Você está na página 1de 7

ABSTRACT: The argument is made that when retrofitting a suspension bridge, the greatest value is obtained

when the engineer considers the fundamentals of engineering, rather than simply following the code. By
'Iundamentals two aspects are considered. One is a thorough understanding of not only structural behavior,
but also how loads, resistances, and safety factors are derived. The other aspect covers the fundamentals of
good design (as opposed to good analysis), in preparing designs that can be constructed during short
occupancies and are tolerant of problems that might arise during the short construction period between
deadlines. The design aspect includes looking for opportunities to reduce maintenance and improve inspection
access, and produce such other benefits as may be desired.
1 INTRODUCTION
This paper reIlects the author`s opinions on what
should be the fundamental considerations for the
refurbishment or capacity increase of a suspension
bridge. The reason for going to the fundamentals is
that very often code requirements are inadequate to
cover a particular situation, so it is important to
understand how code provisions are derived, and
therefore how they might be modified in a controlled
way if modification is appropriate.
There are also fundamental considerations that
the engineer should have in mind while he or she is
designing the retrofit of a suspension bridge. These
include:
What am I trying to accomplish?
How can disruption to traffic be minimized?
What will be easiest for the constructor to build?
What other benefits can be achieved, such as
reduced and easier maintenance, or better access,
or improved seismic resistance?
2 KEY DRIVERS
The key considerations driving the retrofit of a
bridge are, as usual, benefits and costs. The purpose
of this paper is to see how these are best managed
for a suspension bridge.
2.1 Benefits
The main benefits are:
1 Improved traffic flow (more lanes, for example)
which can also lead to increased tolls;
2 Increased safety, both in terms of how the traffic
moves (perhaps wider lanes) and the ability to
take increased loads. Sometimes the loads have
increased, sometimes the bridge was designed for
a loading much less than specified today, often it
has simply deteriorated.
3 Improved access for inspection;
4 Easier and less costly maintenance; and
5 Other benefits that may be desired, such as
improved seismic performance.
2.2 Costs
The four main cost items are:
1 The physical cost of upgrading;
2 The (often hidden) cost to the public caused by
traffic disruption. To this can be added lost
revenue to the operator where tolls are charged.
3 Over-strengthening by being conservative.
Designing a new bridge for an increase in live
load usually adds only a small percentage to the
cost of the structure, but increasing the capacity
of an existing bridge can be extremely costly.
This is why unnecessary conservatism can be
horrendously expensive. The larger the bridge,
the more this is so.
4 Under-strengthening can be the most costly of all,
as it can lead to failure of the bridge and even
loss of life.
Fundamentals of suspension bridge retrofit
P. G. Buckland
Buckland & Taylor Ltd., North Vancouver, BC, Canada

Bridge Maintenance, Safety, Management and Life-Cycle Optimization Frangopol, Sause & Kusko (eds)
2010 Taylor & Francis Group, London, ISBN 978-0-415-87786-2
7
It can be seen that Costs 1 and 2 are related. The
cheapest retrofit will probably coincide with the
maximum traffic disruption. Since ultimately the
user pays the price of retrofit, and also suffers the
loss caused by disruption, an economist could in
theory find a way to balance the two. In the author`s
experience, this is not usually attempted, and it may
be that good judgment and intuition will provide
close to the optimum solution. By the time a bridge
is being retrofitted the traffic volume has usually
built up to the capacity level of the bridge. Therefore
only traffic disruption that occurs at off-peak times
will normally be tolerated.
The best balance between Cost Items 1 and 2 can
be found by careful and creative design, and this
theme will be examined later in the paper.
It can also be seen that Cost Items 3 and 4 are
related, in the sense that either cost will be over the
optimum unless the narrow band is found between
too safe and not safe enough. In other words: how
safe is safe enough? That is the starting point for this
presentation.
3 SAFETY
One can think of safety in several ways. At its most
basic, safety is the margin between the capacity of
the bridge to carry its loads, and the loads imposed
upon it. We therefore need to know three things:
1 the capacity of the bridge;
2 the loading to be applied; and
3 the required degree of separation between the two
(i.e. the safety factor).
It is true that the bridge design codes provide
guidance on these matters, but if there is uncertainty,
design codes err on the side of conservatism, or
should do. And as mentioned earlier, conservatism
may be a very expensive luxury, which can make it
poor engineering. The above three aspects of safety
will be considered in turn.
3.1 The Capacity of the Bridge
Modern analysis software and hardware should be
able to provide a good idea of the capacity of a
bridge, but experience has shown that this is not
always the case.
Inspection of the bridge can detect any weak
points, such as corroded steel, broken rivets and the
like, but even the best inspection does not provide
any information about the stress state of the bridge.
The shape of a suspension bridge is quite
sensitive to any change in its condition. This
characteristic is useful for what we call the
'Analysis-and-Survey method oI determining the
stress condition of the bridge. In this method, a
computer model is created of the bridge as it was
originally built, preferably from as-built drawings or
from design drawings if as-builts are not available.
The model is then altered to reflect any changes that
have occurred since the bridge was constructed.
These changes would include any additional dead
load, structural changes, foundation settlement, and
so on. Once these changes have been made, the
geometry of the bridge will also have changed.
The bridge is then surveyed, preferably in the
early morning before sunrise, with overcast sky, no
wind, and no traffic on the bridge. The survey will
include the profile of the deck from end to end, and
the verticality in both directions of the towers. Once
the computer model has been adjusted to account for
the temperature at which the bridge has been
surveyed, the shape of the bridge in the model
should be the same as that in the computer.
In our company`s examination oI more than 20
suspension bridges, the two have never fully agreed!
This can only be because something has happened to
the bridge that was not modeled because it was not
known.
For example, when this technique was first
applied to the Lions` Gate Bridge in Vancouver, BC,
it was found that the centre of the bridge was about
one meter lower than expected, and the towers were
leaning in towards each other. One of the towers had
settled a few centimeters, and some dead load had
been added to the bridge, but these effects only
accounted for about half of the observed
discrepancies. The only explanation of the observed
discrepancies was that the main cables, which
comprised spiral strands, had stretched over time.
Once this was accepted as the only rational
explanation, these effects could be added to the
others in the computer, and the true stress state was
determined.
In this instance, a cable-bent (side tower) was
found to be seriously overstressed, and remedial
action was taken immediately. This deficiency could
not have been discovered by inspection, but it was
seriously reducing the capacity of the bridge.
3.2 Applied Loading
Dead load can be determined from a weight take-off,
and this may be modified by the findings of the
analysis-and-survey.
The main unknown is the live load imposed by
traffic. Various studies have provided some
guidance on the subject, such as Buckland et al 1980
which provided the basis for the long span loading
in both AASHTO-LRFD and the Canadian Highway
Bridge Design Code CAN/CSA-S6-06.
An extreme case of heavy loads is shown in
Figure 1.
8

Figure 1. Traffic loading on the Blue Water International
Bridge.

Recent research by the Author`s company has re-
examined the multi-lane reduction factors, which are
probably the least known part of the live loading for
long spans. In one recent study it was confirmed that
the multi-lane reduction factors vary with loaded
length, being generally greater for long lengths than
for short lengths, as shown in Table 1. This could
add considerable complexity when trying to apply
the design loading.
Table 1 shows that at long loaded lengths of
about 500 m (1,600 ft.) the reduction factor for 8
lanes loaded is nearly double the measured value for
8 lanes of short loaded lengths up to about 20 m
(70 ft.).

Table 1: Multi-lane reduction factors, recent study
Loaded
length, m
Number of loaded lanes
1 2 3 4 8
6 1.00 0.70 0.60 0.50 0.35
16 1.00 0.65 0.60 0.50 0.35
480 1.00 0.90 0.85 0.65 0.60

For comparison, Table 2 gives the reduction
factors from AASHTO. (Interestingly, AASHTO-
LRFD uses different reductions even though the
traffic is identical.) It can be seen that the AASHTO
factors are conservative, as they should be. But it
can also be seen that using the AASHTO factors for
short loaded lengths is so conservative that it could
result in a large amount of very expensive
strengthening that is in fact not needed.

Table 2: Multi-lane reduction factors, AASHTO
AASHTO
Number of loaded lanes
1 2 3 4 8
All lengths 1.00 1.00 0.90 0.75 0.75

The parts of a suspension bridge governed by
long loaded lengths are the cables, towers and
anchorages. These will be at a maximum when all
lanes are loaded, so in the example given in Table 1
the multi-lane loading would be taken as 0.6 for
these major components.
For some short loaded lengths, however, such as
govern transverse floor beams or trusses spanning
between cables, a multi-lane load factor of 0.6
would produce a design loading of nearly double the
observed load, and a value of 0.35 would be more
appropriate.
This does increase the complexity of the analysis,
but if unnecessarily doubling the design load results
in strengthening of all the transverse floor beams or
trusses, the cost would be enormous. Compounding
the problem, the extra strengthening may increase
the weight of the suspended structure to the point
that the cables, towers and/or anchorages may need
upgrading.
This example makes a very compelling case for
going back to the fundamentals!
4 INSTRUMENTATION
In a recent case, there was concern that the live load
of the A. Murray MacKay Bridge in Halifax, NS,
(Fig. 2) was being over-estimated by the design
code.


Figure 2. A. Murray MacKay Bridge, Halifax, NS, Canada

The possibility of measuring traffic axle loads
directly and doing a study as in the previous
example was considered for the MacKay Bridge, but
it was decided instead to instrument directly the
members of most concern, which were the diagonals
of the stiffening trusses as shown in Figure 3.
Recording of data started at the end of December
2009, and will continue for about a year. This should
indicate the combined effects of both single-lane
loading and multi-lane presence, without identifying
them separately.
9

Figure 3. Stiffening truss, MacKay Bridge
5 SAFETY FACTORS
Having looked at the capacity of the bridge and
determined the loads, the question arises as to what
load (safety) factors to use. LRFD design has a huge
advantage for suspension bridge analysis. The
principle of LRFD (known as Limit States Design
outside the US) is that safety factors are applied
more rationally than they are with Working Stress
Design. This is because a safety factor is really an
ignorance factor: it covers what we do not know
about a load or resistance. It therefore follows that if
we can lessen our ignorance, we can lessen the
safety factor that must be applied. Thus, loads that
we know well, such as dead load, command a small
load factor, and those loads that we know much less
about, such as live loads or wind loads, are given
larger load factors.
This concept is particularly useful for suspension
bridges, which have different components governed
by different types of loads. For example, the main
cables are dominated by dead load, and only 10 to
30% of the total load that they carry is live load,
whereas the stiffening truss or box girder is
essentially unstressed under dead load only, so
almost 100% of the applied load is live load.
Towers are a mixture: the axial load in the tower
legs, coming from the cables, is almost all dead load,
but the displacement of the tower-top is caused by
live load. Thus if one thinks of the P- effect, P, the
axial load, is governed by dead load, and , the
displacement, is governed by live load.
5.1 Loads and Resistances
Let us examine the concept of safety in slightly
more depth, particularly as it relates to load factors.
Think first about the resistance or strength of a
piece of steel or concrete. Figure 4 shows a
distribution of measured values for the yield of
ASTM-A36 steel, which has a nominal yield of 248
MPa. Clearly the actual yield strength of the
material is not what is specified. It is usually greater,
but it could, on rare occasions, be less.

Figure 4. Distribution of measured values for yield values of
A36 steel with a nominal yield of 248 MPa (36 ksi).

Figure 5 shows a more stylized distribution of
nominal strengths of steel or concrete.

Figure 5. Stylized distribution of yield strengths.

Now consider the applied live load.
The bridge design code gives us a design live
load, such as an HS vehicle or a lane load. The
design load is usually not representative of any real
vehicle, but it is intended to represent the effects of
real vehicles. But how accurately do we really know
the effects of real vehicles? And even if we do know
them well, how confident are we that they will not
change in the future? The answer is that usually the
design load gives a greater effect than real vehicles,
but sometimes overloads occur, so conceptually we
have a distribution of load effects as shown in
Figure 6.

Figure 6. Distribution of load effects.

10
Now the two distributions of load and resistance
can be combined, as shown in Figure 7.

Figure 7. Distributions of load and resistance combined.

The distance between the nominal load, L, and
the nominal resistance, R, in Figure 7 is the safety
factor, being the combined effects of load and
resistance factors.
As long as the resistance is always greater than
the load effect, all will be well, but it can be seen
that the two curves overlap, which means that it is
possible for the load effect to exceed the resistance,
which leads to failure. It can also be seen that if the
two curves are further separated, the area of overlap
is reduced and the probability of failure is reduced;
but it can not be reduced to zero if there are no
known limits to the curves, and a quantitative value
of the probability of failure is not easy to derive. If,
however, we produce a distribution of R-L
(resistance minus load), the curve of Figure 8
results.


Figure 8. Distribution of (resistance minus load).

In this case the measure of safety can be seen
more clearly. The probability of failure is the area to
the left of zero divided by the total area under the
curve. A measure of safety (as opposed to
concentrating on the probability of failure) is often
represented by , the safety index. is the number
of standard deviations that the mean of the
distribution is from zero. The greater is, the greater
the safety and the lower the probability of failure.
From this approach, once the target probability of
failure has been set (or the factor established) the
required separation between the nominal (design)
load and the nominal resistance can be defined. The
separation is the combination of the load factor and
the resistance factor, in general: the safety factor.
5.2 Variable safety
The issue of safety has been discussed in some detail
so that we can introduce the concept of varying the
load factors for different members.
Risk (perhaps the opposite of safety) can be
defined as:
Risk = (Probability of an event occurring) x
(Consequences of such an event)
Therefore to have consistent risk (or safety), the
greater the consequences of an event, the lower the
probability we should accept of the event occurring.
Conversely, if the consequences are reduced, the
acceptable probability can be greater.
Consider the tower of a suspension bridge. If it
fails, the entire bridge collapses. Then consider a
steel stringer that supports a steel or concrete deck.
It will probably fail in bending, and the load will be
shed to adjacent stringers, with the result that there
is a temporary dip in the roadway until it is fixed.
The two consequences are completely different, and
it is clearly desirable to have less probability of
tower failure than of stringer failure. In other words,
the safety (load factors) for the two members should
not be the same.
Properly considering these differing requirements
can result in large cost savings during a retrofit. The
subject is well covered in Section 14 of the
Canadian Highway Bridge Design Code (2006) and
its Commentary.
6 PRACTICAL CONSIDERATIONS
The discussion so far has been rather theoretical,
although I maintain that a thorough understanding of
safety is fundamental to safe and economical retrofit
of any major bridge.
There are also, however, a number of practical
matters to be considered.
6.1 Bridge occupancy and traffic disruption
There have been a few cases of suspension bridges
being closed to traffic for months on end while
renovations are made. These are rare, however, and
it is much more likely that construction operations
will be limited to only one or two lanes at a time, or
only at night or weekends, or some combination of
these. The consequences of having short and/or
restricted occupancies for construction are:
1 Any retrofit must be designed to be constructed
during the short occupancies permitted, and shall
be safe for the public between construction
periods;
2 The design must be tolerant of problems that may
be found during the construction occupancy.
Generally these will be unexpected, or they
would have been planned for, and they can
11
include finding that the structure is not as was
expected, breakdown of the contractor`s
equipment, errors in design or construction, and
non-delivery to site of components or
consumables, such as weld rod.
3 The designer`s responsibility is to provide a
design that is as tolerant of unexpected problems
as possible. A simple example is providing shims
in a connection so that if the existing steel is
misaligned, the new steel can be leveled to the
correct elevations, as shown in Figure 9. Here,
the new transverse floor beam (framing in from
the left) was to sit on the old longitudinal girder.
Instead of trying to make the connection directly,
a gap was provided, to be filled by steel shims.
Thus if the elevations or alignments were not as
expected, the shims could be adjusted
accordingly.
Figure 9.Shims between new floor beam and existing girder.

4 The contractor`s responsibility is to provide a
higher level of management than for a typical
project. If work is only to proceed at night, the
contractor has two deadlines per day one to be
ready to start work at a certain time, and the other
to have the bridge ready for traffic on time. This
requires considerably more attention to
scheduling than the conventional project that
must simply meet the one deadline of substantial
completion.
5 An aspect that is too often overlooked is that
most bridge failures occur during construction
by a factor of about a hundred compared to
finished bridges. In the case of bridge retrofit, the
public is using the bridge during the
reconstruction process, which means that the
public is placed at a much greater risk during
reconstruction than it would normally be. The
answer to this is not to increase the safety factors;
it is to increase the quality and quantity of
independent checking of the proposed
construction activities, to a higher level than
normal.
6.2 Maintenance and access
Although the primary purpose of retrofit is usually
to restore or increase live load capacity, the
opportunity should be taken, wherever possible, of
improving access for inspection, and reducing the
cost of maintenance. For example, when the Lions`
Gate Bridge was widened by 47% (Figs. 10 and 11),
the amount of steel surface to be painted, instead of
increasing, was actually reduced by half.

2900
12.2m
1300 2900 2900 1300
STIFFENING
TRUSS
STRINGERS FLOOR BEAM
SIDEWALK
61 GALV
BRIDGE
STRANDS 1
7
16"
(37mm)
13" O.D.
330 mm
WRAPPING
-#9 WIRE
(DOUBLE
GALVANIZED)
YELLOW
CEDAR
FILLERS
TRAVELER
RAIL
75 T-GRID
DECK

Figure 10. Original cross-section, Lions` Gate Bridge

16 762
12 192
STIFFENING TRUSS BRACING
ORNAMENTAL
LIGHTS
HAND
STRANDS
2000
2700 3 LANES @ 3556 = 10668
2000
2700
STEEL
BARRIER
SIDEWALK
CABLE
TRAY
35 EPOXY
ASPHALT

Figure 11. Renovated cross-section, Lions` Gate Bridge

When the Port Mann Bridge (Fig. 12) near
Vancouver, BC, was widened, the opportunity was
taken to add inspection walkways to the outer plate
girders of the approach spans as shown in Figure13.
These have been invaluable in easing inspection of
this ageing bridge prior to its replacement in the next
few years.
12

Figure 12. Port Mann Bridge, BC, Canada


Figure 13. Partial cross-section of Port Mann Bridge approach
spans, original material shown grey, new widening and
inspection walkways in black.
6.3 Other benefits
In addition to improved maintenance, other benefits
may sometimes be achieved with little extra cost,
such as enhanced seismic resistance.
6.4 The devil in the details
Details are typically connections, and field splices
have to be capable of fast connection because of the
limited time available, and they must be tolerant of
misalignment and other potential problems.
Unfortunately, details are also where problems
can be generated. Too often, details are designed in
such a manner that secondary strains are induced,
and fatigue cracking follows a few years later. It is
fundamentally important to consider not just forces
in a connection, but also the structural behavior.
This is particularly true for orthotropic decks, which
are becoming increasingly, and in my view,
unnecessarily expensive, without a commensurate
increase in performance.
REFERENCES
Buckland, P.G., Navin, F.P.D., Zidek, J.V. and McBryde, J.P.,
1980, Proposed vehicle loading of long span bridges,
Journal of the Structural Division, American Society of
Civil Engineers, April, p 915.
13

Você também pode gostar