Professional Documents
Culture Documents
Newmarket Viaduct Deconstruction
Newmarket Viaduct Deconstruction
Newmarket Viaduct Deconstruction
Paper
Newmarket Viaduct deconstruction
Jon Varndell, BEng (Hons), CEng, MICE Introduction
Design Manager – NGA Newmarket Alliance, URS – Principal Engineer The Newmarket Viaduct was constructed in the 1960s and forms
a vital link in Auckland’s motorway network. The dual three lane
Peter Lipscombe, NZCE, BE (Hons), PhD, MIPENZ, CPEng viaduct now carries over 160 000 vehicles per day and is part of
Structure Design Manager – NGA Newmarket Alliance, URS – Principal Engineer the busiest section of road in New Zealand. The viaduct is 690m
long and comprises 16 spans up to 23m high passing over a busy
Ian Billings, BE (Hons), FIPENZ, CPEng
urban retail area and the North Auckland Railway line (Fig 1).
Senior Bridge Advisor – NGA Newmarket Alliance, Beca – Technical Director
The viaduct consists of two post-tensioned, cast in-situ box
Keywords: Newmarket Viaduct, Auckland, New Zealand, Viaducts, Replacing, Motorways,Temporary works,
girders transversely connected at each pier. Constructed by the
Segments, Strengthening, Risk management
balanced cantilever method, it was one of the earliest bridges built
Received: 03/11: Modified: 04/11; Accepted: 05/11
using this method in New Zealand. Deficiencies in the design,
© Jon Varndell, Peter Lipscombe & Ian Billings related to differential temperature effects, were identified during
construction when large (up to 4mm wide) cracks were observed
in the soffit slab. Mitigation measures were put in place and
Synopsis subsequent research carried out on the viaduct was also used for
The Newmarket Viaduct was completed in 1966 and has played a the development of the differential temperature design rules for
vital role in the Auckland motorway network for the last 45 years. bridges.
The viaduct carries more than 160 000 vehicles per day over The combined effects of substantial traffic volume growth and
several key arterial roads and the North Auckland railway line. the vital role the viaduct plays in the security of the motorway
The structure comprises two parallel, balanced cantilever bridges network, led to the decision for the viaduct to be replaced. The
connected at each pier with a post-tensioned portal frame. new viaduct will provide additional lanes and improved seismic
Increased traffic volumes and the limitations of the live load, resistance, edge protection and vehicle load capacity. The key
differential temperature and seismic performance of the structure challenge was that the old viaduct had to be deconstructed and
created the need to replace the old six lane viaduct. This paper replaced with no impact on the peak motorway traffic capacity
describes how the viaduct is being replaced in stages while throughout construction.
keeping the motorway operational throughout. Implemented under The essential link between the design and construction/
an alliance framework, the benefits of the integrated client, design deconstruction methodology, coupled with the uncertainties about
and construction team are discussed. The complex and the old viaduct, and the management of significant risks, resulted
unprecedented challenges of deconstructing this post tensioned in the project being implemented as an alliance; a virtual
viaduct in stages whilst it remains in service are described in detail. organisation comprising the owner (the NZ Transport Agency)
three contractors and four design firms.
This paper describes the project’s unprecedented challenges
and methods used to deconstruct the old viaduct while keeping
the motorway fully operational, and summarises the key mitigation
measures put in place to manage the risks.
1 2 3 4
2 Key construction stages
Stage 1 – Construct the new southbound viaduct, parallel to, and slightly overlapping with, the old southbound viaduct
Stage 2 – Commission the new southbound viaduct and then deconstruct the old southbound structure
Stage 3 – Construct the new northbound viaduct in the gap left by the deconstruction
Stage 4 – Commission the new northbound viaduct and then deconstruct the old northbound structure
4 Old viaduct cross section 5 Old viaduct box girder during construction (circa 1965)
6 New southbound viaduct next to the old viaduct, showing transverse 7 Longitudinal props and shield beam
stabilisation
the cantilevers using a 140m long, 810t, overhead gantry that was – traffic live load.
also used for the new viaduct construction. The deconstruction
progressed back to the pier table, at which point the pier table and The Ultimate Limit State load combinations considered were
pier column were demolished either by hydraulic crushers attached consistent with those derived from the Bridge Manual2 and
to a hydraulic excavator on the ground or cut into small sections AASHTO.
and removed by crane. The viaduct was thus dismantled in the In general, checks on section capacity were undertaken at the
reverse sequence to which it had been built, however, a key Ultimate Limit State only. Serviceability Limit State capacity checks
difference was that the prestressing tendons could not be were undertaken in a few cases but only to determine suitable
progressively destressed because they had been grouted at the limits for alert/alarm values used in monitoring the works. When
time of the original construction. The inability to destress the analysis showed that elements of the viaduct were overstressed,
tendons resulted in a stress distribution within the partially the deconstruction methodology was revised to try and reduce the
deconstructed works quite different from that experienced during loads. If this was not possible, then strengthening was designed.
the original construction. Analysis and control of these built-up Capacity checks were made on the assumption that the tendons
stresses is discussed in the next section. were either fully bonded (as intended in the original construction) or
unbonded (as might happen with poor grouting practice). In cases
Deconstruction analysis where tendon grouting was critical, the more adverse condition
A stage-by-stage deconstruction analysis was undertaken to was assumed to exist unless intrusive investigation was
calculate the expected forces in the viaduct at each stage of the undertaken to prove the quality of the grout.
deconstruction process. This included modelling the reactions
from the 810t overhead gantry used for the deconstruction Temporary works
process, in its various configurations. The gantry was supported Pier brackets
directly over the piers and therefore did not impart loads into the As previously noted, deconstruction required cutting the portal
superstructure. However, the effects of the gantry weight on the frames that supported both superstructures at every pier location
foundations and temporary works were significantly larger than any (see Fig 6). This required cutting the piercap beam and severing
previously applied loads to the structure, requiring strengthening of the post tensioning (comprising six tendons, each with 12 × 0.5in.
some of the foundations. The effects of wind and seismic loads strands) thus compromising the transverse stability of the structure
from the gantry were also very significant and had to be modelled and the support of the outer webs of the box girders.
for each stage of the deconstruction. While properly grouted tendons do re-anchor and therefore not
Analysis of the deconstruction sequence was performed using all of the strength is lost it was considered this should not be relied
three separate computer models and the results combined using a upon for the fundamental vertical support of the box girder webs –
suite of spreadsheets. The first of the models was a 2-D time- structural steel pier brackets were installed to support the load
dependant model that calculated the stress state in the viaduct (see Fig 6). In addition to supporting vertical loads, the pier
allowing for prestressing, creep, shrinkage and dead load effects. brackets also provided anchorage for the gantry holding down
This model was developed after an extensive assessment of the bars and connection points for the transverse props. The brackets,
original structure data (including interviewing members of the each weighing 7t, supported a load of up to 620t. They were
original construction teams). The second model was a 2-D beam clamped to the pier with sufficient force to resist the load by
element model used to calculate wind, seismic and gantry load friction. Friction was enhanced by roughening the surface of the
effects on the viaduct and temporary works. The third model was a concrete column, using weld beads on the steel contact plate, and
grillage model used to assess segment removal and live load 60MPa low-shrinkage grout between the two surfaces. Scaled
actions on the superstructure. trials were undertaken to verify the friction force that could be
The primary load cases considered in the deconstruction were resisted. The coefficient of friction used in design was 0.91, this is
basically the same as those that would be considered for the lower bound residual value after 2mm slip from a series of 16
construction of a new viaduct. tests.
The primary loadings included: At some highly loaded piers a reinforced concrete corbel was
– dead load (including the gantry weight); also added beneath the pier bracket.
– specialist construction equipment (e.g. diamond wire cutters);
– construction live load (typically 0.5kPa); Transverse stabilisation
– accidental release of a segment (derived from AASHTO1); Transverse stability of the cut portal frame was maintained by
– in-service wind loads (based on 15m/s wind speed); transverse props. These comprised twin 762mm OD × 6.4mm WT
– out-of-service wind loads (1000 year return period); tubes diagonally propping the northbound viaduct and a 457mm
– earthquake loads (500 year return period); OD × 12.7mm WT tube supporting the southbound. This
– temperature effects; configuration allowed the southbound viaduct to be deconstructed
8 Temporary steelwork required for pier stability during deconstruction Deconstruction methodology
The typical deconstruction sequence involved removal of the
midspan closure segments, followed by successive removal of the
cantilever segments.
to foundation level prior to deconstructing the northbound.
Midspan closure segment removal
Longitudinal stabilisation Removal of the closure segment at the mid-span location involved
Similar to the transverse direction, longitudinal props were required changing the structural system from a continuous structure, with
in the longitudinal direction to maintain stability when the mid-span its locked-in forces, to a statically determinate one. In the process
closure segments were removed (see Fig 7). The longitudinal props of making the first cut, bending moment and shear was released
comprised 610mm OD × 9.5mm WT steel tubes connected at the and redistributed into the newly created free cantilever. There were
upper end to the pier column, and at the lower end to a specially significant locked in forces from the original construction together
constructed reinforced concrete foundation. The foundations with prestress losses, creep and shrinkage that had occurred since
required four, 90t ground anchors to provide the necessary uplift that time. These locked-in forces had to be released in a controlled
resistance when the prop went into tension. manner to avoid a dynamic load release. This was achieved by
Protection of the longitudinal and transverse props from damage staged cutting and hydraulic jack unloading to decompress the
was a paramount consideration since loss of these had the locked-in forces. Figs 9.1 to 9.4 show the procedure in more
potential to destabilise the pier with consequential loss of the pier, detail. Fig 10 shows the closure segment removal system in use,
9.1 9.2
9.3 9.4
Fig 9.1 A 200mm diameter hole (300mm deep) was cored in the top of each web and a jack was installed in the hole. The jacks
were pressurised to a force equivalent to the estimated force in the top flange of the box girder. A steel beam was
prestressed to the deck above each of the three webs to control any shear movement between the cantilever tips
Fig 9.2 The bottom slab of the box girder was cut, followed by the top slab. The diamond wire saw used for cutting made an 11mm
wide cut; wide enough to release the locked-in displacements in subsequent stages. Each web was then cut from the
bottom of the box girder up to the continuity tendons and from just above the continuity tendons up to the underside of the
jacks. At this point the bending moment across the cut line was carried by a couple between the jacks and the continuity
tendons, and the shear was carried by the small remaining area of concrete surrounding the continuity tendons
Fig 9.3 The horizontal jacks in the top of the webs were slowly and uniformly unloaded to release the bending moment in the
section. The midspan deflected downwards as forces redistributed. This displacement was typically 25mm. The jacks under
the steel beams were preloaded to a force equivalent to the estimated shear across the cut line. Then the continuity
tendons were cut; outer webs first, followed by the central web. The jacks under the steel beams were slowly and uniformly
unloaded to release the shear force across the cut line
Fig 9.4 A second cut was made to isolate the closure segment. The closure segment was then lowered to the ground on strand
jacks. In a few locations it was removed by crane or gantry where the ground underneath was not suitable
10 Closure segment being lowered to the ground on strand jacks 12 Box girder segment fitted with delamination restraints around the top
web/flange junction
with the closure segment being lowered to the ground. support leg.
The magnitude of the locked-in forces was estimated to be quite A primary advantage of using the overhead gantry was that it
large with a significant degree of uncertainty. Furthermore simplified segment removal in areas where access from the ground
temperature difference between the top and bottom flanges was difficult. Segments could be lowered to the ground at places
produced very large bending moments in the box girder that that suited site constraints. Typically, each segment was lowered
changed throughout the day. To manage these effects the critical on to a waiting transporter and carted away for recycling, at a site
activities involved with releasing the locked-in forces were remote from the viaduct. Scrap steel was sold to a local steel mill
undertaken when the temperature difference between the top and and the recovered aggregate is destined for use on an upcoming
bottom flanges was less than 2°C. This typically meant the works Transport Agency project.
were undertaken between midnight and 9.00am.
Strengthening works
Cantilever segments removal Parts of the viaduct had to be strengthened so that it could be
Removal of segments from each cantilever was undertaken with deconstructed safely. This ironic situation arose for three reasons:
the overhead gantry. To minimise the dynamic effects that would firstly to compensate for defects in the original box girder; secondly
result from a sudden release of forces, the hanger bars between to compensate for the weight of the gantry; and thirdly because of
the gantry and the segment were pre-stressed with a force the build up of stresses that could not be released due to the cut
equivalent to the weight of the segment prior to cutting. The tendons redeveloping their bond. These elements are discussed
specified cut sequence was: outer webs then central web – from below.
the bottom up.
To minimise the out-of-balance effects, the segment on the Additional post-tensioning
opposing free cantilever was also stressed up to the gantry in Additional external post-tensioning had to be installed to ensure
preparation for its removal. segments joints remained closed at the quarter span points of the
In the process of cutting along the segment construction joint, spans adjacent to intermediate deck joints. There were six
there was a risk of disrupting the cantilever anchorage in the locations where this was required. Analysis showed that
segment adjacent to the construction joint. If the anchor head had excessively large cracks would have opened at these locations
been damaged, and if the cantilever tendon was poorly grouted, and as a result the shear capacity could have been compromised
the segment adjacent to the one being removed could have been if the external post-tensioning was not installed.
accidentally released. To mitigate this risk, the segment adjacent to
the one being cut was also suspended passively from the gantry Substructure strengthening
on hangers. The load path down the columns and into the foundations passed
Fig 11 shows how the gantry was typically used to deconstruct through the high strength (55MPa) concrete hinge block (plan
the free cantilevers. This method was used to deconstruct down to dimensions 916 × 406mm) at the base of every column. Around
the ±1 segments. Due to the very short pier table, and the large the hinge block there was a 25mm gap to ensure the connection
reaction force from the gantry on the pier table, it was necessary to behaved as a true hinge. At 810t the heavy gantry introduced
leave the ±1 segments on the pier while the gantry launched loads into the substructure up to 40% larger than they had
forward to the next pier. The ±1 segments were then removed with previously experienced with traffic loads. There was concern that
the rear of the gantry while it was supported on the lighter rear the lightly reinforced pad foundations may fail in a brittle manner
under these loads. To reduce this risk and achieve a better load properties. The daily and long term movement of the viaduct
distribution across the foundation, the 25mm gap surrounding the superstructure, expansion joints and piers was monitored using
hinge was grouted with cement grout resulting in a continuous surveying prisms mounted on the structure with a total station
connection. For the piers with piles, the additional bending collecting data automatically over a 24h period. The results were
moment introduced by the continuity would have overstressed the then back-analysed to confirm the findings from the structural
foundations and the gap was grouted only over part of the column analysis.
footprint to minimise moment transfer.
Even with these measures, the piers either side of Broadway Existing post-tensioning
(the longest span) required the foundation pads to be strengthened Three critical aspects of the post-tensioning had to be verified;
to safely carry the gantry loads. This was achieved with a their precise location, competency of grouting and strand
composite concrete overlay, 800mm deep, over the original pad condition. Ground penetrating radar surveys were used to initially
foundation. identify the tendon locations. These were confirmed at key
locations by drilling 25mm diameter holes to expose the tendon
Strengthening to resist excessive prestress ducts. The duct sheath and grouting were removed using a hand
As segments were successively removed from the free cantilever, chisel to expose the tendon strands and an endoscope was used
stresses built up in the remaining structure. This was because the to inspect and record the grout and tendon condition. The tendons
post-tensioned tendons (cantilever tendons and continuity were generally found to be in very good condition and well
tendons) that originally supported the whole superstructure could grouted. It was particularly important not to cut through the
not be progressively destressed as parts of the superstructure cantilever tendon anchorages either side of the closure segments
were removed. Left unchecked, this could have resulted in a and in these cases separate enlarged intrusions were made to
longitudinal shear failure between the web and the deck slab of the expose the anchor heads.
box girder and delamination of the group of cantilever tendons at
the top of the web. Both scenarios had the potential for an
uncontrolled collapse of the superstructure during deconstruction. Ground conditions
The solution adopted comprised the installation of continuous There was a real risk of pier settlement during the deconstruction
concrete strips over and under the web-deck junction. These were due to:
then prestressed to the box girder, both vertically through the deck – the use of the gantry resulted in the maximum pier load being up
and horizontally through the webs (Fig 12). The concrete strips to 40% greater than the viaduct had experienced in its 45 year
generally extended over the first one to three segments each side history;
of a pier and ensured the superstructure remained stable should – the underlying basalt had required extensive grouting during the
shear failure occur. They also effectively confined the grouped original construction to make it perform as a massive block. To
tendons and prevented delamination failure. mitigate this risk the ground conditions were investigated with
extensive drilling and the critical pier foundations were proof
Risk management loaded in advance by stressing of the deconstruction ground
Effective risk management was a critical factor in the successful anchors (by up to 800t) to represent the loading from the gantry.
deconstruction of the viaduct. A comprehensive risk management A fundamental change was also made to the deconstruction
system based on the proprietary Active Risk Management system, sequence and gantry operation to reduce the maximum load on
was used to identify, assess, control and monitor risk. Initially the the piers.
system was used to establish a risk contingency sum for
incorporation into the target outturn cost. During the detailed Trials to validate critical deconstruction activities
design and construction phase more detail was added, particularly Validation trials were carried out for critical elements that were
as new information came to light. Detailed response (mitigation) either unprecedented or where the consequences of failure were
plans were developed and assigned to appropriate single owners significant. Testing of scaled down pier bracket/pier interface pads
throughout the alliance team. Over 80 risks were identified and confirmed the design friction parameters and factor of safety. The
300 response plans developed. This is where the ‘active’ nature of first midspan cut and segment removal were reprogrammed to the
the system came to the fore; as owners were automatically sent southern end of the viaduct – where the viaduct was closest to the
reminders to close out their risk mitigations on time. ground – with temporary back-up propping located under the cut
A project review panel was established to review and challenge location to confirm the performance of the cutting method. This
the robustness of the project team’s identification and mitigation of initial trial also included testing of the delamination restraints to
risk. The panel comprised senior managers from the alliance confirm their effectiveness.
participants who were not directly involved in the delivery of the
project. The panel visited the project every 3 months. Temporary works performance
The key risks and mitigation measures addressed for the A rigorous programme of monitoring was employed to check the
deconstruction are summarised below. forces and performance of the temporary works. Strain gauges
measured the forces in the longitudinal, diagonal and vertical
Existing structure condition props. The ground anchors were re-stressed to compensate for
It was essential for the existing condition of the viaduct to be foundation settlements and the jacks on the pier brackets were
understood sufficiently so that the assumptions made in the checked regularly to verify that they retained the specified force.
deconstruction analysis and temporary works design were based
on an acceptable level of residual risk. Vehicle impact with partially deconstructed viaduct or gantry
The extreme threat for the deconstruction was the risk of a major
As-constructed details vehicle impact with either the partially deconstructed works or the
A detailed review was carried out of previous papers written about gantry. Speed restrictions, backed up by speed monitoring,
the viaduct3, inspection records and structural assessment enforcement cameras and extensive communications, were
reviews. A full set of as-built drawings (in Imperial units) was employed to reduce traffic speeds from the original 100kph to
obtained and the information, including the individual tendon 70kph. Test Level 4 (TL4) barriers were installed to provide a high
profiles, was digitised into a 3D CAD model. Contact was made level of containment. The maximum risk exposure occurred once a
with the original Resident Engineer and Construction Manager, midspan cut had been made and the full balanced cantilever could
with both interviewed to gain a better understanding of issues that be impacted at its tip. To mitigate this risk the cantilever was tied
occurred during the viaduct’s design and construction. back to the new viaduct box using unstressed prestressing
On-site checks were undertaken to confirm the dimension of key strands.
elements and cores were taken to test concrete and reinforcement
Acknowledgments
The authors wish to acknowledge support and contribution of the
New Zealand Transport Agency in the success of this project, as
well as the project participants – Leighton, Fulton Hogan, VSL,
URS, Beca, Tonkin & Taylor and Boffa Miskell.
References
1 American Association of State Highway and Transportation Officials LRFD
Bridge Design Specifications, 4th edition 2007
2 Transit New Zealand Bridge Manual, 2nd edition with amendments up to Please visit http://shop.istructe.org or
and including June 2005. Published by New Zealand Transport Agency
call 020 7235 4535 to purchase your
3 Built, J. W., Smith, A. W., McKinnell, P. B.: ‘The Newmarket Viaduct’; New
ow
Zealand Engineering; 20/12, December 1965, p493-511
copy now!
y n
Bu
Please visit http://shop.istructe.org/or tel: +44(0)20 7235 4535 to purchase your copy now!
Registered Charity with the Charity Commission for England and Wales No. 233392 and in Scotland No. SC038263