Sei sulla pagina 1di 24

Innov. Infrastruct. Solut.

(2017) 2:52
DOI 10.1007/s41062-017-0071-x

TECHNICAL PAPER

Minimizing the use of concrete in tunnels and caverns: comparing


NATM and NMT
Nick Barton1

Received: 4 May 2017 / Accepted: 2 June 2017


Ó Springer International Publishing AG 2017

Abstract For many decades, a tunnelling method has masses where we need tunnels or caverns will lie closer to
been in use which effectively minimizes the use of con- ‘mid-range’ (i.e. closer to Q = 1 which is described as
crete, which should be one of the goals in our CO2-pro- ‘poor quality’). Here we would need combinations of
ducing planet. We call the method NMT (Norwegian corrosion-protected rock bolts and high quality fibre-re-
Method of Tunnelling) and emphasize its ‘single-shell’ inforced shotcrete, with stainless steel or polypropylene
characteristics, to distinguish it clearly from double-shell fibres. We may also need systematic high-pressure pre-
NATM (the so-called New Austrian Tunnelling Method), injection of micro-cement and micro-silica, which may
which is recommended to have (ASG, NATM: the Aus- add 20% to the (low) starting cost of the NMT excava-
trian practice of conventional tunnelling 2010): shotcrete, tion. Written as B ? S(fr) in short-hand, NMT has rock
mesh, lattice girders, rock bolts (if in-rock), drainage bolt c/c spacing in metres and shotcrete thickness in
fleece, membrane, and the final load bearing and often centimetres, as specified by the range of Q values and
steel-reinforced concrete lining, including the invert when excavation dimensions. The details are also affected by
in poor rock conditions. This tunnelling method is the planned use. For instance, at our record-breaking
inevitably several times more expensive, uses many times Olympic cavern of 60 m span (for housing 5400 specta-
the volume of concrete, takes longer to build, and tors or later concert goers), B = 2.5 m c/c ? S(fr) 10 cm
requires at least a ten times larger labour force than sin- were (and remain 25 years later) the stabilizing and per-
gle-shell NMT. The single-shell tunnels for road or rail or manent measures of support and reinforcement. Defor-
hydropower or water transfer, or for large caverns for mation monitoring and distinct element (jointed rock)
storage of oil or food, or for hydropower machine and numerical verification showed 7–8 mm of maximum
transformer halls, can be made stable by judicious deformation in the arch. The moderate Q value range of
application of a well-used ([2000 case record based) the quality of 2–30 (poor/fair/good) and RQD = 60–90
so-called Q-system of rock mass quality estimation. The indicated a well-jointed gneiss, which had only moderate
latter encompasses a rock mass quality scale from 0.001 UCS = 90 MPa compressive strength.
(equivalent to a serious fault zone, where we also may
need a local concrete lining) to 1000 (equivalent to keywords Tunnels  NMT  NATM  Overbreak 
massive unjointed rock) where careful blasting will Shotcrete  Concrete  Collapse
remove the need even for shotcrete. In general, rock

Introduction
This paper was selected from GeoMEast 2017—Sustainable Civil
Infrastructures: Innovative Infrastructure Geotechnology. In view of the above abstracted summary, and as a valid
challenge, it would be interesting to know how mid-
& Nick Barton
European, specifically Austrian NATM (double-shell)
nickrbarton@hotmail.com
designers would have tackled the design of such a large
1
NB&A, Oslo, Norway cavern, and what thickness of concrete would have been

123
52 Page 2 of 24 Innov. Infrastruct. Solut. (2017) 2:52

proposed. The catch question would then be that if they had sufficiently bent into the overbreak. On paper and on the
considered concrete unnecessary, or too complex in such a computer it looks efficient, as no overbreak is ever shown.
large excavation, why is it apparently necessary in exca- The first four parameters in the Q-system: RQD/Jn 9 Jr/
vations of a fraction of the size? NATM is used in a large Ja, representing the fundamentals of relative block size and
number of transport tunnels and metro projects around the inter-block joint friction, have been used for three decades
world, and NATM proponents, not just Austrians, seem to in principal mining countries for mine stope dimensioning,
be of the opinion that this is the best and conventional way and Q is often used for designing mine roadways. The
to construct tunnels. Q value correlates with deformation modulus, P-wave
Commonly, in road tunnels or twin-track rail tunnels, velocity, and with deformation, where D (mm) & span(m)/
NATM designs might specify preliminary shotcrete of Q is the central trend of hundreds of tunnel and cavern
20–30-cm thickness, but actually 40–60 cm or more due to deformation data. It is wise to check ‘plastic-zone-prone’
covering of lattice girders and partial filling of overbreak numerical continuum models, and distinct element UDEC
caused by blasting. Usually, non-uniform, but specified or 3DEC models against such empiricism, using a more
35–45 cm of final concrete thickness may bring the total accurate equation including depth or stress level. This is in
concrete thickness in NATM between 75 and 100 cm. So case of exaggerated joint continuity in the jointed models.
conservatively speaking, we are facing a four to five times The motto ‘a posteriori is more reliable than a priori’ needs
greater use of concrete in NATM than NMT. This is even if to be constantly remembered when viewing the remarkably
we add a 15-cm-thick PC-element free-standing bolted complex equations which may lie behind the ‘a priori’
liner, with outer (hidden) membrane, for the sake of aes- assumptions of a popular and colourful FEM method used
thetic appearance, as in many NMT road tunnels, where by many of the younger generation for assumed tunnel
only 10–15 cm of S(fr) and the systematic bolting is used design. The reality is usually quite different from the
for the permanent stability. The only loading involved in continuum model, and more attention to the usually ani-
the PC-elements is self-weight due to the drained nature of sotropic behaviour of rock masses is required, using
the design. When we pre-inject effectively, the water models with jointing, but having realistic degrees of per-
pressure is pushed several metres into the rock mass, and sistence—not continuous joints from boundary to
bolt holes of 3–5 m length are not expected to leak. The boundary.
latter is the purest version of a single-shell tunnel: pre-
injection ? B ? S(fr) ? (occasionally) RRS (to be
described later). A brief summary of NATM
It will probably be clear for the reader by now that
because of the lesser use of concrete (sprayed or occa- Because of the cost and time involved in construction, and
sionally cast) in single-shell NMT tunnelling, it is neces- the fact that double-shell NATM tunnels and caverns are
sary to have a reliable method of selecting the appropriate used in prominent transport projects, such as motorways,
thickness of the relatively thin shotcrete lining and for metro-stations and high-speed rail tunnels, they are the
selecting the appropriate spacing of the rock bolts, each frequent subject of conference papers and tunnel magazine
used as permanent support and reinforcement of the rock articles. The description ‘double-shell’ is used to confirm
mass. This is where the Q-system comes in, which is a that a final concrete lining will cover the temporary support
method quite unlike any used in NATM, where tunnel phase of (typically) lattice girders, mesh-reinforced shot-
performance classes 1–7 seem to be the closest (Austrian- crete, and rock bolts. There are variations of the latter
based) equivalent. involving fibre-reinforced shotcrete (generally a superior
The Q-system has been used for 40 years (since [11]) choice) and no rock bolts (an adverse choice, though if
for core logging, exposure logging, and tunnel face logging rock cover is absent, this is necessary unless self-drilling
following each blast, each to assist with the selection of the bolts are used). Typical sequences of NATM are shown in
appropriate reinforcement and support classes. In [17] a Fig. 1a, b, from the Austrian Society of Geomechanics [1].
long overdue Q-system update of mesh-reinforced S(mr) to In Fig. 2a, b, examples of the operation of spraying in
fibre-reinforced S(fr) shotcrete was made to reflect its the lattice girders is shown on the left, and the final tem-
widespread use since the 1980s. Remarkably, this updated porary support phase of NATM is shown on the right.
technique does not seem to have come to NATM in any big Moving on to the fixing of drainage fleece and membrane
way, and is seldom recommended by Austrian consultants, and the final concrete lining, shown in Figs. 3 and 4,
not even in their standard: AGS, [1]. S(mr) is considered obviously in different projects, actually on different con-
and frequently seen to be, in a large number of countries, a tinents, we perhaps see why NATM is a serious under-
very inefficient way to cover and support tunnels with taking, deserving some reporting in conferences or
overbreak, as ‘shadow’ results, and the mesh cannot be tunnelling magazines as the investment is high and the total

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 3 of 24 52

Fig. 1 a Top-heading phase of NATM with ‘steel rib, wire mesh,


shotcrete application’ and the rock bolting. This is quite a critical
phase if the choice of tunnelling method is indeed double-shell
NATM. One should not delay with the rock bolting. This is especially
important when reaching jointed rock, when overbreak makes lattice
girders less reliable, and when anisotropic ‘point’ loads may be
experienced, i.e. usually not a part of the more simple design
assumptions. b Benching down, invert, and final concrete lining. This
follows the fixing of a drainage fleece and membrane. NATM is
clearly a labour intensive method which is several times as costly as
the more commonly used single-shell NMT. ‘NATM: the Austrian
practice of conventional tunnelling’ [1] Fig. 1 continued

time of construction is inevitably quite long. In their Introduction to NMT


helpful guide, with the drawings of Fig. 1 directly repro-
duced, we can note that the Austrians title the NATM as To summarise at the start: NMT is single-shell tunnelling
follows: ‘The Austrian practice of conventional tunnelling’ with a much reduced thicknesses of concrete (i.e. shotcrete)
[1]. lining. Besides various important contractual elements (see
All that we see in Fig. 2a, b so far is a temporary sup- multi-author descriptions in [12] it consists of the follow-
port, which is not credited in the final concrete load-bear- ing basic elements:
ing lining, which is delayed by a year or two on many
projects. So this is a hazardous phase in reality, especially 1. Q-system logging for selecting a permanent and
if rock bolts are not and cannot be used due to deep immediately applied support.
weathering. There have been notable collapses of the 2. S(fr) steel or polypropylene fibre-reinforced shotcrete.
temporary support phase of NATM, which will be illus- Typically 5–20-cm range.
trated later in this paper. These are sometimes caused by 3. B (utg)—CT-type with multiple corrosion protection.
unexpected anisotropic loading caused by jointed rock Typically 1.0–2.5 m c/c.
which is suddenly not loading the lattice girders as 4. RRS (rib-reinforced shotcrete arches, bolted) when
expected from the isotropic modeling. needed in low Q rock masses.

123
52 Page 4 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 2 a The temporary support phase of NATM, spraying in the ‘bulges’ on either side are the already completed pilot side drifts. The
lattice girders and mesh in the arch of a metro station cavern. This shotcreting between and into the lattice girders is near the specified
particular arch is unbolted due to the assumed B3 m of rock cover, minimum of 30 cm in places, but the contractor has already used
consisting of fragmented and weathered gneiss. The resulting something like 200% of the specified thickness of S(–), i.e.
temporary support phase is shown on the right. b The top-heading unreinforced shotcrete, due to the need to fill inevitable overbreak
of a 20-m span motorway tunnel, one of two parallel tubes. The (in soil, saprolite, and finally jointed phyllite)

5. Pre-grouting for dry tunnel, displacing water. Protects rock mass description method of [15] is also a source of
environment, prevents differential settlement damage support class selection for quite a large number of single-
to buildings founded on over-lying clays. shell excavations.
6. Free-standing, bolted PC-elements with outer mem- The fact that such single-shell tunnels are constructed so
brane for dry tunnel, but allowing drainage. Gives much faster and more cheaply means that they are less
improved ‘finish’ and lighting for main road tunnels. frequently to be found in the pages of tunnelling journals.
The designers and contractors have probably moved on to
Observant readers will have noticed the ‘more com-
other projects. Note that the world record for a completed
monly used’ reference to NMT in the Fig. 1b NATM fig-
single-shell tunnel is presently 5.8 km in 54 weeks (104 m/
ure caption. This is because single-shell B ? S(fr)
week average progress for a single face mine access tunnel)
(permanent bolting and fibre-reinforced shotcrete) which is
with a peak of 150 m in 1 week. The tunnel section was
most frequently chosen with the help of the Q-system, is
38 m2, and it was driven in coal measure rocks (by D ? B)
such a common method in hydropower tunnels, mine
and needed some B ? S(fr). Smaller drill-and-blast tunnels
roadways, hydropower caverns, and countless road and rail
with minimal need of any (single-shell) support have been
tunnels in many countries. Of course the alternative RMR
driven at a peak of 165 m in the record-breaking week.

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 5 of 24 52

Fig. 3 a In these photographs, we see the difficulty of drainage fleece welds per 1 km of tunnel are needed in a large tunnel. If there are flaws
and membrane fixing due to the effect of overbreak (left) and the in this welding work, and if the membrane leaks in a specific location,
complexity of a cavern excavation corner (right). Finally comes the it is actually impossible to locate where the repair is needed. In the
concrete lining, certainly a challenge for the 3D shuttering in a metro case of a pre-injected NMT tunnel, an eventual leakage point is
cavern, when not choosing the far simpler single-shell NMT. readily seen and repaired in the correct location both during
Recently, S(fr) was even used for stations in London clay in the construction, or afterwards. This is because the water cannot flow
major Crossrail project. However, it was very thick and without bolts. behind bonded shotcrete, as it can behind a smooth and slippery
b It can be readily estimated that approximately 15 km of membrane membrane-on-concrete interface

Both these records belong to Norwegian contractors, short section of tunnel, such as B 1.6 c/c ? 12 cm of S(fr).
respectively, LNS and Veidekke. This is completely applied before more than two or three
Figure 5 illustrates the principal stages of NMT. Note rounds of advance.
the central activity ‘Q-logging’ which now has a formal There are a number of compelling reasons why Nor-
‘owner’s half-hour’ for their engineering geologist to map wegian and Swedish contractors were quick to adopt wet
and come to agreement (if possible) with the contractor’s process S(fr) from the end of 1970s. (First seen by the
engineering geologist, on the required support class for the writer in a hydropower cavern in western Norway in 1979.
(usually) 4–5 m of rock mass that has just been exposed by First used in a road tunnel in Norway in 1981). Besides the
the last drill-and-blast tunnel advance. They agree not on volumetric 15–25 m3 per hour application rates, the pro-
the temporary support, which the contractor may add for duct gives steel-reinforced concrete in one pass, and can be
immediate safety, but on the permanent support for that built rapidly to desired thickness when using non-alkali

123
52 Page 6 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 4 The final completed concrete lining stage of the Channel chalk marl. Nevertheless, drainage holes were urgently needed in this
Tunnel sub-sea (cross-over) cavern on the right. The temporary critical phase of the project. Note the several metres distance to the
support phase of the same project does not fill one with confidence face and the ineffectiveness of a very broad mesh that attracts
about the effectiveness (or need) of lattice girders in the mostly dry corrosion and does little to strengthen the shotcrete

Fig. 5 Some details concerning


NMT. Tunnels are dry, drained,
and PC-element cladded (a free-
standing ‘drip-shield’) if
required for road or rail use.
(‘Pigging’ = scaling). These
details date from a 20–25-year-
old Norwegian contractor
brochure, and have remained
basically unchanged, as the
resulting tunnel, even with the
optional free-standing cladding-
and-membrane, is
approximately  of the price of
a truly double-shell NATM
tunnel. See for instance Barton
[4]

accelerators. It has low permeability, there is no ‘shadow’ which they do not seem to have experience with—when
[as with S(mr) seen in Fig. 5b]. It fills overbreak perfectly there is no overbreak. These very surprising recommen-
(Fig. 6) and due to the fibres, cures better even if applied dations are counter-intuitive and diametrically incorrect.
too thinly on convex ‘corners’ or noses. One only needs to inspect Fig. 6 to appreciate this (Fig. 8).
Figure 7 illustrates, to an unusually extreme degree, Figure 9 is designed to be self-explanatory regarding the
why S(fr) is preferred to S(mr) because of exceptional importance of Jn/Jr C6 to explain whether overbreak is
overbreak. In the examples shown, a very ineffective mesh likely to occur. This has by now been used several times by
and therefore an ineffective composite S(mr) would result contractors in claims situations, because when overbreak is
from attempted use of the NATM recommended S(mr). inevitable, but is not predicted during site investigations,
Long ago, close to 40 years ago, it was realized how big then the contractor will be forced to use a significantly
were the advantages of using S(fr). It is strange indeed that larger volume of shotcrete or final concrete than was
even today, Austrian consultants can recommend designing specified, due to the need to fill the overbreak. This last
tunnels with S(mr) when there is overbreak, and S(fr)— comment applies strictly to projects where NATM is the

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 7 of 24 52

Fig. 6 Top-pair S(fr) in a pre-


injected high-speed rail tunnel
driven in shales and limestones.
Note the first layer of 5 cm of
S(fr) applied close to the face,
likewise the permanent CT-
bolts, also applied close to the
face. Bottom pair The old-
fashioned method of temporary
support S(mr), still
recommended in NATM. This
results in great difficulties for
the contractor, if significant
overbreak. See Vandevall [20]
concerning the sketches

Fig. 7 Examples of major


overbreak to emphasize what
may happen when there are
three well-developed joint sets.
As seen on the next page, it
depends on the Q-parameter
ratio: Jn/Jr. The overbreak is at
a metro project and in the pilot
top-heading of an incorrectly
orientated set of parallel
hydropower caverns

design method. In the case of single-shell NMT, the CT-rock bolts


increase in shotcrete volume will be related with the
‘roughness’ (or increased length of the perimeter, as drawn The second component of NMT, or the first if written as
in a tunnel or cavern cross-section). It is not necessary in B ? S(fr) is the systematic bolting, which is a very
NMT to bring the perimeter back to its more circular (or important part of NMT because of the need to have a
D-shaped) origin, by filling the volume created by over- corrosion-protected solution, since there will never be a
break with concrete. We will illustrate this later, when final concrete lining to rely on. Figure 10 illustrates, in the
presenting the Q-system design chart for choosing the form of a short demo length, the principles of the PVC-
relevant quantities of B ? S(fr). sleeved CT-bolt, which was developed by Ørsta Stål more

123
52 Page 8 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 8 Some details of the S(fr) robotic shotcreting trucks, from location at least half-way up the boom for better control of the work.
AMV, as advertised more than 20 years ago, and of course still in A detail to note in the 20–25-year-old S(fr) ‘menu’ is that modern
widespread use in Norwegian transport tunnels. Note operator steel fibre dosages require just 25–35 kg/m3

Fig. 9 Reasons why the Q-system ratio Jn/Jr (number of joint sets/joint roughness) is important for determining if there will be overbreak or not,
despite careful blasting. Now used in some tunnel claim situations. Barton [5]

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 9 of 24 52

Fig. 12 English consultants finally designed sprayed concrete linings


(SCL) for some Crossrail stations in London clay. The writer was part
of a Norwegian ‘delegation’ visiting the Jubilee Line Extension in the
1990s, and of course suggested ‘why not S(fr) in the hard London
clay’ for the long-in-coming Crossrail project? In [9] we discussed
NATM contra NMT, and also suggested combining the best
components of each.

into the inner annulus, displacing air and any water as it


flows around the wedge-lock anchor and then down the
outer annulus to the exit tube. Bolts can be fully grouted
after S(fr) application if desired, using a short extension
tube. The joint or crack seen in the top-right photo is
typical when bolts are installed in jointed rock. Deforma-
tion with tunnel advance would be expected to crack the
Fig. 10 The principles of the CT-bolt, which has multiple layers of
protection against corrosion. Lengths could be 3, 4, 5 or 6 m. Note the usual single thickness of grout which could start corrosion
cross-section with the outer (only) cracked grout annulus, the normal if there was oxygenated water flow. The CT-bolt has four
fate of conventional bolts layers remaining (Figs. 11, 12).

RRS contra lattice girders

In the Grimstad and Barton [8] and Barton and Grimstad


[17] updates of tunnel and cavern support methods using
the Q-system, we changed the earlier (1974) S(mr) (mesh-
reinforced) Q-support recommendation to the modern wet
process S(fr) (fibre-reinforced) method, because S(fr) had
already been widely used for more than a decade. At this
time, attention was also drawn to the monitored results
shown in Fig. 13. Ward et al. [21] had monitored different
tunnel support methods in an experimental tunnel in
mudstones, at the Kielder project in NE England. The
obvious advantages of immediately combining B ? S (not
Fig. 11 Reisseck II pumped storage, Austria. This cavern photograph even mesh- or fibre- reinforced shotcrete) were very clear
indicates that even in Austria, someone decided to use ‘single-shell’ reasons why steel sets (or lattice girders) had never/will
S(mr) ? B in this large machine hall. Similar decisions could/should never be intended components of Q-based NMT support.
be made about smaller tunnels, if one had serious concern about The reasons are multiple:
unnecessary concrete use. Of course S(fr) would be preferable, as then
the concrete quantity is minimized and acts more efficiently 1. It is difficult to make good contact between the tunnel
than 25 years ago. It is now marketed also by Dywidag. (or cavern) perimeter and the lattice girder, especially
The steel bolt is galvanized, epoxy-painted, has an inner if there is deep overbreak.
annulus of grout and an outer one, as the grout is pumped

123
52 Page 10 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 13 Avoidance of steel sets


(or lattice girders) remains an
important advisory for NMT Q-
system users. This suggestion
has not always been followed,
even in Norway, and one may
wonder at the confidence of
those using unbolted free-
standing structures. There have
been some dramatic failures
when unexpected loads have
resulted from unstable rock
masses, and the isotropic
loading assumptions, as might
be reasonable from soils, fail
completely due to anisotropy

Fig. 14 Left the principal of RRS. This is a much sounder scientific (16 mm bar) ribs, prior to spraying thick individual RRS (rib-
method than unbolted, deformable lattice girders. Right a road tunnel reinforced shotcrete) arches. The left-hand figure from [3] shows the
through a section of faulted rock. Note systematically bolted important sequences involved

Fig. 15 Left Bolted RRS arches used in western drift of a 28-m distance are temporarily supported on slim columns (see greater detail
station cavern, with extensions needed when gradually removing the in Fig. 16). If they are bolted, and with smooth foundation S(fr) then a
pillar (not shown, on right side) below a deep clay-filled valley in good result might be achieved. Similar volumes of S(fr) in both cases,
downtown Oslo (National Theatre Station). Right lattice girders in a but large difference in safety level if lattice girders were unbolted, as
part of the 35 m span Holmestrand Station. The ‘half’ girders in the is normal

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 11 of 24 52

2. Rock mass deformation is needed to make ‘solid’ but


only local contact with the steel. Rock mass strength
may be reduced in the wrong places as a result.
3. When load starts to be applied by less stable parts of
the excavation perimeter, the footings of the lattice
girder will inevitably deform by a finite amount.
4. Strain and resulting stress must build up in the steel
bars of the lattice girder for it to finally apply
resistance to further deformation.
One could roughly estimate that 25–50 mm of defor-
mation might occur in this collective straining of the rock
girder combination, which as mentioned, also involves the
(elephant) footing with enlarged area for spreading load
when in soft ground.
The NMT alternative to lattice girders or steel sets is
RRS, which means (steel) rib-reinforced shotcrete arches.
The principles of this method are illustrated in Fig. 14,
showing the arrangement of the reinforcing bars, which at
16 mm diameter are neither too heavy (collectively) nor
too stiff. In fact, if a strong overbreak dominates in some
part of the periphery of the tunnel, the preliminary and
smoothing shotcrete arch can allow for some bending into
the overbreak. This is defensible because the RRS is bolted
at 1 m intervals around the arch. Figure 15 shows an
example (left).
In Fig. 15, one may note the possibility of bending RRS
into architectural shape or into major overbreak, as the
Fig. 16 The largest underground railway station ever built in rock, 16-mm diameter bars are easier to handle than, for
with a span of 35 m is to be found parallel to this steep scarp in
instance, the typical 25 9 25 9 32 mm, or
Holmestrand, in SE Norway. A surprising (for Norway) use of lattice
girders reportedly bolted in this case. A final free-standing bolted roof 20 9 20 9 25 mm three-bar triangular lattice girders. The
of steel plates was used in place of cast concrete ‘location’ of the extra secure RRS support and reinforce-
ment option is shown by the white arrow on the poor rock

Fig. 17 The provision of RRS


(not the inflexible and unbolted
lattice girders) is for ‘area 8’ in
the Q-support diagram (see
arrow). This chart is the [17]
update: changing the 1974
recommendation of S or S(mr)
to the more modern S(fr).
Besides the RRS dimensioning
seen in Fig. 18, there are minor
changes to the lesser thicknesses
of S(fr) due to increased
conservatism over the years, and
perhaps due to some sub-
standard thicknesses which
could have prejudiced curing

123
52 Page 12 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 18 The updated Q-support chart first published by [16]. The figure) is estimated from: L = 2 ? 0.15 SPAN/ESR (m). For walls
details of RRS dimensioning given in the ‘boxes’ in the left-hand side L = 2 ? 0.15 HEIGHT/ESR (m). For large caverns with eventual
of the Q-support diagram were derived by a combination of cable anchors, the factor (2 ? 0.15x) is replaced by 0.4x and 0.35x,
empiricism and some specific numerical modeling by a small team respectively, for the length of anchors in the arch and walls. Note that
of former NGI colleagues. Note RRS dimensioning example ‘box’ at HEIGHT refers to the full excavation height. See updated ESR values
the bottom of this figure. Note from [11] that bolt length (right side of (Table 1), which follows on the next page

quality side of the Q-support chart of 1993, in Fig. 17. Note Barton and Grimstad [9], and are shown in Fig. 18. For
the ‘RRS ? B’ and the recommended range of S(fr) instance, with a span = 12 m, and Q = 4 (‘fair’), the
(general) thickness varying from 15 to 25 cm when there above Q-support chart from 1993 recommends best quality
are low Q value conditions. Locally, as seen in Figs. 14 and S(fr) of 5 cm thickness and B (CT-type) c/c 2.1 m. An
15, the RRS has greater thickness in each arch. The extra 1 cm of S(fr) (total 6 cm) is suggested in the most
requirements, which are tunnel-size dependent, are shown updated chart (Fig. 18), but unchanged bolt spacing of
later, in a recently updated Q-support chart from Grimstad, 2.1 m c/c. Note that the ESR (excavation support ratio)
reproduced with many other Q-system details, in Barton safety number (see Span/ESR = equivalent dimension, left
and Grimstad [9]. side of figure) is given in Table 1 and illustrated in Fig. 19.
Note that today’s minimum S(fr) thickness has been Figure 20 shows a completed NMT-style rail tunnel
recommended as 5 cm thickness. Small adjustments in the from neighbouring Sweden. The moderate to very good
most updated chart, developed by Grimstad, are given in lighting available when this tunnel was photographed

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 13 of 24 52

Table 1 In the central column, ESR (1994) values in use in the nineties [9] are tabulated
Type of excavation ESR ESR (2014)
(1994)

A Temporary mine openings, etc. ca. 2–5 ca. 2–5


B Permanent mine openings, water tunnels for hydropower (exclude high-pressure penstocks), pilot tunnels, 1.6–2.0 1.6–2.0
drifts and headings for large openings, surge chambers
C Storage caverns, water treatment plants, minor road and railway tunnels, access tunnels 1.2–1.3 0.9–1.1
Storage caverns
1.2–1.3
D Power stations, major road and railway tunnels, civil defence chambers, portals, intersections 0.9–1.1 Major road and rail
tunnels 0.5–0.8
E Underground nuclear power stations, railway stations, sports and public facilities, factories, major gas 0.5–0.8 0.5–0.8
pipeline tunnels
Some updates recommended today due to the demand for greater safety are shown on the right (2014). Note the use of italics to emphasize no
change from 1994 to 2014

Fig. 19 The workings of ESR, for modifying SPAN to equivalent of RMR, the ‘unsupported span’ is the longitudinal distance from the
span. The way ESR modifies the equivalent span is shown by the face to the nearest support or reinforcement. These two ‘spans’ are
sloping lines, assuming ESR = 1.6 marks the unsupported boundary sometimes confused, due to interest in using the ‘stand-up time’
(for hydropower) [2]. Note that ‘unsupported span’ in the Q-system chart developed by Bieniawski [15]
refers to the width of excavation. In Bieniawski [15] concerning use

allows one to see the ‘minimalist-but-sufficient’ S(fr) use in ‘complete ring of concrete’ and their well-practiced ability
NMT-style support (the Q-system was used on this project to use thick-walled cylinder formulae.
for dimensioning), compared to the 75–100 cm that may In the case of the NMT-style tunnels illustrated, the
typically be the total shotcrete and concrete thickness in a initial support and reinforcement is finalized while pro-
full-blown NATM tunnel. Thousands of case records gressing in the tunnel. There is no unsafe, temporary sup-
document the sufficiency of the Q-based support. port phase, waiting many months or even 1 or 2 years for a
Numerical modeling with joint sets represented (in final (and finally safe) concrete lining, as in NATM. Risk is
UDEC-BB: BB are the non-linear constitutive laws for the high in this phase of NATM. Risk is not high when prac-
joints), and structural elements to represent the thin and ticing NMT, if Q-dimensioned support and reinforcement
strongly ‘undulating’ shotcrete surface (as illustrated in the (and good quality products) are used. In this connection,
above photos) indicate that bolt forces (where they cross note that Dramix cold-drawn steel fibre, the typical
and reinforce the jointing) are significantly reduced by the 0.5 9 35 mm variety, is stiff enough ‘to draw blood’ (lit-
addition of the shotcrete from the start of modeling. The erally) if scratched across the back of your hand. A ‘plas-
roughness and intimate filling of the overbreak by the S(fr) tic’ bending variety seen in some cut-price tunnelling,
is the reason for great effectiveness, even if bonding is though cheaper to purchase, may make the tunnel much,
artificially deleted. Traditionalists would no doubt miss the much more expensive in the long run. Do not cut corners

123
52 Page 14 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 20 Left an NMT main road tunnel in preparation in Norway. The common way to complete this twin-tube road tunnel. Right a railway
rock mass is highly jointed, causing marked overbreak in this tunnel in Sweden (Botniabanan) where there is also strong overbreak
location, but the arch is well-stabilized by the CT-bolts @ 1 m c/c due also to Jn/Jr [ 6. Q-system designed B ? S(fr) for permanent
(approximately) and by the rough undulating 5–10 cm of S(fr), support. Note permanent installations, prior to track-laying. Neither
perhaps not yet completed on walls with scaled blocks still shown. A tunnel has been high-pressure pre-injected, hence the marked
final free-standing PC-element liner with outer membrane will be a overbreak in both

when seeing a high unit price. The tunnel will be cheaper if


the high unit price is appreciated—for its better quality and
hence tunnel performance enhancing properties. This
applies especially to fibres, micro-silica, and corrosion-
protected (e.g. CT) bolts. If polypropylene fibres are cho-
sen, make sure rough (‘miniature-rope-like’) fibres are
used: they will bond properly, unlike cheaper and smooth
floor-slab fibres seen in some cut-price tunnelling.

Water-proofing methods in NMT

Although the huge cumulative lengths of hydropower


tunnels worldwide (with 3500 km in Norway alone)
mostly do not have strong water-proofing demands, as
often driven in mountainous regions with limited popu-
lation and environmental demands, there are numerous
tunnels (road, rail, metro, water transfer and also
Fig. 21 A typical drained-but-dry NMT tunnel with permanent
hydropower in populated regions) where inflow criteria support provided by the Q-system designed B ? S(fr). In this case,
are strict, or should have been stricter. This is because of the (road) tunnel has a permanent PC-element and outer membrane
passing under nearby housing, villages, towns, and of drip-shield. Modified from Kveldsvik and Karlsrud [19]
course sub-city metro locations, with over-lying com-
On the environmentally sensitive subject of tunnels
pressible clays or other sensitive soils. There are also
where pre-injection was needed but ineffectively per-
green areas, woods, and lakes for recreation, where
formed (for instance, due to insufficient drilling positions
inflow criteria may need to be strict, even as stringent as
on a TBM), the writer knows of a hydropower project
1–4 l/min/100 m of tunnel. When such criteria are not
where [2000 houses have been damaged and wells drawn
met, there may be permanent groundwater drawdown,
down, even as far as 1–3 km distant from a several hundred
settlement damage to buildings, and even drained lakes
metres deep tunnel. A jointed rock mass will often be
and damaged woods and forests.
highly anisotropic, especially as regards permeability [10],

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 15 of 24 52

Fig. 22 Mounting of 9 tons arch and wall PC-element ‘drip-shields’ injection were not sufficiently successful. Note that the 15 cm
with outer membrane visible in left-hand photo (white sheeting). thickness of the PC-elements added to the Q-system stability needs of
These elements were needed due to insufficient pre-injection, which S(fr), of 10 cm, makes a total 25 cm of concrete thickness
was rushed due to pressing deadlines. Extensive efforts with post-

so the effect of inflow to a tunnel on the surroundings is


extremely difficult to predict, and most isotropic perme-
ability assumptions will need to be revised, sometimes
drastically. However, we can start with tunnels where
drainage is allowable, and show the contrasting NMT and
NATM solutions.

Dry but drained

Figure 21 illustrates a dry-but-drained single-shell NMT


tunnel, where the free-standing bolted PC-elements act as a
permanent drip shield. Insulation against frost for the early
chainage from each portal is also shown. In this case, there
are no environmental concerns outside the tunnel. If there
were, pre-injection would have been used as the first
activity when driving the tunnel. Pre-injection will be
described later. Of course it is extremely important, espe-
cially if it has been carried out ineffectively, due to limited
understanding of the need for using micro-cements and
micro-silica and high pre-injection pressures. Figure 22
illustrates the fixing of bolted PC-elements in a large[1002
m double-track high-speed rail tunnel. These can be
mounted at a rate of 900–1000 m per month when using
efficient vacuum erectors.

Pre-injected and almost dry


Fig. 23 Top a Norwegian contractor drawing (original source
unknown) of the pre-injection umbrella principle. Bottom pre-
This sub-title is so named ‘almost dry’ to make allow- injection of an NMT tunnel, with up to 70 holes in a high-speed
ance for the fact that some pre-injection designs are not 110 m2 (electric) twin-track rail tunnel, driven in shales, limestones
sufficiently stringent. For instance, prioritizing the outer and numerous volcanic dykes. Pressures from 5 to 10 MPa [6, 7]

123
52 Page 16 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 24 Top a pre-injected single-shell NMT-style metro tunnel in water, and can be sprayed with the same robotic equipment as the
Switzerland. Damp patches in the shotcrete have been treated with a S(fr). The plate-loading test results indicate that the presence of the
sprayed membrane in a sandwich, using BASF 345 technology, as membrane is positive for two reasons: improved S(fr) toughness
described by Holter and Nymoen, [18]. Note the use (in this case) of (greater fracture energy or area under the curve), and the expected
polypropylene fibres. The membrane is formed from powder and water proofing

environment, with an ‘allowable’ 8–12 l/min/100 m Relative Q values for cost, overall time and cycle
limits on inflow to (nearly) protect green areas, may time
allow damaging dripping in the tunnel instead of prior-
itizing the inner environment of the (electrified) rail Because of the frequent use of the Q-system for mapping
tunnel with the more stringent 1–4 l/min/100 m. It is the and documentation of tunnelling conditions, studies have
latter which is needed for achieving a virtually dry been made of the relative cost and time involved in NMT
tunnel. For a shallow (&50 m deep tunnel), this may be single-shell tunnelling. A study involving some 50 km of
equivalent in achieving a near-tunnel permeability of Norwegian (mostly) and Swedish tunnels, using contractor
better than 10-8 m/s (roughly equivalent to 0.1 Lugeons) data, was reported by Roald and presented in a joint pub-
and approaching 10-9 m/s (roughly equivalent to 0.01 lication [14]. Figure 25 shows why NMT is as cheap and
Lugeons (Figs. 23, 24). fast as it is, in relation to NATM. This is because the Q-

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 17 of 24 52

Fig. 25 Relative time (left) and cost (right) of tunnel construction in relation to the Q value, according to a 50-km survey of tunnels carried out
by Roald, and published as Barton et al. [14]. NATM methods partially ignore the real needs

Fig. 26 The results of relative cost in relation to Q, in an independent of frequent borehole cores, give the possibility of direct costing. Note
application of Q-based support and reinforcement, including the that with pre-injection, the argument is made that many Q-parameters
mostly much reduced wall treatment. Spans of 7, 12 and 15 m are improved. If this is accepted, then the low-Q ‘tail’ of the Q value
tunnels, or cross-sections 50, 90 and 130 m2 excavations are shown, distribution can be expected to partially ‘move-to-the-right’, therefore
in relation to motorway ramps and main tunnel tubes. The relative partially removing the higher costs
frequencies (in metres) of specific Q value ranges, based on logging

system recommendations for B ? S(fr) ? RRS are used in scientifically based method that NATM. The latter passes
the one case, and of course not used and with rather dif- through a very questionable and potentially unsafe tempo-
ferent consequences in NATM (refer to Fig. 1a, b). NATM rary support phase with deformation poorly controlled.
is ‘well up these curves’ because the ‘heavy double-shell Indeed, the monitoring of deformation with NATM is an
support’ is used also when Q [ 1. This is wasteful. important and integral part of the method. When deformation
It should be clear from the above relative cost and relative time histories have slowed sufficiently, it is judged to be time
time exercises that NMT is a far more sustainable and to have the final concrete lining installed. In fact, the design

123
52 Page 18 of 24 Innov. Infrastruct. Solut. (2017) 2:52

of the concrete liner is supposed to not take credit for the


temporary support of S(mr) ? B ? lattice girders because it
is assumed that these will eventually degrade, since they
were not specifically designed to last. This is a waste of
resources, and combined with the three to four times greater
thickness of concrete (than NMT); the process seems to be
overdue for serious review (Figs. 26, 27).
If large caverns can be permanently supported and
reinforced with good quality and corrosion protected
B ? S(fr), even in Austria as illustrated in Fig. 11, then
why not tunnels that is a small fraction of the cross-sec-
tion? For instance most of the world’s large caverns of
600–1000 m2 cross-section utilize B ? S(fr) in an NMT-
style single-shell format. Why should tunnels of
100–150 m2 need to be so expensive in relation to the
Fig. 27 Cycle time (drilling blast holes, loading with explosives, caverns? Is the planned use—which is of course so dif-
blasting, waiting for gasses to clear, scaling, geological inspection,
mucking, reinforcement and support) as observed by Grimstad in the
ferent—sufficient cause. The world’s many hundreds of
Fodnes road tunnel, which has a cross-section of 50–55 m2. For hydropower caverns, some 3 9 800 in number, have 800
comparison, the cycle time for labour intensive temporary support machine halls with personell present and a lot of very
methods in a hydroelectric project in India is also shown in relation to expensive machinery.(The 800 transformer halls are also
the logged Q values. A final concrete lining will add to the differences
in time and cost of NATM compared to NMT
extremely important caverns, likewise the surge chambers,

Fig. 28 The ‘design-and-execute’ tunneller’s desk of drawers, used by Barton [3] to summarise key elements of NMT for an international
readership. Table 2 gives a summary of the ‘content’ of each drawer, using added connecting text

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 19 of 24 52

Table 2 An expanded text to explain the NMT abbreviations in the ‘drawers’ in Fig. 28
Rock mass characterization using the six Q-parameters. A relationship Site investigation using seismic refraction, radar, cross-hole VP or Edyn
between Q and VP and deformation modulus M is indicated, using Qc, tomogram, or attenuation tomogram. (Note Qseismic = 1/attenuation
rc, matrix porosity n% and depth H (m) or stress level is numerically close to modulus M in GPa)
Support design measures consist of none, sb, B, B ? S, B ? S(fr), Numerical verification of support designs using codes such as UDEC,
RRS, CCA. (Untensioned grouted Butg bolts, tensioned resin end- UDEC-BB, UDEC-Sfr, less frequently FLAC, FLAC-3D and 3DEC.
anchored bolts, and CT bolts). Also may use spiling, drainage, pre- Relevant parameters JRC, JCS, ur, Kn and Ks, Emass. Traditional
injection, and freezing c ? u, etc., for continua
S(fr) robot technology using Portland cement, silica fume, plasticizer, Norwegian tunnel contract system uses a flexible contract, with unit
superplasticizer, aggregate and non-alkali (low) accelerator. Steel prices for all possible measures in tender documents: use the motto
fibre: EE 20–25 mm (previously), Bekaert 30–35 mm/0.5 mm. ‘expect the unexpected’. Norwegian owners burden the risk, and
(Today: also PP fibre Barchip 48 mm, 0.4/1.4 mm) tunnels are cheaper
Rapid advance due to wet process S(fr) shotcrete, gives low rebound Low cost and less conflicts, permanent single-shell support compared
and improved environment to double-shell NATM
Many of the features mentioned were used in the Gjøvik cavern

each for very different reasons.) NMT is a scientific and and ‘point’ loading effects, which are poorly resisted by
well-documented method, as demonstrated in Fig. 28, with unbolted lattice girders. This is because the designers
the ‘designer’s desk’ (from [3]) (Fig. 29). usually have made no allowance for non-uniform loading
in their continuum modeling. It is much easier to ignore the
possibility. An example of uniform isotropic optimism is
Failures during temporary support phase: NATM shown in Fig. 32.
lessons The unfortunate reality, easier to explain after the event
than before, was that the vertically jointed phyllite that was
With overbreak never drawn and never numerically encountered after the soil and saprolite, should have been
modeled, the designers of NATM tunnels are free to specifically modeled, because it caused anisotropic loading
imagine that they have a final concrete lining that is quite unlike the isotropic conditions that had been assumed
perhaps even in overall compression, with little problem by the designers (Fig. 33).
of bending stresses such as in the nice example shown in In addition, there was a challenging ‘triggering
Fig. 30. The overall volume of shotcrete, first for the event’—a deeply weathered dike—eventually intersect-
initial lining, second for the final lining, can appear to be ing the tunnel at an acute angle, and causing a pre-
a slim 50 or 60 cm in total. The reality is seldom as liminary fall of ground. Figure 34 shows a ‘failure-in-
illustrated. First in the soil section, where loading might progress’ UDEC model with the sub-vertical structure
have approached that provided by an isotropic contin- of the phyllite modeled for the first time. The lattice
uum, there will be unwanted overbreak, fall-of-sand, girders’ resistance is overcome and plastic hinges are
etc., between each lattice girder, before getting control. then modeled. A complementary 3DEC model, also
In the (eventual) saprolite and deeply weathered rock in performed by Dr. Bandis, indicated that retrogressive
the next tens of metres of tunnel (assuming entry into a failure all the way back to the portal would occur, with
hill), there will be the probability of larger individual the boltless rock and the unbolted lattice girders over-
occurrences of overbreak. A measured statistic on the come by a progressive failure. Figure 35 illustrates the
thickness of the lattice girder-covering and overbreak- dramatic result of 140 m of failed tunnel in the centre
filling temporary support shotcrete is shown in Fig. 31. of the helicopter photo. Just glimpsed to the left is the
When the tunnel reaches rock, if the same lattice girder parallel tube which failed, all 140 m of it, some months
and shotcrete, and no rock bolts method are continued, later. The second top heading that failed, showing signs
there will be the probability of some individual deeper of shearing across the lattice girders, some 100 m into
sections of overbreak, if the Q-parameter ratio Jn/Jr (joint the tunnel, completely failed some months later, making
sets/joint roughness) is C6. The result can be anisotropic a total of 280 m of collapsed tunnelling. The severely

123
52 Page 20 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 29 The largest span NMT project, the 62 m span Gjøvik Olympic cavern in Norway. Note moderate quality of the jointed gneiss and the
local overbreak of [1 m. No lattice girders, no concrete, so no problem. Drip-shield, drained but dry [13]

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 21 of 24 52

Fig. 30 An ideal looking double-shell NATM tunnel design, with rather different from the reality. For instance, note the initial lining
initial lining, water-proofing lining (usually with drainage fleece) and thickness statistic from the first 140 m of a tunnel in Fig. 31
final lining. The nice uniform thickness all-in-compression liner is

The writer has earlier argued that despite the fact that
lattice girders (and steel sets) are fabricated with the
hardest of materials, the overall response is ‘soft’ because:
1. It is difficult to make continuous contact between the
tunnel (or cavern) perimeter and the lattice girder when
there is Jn/Jr C6 causing the overbreak.
2. Rock mass deformation is needed to make ‘solid’ but
only local contact with the steel. However, the steel
structure is not yet stiff, so deformation continues.
3. Rock mass strength may be reduced in the wrong
places as a result of any significant (several mm?)
deformation.
4. When this loading starts to be applied by the less
stable parts of the excavation perimeter, the footings of
the lattice girder will inevitably deform then stiffen.
Fig. 31 An undesirable statistic concerning the temporary lining
shotcrete thickness in a large 20 m span NATM tunnel. The shotcrete 5. Strain and resulting stress must build up in the steel
was used to fill overbreak and to cover the lattice girders. The bars of the lattice girder for it to finally apply
designed minimum thickness was 30 cm. A great majority of the data resistance to further deformation of the tunnel
(from the first 140 m of tunnel) lies in the region of 40–60 cm
perimeter.
thickness, with too many cases[60 cm. One may assume 75–100 cm
total thickness of concrete (sprayed or cast) when adding another It is reasonable to expect that wedges of unstable rock,
35–40 cm
with low-roughness (low Jr) joint boundaries, will behave
adversely as the tunnel is advanced, if without bolting, due
challenged and initially neat-looking portal area, and a to meeting only ‘soft’ resistance from the optimistically
major tension crack in the terrain, with rotation of over- located lattice girders (Figs. 37, 38).
lying ground, is illustrated in Figs. 35 and 36.

123
52 Page 22 of 24 Innov. Infrastruct. Solut. (2017) 2:52

Fig. 32 The assumption that causes least work and gives perfect- optimistic modeling was one of the major causes of a dramatic double
looking symmetry, is to ignore any anisotropic loading, and to ignore collapse in the temporary support phase of a major NATM motorway
any peculiarities of the (possibly sloping) terrain. This colourful, tunnel project illustrated on the next pages as a lesson

Fig. 34 A jointed UDEC-MC (Mohr–Coulomb shear strength)


model, with sub-vertical jointing and a weathered dyke each
challenging the capacity of inadequately designed and inadequately
dimensioned lattice girders. (Dr. Bandis)
Fig. 33 Top an elasto-plastic isotropic UDEC continuum model,
showing displacements, which are largest in the first (right-hand)
tunnel model. Magnitudes are in the range of 40–60 mm which is
cause for concern. Bottom when the actual initial presence of the
strongly sloping ground surface was included, as of course it should
have been by the designers, quite different and adverse behaviour was
indicated. (Modeling by Dr. Stavros Bandis, on behalf of the author)

123
Innov. Infrastruct. Solut. (2017) 2:52 Page 23 of 24 52

Fig. 35 Fortunately the tunnel crews had a lunch break when the
total collapse of 140 m of NATM temporary top-heading support
occurred, due to over-loading and retrogressive failure back to the
Fig. 37 The collapse of the second tube caused rotation of the road
portal. The latter was initially excavated in the deep soil and saprolite
and trees, and telegraph pole, and wide but subsequently soil-filled
and showed significant early deformation. Note slope
tension crack formation

Fig. 38 Although this metro station cavern failed for very special
reasons (a hidden 10–11 m high ridge of rock preserved in the soil
due to differential weathering, which fell 10 m to the cavern floor), it
is a sober lesson to see robust 25 9 25 9 32 mm bars constituting
the (unbolted) lattice girders bent or broken in tension. Most of what
Fig. 36 Failure of the second tube, also excavated some 140 m, we see in Fig. 2a collapsed due to the irresistible loading
followed some months later, making a total of 280 m of collapsed
tunnel. We can conclude that the main cause was to assume uniform
loading of the temporary lattice girders, when in reality there was an 3. With NMT there is no unsafe, temporary support phase
anisotropic weak (phyllitic) rock mass and a neglected dike ready to
challenge the traditional temporary support, once leaving the soil and
relying on deformable, unbolted lattice girders, waiting
saprolite near the portal many months or even 1 or 2 years for a final (and
finally safe) concrete lining, as in NATM. Risk is high
in this phase.
Conclusions

1. NATM and NMT tunnels are based on radically dif-


ferent use of resources, particularly regarding concrete References
(or shotcrete) thickness, and weight of steel. Cost and
time differences may be 4:1 and 2 years added con- 1. Austrian Society for Geomechanics (2010) NATM: the Austrian
practice of conventional tunnelling
struction time.
2. Barton N (1976) Unsupported underground openings. Rock
2. In the case of NMT tunnels, the initial support and Mechanics Discussion Meeting, Befo, Swedish Rock Mechanics
reinforcement is finalized while progressing the tunnel, Research Foundation, Stockholm, pp 61–94
with a much smaller (1:10?) labour force and more 3. Barton N (1996) Investigation, design and support of major road
tunnels in jointed rock using NMT principles. Keynote Lecture,
mechanization, especially the high-capacity S(fr)
IX Australian Tunnelling Conf. Sydney, pp 145–159
robots.

123
52 Page 24 of 24 Innov. Infrastruct. Solut. (2017) 2:52

4. Barton N (2002) Some new Q-value correlations to assist in site measured performance of the 62 m span Norwegian Olympic Ice
characterization and tunnel design. Int J Rock Mech Min Sci Hockey Cavern at Gjøvik. Int J Rock Mech Min Sci Geomech
39(2):185–216 Abstr 31(6):617–641
5. Barton N (2007) Future directions for rock mass classification 14. Barton N, Buen B, Roald S (2001) Strengthening the case for
and characterization—towards a cross-disciplinary approach. grouting. Tunn Tunn Int Dec 2001:34–36/Jan 2002:37–39
Invited lecture. In: Proc. of 1st US-Canada Rock Mech. Symp., 15. Bieniawski ZT (1989) Engineering rock mass classifications: a
Vancouve complete manual for engineers and geologists in mining, civil and
6. Barton N (2012a) Defining NMT as part of the NATM SCL petroleum engineering. Wiley, New York
debate. In: Wallace S (ed) TunnelTalk, p 4 16. Grimstad E (2007) The Norwegian method of tunnelling—a
7. Barton N (2012b) Assessing pre-injection in tunnelling. Tunn J, challenge for support design. XIV European conference on soil
pp 44–50 mechanics and geotechnical engineering. Madrid
8. Barton N, Grimstad E (1994a) The Q-system following twenty 17. Grimstad E, Barton N (1993) Updating of the Q-System for
years of application in NMT support selection. 43rd Geome- NMT. In: Kompen, Opsahl, Berg (eds) Proc. of Int. Symp. on
chanic Colloquy, Salzburg. Felsbau, 6/94:428–436 sprayed concrete—modern use of wet mix sprayed concrete for
9. Barton N, Grimstad E (1994b) Rock mass conditions dictate underground support, Fagernes, 1993. Norwegian Concrete
choice between NMT and NATM. Tunn Tunn 29(10):39–42 Association, Oslo, pp 46–66
10. Barton N, Quadros E (2015) Anisotropy is everywhere, to see, to 18. Holter KG, Nymoen B (2009) Permanent waterproof tunnel lin-
measure and to model. Rock Mech Rock Eng 48:1323–1339 ing based on sprayed concrete and spray-applied double-bonded
11. Barton N, Lien R, Lunde J (1974) Engineering classification of membrane. In: Proc. of ITA Congress, Helsinki
rock masses for the design of tunnel support. Rock Mech 19. Kveldsvik V, Karlsrud K (1995) Support and water control in
6(4):189–236 Oslo. World Tunn 8:167–171
12. Barton N, Grimstad E, Aas G, Opsahl OA, Bakken A, Pedersen 20. Vandevall M (1990) Dramix—tunnelling the world, 1991 edition.
L, Johansen ED (1992) Norwegian method of tunnelling. WT NV Bækert S.A, Belgium
Focus on Norway, World Tunnelling, June/August 1992 21. Ward WH, Todd P, Berry NSM (1983) The Kielder experimental
13. Barton N, By TL, Chryssanthakis P, Tunbridge L, Kristiansen J, tunnel: final results. Geotechnique 33(3):275–291
Løset F, Bhasin RK, Westerdahl H, Vik G (1994) Predicted and

123

Potrebbero piacerti anche