PTI Journal Technical Session Paper

MULTI‐TUBE SADDLES FOR STAY CABLES

By
TOMMASO CICCONE AND
ANDREA CASTIGLIONI DI CARONNO

Authorized reprint from: October 2018 issue of the PTI Journal

Copyrighted © 2018, Post‐Tensioning Institute
All rights reserved.
 TECHNICAL SESSION PAPERS

MULTI-TUBE SADDLES FOR STAY CABLES

BY TOMMASO CICCONE AND ANDREA CASTIGLIONI DI CARONNO

Saddle systems for stay cables are at the moment one a consequence of the sensible savings of material provided
of the preferred solutions in the field of cable-stayed bridge in the pylon erection. As a matter of fact, the absence of
design across the world, and this trend seems to increase day anchorages at pylon level allows the designer to reduce the
by day. overall dimension of the tower, making it more slender
A stay cable saddle system allows the bundle of tensile and, as a consequence, architecturally elegant and graceful.
elements to run continuously from one anchorage to the other, On the other hand, a stay cable saddle is a complex
both at deck level, without using a terminal anchorage at structural system whose correct behavior is, in many cases,
pylon level. This provides a remarkable reduction in the overall intrinsically nonlinear and strictly dependent on a proper
dimensions of the pylon and, as a consequence, material savings detailing of all its components. Moreover, inspectability,
as well as aesthetically pleasing architectural shapes. maintenance, and replaceability of the stay cable system
The most advanced systems currently available for can be more complicated when saddles are present.
parallel strand stay cable systems are multi-tube friction A careful design of the saddle and an accurate instal-
saddles, where each single strand runs into an individual tube lation procedure are the two main keywords for a safe
and the stay cable differential force is transferred by friction. behavior of the cable-stayed bridge.
The structural behavior of a specific saddle device is a very
complex topic involving many mutually dependent aspects. DESIGN HINTS
The development and validation of saddle systems is usually In general terms, a stay cable saddle is a structural
performed in design assisted by testing, a procedure which is device that allows continuous deviation of the tensile
making its way in the civil construction field. elements from the deck through the tower and back to
An extended testing campaign has been performed the deck, as well as transferring the stay cable force into
to investigate the saddle behavior. Tests have been the pylon. The most advanced systems available by now
carried out, taking into account current developments of are multi-tube saddles, where each single strand runs into
international recommendations, such as PTI DC45.1-12 an individual pipe and the differential stay cable force—
and fib Bulletin 30. Friction, fatigue, and static behavior that is, the difference between the forces acting at saddle
have been assessed through full-scale tests executed in the ends—is resisted by friction.
Structural Laboratory of Politecnico di Milano (Italy). The first key point to be considered when designing
This paper describes the results of these tests and the main a saddle is of course its geometry, which in many cases is
design considerations that follow. dictated by the configuration chosen for stay cables. It is
clear that the stress level in the cable is strictly influenced
INTRODUCTION by the saddle. In other words, the uniform tensile stress
Multi-tube saddle systems have become usual distribution produced by permanent and live loads over
solutions in cable-stayed and extradosed bridges, mainly as the stay cable cross section is locally disturbed by bending
stresses induced by the geometrical curved configuration
PTI JOURNAL, V. 14, No. 1, October 2018. Received and reviewed under of the saddle. Such bending stress distribution is difficult
Institute journal publication policies. Copyright ©2018, Post-Tensioning Institute.
All rights reserved, including the making of copies unless permission is obtained
to be properly predicted due to the intrinsically nonlinear
from the Post-Tensioning Institute. Pertinent discussion will be published in the interaction among wires. Many formulas are available
next issue of PTI JOURNAL if received within 3 months of the publication. in literature to compute the bending stress developing
PTI JOURNAL | October 2018 17
TECHNICAL SESSION PAPERS

in a curved rope. However, different formulas often lead to duced formulas is summed up in the following table. It can
completely different stress values, which are in many cases very be seen that, regardless of the approach used, the bending
high with respect to what is observed on actual stay cables. stress is often not negligible (Table 1).
The maximum bending stress in a bent rope can be The accurate estimate of the maximum bending stress
computed assuming the same behavior of a beam to which in a strand is a complex topic out of the scope of this paper.
a curvature is imposed. Let E, d, and D be the elastic It is worth noting that bending stresses may dramatically
modulus of the material, the diameter of the rope, and affect the behavior of a bent stay cable and, therefore, must
the diameter of the curve, respectively. Thus, the so-called always be taken into account.
Reuleaux’s formula is obtained Secondly, interaction between strand and pipe may
produce fretting fatigue phenomena, leading to significant
d damage of the strand and even breakage. Proper saddle
σb = E (1) detailing must be carried out to avoid any unintentional
D
contact, which may lead to sudden and brittle failure of the
strand.
Equation (1) provides the maximum bending stress The design of a multi-tube saddle must investigate all
inside a rope or a strand under the assumption of perfect of the main aspects characterizing its response, which can
interaction between wires composing the rope or the strand. be summed up as follows:
However, this behavior is only valid when stresses are small. 1. Mechanical performance:
For increasing bending stress, the external wires may slip a. Transfer of differential stay cable forces (by
and the bending stress reduces abruptly. friction in a multi-tube friction saddle);
In the specific case of spirally coiled wires, Timosh- b. Fatigue strength; and
enko’s formula provides the bending stress as a function of c. Static efficiency.
the strand lay angle α. 2. Functional performance:
a. Corrosion protection; and
d 2 cos α b. Replaceability.
σb = E (2) Friction influences the force transmission from the
D 2 + ν cos2 α
stay cable to the pylon and may remarkably affect the
saddle device’s structural efficiency. As a rule of thumb,
In literature, plenty of formulas have been proposed to the higher the friction, the greater the horizontal load that
correct the first one, taking into account the actual working the saddle can transfer to the pylon. Thus, the friction
condition of a rope. For instance, Bach’s formula, used for coefficient developing at the interface between strands and
ropes, introduces an effective elastic modulus E0 = 3/8E. pipes becomes an important design parameter required by
designers to control the maximum unbalanced load acting
3 d upon the deck.
σb = E (3) The differential force transferred by the cable shall
8 D
then go from the saddle to the pylon. To this purpose,
shear connectors (studs or plates) are usually used to
Carstarphen’s formula modifies the previous formula improve the stress transfer to concrete, which otherwise
using an effective elastic modulus equal to E0 = 0.44E. would just rely on bond.
The width of the saddle should be designed in such a
d way that pressure transmitted to the underlying concrete in
σ b = 0.44 E (4) the most severe load combination would be small enough
D
to avoid any concrete failure.

Taking into consideration a standard seven-wire, CODE REQUIREMENTS

15.7 mm nominal diameter strand with 5.2 mm nominal A base assumption of all main international
diameter external wires helicoidally coiled around the recommendations is that a stay cable system with a saddle
central king wire and a saddle with a 3000 mm radius, the shall exhibit the same performance level of a standard stay
maximum bending stress for each of the previously intro- cable system, anchored at pylon level with anchorages.

18 October 2018 | PTI JOURNAL

 TECHNICAL SESSION PAPERS
Table 1: In addition, current recommendations do not explain
Bending stress σb whether or how fatigue may affect the friction coefficient,
MPa %GUTS even though a slight reduction in friction, due to progressive
Reuleaux 510 27 system use, seems quite reasonable.
Timoshenko 507 27 Moreover, further and detailed information should
Bach 191 10
also be provided on testing conditions, saddle deviation
angles to be tested, and load levels. However, a need for
Carstarphen 225 12
recommendations improvement on the aforementioned
topic is well known. More complete requirements are
This refers to both the mechanical performance and the now under discussion and will be introduced in the next
functional ones—that is, the saddle system has to ensure versions of the main international recommendations on
corrosion protection and replaceability of any individual cable-stayed bridges.
strand of the bundle. Both PTI and fib recommendations require carrying
The analysis of the structural behavior of a specific saddle out a tensile fatigue test, which aims to validate the stay
system is a very tough task and involves many independent cable behavior in service conditions. The test is carried out
aspects. For this reason, it cannot be approached from the over a specimen having at least 2 m free length, installed
theoretical point of view only. Development and validation with a 30-degree inclination against horizontal and
of stay cable saddles is usually carried out in the so-called subjected to a 10 mrad angular deviation at saddle exits.
design assisted by testing. Actually, testing plays a crucial The saddle must be moved vertically for 2 million cycles
role in the evaluation of the structural and functional between a lower and an upper configuration corresponding
response of these devices. to a stress in the stay cable equal to 45%GUTS-Δσ and
Design, testing, and installation of saddles are generally 45%GUTS, respectively. The general test procedure
performed according to either PTI DC45.1-12 (2012) or presented by the two standards is precisely the same,
fib Bulletin 30 (2005). These are the two main international but fib Bulletin 30 is a little more stringent. As a matter
recommendations addressing stay cable systems. of fact, fib recommendation introduces a stress range for
fib Bulletin 30 does not have any specific requirement parallel strand stay cable of 200 MPa, greater than the one
about friction, while PTI DC45.1-12 clearly requires that reported by PTI DC45.1-12, equal to 110 MPa plus 35
the friction coefficient is assessed through testing. However, MPa through vertical movement of the saddle. Acceptance
testing details and conditions are not specified. criteria are the same for both standards (2% maximum wire
According to EN 1993-1-11, Section 6.3.2, the friction fractures) while, for small-sized cables, PTI DC45.1-12
phenomenon in the saddle, assessed through the belt fric- accepts a slightly greater number of wire fractures than
tion equation (Capstan equation or Eytelwein’s formula), fib Bulletin 30(three instead of two). In addition, both
should prevent slippages. To this purpose, the following standards require that no cracking or fracture develop in
condition should be met the anchorages material.
For these reasons, a stay cable saddle fatigue test
µθ carried out achieving all fib requirements also satisfies the
F  γ (5) PTI requirements.
max  Ed1  ≤ e M , fr
 FEd 2  At the end of the fatigue test, the static efficiency of the
stay cable system, linked up with its behavior at the ultimate
where FEd1 and FEd2 are the design values of the maximum limit state, must be tested. fib Bulletin 30 requires the
and minimum force, respectively, on either side of the performance of a tensile static test in which the stay cable
cable; µ is the coefficient of friction; θ is the angle (in is symmetrically loaded to reach the maximum attainable
radians) of the cable passing over the saddle (deviation stress. The stay cable shall develop a minimum tensile force
angle); and γM,fr = 1.65 is the partial safety factor for fric- equal to 95% GUTS or 92% AUTS, whichever is greater.
tion (resistance factor).
It is reasonable to define the friction coefficient as a TESTING CAMPAIGN
mean value, averaged over all the tubes for any specific The TENSA multi-tube saddle system TSS-T is a
saddle device. However, the recommendations do not steel welded box member, provided with several parallel
provide any information on this topic. steel pipes embedded into a high-strength cement mortar
PTI JOURNAL | October 2018 19
TECHNICAL SESSION PAPERS

(Fig. 1). The inner surface of each tube is covered with a and the loading condition. Taking into account all of
special high-friction compound. the actual inclinations that a stay cable can have on a
To assess the behavior of the TSS-T saddle system, bridge, the friction coefficient was measured on saddles
a wide testing campaign was developed over a period with deviation angles ranging between 30 and 150 degrees
of almost 3 years. Most of the tests were carried out at (Fig. 2 and 3).
the Structural Engineering Laboratory of Politecnico Stay cables were subjected to different initial loads
di Milano, following recommendations described smaller than 50%GUTS, which is the maximum design
previously and introducing additional improvements in working load of the stay cable. To properly evaluate
testing protocols. the effect of the actual force in the cable, friction was
Friction tests were performed over several TSS-T measured on the same saddle for several load levels (20%,
saddle systems, varying both the geometrical configuration 25%, 30%, and 35%GUTS).
As expected, tests proved that neither the saddle
deviation angle nor the stay cable initial tension significantly
affect friction and provided friction coefficients ranging
between 0.35 and 0.42, with an average value equal of 0.40.
To further analyze the saddle capacity, a full-scale
tensile fatigue and static test was performed over a seven-
strand saddle system (Fig. 4 and 5).
To this purpose, a 10-ton steel frame was designed,
manufactured, and assembled in the laboratory to resist
all of the loads expected for the test, with the stay cable
Fig. 1—TENSA multi-tube saddle system TSS-T. configuration as close as possible to the one actually
adopted on site. Expensive and dedicated equipment was

Fig. 2—Setup for friction test over a multi-tube saddle with remark- Fig. 3—Setup for friction test over a multi-tube saddle with small
able deviation angle. deviation angle.

20 October 2018 | PTI JOURNAL

 TECHNICAL SESSION PAPERS

used to stress the stay cable and to impose the vertical

movement of the saddle. Moreover, all the components
generally used in the free length, transition length, and
anchorage zone of the stay cable—that is, deviator disks,
wax boxes, and so on—were included in the test.
The real working condition on site was simulated
embedding the saddle into a concrete block representing
the bridge pylon. In addition, after the stressing phase, the
saddle was filled with a protective material able to provide
adequate corrosion protection to the strands.
Tensile fatigue test was carried out at 1.20 Hz mean
frequency and lasted approximately 20 days. No wire
fracture or damage in the saddle and in any component
of anchor heads occurred during testing, as confirmed by
a system of accelerometers that the laboratory installed
on the sample to continuously monitor the cable state
and, subsequently, by a visual examination carried out
during dismantling. The testing fully met acceptance
criteria introduced by fib Bulletin 30 and PTI DC45.1-12,
and the system exhibited an excellent performance with
respect to fatigue loading.
During the subsequent static tensile test, the
seven-strand stay cable was loaded up to the maximum
attainable force and a load corresponding to 97.5% of
GUTS and 96.2% of AUTS was recorded. The load
level reached totally satisfied the acceptance criteria
of fib Bulletin 30. A significant stay cable elongation,
equal to approximately 1.90%, was reached in the
final test stages. Because it corresponded to the testing
rig hydraulic jack’s end of stroke, the load level was
not further increased and, therefore, the test
was stopped. Fig. 4—Setup for tensile fatigue and static test over a seven-strand
At the end of the static test, no evidence of damage multi-saddle (front view).
of the saddle system and anchorages was detected. More-
over, after almost 1 month of direct exposure to atmo-
sphere, no corrosion was noticed over any component of
the tested system.
Other friction tests were performed at the end of
the static test to estimate the effect of both fatigue and
the strand’s large strain on the friction coefficient. Values
ranging from 0.25 to 0.39 were observed.
Such a decrease in friction is due to the remark-
able use of the system—that is, pipe’s surface and
high friction compound—produced by the fatigue test
and the subsequent static test. Thus, the measured
friction coefficients should be considered as minimum
values for ultimate limit state—that is, lower bound
of the actual friction coefficient that the saddle is able Fig. 5—Setup for tensile fatigue and static test over a seven-strand
to provide. multi-tube saddle (side view).

PTI JOURNAL | October 2018 21

TECHNICAL SESSION PAPERS

The high performance of the system detected during

the tensile fatigue and static tests over the seven-strand
saddle system proved that saddle devices are perfectly able
to carry fatigue loading—corresponding to traffic loads on
the bridge—and, moreover, can be designed not to affect
the stay cable performance in terms of static efficiency.
However, the tested saddle was fairly small compared to
what is often required and supplied on site.
The increase in dimensions may dramatically affect the
actual structural behavior of a component due to an inevi-
table size effect. Additionally, a wider size means higher
load levels as well and, in many cases, important deviation
of the bundle.
For all these reasons, another tensile and static test
Fig. 6—Setup for tensile fatigue and static test over a 37-strand was performed on a 37-strand saddle system, again in
multi-tube saddle. the Structural Engineering Laboratory of Politecnico
di Milano (Fig. 6). A dedicated 25-ton steel structure
was designed, produced, and assembled. The impressive
load level required by testing according to fib Bulletin
30 required special dedicated fatigue and tensile equip-
ment with high capacities and an accurate control
system (Fig. 7). As previously mentioned, the necessity
to build a testing setup as close as possible to the actual
site configuration in a laboratory requires imposing
important deviations to the bundle. Special attention
must be paid to this point because it can dramatically
affect the system performance.
A tensile fatigue test was carried out at 1.55 Hz
average frequency and lasted approximately 15 days.
The system of accelerometer installed to monitor the
cable during fatigue cycles recorded several events
during the fatigue test. However, as confirmed during
dismantling, only two wires broke due to fatigue, while
others were stabilizing movements from the massive
steel structure. Compared to the total amount of 259
wires, the loss of stay cable cross section due to fatigue
was approximately 0.77%, well below the 2% limit
required by fib Bulletin 30. This first important result
confirmed the good behavior of the stay cable system
deviated by a saddle device.
During the subsequent static tensile test, the 37-strand
stay cable was again loaded up to the maximum attainable
force and a load corresponding to 95.9% GUTS and 93.2%
AUTS was recorded. The load level reached totally satis-
fied the acceptance criteria of fib Bulletin 30. Due to the
notable force imposed to the cable, the test was stopped
for safety reasons to avoid any further dangerous breaks.
Fig. 7—Special dynamic jacks for fatigue testing and saddle block A cable elongation equal to approximately 1.53% was
positioning. reached at the end of the test.

22 October 2018 | PTI JOURNAL

 TECHNICAL SESSION PAPERS

During dismantling, two strand failures were Tommaso Ciccone is TENSA Technical Director for
detected. The first one was located at saddle exit, post-tensioning and stay cable systems. He has 17 years
where the resisting cross section of the strand was of experience in the field and he is an active member of
reduced due to fatigue wire break. The other failure PTI Committees DC-45, Cable-Stayed Bridge; M-50,
was recorded in the deviation length of the stay cable, Multistrand Tendon; and CRT-70, PT System Qualifica-
close to the deviation system. No evidence of damage tion Testing and Certification Committee. He is also an
of the saddle system and anchorages was detected— active member of fib TG5.5, cable-supported structures.
just some negligible steel powder was observed He received his MS in civil structural engineering from the
on several strands due to an incipient fretting Politecnico of Milan (Italy) and an Executive Master in
fatigue phenomenon. Business Administration (MBA) from MIP – Politecnico
The testing campaign provided a widespread and of Milan (Italy).
complete insight into the structural and functional Andrea Castiglioni di Caronno is a Project Engineer in the
behavior of multi-tube saddles, accounting for several Technical Department of TENSA, where he has worked
working conditions. Particularly, the TSS-T saddle on stay cable system and post-tensioning system develop-
system matched all the existing recommendations ment, design, and testing, with a major focus on multi-tube
requirements—that is, providing a structural perfor- saddle system analysis and validation. He received his BS
mance equivalent to what is expected for a traditional and MS in civil structural engineering from Politecnico of
stay-cable system, and gave up satisfactory responses for Milan (Italy) in 2015. He is also a member of fib.
the conditions not fully described and detailed in the
current recommendations.

CONCLUSIONS
A wide and expensive testing campaign was carried
out during the last 3 years to develop performing multi-
tube saddle systems for stay cables, able to fulfill all Share Valuable
the requirements introduced by the main International
Recommendations. Tests provided a lot of important
Industry Knowledge
and satisfactory results, which gave a complete and
clear view of the structural behavior of these devices. Gain recognition and share your case studies,
The TENSA multi-tube saddle system proved to be research, innovations, and expertise with the
adequate to withstand all of the severe loading conditions post-tensioning industry. Submit a session
required by testing, providing at the same time enough presentation for upcoming PTI Conventions.
friction to avoid any strand slippage and, also, ensuring a
suitable level of strand corrosion protection as required E-mail miroslav.vejvoda@post-tensioning.org
during testing by fib Bulletin 30 and PTI DC45.1-12. The for more information.
good results obtained in all the performed tests provided
good confidence in the use of multi-tube saddle systems
within cable-stayed bridges.

References
1. PTI DC45.1-12, “Recommendations for Stay
Cable Design, Testing, and Installation,” Post-Tensioning
Institute, Farmington Hills, MI, 2012.
2. fib Bulletin 30, “Acceptance of stay cable systems
using prestressing steels,” International Federation for
Structural Concrete (fib), 2005.
3. EN 1993-1-11, “Design of steel structures; Part
1-11: Design of structures with tension components,”
CEN, 2006.
PTI JOURNAL | October 2018 23