Design and construction of an immersed concrete tunnel using an
integrated dock facility
C. Bauduin
BESIX, Brussels, Belgium
University of Brussels, Belgium
P. Depuydt
BESIX, Brussels, Belgium
ABSTRACT: The construction of the A73 highway (the Netherlands) involved the construction of a 2.4 km
Jong, 2*2 lane tunnel to mitigate the impact of traffic on the city of Roermond and to cross the 1 km wide val-
ley of the river Roer. Given the ground conditions (mainly dense sand and gravel), required cross section and
tunnel depth, an immersed tunnel appeared to be the most economic solution to cross the valley. However, a
facility to permit precasting of the RC tunnel elements needed to be established. The small depth and width of
the Roer prevented transportation of tunnel elements by use of the river and dewatering was not permitted due
to the associated environmental impact. The analysis of the geotechnical data indicated the presence of a5 m
thick local loam layer over a length of 350 m along the tunnel alignment, located at the eastern part of the
Roer valley. This impervious layer offered the opportunity to excavate a 350 m long dock along the axis of
the tunnel between temporary anchored sheetpiles that were installed into the loam layer. Two precast tunnel
elements of approximately 158 m long could be constructed in this dock. A trench was excavated between
and temporary sheetpiles were installed 5 m outside the future tunnel location, thus permitting the transporta-
tion and immersion of a total of four such tunnel elements in two installments from the dock to their final lo-
cation, The remaining part of the tunnel was constructed in what was previously the dry dock after the immer-
sion of the elements. This paper describes how the concrete structure was designed and specified (concrete
weight and tolerances) with allowances for the specific geotechnical and hydraulic conditions (water depth,
concrete weight and freeboard, water level management in the trench, excavation depth and uplift of the im-
pervious layer etc.). The paper describes the behavior and provisions of the concrete structure on the gravel
bed foundation, which was preferred to sand flow to minimize the risks of liquefaction as the area is mode-
rately seismic,
1 INTRODUCTION
As part of the new 40 km A73 Highway in the south
of the Netherlands, the Ministry of ‘Transport Public
Works and Water Management (Rijkswaterstaat)
planned to build a 2*2 lanes, 2.4 km long tunnel,
parily adjacent to the city of Roermond and partly
under the 1 km wide valley of the river Roer. The
tunnel solution was selected in order to minimize the
impact of the highway in the urban area and to pre-
serve the natural environment of the Roer valley.
The Ministry prepared the tender for the funnel
works as a Design and Construct contract and the
client’s specifications were stated on a rather ab-
stract level, leaving a large degree of design freedom
to the competing contractors. In 2004 the contract
was awarded to the lowest bidding contractor that
conformed to the client’s specifications. The design
and the construction were the contractor's respon:
bility, under process supervision by the Ministry.
The contract was awarded to Besix, the works were
executed by Besix-Strukton Betonbouw JV, the de-
sign was undertaken by Besix Design Department
and the preparation, design and execution of immer-
sion operations was undertaken by Mergor (part of
Strukton Betonbouw).
4
Fig. 1, Location of the tunnel.Severe environmental constraints applied to the
design and construction of the tunnel:
= To preserve nature and to avoid settlement of
neighboring structures, with no permitted changes
to ground water levels
~ The construction should not influence the course
of the River Roer and not reduce the discharge
capacity of the Roer Valley in case of flooding
= Vibrations and noise should not adversely impact
the natural environment or cause hindrance or
damage to neighboring structures and persons.
The ground conditions are characterized by me-
dium dense sand and sandy-loam top layers, overly~
ing dense to very dense, highly pervious sand and
gravel layers, Stiff loam layers with low permeabi-
ity were found locally between these layers.
oe ew ST le ew ol
= SNSUENE Arne
ws EE]
Doainm
Fig. 2. Typical CPT in construction pit.
The discharge of the River Roer is governed by
rainfall and the melting of snow in upstream regions
‘The water level is mainly governed by the discharge,
the water level in the Meuse and by the opening or
closing of the upstream dams, In normal discharge
conditions, the Roer is approximately 15 m wide and
4m deep; in extreme conditions, however, the valley
is flooded over its whole width between the winter
dikes (approximately 1 km). The water level varies
from 18.75 m NAP (minimum level) to 20.74 m
NAP (return period: 1/10 years), 21.55 (retum pe-
riod: 1/100 years) which corresponds to a river
width of I km. The design water level was 22.44 m
NAP (return period 1/10000 years). ‘The maximum
top ground level is 20.5 m NAP. The ground water
piezometric head reacts moderately to variations in
the river level.
The geometric requirements of the tunnel were:
= The cross section was to consist of two tubes (two.
catriageways cach) separated by a central emer-
‘gency evacuation tunnel (fig. 3).
= Internal free height requirements: 5.0 m (inclu-
ding an allowance for electro-mechanica! installa
tion) and width between the wails: 10.20 m
minimum to be increased to allow for construc-
tion tolerances. The height between bottom and
top slab is 5.6 m to allow for ballast concrete,
road pavement construction and settlement tole-
rances (fig. 3)
= Specific to the tunnel in the Roer Valley: the top
level of the roof slab was to be located at 14.85 m
‘NAP maximum over approximately 200 m width
of the valley in order to avoid adverse effects of
the tunnel on the groundwater flow. The overbur-
den combined with the design water level go-
verned the maximum values of the actions for de-
signing the tunnel cross sections (fig. 4).
— The design specifications required that a mini-
mum contact stress of 5 kPa due to the absence of
overburden over approximately half of the tunnel
Tength to accommodate possible future natural
changes of the river bedding be included in the
design. This governed the volume of structural
and ballast concrete that was required for the in
temal volume of the tunnel, given the unit weight
of the conerete and the amount of reinforcement.
~ The concrete volume and internal net open vo-
lume determined the free board. The smaller the
free board, the easier the immersion and ballast
exchange operations.
1260
Fig. 3. Tunnel cross section TEL/TE2, concrete pouring stages.
2. GENERAL DESCRIPTION OF THE,
TUNNEL
Four main areas can be identified (fig. 7):
— An open access ramp at the western side was de-
signed as a polder construction formed by a PE
sheet placed underwater on the bottom and the
slopes of a deep dredged open trench, Backfill on
the sheet compensated for uplift ground water
pressure.
— Piled open ramp and tunnel section, west of the
Roer valley. The deepest part of the access ramp
and a 60 m long part of the tunnel west of the
dike, including the main technical building and
sump pit, were designed as reinforced concrete
structures supported by GEWI piles and con-
structed between permanent anchored sheet piles
or combi-walls. A temporary unreinforced con-crete floor was designed below the RC structure
between the sheet piles and anchored by the
GEWI piles to avoid dewatering. This design was
chosen because the required width in the trench
for the PE sheet was not available and also due to
the complexity of the structures to be built.
— Crossing of the valley, as far as the eastern dikes
(see chapter 3).
— An urban section of the tunnel, which is located
partly under the urban ringroad to the south of
Roetmond; the alignment being crossed at ground
level by local access roads at several locations.
The urban part was designed as a cut and cover
cast in-situ reinforced concrete structure,
3. TUNNEL CROSSING THE VALLEY
‘The design for the crossing of the Roer Valley was a
key element to the project due its very important
economic impact on the total construction cost.
Several solutions complying with the fundamen
tal requirements were predesigned and evaluated in
terms of cost and risk. Immersed tunnel solutions
appeared to be the most appropriate because they
‘would require no piling works and construction ma-
terials could be used efficiently provided that:
~ the trench could be constructed using temporary
facilities such as recoverable sheet piles
an efficient and economic solution for the area for
precasting the immersed elements could be found
while preventing the need for dewatering
The trench along the tunnel alignment was sup-
ported by temporary anchored sheet pile walls lo-
cated 5m beyond the planned tunnel, thus permit.
ting recovery of the sheet piles after completion of
the works.
‘The use of the Roer to transport elements was not
feasible as it has insufficient depth and width. Care-
fl analysis of the ground conditions indicated the
presence of a 5 m thick loam layer along the tunnel
alignment at the eastern 350 m of the valley, the bot-
tom of the layer being at approximately -2 m NAP.
1
Table 1.
Parmer Ti fiveraby afiess Tntiverbly aes Torani il condition or
esgn parameter
Deep excavation oval ‘Wate depitvequiedinRoertoen | Uplittleam layer Disignporawar
consiuton pit ble Noting of TE Dimensions reining stusture of pit
“Tishnes of lamlayer | Thick | Excavation dept (higher piles “Tobe determined by etal
) ee ground investigation _|
Teal nl Haight Haigh | Stength of roof ana testom sb | Required water depth for loating condi | Determined by te height, bale
Allowable thickness otallasteon- | ous (ative feels between the Roce | lst thikess and rootbtiom
cree aud te depth of excavations) sab thickness
‘Verte tolerance for immersion and
__| acewacy of prediciedsotlemens
“oun toed ioe Tow | Required water dep for oaing | Swengh ofvoot and baton sab ‘Deerined by fee eign (Oe
conditions (relive levels between the | Allowable thickness okaastconcete_| sign requirement), allasthick-
[Roe and the depth afexcavations) | Venicl tolerance for immersion and ae- | ness and oofbottom sab
curacy of peditdsetlements thickness
‘Waar level Rose High | Plating coolions (ow fee Bord) Vasile, slightly pedicle,
(Clearance under TE manageable
Waterevel Row Tow Regus high ee bowed ‘Variable sighly edie,
sumnageable
ee board TE Taw | Foating conditions Reqd water depih footing coral | Design paramsi,
Ballastquatiy tions (elaine level between he Reer | minimum tee bosd (0.2 mto
Egil afer ierson sd the depth ofeseavations)toenable | erable lotng of TE
foating
Fre board TE THigh | Respird depth oF constustion pe | Inmerson (igh vem of ballast wa | Design parame, ied by
eguited depth wench ) strength egucerent for ap-
Required minimum evelofRoerto | Exchange of water allast to permanent | plied load
able Noting of TE tall
Deptraftinneisiement | Follows | The smaller the dept, he snlle the “Allows optiniztio of?
infval postion (us| tom vent- | floor and rf thickness thar otal ele ‘Wess
maximum alae oforer- | callin: | ment height
‘burden on of) ent
‘Ceaance fo ballast, High | Upliinabeence cf overburden | Total wane eight ‘Design pavantr Hinting de
const in tunnel Allowable uncertainty ox settement ition to wravel bed
‘Weight of reinforced Tow | High fee tourd, ds requved dept | Eifetive contacts underbotton | Design parameter
cencite ee slab
‘Maximum tee oftuw | Designs eight ofthe tunnel (ming evel of | Design reqaerent oi
aclroof afer immersion | urement ‘uel 008) size effets of tunel on
round wate ow
Wooverbadenontun | Designs ‘Minima hekners ofall eonerete | Design equuement
elo! qltement vodeaey)This geotechnical feature was the key to the de-
sign of the construction area for the precast RC
elements of the immersed tunnel, as it permitted
the construction of a construction pit along the axis
of the planned tunnel, ie. in the same alignment as
the trench, without the need for dewatering pro-
vided that the retaining wall could penetrate suffi
ciently deeply into the impervious layer,
‘This construction pit was used to construct two
tunnel elements (each of which was approximately
135 and 158 m long), after which the pit was inun-
dated and the tunnel elements were floated out of
the pit and subsequently immersed (fig. 4). The pit
‘was then dewatered and used again to construct the
next (Wo tunnel elements which were similarly
transported and immersed at their final location.
Finally the pit was dewatered again to construct a
remaining part of tunnel approximately 350 m long
as a cast in-situ cut and cover tunnel using the
framework of the immersed parts. After backfilling
the french and the construction pit, the sheetpiles
located adjacent to the trench and the construction,
pit were removed.
In order to achieve the conceptual design de-
scribed above, it was necessary to successfully
balance a number of oppositely interacting pa-
rameters, as shown in table 1.
Analysis of the interaction between these pa-
rameters has indicated that, for a given level of the
underside of the loam layer and thus the allowable
exeavation depth, the feasibility of the concept was
governed by:
= The required tunnel height for resistance with
full overburden and uplift safety without over-
burden, including tolerances on settlement
= The required water level in the Roer
— The ability to predict the unit weight of the rein-
forced concrete adequately and the control of
this value during construction, in addition to the
control of dimensions during construction
Several scenarios were analyzed, which finally
Jed to the following design (fig. 4):
— For the deepest and thus most heavily loaded
elements 1 and 2, the required slab thickness
was too large to be compatible with the pro-
bable low water levels in the Roer and may
mum allowable excavation depth in the con-
struction pit. It was decided to install a lock
wall in the trench which allowed the water level
in the construction pit and the adjacent part of
the trench to be controlled independently of the
water level of the Roer. The elements were then
towed to the deepest parts of the trench and fi-
nally the lock was opened afler equating the two
water levels. The elements 1 and 2 were de-
signed for a free board of 20 cm to 50 cm de-
pending on the assumed tolerances on concrete
weight and dimensions.
— For the two highest and thus less loaded ele-
ments 3 and 4, e design with a high free board
(greater than 1 m) was adopted, as the possi
ity of using the lock wall no longer existed fol-
lowing the immersion of elements 1 and 2
— After immersion, the fourth element was strut-
ted and preloaded under water against the longi-
tudinal walls of the construction pit in order to
avoid loss of compression in the Gina profile,
which could occur during the subsequent dewa-
tering of the construction pit which was neces-
sary for the construction of the final in-situ part
of the tunnel under the Roer Valley
The structural design of the tunnel and temporary
structures was undertaken so that the aforemen-
tioned requirements were accommodated,
Cet -4 CONCRETE
Conerete grade C28/35 was selected as:
= Its spalling behavior is more favorable when
‘compared with concrete of a higher strength
— Because it develops manageable internal tem-
peratures in its initial hardening stage (hydra-
tion process). In order to avoid cracking, it was
specified that the maximum tensile stress in the
hardening concrete during the hydration process
‘was not to exceed half of the average instanta-
neous tensile strength. FE heat and stress caleu-
lations were performed for the selected concrete
mix to check this criterion and to determine the
amount of water cooling that was necessary to
concrete the pouring stages (stage 1: floor slab,
stage 2: internal walls; stage 3: extemal walls
and roof).
‘The specified design life for the tunnel is 100
years, Durability of the structure was achieved by
‘adopting a conerete cover of 50 mm, specifying
‘environmental class XD3/XF4, using CEM Ill ce-
ment with a minimum slag content of 50% and a
maximum water penetration of 20 mm.
The value and range of variation of the unit weight
of the concrete are of large importance for the im-
mersed tunnel. Values were specified on the initial
concrete mix type tests. The unit weight of the
concrete (y= 22.85 kNim’ - 23.46 kN/m?) and the
as built dimensions were measured continuously
during the construction in order that they remained
within the acceptable limits for successful floating
and immersion operations,
5 FOUNDATION DESIGN
Due to the seismic risk in the Roermond area, a
gravel bed foundation was preferred to classical
sand flow in order to minimize the risk of liquefac-
tion during an earthquake event, This foundation
consists of nine ridges, 1.65 m wide at their top
level and 0.5 m high per 22.6m long tunnel section
(fig. 5).
“The interaction between the foundation and the
tunnel was determined by the stiffness of the
ridges and by the uniformity of their top level. The
Joad-settlement relationship of a single ridge was
established using FE geotechnical calculations as-
suming @ hardening soil model for the natural
ground and Mohr-Coulomb behavior for the gravel
ridges. ‘The behavior that was predicted for the
ridge was confirmed by plate load tests that were
undertaken on full scale ridges which were con-
structed using the intended gravel. Deviations of
the actual top levels of the ridges from the theo-
retical levels provoked non-uniform reactions
which had to be accounted for. Deviations of the
top levels may be stochastic or systematic, but are
not known during the design. They interact with
the structure differently depending on the scale
(area over which local deviations are averaged) of
the deviation:
~ Stochastic deviations act at a local level; par-
ticularly at “stiff points” as protruding ridges in
the span of the bottom slab are unfavorable
when compared with a uniformly distributed
soil reaction. These effects have been mitigated
by reducing the thickness of the lower face of
the bottom slab in the centre of the span below
the tunnel cells and by the design methodology
He BB B42,
ta
"ec aN a aa ac oon
Fig. 5. Plan view ridges for one TE and detail of
izes.
— Systematic deviations over areas of quarter to
half of the area of a section mainly act at a
global level in a single tunnel section provoking
torsion or cross spanning of the tunnel from one
outer wall to the other
— Systematic deviations over areas that are ap-
proximately the area of a whole tunnel section
(22.6m*27.64 m) result in the shear forces be-
ing transferred fiom the poorly supported sec-
tion to the well supported section
Separate non-linear analysis models were con-
structed to simulate each of the types of aforemen-
tioned deviations. In all of these models, the non-
linear behavior of the spring and possible gaps be-
tween the top of ridges were modeled. A large
number of calculations have been undertaken with
these non-linear models in order to develop a full
understanding of the effects of the assumed devia-
tions. These calculations led to envelopes of inter-
nal forces that were used to construct equivalent
deterministic simple 2D-models which were then
used for routine design, of which the results hadbeen proven to be conservative when compared
with the envelopes that were obtained from the
non-linear models.
The different simplified models that were used
for the design have been incorporated into the
permanent works by introducin;
~ Specifications for the placement of the ridges
— Acceptance criteria for the gravel bed as placed
in-situ, by introducing limits on the measured
deviations of the as built levels when compared
with the theoretical.
The flow diagram shown in fig 6 indicates the
design steps that were followed to establish the
possible effects of any deviation, to convert these
models to casy-to-handle routine design models,
and to specify allowable mean deviations of the
ridges over areas of different sizes,
‘Sepia fone dvs oop ee and sot
conta sores || “Senmesarazsen || secmeaeonyze ce
tons
Things of ght eos ot
eter see er oe
fonveryterotsertaans
2fotbea roves nr
‘Sep} Snpiied meds talloaine design stantondinaaceatom |
“atldjon model emelopes
Tee equa 7 wae wi ote | [ Tavalnostae tama
cones Ryarrreuin arenes
Sorotuttenrcuerwatswesa.c.e | | mamatueemtcotswces
oie Siocewenceee
en atnabie devatong top aes
cetomentaras a6, || ousransiocen || haratsen nee
SreerimacetanSoem || Cletmsceeaon |] Cleanses
Fig. 6. Flow scheme of design and specification of gravel bed
‘The non-linear models were used to predict the
most probable value of settlement due to backfill
after immersion as well as the upper and lower
its of these settlements (as approximated 5%
and_95% ranges). The theoretical level of the
‘| ROER
Fig. 7, Overall view of the tunnel and design solutions adopted,
gravel bed has been corrected on the basis of the
calculated most probable values of settlement
G5 mm in the deepest section reducing to 5 mm in
highest section), After backfill, the deviation be-
tween the as built level and theoretical level was
for 90% of the measurements less than 5 mm, ex
treme deviation was 15 mm.
6 ACCEPTANCE PROCEDURE FOR
GRAVEL BED
The ridges were placed by an underwater scrader
that was specially developed for the project. The
top level of the ridges was measured in-situ by use
of a multi-beam, Transforming the measured data
into a calculated average level per meter length,
the level of the ridges and the levels of the as-
sumed areas in the design could be easily checked
against the acceptance criteria, If the three crite
(lig. 6) were met, the foundation bed was accepted.
If one or maybe several criteria were not satisfied,
the actual measured in-situ ridge levels were intro~
duced in the relevant calculation models to check
the design (slab reinforeement, shear force in shear
key, etc.) for the actual as built situation. If neces-
sary, profile of the gravel bed had to be adapted.
7 CONCLUSION
Strict client's specifications on environmental con-
ditions and geometric boundaries were applied in
the design and construction of the tunnel.
Considering the local ground conditions, an
immersed tunnel was the optimum design solution
in terms of costs and risks. Integration of design,
construction methods, specifications and monito-
ring was the key issue for a successful project.
‘The consttuction pit was started in November
2004 and tunnel elements were immersed in May
and October-November 2006. The project was de-
livered in November 2007.