds of each lled with concrete. A second
component, essential to the integrity of the
structure, is also hidden by the steel skin; a
series of large, steel-coated spheres, free to
roll in oversized padded boxes, are located
inside each spire. The purpose of these
impact dampers is to stabilise the motions
of the spires in high winds. This article
provides an overview of the spire structures
and focuses on the challenges encountered
during the design, development and test of
the damping devices.
Load demands on the foundations
anchors, baseplates and caissons. Early
in the evolution of the memorial it was
decided that the spires would embody their
true structural expression. No architectural
cladding would shroud the monument. The
stainless steel structural skin itself would
instead be ground smooth and ush, beadblasted, polished, and etched under an
exacting litany of procedures to ultimately
become the face of the monument with full
penetration welds specied for all spire plate
connections to improve fatigue performance.
The spires were fabricated off-site in 12m
segments. Each spire was pre-assembled
in thefabrication shop to ensure satisfactory
alignment prior to delivery to site (Fig. 3).
During erection, the lower two-thirds of each
spire were lled with high-strength (80MPa)
concrete to increase their stiffness (Fig.
4). The composite concrete sections are
heavily reinforced to further increase their
bending stiffness; the total amount of internal
reinforcement gradually reduces with the
height of each spire. Each piece of internal
reinforcing steel was mechanically spliced
to provide the most secure and direct load
path through the body of the spires. Above
the concrete zone, there is a transition zone
of concrete piers, or ngers, located in
the corners. Above the transition zone, the
segments are hollow with internal stiffeners.
The spires are founded on deep belled
caissons in order to evenly distribute their
overturning forces. The caissons are grouped
in pilecaps 10 caissons for the tall spire,
eight for the medium spire, and six for the
short spire. The three pilecaps are in turn
joined together by triangulated grade beams
that ensure that the foundation system is
engaged as a monolithic whole. Each spire

20/4/12 13:23:14

16

TheStructuralEngineer
May 2012

Project Focus
US Air Force Memorial

N Figure 3
Pre-assembly of spires
during fabrication

terminates in a base plate with an anchor


plate deeply embedded in the pilecap below.

N Figure 2
Thunderbird F-16s
executing bomb-burst
manoeuvre

E Figure 5
Schematic of impact
damper and location in
monument

Wind induced motions


Tall slender structures such as the USAFM
have an inherent susceptibility to wind
induced vibrations. As the monuments
function is purely aesthetic, provision in the
design for controlling wind induced motions
had to be particularly sympathetic to the
structures appearance. In practice this
meant there was little scope for reducing
the slenderness to any signicant degree
or making a favourable adjustment to the
aerodynamics.
The USAFM spires have a triangular
cross section; such cross-sections have
a vulnerability to the aeroelastic instability
known as galloping. Across wind galloping
is a phenomenon which occurs for noncircular sections in which, for certain
oncoming wind directions, the uctuating
transverse aerodynamic load (the lift) acts
with structural motion rather than against
it. Under such conditions, energy is added
into the dynamic system by the wind
with potential to develop large unstable
deections. The wind tunnel tests conducted
at an early stage of the design conrmed
that the spires could indeed gallop in their
fundamental sway modes at winds well
below the site design wind speed. This
effect needed to be addressed to reduce
the predicted deections and associated
loads on the structures. The galloping was
predicted in a number of specic oncoming
wind directions. The spires have equal
bending stiffness about their two principal
bending axes and were therefore predicted

TSE5_14-18_USairforce.indd 16

to have the same fundamental natural


frequencies in all sway directions.
The expectation was therefore that the
galloping excitation would excite the same
modal frequency for all the vulnerable wind
directions. The large deformations resulting
from galloping could result in strength
or rapid fatigue failure of the structures.
A common solution for structures with a
potential galloping problem is to adjust their
aerodynamic prole such that the structure
remains inherently stable.
However, given the constraints on the
appearance and slenderness of the USAFM,
galloping had to be countered by modifying
the structural dynamics or, more specically,
by increasing the damping. Although added
damping was essential, the structural form of
the spires restricted the number of damping
options which could be implemented. For
example, after installation, access to the
voids in the top section of the spires would be
restricted. Hence this precluded the use of
tuned vibration absorbers as there would be
no access for initial tuning or any subsequent
maintenance. It is also questionable whether
devices which rely on accurate, long-term
tuning should be used to ensure structural
integrity in concrete and composite
structures where there is potential for longterm shifts in the natural periods. Similarly,
there is limited scope for using viscous or
friction dampers on monolithic structures
such as the spires, where there is an absence

of points in close proximity which exhibit


signicant relative movement. Such dampers
work by resisting the relative velocities
between different parts of the structure.
The USAFM solution was to design and
install a set of impact dampers or ball-inbox dampers within each of the spires (Fig.
5). Each damper consists of a 750kg, 0.5m
diameter sphere, free to roll in a steel box.
The spheres are steel coated lead to provide
enough mass but with reduced size, the
internal vertical surfaces of the boxes are
coated with synthetic rubber. When the spires
vibrate, collisions occur between the inside of
the box, which moves with the structure, and
the free rolling ball. Each collision dissipates
energy and thereby provides damping. The
tallest, 80m, spire has a stack of six damper
boxes, the middle spire four and the shortest
spire three. The dampers are installed
approximately three quarters of the way up
each of the spires adjacent to the concrete
transition zone.
Impact dampers are generally only practical
for modest levels of additional damping;
say a maximum 2 to 3% of critical. But by
calculation, each of the spires required only in
the order of 1% of critical damping to ensure
stability (i.e. prevent build up of excessive
deections). Impact damper efficiency, in
terms of added damping per mass of damper,
is lower than for a tuned vibration absorber
and in addition the noise of impacts can be
unattractive for some applications. However,

23/4/12 15:22:02

www.thestructuralengineer.org

17

W Figure 4
Overview of
structural
scheme

E Figure 6
Schematic
of dampers
under mode
of optimum
operation

E Figure 7
Plot of
damping
against
amplitude
for impact
dampers on
82m spire
derived from
laboratory
tests

impact dampers have other characteristics


which increased their suitability for the
USAFM, principally:
- Provided the metallic and elastomeric
components are designed for fatigue, impact
dampers should be service free for the
design life of most structures.
- Impact damper effectiveness is insensitive
to shifts in the structures natural frequency.
Thus, the impact dampers could be
pre-installed in the steel sections prior to
erection and the subsequent restricted
access could be managed by appropriate
design.

Impact damper design considerations


The performance of impact dampers
depends principally on three dimensionless
parameters:
- The ratio between the damper ball mass
and the mass associated with the structural
mode/s of vibration being damped. The
additional damping developed increases
linearly with this ratio.
- The coefficient of restitution describing the
energy dissipation in each collision.
- The gap ratio, which is the ratio of the
amplitude of structure motion, at the location
of the dampers, to the clearance between
the ball and enclosing box in which it is free
to roll.
The gap ratio has to be correctly sized if
impact dampers are to work efficiently. If the
clearance is too great, collisions between
the ball and box are intermittent and the
energy dissipation per cycle of structural
sway motion is relatively low. If the gap is
under-sized, multiple collisions may occur

TSE5_14-18_USairforce.indd 17

per cycle but the collision velocity, and


consequently energy dissipation, is low.
Optimum performance of impact damper
is achieved by sizing the system so that,
under design operating conditions, two
collisions occur per cycle of structural
oscillation and the ball and structure are
approximately 90 out of phase as illustrated
in Fig. 6. For operation in this mode, the
gap is typically a half to three times the
amplitude of motion of the structure at the
location of the box. Maximum damping is
achieved when this ratio is in the range
two to three. The existence of these
different modes of operation means the
damping offered by impact dampers is
highly non-linear i.e. it depends strongly on
the amplitude of structural vibration. The
theoretical variation of additional damping
to amplitude of vibration is illustrated in Fig.
7. As a consequence of this specic range
of amplitudes for which impact dampers
are most effective, it was important in the
case of the USAFM to ensure structural
stability was maintained at the peak spire
displacements, which could be expected
to precede galloping. These displacements
were taken to be the peaks from buffeting
of the spires derived from the wind tunnel
test data. Buffeting is the build up of
displacements due to sequential gust
loading at a structures natural frequency.
Although the proportion of energy
dissipation during each collision governs
the damper performance to some degree,
there is limited scope for tuning this as part
of the design process because suitable
commercially available elastomeric materials

tend to have a fairly narrow range of impact


dissipation characteristics. The principal
concern in designing the impact pads for
the spires was limiting their peak strain to
a level below which the elasticity of the
synthetic rubber did not degrade. To assist,
damper pad assemblies were clad with a
stainless steel contact plate to prevent
erosion or tearing of the synthetic rubber
during collisions. For the fundamental modes
of vibration of cantilever structures (such
as the spires), damping devices operate
more efficiently if they are higher up the
structure where the mode shape value and
motions are greater. In the USAFM there
was insufficient space for the dampers
near the spire tips due to their taper. The
dampers therefore had to be located at
a lower elevation, approximately three
quarters of the way up, with the reduction in
effectiveness compensated for by additional
damper mass. Moreover, the geometry did
not allow sufficient conservatism with steel
spheres; hence, lead was favoured with a
steel casing to prevent deformation during
impact.
Theoretical predictions of impact damper
performance can be derived for a few
idealised cases. However to obtain a more
complete understanding of the design
sensitivities and ensure a robust system,
a transient simulation of the structure and
damper system was developed using rigid
body dynamics and modal vibration theory.
The simulation tool was used to nalise the
geometry of the impact dampers prior to
detailed mechanical design of the boxes and
damper pads. This simulation represented

20/4/12 13:23:40


Project Focus
US Air Force Memorial

TheStructuralEngineer
May 2012

18

W Figure 8
Laboratory test
set up at State
University New
York

E Figure 9
Plot of
predicted
upper and
lower bound
energy
dissipation
and energy
dissipation
recorded in
laboratory
tests

Spire
height

Measured
natural
frequency
(Hz)

Measured
modal
damping (%
critical)

Target
damping
(% critical)
[based
on site
measured
mode
shape]

61m

1.00

2.4

0.90

70m

0.79

1.8

0.95

82m

0.57

1.2

0.82

Table 1 Summary of site test measurements

the galloping effect by applying a negative


damping constant to the structure. One
phenomenon for which the simulation
was particularly helpful was in assessing
the effect of higher frequency modes on
vibration of the whole system. When impacts
occur in the dampers, energy is not only
transferred to and from the fundamental
mode for which the dampers are intended
to be active, it is also transferred between
the fundamental mode and higher modes. In
the case of the USAFM, the modes above
the fundamental (measured frequency =
0.57, 0.70 and 1.00Hz for the tall, medium
and short spires, respectively) which
receive energy from the impacts, are
resistant to galloping as they have positive
aerodynamic damping at high wind speeds.
They therefore tend to dissipate the energy
transferred to them adding to the overall
stability of the spires.
The numerical analysis was able to
demonstrate that the proportion of input
energy which transferred to higher modes
during build up of the fundamental bending
mode would only be of the order of 10%.
Nevertheless, this level of transfer was
sufficient to increase conservatism in the
design without leading to concerns about
the stresses induced by any vibration in the
higher modes.

Testing
Two programmes of testing were conducted
on the impact dampers: laboratory tests were
performed on a single damper unit prior to its

TSE5_14-18_USairforce.indd 18

installation in the largest spire and site tests


were performed on the completed structures
to conrm that the required levels of damping
had been achieved.
The laboratory tests, performed on a
shake-table at State University of New York
at Buffalo, were programmed to mimic the
sway of the spires during high winds. The
damper box was mounted on the shake
table via load cells such that the forces
the damper applied to the structure during
sway motions could be recorded and the
energy dissipated calculated. A photograph
of the experimental set up is shown as Fig.
8. The accumulated energy dissipation from
one of the laboratory tests and equivalent
results predicted by the simulation tool are
illustrated in Fig. 9. These, and other well
correlated results, provided condence in the
design procedure which had been employed
and in the robustness of the design. Acoustic
tests were also conducted at the laboratory;
these demonstrated that the noise of the
impacts would be indiscernible over the
background noise at the site. Site tests
(summarised in Table 1) were conducted prior
to the opening ceremony for the memorial
when the spires were structurally complete.
In these tests, the spires were dynamically
excited in their fundamental nodding mode
by a hydraulic system connected to the tip
of each spire via a vertical steel cable. Once
a prescribed amplitude of motion had been
attained, equal to the maximum dynamic
displacement which could precede galloping
conditions, the loading was removed so
allowing the oscillation of the spires to decay.
Monitoring of the decay time histories then

allowed calculation of the damping in the


fundamental mode of each of the spires.
In practice, the combination of inherent
damping in the spires and damping provided
by the impact dampers was observed to
restrict the achievable amplitudes of motion
of the spires. Nonetheless, the tests provided
condence that the design levels of damping
were comfortably exceeded in all three
spires.

Conclusions
The slenderness and form of the United
States Air Force Memorial provided
considerable engineering challenges.
The design and installation of impact
dampers allowed the architects vision
of the monument to be realised without
compromise. The performance of the impact
dampers was veried by site test and no
abnormal wind induced behaviour of the
structure has been observed over the last
ve years. The result is a striking monument
betting its signicance and location. The
memorial was accepted by President George
W. Bush on behalf of the USA in October
2006.

Daniel Powell, MA, MSc, CEng,


MIMechE
Associate, Arup, London, UK

Patrick McCafferty, BSc, MEng, PE,


MASCE
Associate, Arup, Boston, USA

Bibliography
E1

Simiu, E. and Scanlan, R. (1996) Wind Effects on Structures: Fundamentals and


Applications to Design, 3rd ed. New York: John Wiley & Sons, pp. 230-236

E2

Caughey, T. and Masri, S. (1966) On the stability of the impact damper, J. Appl.Mech.,
33 (3), pp. 586-592

E3

Argiris, L., Jackson, A., McCafferty, P. and Powell, D. (2007) Engineering the United
States Air Force Memorial Washington DC, Arup J., 41 (1), pp. 24-31

23/4/12 15:22:21

También podría gustarte