Understanding Cable Tension Mechanics
Understanding Cable Tension Mechanics
STRUCTURES
6 . 1 CABLES
The most elementary of all structural elements, and thus the possibility of buckling. Bridges with single spans longer
most easily understood, is the simple tension cable. Cables than a full mile are now common due to the combined
have been used for thousands of years, beginning with the strength of steel and efficiency of pure tension (Figure 6.1).
form of simple ropes made of natural plant fibers and ani- As introduced in Chapter 3, it is very easy to visualize
mal hairs, to the present day where the properties of the load-carrying action of a simple cable can·ying a weight
advanced materials have made the long-span bridges of the when suspended along its length. This type of behavior is
modem world possible. The high tensile strength of steel, in very familiar to almost everyone and clearly produces pure
particular, combined with the efficiency of simple tension, tension in the cable. Loads acting perpendicular to a cable
makes a steel cable the ideal structural element to span spanning a distance will always produce pure tension; how-
large distances. ever, the magnitude of the tensile force can vary throughout
Cables are flexible because of their small lateral dimen- the cable length, depending on the placement of the loads.
sions in relation to their length. This flexibility indicates that In order to understand the mechanism of how an initially
in a cable there is a limited resistance to bending. As intro- horizontal cable supports vertical loads, consider a cable sus-
duced in Chapter 5 (and further discussed in Chapter 7), pended between two fixed points located at the same level
bending is a complex structural action that involves simulta- and carrying a single load at midspan (Figure 6.2). Under the
neous tensile and compressive stresses within a member action of the load, the cable assumes a symmetrical, t:Iiangu-
cross section. In the case of cables, uneven stresses due to lar shape, and half the load is carried to each support by sim-
bending are prevented by flexibility. The tensile load in a ple tension along the two halves of the cable.
cable is thus evenly divided among the cable's strands, per- The characteristic triangular shape acquired by the
mitting each strand to be loaded to the same safe, allowable cable is referred to as the sag: the vertical distance between
stress. This behavior makes cables the singularly most effi- the supports and the lowest point in the cable. Without sag
cient structural form possible. All of the material of a cable the cable could not carry the load. We see from the vector
can be effectively used to carry loads, and there is no free body diagram (FBD) of Figure 6.3 that without this sag
{, t :
L = Span
Th7l
, " - ... (Varies with sag)
\ I
f = sag - - T Ty= W/2
II
Figure 6.2 Symmetrical load on cable (above left)
A single load, W, placed at the midspan of a suspended cable will naturally deform the cable into a triangular shape. The vertical distance of the cable
measured from the highest to lowest points along its length is referred to as its sag.
Figure 6.3 Internal Cable Force Actions (above right)
This illustration demonstrates the internal forces on the cable of Figure 6.2 as a free body diagram with the upper right joint isolated for clarity.
The tension force on a cable structure can be resolved into horizontal and vertical components. The combination of the two force components, Tv & Th, have the
same net effect on the structure as the actual force, T, but facilitate a clearer understanding of the horizontal and vertical equilibrium of the forces. Note that in
both horizontal and vertical directions, every force vector is balanced by an equal and opposite vector to maintain static equilibrium-were this not the case, the
structure would be in motion.
For a symmetrically loaded cable, it is evident by inspection that½ of the total vertica l load Wis carried by each support. The vertical components of the
forces must therefore each carry½ of the load W. The inclination of the cables generates a horizontal inward pull, Th, on the supports, each of which must
be equal and opposite in direction to maintain horizontal equilibrium. The magnitude of Th, however, depends entirely on the sag distance f, as illustrated in
Figures 6.4a and 6.4b.
the cable tension forces would need to be horizontal, but A simple experiment can bear this out. If the reader
purely horizontal forces cannot balance a vertical load - holds a thread in each hand and attaches a weight at the mid-
this would not fulfill the requirement for linear equilibrium dle, one may sense physically that the string develops no
in the vertical direction. The inclined pulJ of the sagging horizontal pull when the fingers are touching, while an
cable on each support may be split into the equivalent of two increasing pull is developed as the hands are moved apart,
components: a downward force equal to half the load and a thus decreasing the string sag. The pull may be shown to be
horizontal inward pull. Due to the horizontal action of this inversely proportional to the sag: Reducing the sag by half in
inward pull, if the supports were not fixed against horizontal fact doubles the pull (see Figure 6.4). For vertical equilib-
displacements, they would move inward under its action, rium to be maintained, the vertical pull on the hands is
and the two hal ves of the cable would become vertical. always equal to one half of the load and is independent of
, ., _r • Thi
<W/2
1
A I \
~ -~, -,
1I
\ , _ .,,. , T (Smaller TY = W/2 T (Larger Ty= W/2
force) force)
f = Deep Forces in upper right cable Forces in upper right cable
(b)
(a)
the sag. Hence, as the sag decreases, the tension the cable than the tensile strength of the thread. In actuality, if we look
exerts on the hands increases because of the increase in the at the graphic vector equilibrium of the FBD (Figure 6.4b),
horizontal, not vertical pull. If the thread used is weak it is seen that the horizontal component becomes infinitely
enough, there comes a point when it snaps, indicating that, long as the angle decreases to zero, indicating that it is tech-
as the sag diminishes, the tension eventually becomes larger nically impossible for a completely horizantal tension ele-
ment with no sag to carry a vertical force.
The cable problem just considered raises an interest-
L
ing question of economy. A larger sag increases the cable
T
71
length, but reduces the tensile force in the cable and, hence,
l 45o\ /
allows a reduction of its cross section; a smaller sag
reduces the cable length, but requires a larger cross section
f = U2 / because of the higher tension developed in the cable .
Hence, the total volume of the cable material, the product
of its cross section and its length, is large for both very
•
T = TvV2 Tv = W/2
1.41W/2 small and very large sags and must be minimum for some
intermediate value of the sag. The optimal or "most eco-
nomical sag" for a given horizontal distance between the
Figure 6.5 Optimal Sag
supports turns out to be one-half the span, and corresponds
For a load placed at midspan, the optimal sag is one-half of the span
to a symmetrical, 45-degree-triangle cable configuration
distance, L. This results in a cable inclination of 45 degrees, and so both
the horizontal and vertical components of the cable force will be equa l
with a horizontal pull equal to half of the load (Figure 6.5).
to half of the applied load. Although the force magnitude will be higher If the load is shifted from its midspan position, because
than a cable with greater sag, the length of the cable will be longer, of its flexibility the cable changes shape, and adapts itself to
thereby requiring more total material. A sag distance of one half the span carrying the load in tension by means of straight sides of dif-
represents an optimal ba lance between the amount of force in the cable an ferent inclinations (Figures 6.6a). The two supports develop
the total amount of material required in the cable. different ve1tical reactions, but equal horizontal pulls since
7'
T (right)
(Smaller force)
T v (right) = 1/3W
T (left)
T v (left) = 2/3W (Larger force)
(a)
Figure 6.6a
For a single, asymmetrically placed load, free body diagrams for the upper left and right cable ends clearly illustrate that the force magnitude in the cable
is not equal on either side of the load, the left side being greater than the right. For a load positioned at 1/3 the distance from the left support, the vertical
reaction components will be 2/3 of the total load on the left side and 1/3 the tota I load on the right side. Since no other horizontal forces are present, however,
the horizontal components on either side will be equal and opposite one another, the actual magnitude varying with the amount of cable sag.
{b)
(c)
Figure 6.6b
Two symmetrically placed loads on a cable will reshape the cable into a Figure 6.6c
three-sided polygon. Additional loads will continue to reshape the cable into a closer
approximation of a curve as the straight sides become smaller and smaller.
CHAPTER SIX Tension and Compression Structures 71
the cable forces must always be in equilibrium in the hori- If the equal loads are distributed evenly along the curved
zontal direction. The value of the horizontal pull differs from length of the cable, rather than horizontally, the funicular
the value for a centered load, but still varies inversely as the curve differs from a parabola, although it has the same general
sag. The reader may sense this by shifting the load along the configuration: It is a catenary, the natural shape acquired by a
hand-supported thread. Because the vertical components on cable of constant cross section or a heavy chain (catena in
each side of the cable are different, the resultant total ten- Latin) under its own weight, which is uniformly distributed
sion forces on each side of the load are also no longer equal along its length (Figure 6.7 b). The optimal sag for the cate-
to one another. This fact may at first be counterintuitive, yet nary is about one-third of the span; for such sag ratio, the cat-
can be clearly demonstrated by a graphic FBD at the upper enary and the parabola are very similar curves (Figure 6.7 c).
left and right joints (Figure 6.6 a). A cable carrying a combined load of its own self-weight
If the single load now is replaced by two equal loads plus a load uniformly distributed horizontally acquires a
placed on the cable in symmetrical locations, the cable again shape that is intermediate between a catenary and a parab-
adapts itself and carries the loads by acquiring a new con- ola. This is the shape of the cables in the central span of
figuration, this time with three straight sides, the middle of suspension bridges, which carry their weight and that of the
which is horizontal (Figure 6.6b). As the number of loads is stiffening trusses on which the roadway is laid (Figure 6.8).
increased, with each new loading the cable again acquires a A cable would span the largest possible distance if it
new shape in response to new equilibrium configurations. could just carry its weight, but break under the smallest
Each of the new shapes will have straight sides between the additional load (which is to say that under its own dead load
loads and changes in direction at the points of application of it is stressed to its limit). Assuming an optimal sag-span
the loads. This process can continue, and each time the cable ratio of one-third to minimize the weight of the cable, it is
will readjust with a new form for each additional load. found that such a steel cable with a strength of 200,000
It is important to note that for each configuration of pounds per square inch (1.38 GPa) could span a distance of
loading, there is one and only one shape that the cable 17 rrules (27.4 km)! This maximum distance is independent
will take. This characteristic shape acquired by a cable of the cable diameter, since both the weight of the cable and
under concentrated loads is called a funicular polygon the tension in it are proportional to the area of the cross sec-
(Figure 6.6c). The name is derived from the Latin word tion. Obviously, actual cable spans are built to carry loads,
funis for rope, and the Greek words poly for many and which are usually much heavier than the cables themselves;
gonia for angle. The funicular form is the natural shape hence, cable spans are much shorter than the limit span of
required to carry loads in pure tension or (as will 17 miles. The longest North American suspension bridge to
be shown) pure compression and therefore represents the date (2015), the Verrazano Narrows Bridge at the entrance
optimum structural form that can be achieved by a unidi- of New York harbor, spans 4260 feet (1,298m) between the
rectional spanning structure. It is for thi s reason that the towers, a length over two-and-a-half times that spanned by
longest-spanning bridges are supported by cables. the Brooklyn Bridge (which opened in 1883).
As the number of loads on the cable continues to increase, The Verrazano Narrows Bridge reigned as the longest in
the funicular or string polygon, acquires an increasing num- the world for 17 years from 1964 until surpassed in 1981 by
ber of smaller sides and approaches a smooth curve. If one the Humber Bridge in England, with a span of 4626 feet
could apply an infinite number of infinitesimally small loads (1,410m) between towers. The Akashi-Kaykio Bridge in
to the cable, the polygon would become a funicular curve. For Japan (fig 6.1), completed in 1998 with a span between tow-
example, the funicular polygon for a large set of equal loads ers of 6,532 feet (1,991m) presently has the longest main span
evenly spaced horizontally approaches a well-known geo- in the world. Such spans are rapidly approaching limiting val-
metrical curve, the parabola (Figure 6.7 a). The optimal sag ues, beyond which it is not practically conceivable to go with
for a parabolic cable equals three-tenths of the cable span. the types of steel available today. Only an improvement in
lI
1111,,11111 I I Irn fr( \_Catenary
Figure 6.9 Proposal for a bridge across the Strait of Gibraltar by Prof. T.Y Lin
With a main span of 16,400 ft (5000 m), the bridge would be 2.5 times longer than that of the Akashi Kaikyo bridge of Figure 6.1. This incredible span would
be made possible by a hybrid structural system of cable-stayed structures at the masts that in turn support suspension cables, in turn supporting a deck of
high strength but light weight glass fiber. At the present time (2015), the bridge remains an unrealized proposal. The obstacles to be overcome, particularly
with supporting piers plunging nearly 3300 feet (1 kM) in water, are enormous.
Photo courtesy of Deborah Oakley
cable strength will make substantially larger spans possible Stiffening trusses are usually rigid in the direction of the
with a single suspension cable, although hybrid designs that bridge axis, but less so in a transverse direction: Large dis-
combined suspension cables with cable-stayed structures can placements of suspension bridges caused by lateral winds
further extend the range, such as the proposed span at the can be substantial. Moreover, the long roadway and the shal-
Strait of Gibraltar (Figure 6.9) by Dr. T.Y Lin. low trusses constitute a thin ribbon, so flexible in the verti-
Given the inherent efficiency and economy of steel cal direction that it may develop a tendency to twist and
cables, one may wonder why they are not used more fre- oscillate vertically under steady winds (see Section 2.6).
quently in smaller structures. The limitations in the applica- Modem suspension bridges are made safe against such dan-
tion of cables stem directly from their adaptability to gers by the introduction of stiffening guy wires (like those
changing loads: Cables are unstable, and (as introduced in used in the Brooklyn Bridge in New York, designed in 1867)
Chapter 4) stability is one of the basic requirements of struc- and by an increase in the bending and twisting rigidity of
tural systems. The trusses hanging from the cables of a sus- their roadway cross section.
pension bridge have the purpose not only of supporting the In so-called stayed bridges of the "harp" or "fan" type
roadway but also of stiffening the cables against motions (Figures 6.10 and 6.11), the guy wires- or stays- have the
due to changing or moving loads. double role of supporting the deck and of stabilizing it.
CHAPTER SIX Tension and Compression Structures 73
Cable Saddle
anchorage for the suspension cables of New York's
Verazzano Narrows Bridge, which opened in 1964. The Splayed Cables
main suspension cables actually comprise numerous
bund les of smal ler cables, each of which in turn
are made of multiple single wires. The cables are
Cable Anchor Girder
splayed within the anchorage to permit each wire
bund le to have a connection within the anchorage
structure. The enormous mass of the concrete
anchorage counterbalances the tension from the main
suspension cable.
ratio leads to very tall, expensive towers. Compression in development has brought about a number of new solutions
the towers, bending in the deck (often comprised of trusses), in which tensile cables are the basic element in what may
and shear in the anchorages are essential to the stability and otherwise be a complex structural system. As noted above,
strength of the tensile cables of suspension bridges. In self- the relative flexibility of cables is a primary consideration.
anchored bridges the cables are anchored to the ends of the There are numerous design approaches, each with different
stiffening trusses or beams, thus compressing them, while aesthetic qualities and fit with the architectural require-
the ends of the trusses are anchored to the piers to develop ments, but each design solution is essentially aimed at stabi-
downward reactions (see Figure 6.14). lization of the cable system.
Cable bridges are essentiaJly unidirectional spanning The most simple tensile roof design consists of a series
structures, which is appropriate for bridges . But this does of cables hanging from the tops of columns or buttresses, as
not mean that the use of cables is constrained only to one though taking a number of suspension bridges and placing
direction . On the contrary, complex structures involving them side-by-side. Unlike the massive anchorages of a sus-
cables at multiple angles that cover a surface as well and pension bridge, however, the columns must be capable of
othe r tensile structures are frequently used in building significant bending resistance if the cable terminates at the
structures, and are considered in the next section. column, due to the cable's large horizontal force component.
An alternative design solution that prevents this column
bending is to pass the suspension cable over the tops of com-
6.2 CABLE ROOFS pressive struts (i.e., columns) and anchor them to the ground,
The exceptional efficiency of steel cables suggests their use instead of terminating the cable at the column. In either
in the construction of large roofs such as sports stadia and design strategy, straight beams or plates then connect the
transit terminals. This relatively recent twentieth century parallel suspension cables, thus creating a polygonal or
CHAPTER SIX Tension and Compression Structures 75
inverted barrel roof surface (Figure 6.15). The simplicity inclined posts supported by shorter compression struts that
and low cost of this suspension-bridge scheme would make provide overall lateral stability to the building. The horizontal
it popular, but for the fact that the straight elements connect- roof consists of reinforced concrete beams and slabs prefabri-
ing the cables are usually light and tend to oscillate or "flut- cated on the ground and is suspended from the cables by
ter" under the action of wind. To avoid flutter, the roofing means of wire hangers. The dead weight of this relatively
material must be relatively heavy, or the cables must be sta- heavy roof acts as a stiffening truss. It provides the necessary
bilized by guy wires or stiffening trusses. stabilizing force for the suspension cables, plus resists their
The suspension-bridge principle was directly adopted in tensile anchorage force though in-plane compression. This
a structure designed in 1961 by Nervi and Cover for an Italian brilliantly simple structure permitted the economic roofing of
paper manufacturing plant (Figure 6.16). This structural al0,000 square feet (930 m2) area without a single intermedi-
scheme was an ideal fit with the architectural program because ate support or interior column. The peripheral walls are inde-
of the linear nature of the paper-malting process. Unlike most pendent of the roof structure, and wind pressure against these
suspension bridge designs, this is a self-anchoring structure exterior walls is resisted by ve1tical columns. Although verti-
(similar to the much later bridge of Figure 6.14), which bal- cal, these act more like beams than columns, since most of
ances the cable tension with internal compression in the roof their structural action is in bending.
structure rather than directing the tension to large anchorages Regarded as a twentieth century architectural master-
in the ground. The reinforced concrete towers of each roof piece, Eero Saarinen's 1962 Dulles Airport Terminal near
section (100 feet (30.5 m) wide and 830 feet (253 m) long) are Washington, D .C. (Figure 6.17), is similar in structural
76 CHAPTER SIX Tension and Compression Structures
principle to Nervi 's paper mill, but consists of multiple par- parabolic arches simultaneously support a series of main
allel cables. The roof structure is a heavy concrete slab sup- suspension cables and provide for cable tensioning by their
ported directly by the suspension cables; however, it does deadweight (Figure 6.19). Stabilization of the main down-
not resolve the tensile force of the cables. Here instead, the ward-sagging cables is obtained by upward-curving cables
massive concrete pylons supporting the roof are angled out- at right angles to, and on top of, the primary cables-this
ward, their dead weight and tendency to outwardly overturn then creates a mesh for support of the roof structure. The
counterbalances the horizontal pull of the roof cables. roofing consists of lightweight corrugated metal plates per-
Designed at the start for expansion, the terminal was more manently anchored to the cable mesh. The concrete arches
than doubled in length in the late 1990s by adding addi- are themselves stabilized by ve1tical columns. By means of
tional columns, cables, and roof structure matching the compression along their curvature, the arches resist the
original design. inward pull of the main cables. The surface thus defined by
A related yet different principle was used by Saarinen in the cables resembles a saddle and is more stable under wind
the Yale University Ingalls skating rink in 1958 (Figure 6.18). loads than a barrel-shaped roof. Notwithstanding the shape
In this design, the cables are suspended perpendicularly from of the roof, in order to avoid flutter of the lightweight cor-
each side of a central concrete arch, which has an inverted rugated plates, it was still necessary to stabilize the roof
curvature at the ends. The cables hang with a natural down- panels by guy wires connecting a number of internal mesh
ward sag, and the outer ends are anchored to the rink's heavy points to the outer vertical col umns. The interior of the
peripheral walls that are curved outward in plan. The roofing structure is completely column free, and the essentially
material is wood: Unlike Nervi 's paper mill structure or the free-floating roof allows for abundant daylight through the
Dulles Airport Terminal, its relatively lightweight does not perimeter walls.
stabilize the cables entirely. The stabilizing is therefore In designing the roof for the 1956 Cilindro Municipal
accomplished by upward curving tension cables that pull stadium in Montevideo, Uruguay, Leonel Viera invented an
downward against the primary suspension cables. The cen- inexpensive and stable system that was well suited to cover
tral concrete spine was too slender to be laterally stabilized large circular areas (Figure 6.20). In this roof, a series of
by the roof structure alone, and so several additional straight radial cables connects a lower, central tension ring of steel to
cables tie it directly to the perimeter walls-a slight conces- an outer compression ring of concrete. The outer ring resists
sion to the architectural desire of making the spine as slender all the tensile force of the cables through circumferential
as possible. compression, and is supported by a cylindrical exterior con-
The 1952 solution by Mathew Nowicki for the roof of crete wall. With all tensile forces resolved within the roof
the Raleigh, North Carolina, arena illustrates an early structure itself, the exterior walls were required to only carry
implementation of the concept of interlocking counter- vertical loads, and thus were relatively thin. The roof deck-
curved cables. The building profile is dominated by two ing consisted of a large number of prefabricated, wedge-
large inclined, intersecting concrete arches. In a manner shaped concrete slabs, which were supported on the radial
similar to the Dulles Airport Terminal pylons, these massive cables by the hooked ends of their own reinforcing bars.
CHAPTER SIX Tension and Compression Structures 77
---- - The genius of this roof structure was due as much to its
construction technique as to its structural concept. Here, to
reduce cable instability, during construction the slabs were
loaded with a ballast of bricks or sandbags after placement,
Concrete which temporarily over-tensioned the cables. The resulting
panels stretch of the cables created gaps between the wedge-
shaped slabs. These radial and circumferential gaps were
subsequently filled with cement mortar. Once the mortar
Ballast weight had set, the entire roof became a monolithic concrete "dish."
When the temporary ballast was taken off the dish, the
cables tended to shorten, but were prevented from doing so
by the monolithic concrete roof in which they were embed-
ded (see also Section 12.7). The inverted roof was thus pre-
stressed by the cables, and showed little tendency to flutter.
Similar roofs, like that of Madison Square Garden in New
Stage2 York City and the Oakland-Alameda County Coliseum
Gaps filled with Arena (Figure 6.21), have been successfully and economi-
cement grout cally built in the United States and elsewhere on this prin-
ciple (though with different construction technique), which
Viera also applied to suspension bridge design.
The dish shape of downward sagging cable-supported
roofs presents an architectural challenge of how to address
Stage 3 the natural tendency for such shapes to collect rainwater. In
(Completed roof) the case of the roof of the Dulles Airport Terminal, three
Radial large sculptural elements dominate the interior roof surface
Compression
ring
of the terminal, which conceal vertical drainpipes. In the
case of arenas, such centrally located drainpipes would be
completely counter to the need for having an open clear span
in the first place. The drainage of Viera-type roofs is obtained
by the somewhat inelegant solution of pumping the rainwa-
ter to drain pipes located on the outer rim of the roof. In the
Tension ring
Madison Square Garden and Oakland arenas, structures
with very low-slope roofs housing offices and mechanical
rooms were constructed atop the downward curving main
Figure 6.20 The Cilindro Municipal, Montevideo, Uruguay, by Leonel Viera roofs. Rainwater on these slightly sloped roofs drains natu-
In order to provide permanent rigidity in the essentially simple cable- rally by gravity to the perimeter of the structures, thus avoid-
supported roof structure, Viera's construction process innovation was to ing the need for mechanical pumping.
hang wedge-shaped tiles on cables with gaps in between. The tiles were then Unlike the stadium or terminal roofs described above,
covered with additional ballast weight, which caused the cables to stretch. in the design of airplane hangars it is essential to provide
Cement grout was filled within the gaps and the ballast was removed after large front doors that slide open smoothly. These sliding
curing. The lessoning weight caused the cables to shrink, but with this doors cannot be supported by columns and are usually
movement prevented by the now cemented roof, a permanent compressive hung from a top beam, longer than the hangar opening. The
prestress in the roof surface was created. The roof structure functioned for structural problem to be solved for this type of building
54 years until a fire inside the stadium caused its collapse in 2010. The consists in covering a rectangular area having columns on
remaining structure was subsequently demolished.
r7';
::
:''i Compression mast
''
''
''
''
'' ''
Lr" u-
125' 38 m
not more than three of its sides. A solution, realized by A related solution was adopted by Tippets-Abbett-
Ammann and Whitney in 1955 for the TWA Maintenance McCarthy and Stratton in roofing the 1960 Pan-American
Hangar at the Philadelphia Airport, used a scheme in which Worldport terminal at John F. Kennedy International Airport
a cantilevered span is equivalent to one half of a stayed in New York. If one could imagine taking the cross section of
bridge (Figure 6.22). The space between "tower" and the Philadelphia hangar, and rotating that section about a cen-
"anchorage" was used for offices or shops, while the cable- tral point, this essentially describes the Pan-American termi-
supported cantilevered roof consisted of "folded-plate" nal concept, though the form was elliptical versus circular in
reinforced concrete construction (see Section 10.8). The plan. Considered revolutionary in its day, the roof consisted of
limited weight and high resistance to bending and com- reinforced concrete plates supported by radial beams, an outer
pression of folded plates, together with the tensile strength elliptical ring and an inner elliptical ring (Figure 6.23). The
of cables, make such structures capable of economically beams are cantilevered from the outer ring, and supported by
cantilevering as much as 125 feet (38 m). cable stays running over compression struts at the outer ring,
capacity was soon exceeded and subsequent ,Concrete F¢undations Tension Anchors!
I I
additions detracted from the purity of its I
Valley Cable
Compression Strut
outer compressive ring of reinforced concrete (Figure 6.27). under 800 feet (244 m) it is presently (2015) the largest
The vertical pipes, sheathing the continuous cables, are sup- cable-supported domed stadium in the world. The 2012 La
ported by the tensed diagonals, and kept vertical by concen- Plata dome (Figure 6.28) is the latest iteration of the Levy-
tric steel rings connected to their foot and inclined stays type "Tenstar Dome."
connected to their top of concentric steel rings. Radial- Pneumatic roofs present one of the most interesting
tensed cables, running between the compression and tension applications of cables to the reinforcing and stiffening of
rings along the vertical pipes' bottoms, constitute the lower membranes. They consist of air-supported or air-inflated
chords of the trusses. Similar radial cables (the "ridge plastic fabrics stretched over a network of cables and can
cables") running along their tops constitute the upper chords span hundreds of feet. Geiger and Levy-style tensegrity
of the trusses. The prestress in the radial ridge cables domes can also be considered as membrane roofs supported
increases the tension in the lower chords due to the roof by trusses rather than air pressure: Elimination of the fan
loads, and is high enough to entirely cancel the compression system and revolving doors (needed to maintain air pres-
in the upper chords due to these loads. A third set of pre- sure) reduces the cost of these roofs in comparison with that
stressed radial cables (the "valley cables") runs from the of pure pneumatic roofs . Chapter 11 covers these membrane
compression to the tension ring along the "valleys" formed structural types in detail.
by the intersections of the inclined stays at the concentric The tensile solutions mentioned above are just a few
steel rings. All the components of the structure can be of the most recently adopted to roof large areas. Whereas
bought "off the shelf." The undulating roof surface is a the largest span covered by a compressive roof to date
membrane of silicone fiberglass fabric wrapped over and (2015) is the 715 feet (218 m) C.N.I.T. double concrete
attached to the ridge and valley cables. Since the tension shell in Paris (see Section 12.7), tensile roofs can easily
ring is at a higher level than the compression ring, these span much larger distances. In view of the exceptional
roofs are shallow tensile domes similar in principle to "bicy- potentialities of cable roofs, it is to be expected that their
cle-wheel" roofs. The 1992 Georgia Dome by Weidlinger's use will increase with the increasing dimensions of future
Matthys Levy uses a similar principle, and at a span of just areas to be covered.
CHAPTER SIX Tension and Compression Structures 81
J
6.3 TRUSSES is placed between the two supports, this horizontal move-
ment will be prevented, and the tensile force will be bal-
At the beginning of this chapter, a flexible cable support- anced by compression within the strut (fig 6.29c). This is
ing a load at midspan was found to be a pure tensile struc- the simplest form of the structural type known as a truss.
ture (figs. 6.3 and 6.29a). Structural stability of this basic The triangu lar shape described by the cables and compres-
cable system requires that the ends be fixed to supports sion strut is refe1Ted to as a panel- larger trusses described
that prevent movement of the cables. If one support, for below are multipanel trusses. The outer structural elements
example, were on a roller, the tensile pull on the cable of the panels are referred to as chords of a truss. In these
would cause an immediate collapse by horizontal move- initial examples, the chords are horizontal, but sloping
ment of the roller (Figure 6.29b). If instead a wooden strut chords are common as well.
Consider now the structure that is created by inverting
the original cable structure and strengthening its inclined
H = Horiz. Pull sides to make them capable of resisting compression. The
"negative sag," or rise changes the nature of the stresses, and
the inverted cable becomes a pure compressive structure
Tv = W/2 (Figure 6.30).
T T The load at the top of the truss is channeled by the com-
pressed struts to the supports, which are acted upon by down-
ward forces equal to one-half the load and by outward
horizontal thrusts. By inverting the structure, therefore, all of
the forces are exactly equal in magnitude, but opposite in
direction, to the original tensile truss. The horizontal thmst of
the inverted truss can be absorbed either by compression in
buttresses of a material such as stone, masonry, or concrete,
(a)
.......
\ _ _ Tensile
steel rod
(b) (c)
Figure 6.29b Weight on cable with roller support Figure 6.29c Tensile truss with compressive strut
If the rigid anchorage of the hanging cable is replaced on one side by a The horizontal pull in the hanging cable is resisted by compression
roller connection, a collapse mechanism results due to the tension of the developed in the strut, instead of transferring this force to the support. This
cable. It is an inherently unstable structure. is the most simple form of a truss, consisting of a single panel.
82 CHAPTER SIX Tension and Compression Structures
/ (b)
w
l
Figure 6 .34 Truss with Compression Diagonals
Inverting the direction of the diagonals of the Pratt truss in Figure 6.33 will also cause the internal forces to invert. Providing the same equal load at each panel
point as the Pratt truss, the diagonal members switch from tension to compression members, and the shorter vertical members switch from compression to
tension. Created by the designer William Howe in 1840, the construction typically combined heavy wood diagonal members with light iron vertical members,
which simplified construction compared to the all-wooden trusses of the era. The dotted lines represent theoretical zero-force members. If all loads are applied
at the lower panel points, the horizontal and vertical end members may be eliminated from the structure altogether since they carry no force. This is the form of
many railway and highway bridges of the late 19th and early 20th centuries.
force members. This vertical member would be necessary, created bending forces on members that were designed for
however, to support a load placed at the upper joint, making only tension or compression.
the formerly zero-force member a compression member. In Figure 6.33, we can see that merely reversing the
If the smaller truss of Figure 6.32a is itself supported by inclination of the interior diagonal members causes these
a larger cable and strut structure (Figure 6.32b), then a bars to work in compression, and the vertical members in
longer-span truss of six panels is created. Note that the tension, much as the single panel compressive arch of
"doubling-up" of the compression member of the top chord Figure 6.30 is a mirror of a hanging cable. The reader may
is reflective of the increase in force in this member. If the wish to diagram the conceptual assembly of this truss type
six-panel truss is in tum supported yet by another cable and in the same manner as the diagrams of Figure 6.32a-c to
strut structure (Figure 6.32c), then a still longer truss con- follow its load-channeling mechanism. In this truss- under
sisting of 12 panels is created. Again, the additional horizon- this specific loading- the end verticals and last upper chord
tal top strut and bottom cable elements in the larger structure bars are unnecessary and are thus theoretical zero-force
are reflective of the increase in force in the members toward members. Similarly, the end verticals and the last lower
the center of the span. A complete Pratt truss is created in chord bars would have been unstressed ( and thus redun-
this conceptual process as illustrated in Figure 6.33. dant) if the truss of Figure 6.33 had been supported at the
The bars of trusses are considered hinged at their con- upper left and right joints. The scheme of Figure 6.34 is
nection or panel points. Because they are hinged, they are commonly used in highway bridges, and that of Figure 6.33
free to rotate at their ends and do not develop bending if in railroad bridges, where the truss is often below the road-
properly loaded. If, as is desirable, the loads on the truss way. Additional insight on truss behavior may be gained by
are applied only at the panel points, and the dead load of analogy with beam behavior, as shown in Section 7 .2.
the truss members is negligible, all the truss members are Upper chord members, as well as interior verticals and
only tensed or compressed. When loads are applied diagonals, may buckle under compression unless properly
between the panel points and the dead load is not negligi- designed. One of the first studies on the buckling of trussed
ble, the truss bars develop some bending also (see Sec- bridge structures was prompted by the failure in buckling of
tion 7.4). Therefore, for maximum efficiency, the heaviest a Russian railroad bridge at the end of the nineteenth cen-
loads should be located only at panel points. Numerous tury. Since, moreover, moving loads may produce either ten-
structural failures have occurred through loading that sion or compression in the same member, depending on the
84 CHAPTER SIX Tension and Compression Structures
i
w w
Figure 6.36 AGable Truss
Figure 6.35 AWarren Truss Commonly used in residential construction, this truss form provides for a vaulted
Named for its inventor, James Warren, who patented it in 1848. Well suited ceiling. The angle between the upper and lower chords at the supports is limited
for prefabricated construction, there are no vertical members and all by large forces developing as the angle becomes smaller. At small angles, very
diagonals are of the same length. The diagonals must be designed for the high shear forces develop at the connection that must be transferred between
maximum compression force, however, since a heavy moving load may each wood member. Typically, bolting and/or a variety of metal connector pieces
cause a tension member to become compressive, or vice-versa. are used in wood construction to augment the wood itself.
load location, some trusses are built with both tension and
compression diagonals, so that loads may be always carried
by tension diagonals with purposely slender compression
diagonals that are simply considered inactive when buckled.
The combinations of tension and compression bars
capable of producing practical trusses are extremely varied.
Figures 6.35 and 6.36 illustrate two of the many forms of
trusses possible.
The bars of a truss are joined by being bolted or welded
to a "gusset plate" at their panel points (Figure 6.37). In
either case, the restraint against relative rotation produced
by the gusset plates transforms the truss bars from pure ten-
sion or compression members into elements developing a
minor amount of additional bending and shear stresses.
These so-called secondary stresses are considered in
Figure 6.37 Truss gusset plate
Section 7.4.
Trusses are used in bridge design to span hundreds of feet
Acommon type of connector in a truss is known as a gusset plate. Agusset
plate is a means of physically connecting members while transferring large
between suppo1ts. They may be cantilevered from piers and in
forces between them. They may be used in a decorative manner, such as turn cany other simply supported trusses (Figure 6.38).
this heavy timber truss shown being fabricated, or may be hidden within the Bridge trusses with curved top chords behave very much like
member thickness in so-called knife plate connections. Only the bolts are visible suspension bridges (Figure 6.39). Parallel trusses are com-
on the outside of the members in the latter case, and sometimes even these are monly used in bridges and in steel design to cover large halls
countersunk and the holes plugged for a completely smooth appearance. (Figure 6.40). Open web joists are "off the shelf' light steel
Photo courtesy of Deborah Oakley
Figure 6.38 The Forth Railway Bridge, at the Firth of Forth, Scotland
This bridge, consisting of two main spans of 1,710 feet (520 m) each, was inaugurated in 1890. Designed by engineers Sir Benjamin Baker and Sir John
Fowler, it is constructed as three double-cantilever spans, each supported at their centers by enormous stone and wood caissons sunken into the riverbed. The
two outermost cantilever arms are separated from the central structure by much shorter and smaller trusses, simply supported by the cantilever arms of the
center and outer structures. The span is thus increased without adding even more depth to the already enormous main cantilever trusses, allowing ships to
pass freely. It remains the second longest cantilever bridge in the world even after more than 125 years of continuous service.
Photo: Ramonespelt/Fotolia
CHAPTER SIX Tension and Compression Structures 85
=•~•>;&f
-~t.1
I _;
.._~ I .,
I
Figure 6.39 The Macombs Dam Bridge, Spanning the Harlem River in New York
Opened in 1895, the Macombs bridge is a double canti lever supported from a centra l pivot point, referred to as a swing bridge, since the entire bridge can
pivot about the central support to allow ships to pass on either side. The curving shape of the top chord reflects the nature of the increasing forces toward the
central support, and in fact bears resemblance to a cable suspension bridge design.
Photo courtesy of Davie Fried
Figure 6.40 AParallel Chord Trussed Roof Figure 6.41 An Open-Web Steel Joist
Rows of para llel chord trusses are a common and practical method of Basically a premanufactured truss, the open-web steel joist (OWSJ)
creating roofs or floors over long spans such as auditoriums, concert is a mainstay of efficient modern construction. Fabricated in a range
halls, stadiums, and so on. Smaller spaces may utilized preengineered/ of standard depths, lengths, and capacities, designers can select the
prefabricated members selected from a catalog, based on span and load required member type from a catalog to suit the span and load capacity
requirements. Larger spaces (and thus longer clear spans) commonly requirements of a specific design.
feature custom fabricated trusses designed for a single specific project.
trusses used to span small distances, either as roof or as floor compressive structure. It was developed in the shape of a
structures (Figure 6.41). Vertical trusses are used in high-rise half-circle by the Romans (the greatest road builders of
steel buildings to stiffen their frames against wind and earth- antiquity) to span large distances. An aqueduct supported
quake forces (see Section 8.3). by a series of semicircular arches survives today in Sego-
Although not as popular today as they were in the via, Spain (Figure 6.43). Arches [Link] also used in a variety
1800s, at the beginning of the structural steel era, trusses of shapes to span smaller distances. The arch is one of the
are still to this day one of the most essential components basic structural elements of all types of architecture.
of large structures. The ideal shape of an arch capable of supporting a given
set of loads by means of simple compression may always be
found as the overturned shape of the funicular polygon for
6.4 FUNICULAR ARCHES the corresponding tension structure: It is by this method
Returning once more to the pure funicular forms intro- that the Spanish architect Antonio Gaudi determined the
duced at the start of this chapter to consider the inverse of form of the arches in the Church of the Sacred Family in Bar-
the tensile form: the funicular arch. The parabolic shape celona and many of his other structures (see Figure 14.15).
assumed by a tensed cable can-ying loads uniformly dis- The funicu lar polygon for a set of equal and equally
tributed horizontally may be inverted to give the ideal spaced loads converging toward or diverging from a. com-
shape for an arch developing only compression under this mon point is a regular polygon, centered about this point
type of load (Figure 6.42). The arch is an essentially (Figure 6.44a). At the limit, when the loads become infinite
86 CHAPTER SIX Tension and Compression Structures
in number and infinitesimally small, the funicular polygon The channeli ng of vertical loads along a curve by
becomes the funicular curve for a radial inward pressure or means of compression is today such an elementary con-
outward suction, and is a circle (Figure 6.44c). Since arches cept that it is hard to realize how slow its evolu tion has
can be funicular for only one specific set of loads, the circle been. Practically unknown to the Greeks, it produced the
cannot be the funicular curve for a vertical load uniformly daring and lovely lines of the Gothic cathedrals as well as
distributed either along the horizontal or along the circular those of modern bridges (see Section 8.4). The ultimate
arch. Hence, the Roman arch develops some bending and, consequences of the use of curved structural elements are
although a compression structure, is not uniformly com- even more recent and interesting: They are considered in
pressed across its depth. Chapters 11 and 12.
The shape of a masomy arch is usually chosen to be the
funicular of the dead load. But whenever an arch is used in a
bridge, a variety of moving loads must be assumed to travel KEY IDEAS DEVELOPED IN
over the arch, and a state of stress other than simple com-
pression is bound to develop: The arch also develops some
THIS CHAPTER
bending stresses. A cable can caITy any set of loads in ten-
• Cables are flexible structures that support loads in pure tension.
sion by changing shape; a rigid arch cannot change shape • Because of their flexibility, cables change their shape under
and, hence, cannot be funicular for all the loads it is sup- moving loads.
posed to carry. The stability of the arch implies lack of • A cable hung from two supports and loaded by a concentrated
adaptability. The relative importance of bending stresses in force in the middle will form an inverted triangular shape; the
arches is considered in Section 8.4. sides of which will be straight lines.
CHAPTER SIX Tension and Compress ion Structures 87
• The inclined reactions at the support will match the cable angle, QUESTIONS AND EXERCISES
and can be described by two components: one vertical and one
horizontal. The vertical components balance the vertical load 1. Using a piece of string approximately 12" (30 cm) long, hold
by carrying a portion of the load on each end, while the hori- the string between the thumb and forefinger of each hand while
zontal components prevent the cable from collapsing inward. suspending a small weight of about one pound (4.5 N) from the
• The horizontal components are always equal and opposite to string (a sturdy coffee mug will work for this). Observe the ten-
each other, and the magnitude is inversely proportional to the sion felt in the string through the fingers. Notice how the force
sag of the cable. The smaller the sag of the cable, the greater decreases as the hands are brought together and the string gets
will be the horizontal force components at the supports. deeper, and how the force increases as the hands move apart.
• When additional loads are applied, the cable readjusts with Is it possible to make the string completely flat with enough
straight lines between loads and supports. The vertical reactions tension? Why or why not?
add up to the total load; the horizontal components will increase 2. Tie the weight (e.g., a coffee mug) at a point not in the mid-
but remain equal and opposite to each other. dle at the string (it needs to be tied to prevent sliding on the
• Under an evenly distributed vertical load, the cable forms a stiing). Lift the string again with two hands and feel the differ-
smooth parabolic shape. ent forces on your hands. Which one is greater? Why?
• Suspension bridge cables are parabolic. 3. Take a flexible ruler, bend it into a shallow arch end place the
• The cable ends of suspension bridges are typically embedded ends between two heavy books. Press down on top of the arch.
into heavy buttresses (anchorages) in order to supply the nec- Why do the books move?
essary horizontal reaction. Because of the inherent instability 4. Observe spanning systems in buildings and bridges. This
of cables, stiffening trusses are used to stabilize suspension is ideally done in person, but many structural images can
bridge cables. now be found on the Internet. Which structures use sus-
• Cables supporting roofs are laid parallel to each other or are pension cables, which use stay cables, which are trusses,
arranged radially. In either case, the cable ends are anchored or and which use arches? Is one type more common than the
are held by a compression ring. others? How does it appear that the spanning distance or
• When a cable that supports a load at midspan is turned upside usage of the structure impacts the choice of the structural
down, it becomes a compressive structure. Its sloping sides are system?
in pure compression, and horizontal buttress reactions prevent
the ends from moving.
• If a third horizontal, tension member is added to replace the hori-
zontal support reactions, a basic three-bar truss panel is created. FURTHER READING
• Several panels connected to each other may form larger and
more complicated trusses. In ideal trusses, the members are in Allen, Edward and Zalewski, Waclaw. Form and Forces: Designing
either pure tension or in pure compression. Efficient, Expressive Structures. John Wiley & Sons, Inc. 2009.
• Cables loaded at several points when turned upside down (Chapters 1 and 2)
become arches, with their members in compression. Sandaker, Bjorn N., Eggen, Arne P., and Cruvellier, Mark R. The Struc-
• Stone and concrete arches that can support compression have been tural Basis ofArchitecture, 2nd Edition. Routledge. 2011. (Chapter 10)
built since Roman times as bridges, aqueducts, and cathedrals.
Schodek, Daniel and Bechthold, Martin. Structures, 6th Edition.
• The longest-spanning arch bridges of today are made of struc-
Pearson Prentice Hall. 2008. (Chapter 5)
tural steel elements
• Circular arches such as bicycle wheels and roofs of similar shape
are in circumferential compression and are loaded by radial ten-
sion members or even membranes.