Sie sind auf Seite 1von 27

Nick Barton & Associates Rock Engineering

RIO MANTARO HYDROELECTRIC PROJECT

INDEPENDENT OPINIONS ON VARIOUS ASPECTS OF THE


POWERHOUSE CONSTRUCTION AND TUNNELLING AT THE
CERRO DE AGUILA PROJECT

Contract: Astaldi/GyM Consorcio Rio Mantaro

Contact: José Alves Perreira Neto

March 2014

Fjordveien 65 c, R. Miguel de Almeida Prado 20, Butanta,


1363 Høvik, Norway Sao Paulo, SP 05578 040, Brazil
T/F +47 67 53 15 06 T/F +55 11 3722 0889
nickrbarton@hotmail.com nickbarton@uol.com.br
Nick Barton & Associates Rock Engineering

Table of Contents

1. INTRODUCTION......................................................................................................3

Cavern orientation confirmation………………………………………………………....4

2. CAVERN BOLT LENGTH-AND-SPACING OPTIMIZATION………………………...4

3 CAVERN CRANE BEAM DESIGN OPTIMIZATION………………………………….9

4 BLASTING ROUNDS and SUGGESTIONS for UNSUPPORTED SPAN …………11

Blasting responsibilities and overbreak………………………………………………..12

5 GROUTING ‘TIMING’ FOR MECHANICALLY-ANCHORED BOLTS IN


DIFFERENT ROCK CLASSES…………………………………………………………12

6 SHOTCRETE APPLICATION, THICKNESS, CLASS A and B……………………..14

7 FIBRE-TYPE, DRAMIX OR POLYPROPYLENE………………………………….. …15

8 HEADRACE STRESS FRACTURING – DESTRESSING PLANS………………….16

9 SUPPORT AND REINFORCEMENT MEASURES TO RESIST ROCK BURST ….21

10 COMMENTS ON TUNNELLING METHODS THROUGH TRANSITION ZONE……23

11 DEFORMATION EXPECTED IN POWERHOUSE…………………………………….26

12 PERMANENT SLOPES AT THE SITE (pending)

REFERENCES……………………………………………………………………………..27

2
Nick Barton & Associates Rock Engineering

1. INTRODUCTION

The writer was requested to provide consultancy services to Astaldi, through the Consorcio
Rio Mantaro, by project manager Jose Alves, on 27th November 2013. Site visits of a few
days duration, at two to three months intervals were proposed. The first of these took place
from 9th February 2014 (6 am start from home in São Paulo on Sunday) to 14th February (6
pm return home in São Paulo). The trip started one day after arrival from Europe, and
terminated one day before departure to Europe.

Travel/airport waiting time was 13 hours (plus 2 hrs time-change) on 9th Feb to Pampas and
4 hours from Pampas to the powerhouse end of the project on 10th Feb. (total charged at
1/3 x 19 hrs). Return on 13th to Lima /and to São Paulo on 14th Feb was 10 hours door-to-
door on 13th, and 11 hours (minus 2 hours time-change) on 14th Feb. (charged at 1/3 x 19
hrs). The consultant was on site for an effective 2 ½ days, with site visits mostly on 10th
afternoon (underground) and 12th afternoon (permanent rock cuttings for roads), and had
time for several meetings and preliminary report writing.

A preliminary report was sent electronically on 15th February, prior to return to Europe. This
addressed the following requested priority topics. They are approximately repeated in early
parts of the present report:

1. Cavern orientation confirmation


2. Cavern bolt length-and-spacing optimization
3. Cavern crane beam design optimization
4. Blast length and unsupported span length in different rock classes
5. Grouting ‘timing’ of mechanically-anchored bolts in different rock classes

3
Nick Barton & Associates Rock Engineering

Figure 1. For the benefit of the writer, with many similar projects during several years, the main
underground construction aspects of the Rio Mantaro are summarized in these two photos.

Cavern orientation confirmation

The machine hall and associated parallel and perpendicular intersecting excavations have
been oriented with attention payed both to rock mass structure and to major rock stress.
The foliation is the obvious dominant feature, and the cavern is optimally oriented nearly at
right-angles. The cavern is only at some 350 m depth, so despite some discussion about
over-coring stress estimates in relation to modulus, the orientation with respect to stress
would hardly be the dominant issue. But in fact, both the orientation with respect to
structure (approximately perpendicular to the cavern axis) and the orientation with respect
to the major stress (approximately perpendicular to the cavern axis) are close to optimal.

2 CAVERN BOLT LENGTH-AND-SPACING OPTIMIZATION

There is the possibility of some significant cavern support optimization. This is partly in
response to empirically-based norms, and also in response to the quite high rock mass
qualities observed by the writer during the site visit, and already logged in the cavern by the
consultant, Lombardi. This favourable impression includes some sections in cavern access
tunnels with numerous half-rounds visible (e.g. Figure 2) and the need for only spot-bolting,
which in places approach systematic bolting. No shotcrete is used or needed in these
sections. As far as is understood, Q-values have been logged in the present cavern
excavations which lie in the range of 5 to15 (approx.). This is considered to be on the
conservative side.

4
Nick Barton & Associates Rock Engineering

Figure 2 Examples of the rock mass quality. The Q-value usually exceeds 50 when so
many half-rounds are visible.

However conservatism does no harm when seismic design is considered. As will be shown
later, the rule-of-thumb use of 2 x SRF (giving Qdynamic = ½ Qstatic) guarantees a 25%
increase in support capacity in case of the passage of dynamic waves (See Barton, 1984

5
Nick Barton & Associates Rock Engineering

and Figure 4). This has been used successfully in the past, including in a much larger
cavern in a more seismic region (Taiwan). It has also been ‘put to the test’ in a major
earthquake in this fault-and-clay infested cavern (Barton,1994), and performed well, i.e.
reportedly prevented shotcrete cracking as a result of seismic shaking.

The Q-system case record precedent practice data base, which now includes more than
2,000 cases (210 in 1974, 1050 in 1993, about 800 in 2002) suggests 5 m long bolts for 20
m spans. This is shown on all published Q-support diagrams from these different periods,
and is confirmed by the empirical formula L (arch bolts) = 2 + 0.15 SPAN/ESR (m). ESR = 1.0
is the high security number for powerhouses, used during the last 40 years. Checking:
2 + 0.15 x 20/1 = 5m.

Concerning cavern walls, the whole cavern height is used for recommendations of bolt
length. The empirical formula is L (wall bolts) = 2 + 0.15 HEIGHT/ESR. With the 38 m height
rounded to 40 m, this formula suggests 8.0 m length, exactly as designed. So concerning
the optimal length of systematic bolting in the cavern, 5 m is recommended in the arch
(shorter than at present) and 8 m in the walls (the same as at present).

Concerning the systematic spacing of these rock bolts, there will be a range of possibilities
in relation to an assumed rock mass quality round-figure range of 4 to 20. There will also be
a range of S(fr) thickness varying with this range of quality. To help with the communication
of these possible bolt spacings and shotcrete thicknesses, which of course will be simplified
to minimize variation, it is helpful to refer to Figures 3a and 3b.

As can be noted from both the figures, provided the shotcrete S(fr) quality and washing
routines are acceptable, the thicknesses of S(fr) may vary from 5 to 7 cm in the arch and 6
to 10 cm in the wall. Correspondingly, the bolt c/c in the 20 m span arch of the cavern can
vary from 2.1m (Q=4) to 2.5m (Q=20).

Concerning wall spacing of rock bolts it is necessary to refer to Tables 1a and 1b. Using the
conservative Qw = 2.5 Q, one can observe that a suitable mid-range bolt spacing in the
walls should be increased from the present 2.0 m to 2.5m. This is because the upper
rectangles in Figures 3a and 3b need to be shifted to the right, to Q = 10 to 50, in place of 4
to 20 (due to the x 2.5 empirical correction).

A simple ‘standard support’ (on the conservative side ) for arch and walls could therefore be
the following: ARCH: B c/c 2.3 m, L = 5m, S(fr) 6 cm. WALLS: B c/c 2.5m, L= 8m, S(fr)
8cm. Note that 5 cm of well bonded shotcrete with 1 MPa bond strength resists 100 t/m 2.

6
Nick Barton & Associates Rock Engineering

Figures 3 a and 3b. The Grimstad and Barton, 1993 and Grimstad 2007 Q-support charts. Cavern
‘location’ rectangles have been placed at Q = 4 to 20, and SPAN = 20m, and HEIGHT ≈ 40m. As
described in the text, the WALL rectangle is often placed to the right, if the Q-value is sufficient, as
walls traditionally have wider bolt spacing and less thick shotcrete.

7
Nick Barton & Associates Rock Engineering

Figure 4 Seismic design ‘rule-of-thumb’ for allowing a 25% increase in loading capacity as a result
of the passage of dynamic (as opposed to the static) stresses acting around the cavern. Note in this
connection the very high capacity of the 32 mm bolts. Already there is built-in conservatism,
because the required support pressure capacity is not high in Q = 4 to 20 qualities.Barton, 1984.

Table 1a and b. The empirical rules for bolt lengths and wall support. From Barton et al., 1974.

In Figures 3a and 3b, it has been conservatively


assumed that Qw = 1.0Q – in other words the same
basis for wall support as arch support. In reality wall
support should be significantly less (due to Qw ≥ 2.5Q.
This shifts the upper ‘rectangles’ in Figures 3a and 3b
over to the right. It is normal to have less wall support
unless there are (wedge-) stability reasons for heavier
support. Note that anchors are not usually needed in
caverns excavated in these fair to good rock qualities.

8
Nick Barton & Associates Rock Engineering

3 CAVERN CRANE BEAM DESIGN OPTIMIZATION

The present ‘design’ appears very strange to several engineers who have seen it, as gently
downward-oriented bolts and cables do not optimize the intended function of bolts and
cables, putting them in pure shear and lower ones even in slight compression. This opinion
is given because the optimistically drawn ‘ledges’ will be almost impossible to guarantee
without ‘sculpting’ (low-energy, tight-hole spacing) blasting in the spring-line beneath the
arch. Therefore the downward component of self-load and the moving crane/turbine load
will need to be compensated by an upward force component, not a downwards one.

In Figure 5 some lecture over-heads are reproduced showing the theory (of related wedge-
in-cavern support) and bolt capacity optimization. In Figure 6 and a sketch of a practical
crane-beam solution/recommendation is also given. Bolts or anchors under combined shear
and tension, and inclined at mobilized friction angles (35 to 45°) are known by testing and
theory to have much higher capacity than those in pure shear. The latter even risk being
damaged in case there was ‘under-design’ of capacity (which of course there is not).

Concerning design capacity, the 18m ‘cone’ concept for supposedly resisting ‘pull-out’ is an
imaginary concept that is not accepted by those with rock mechanics shear strength
experience. It is of course necessary to avoid (i.e. to sufficiently anchor) actual physical
wedges in the wall behind such a crane beam, if such were found. A large-scale field
experiment in jointed shale using 1m, 2m, and 4m long bolts failed to reach any measurable
deformation with anything but the 1 m long 32 mm bolt. Bond failure occurred at 48 tons,
but with only 1mm of total deformation, despite the presence of two parallel joint sets that
defined blocks expected to pull out or at least ‘yield’ in the loaded direction. The blocks did
not deform due to slight shear-induced dilation and therefore increased normal stress. The
same will occur in case of ‘blocks’ beneath the crane beam.Consequently, and to be very
conservative, 2 and 4 m long bolts at 1 m spacing, in a rock several times stronger than the
above referred shale, tensioned and fully grouted, and with the site-norm of 32 mm
diameter, are going to provide gross over-capacity in relation to a 150 m vertical live load.

9
Nick Barton & Associates Rock Engineering

Figure 5 a and b. Concerning the optimal support of an unstable wedge in the wall of a cavern, the
force diagram in a) shows ‘closure’ when the anchors or bolts are inclined (upwards) so that the
reinforcement is under combined shear and tension. In pure shear (or even some compression) as
presently ‘designed’, bolt or anchor capacity is minimised. Numerous shear tests of bolted blocks in
large-scale direct shear tests show (such as this from Bjurstrøm, 1974) an optimal capacity when
‘alpha’ is about 35 to 40°, i.e. the mobilized friction angle.

Figure 6 A sketch showing the principle of crane-beam anchoring sufficient to give a factor of safety
of at least 2.5. (Note: > 400 tons/m on both sides).The impractical (and nearly impossible) rock
ledge will not be available in practice due to inevitable blast damage and sometimes also the
presence of adverse joint orientations in the ‘wrong’ (unwanted) locations. It is therefore best to
make a design excluding such features. The proposed principles of design sketched here guarantee
a high factor of safety, and of course will be much easier (i.e. also much faster) to construct than the
present unusual solution.

10
Nick Barton & Associates Rock Engineering

4 BLASTING ROUNDS and SUGGESTIONS for UNSUPPORTED SPAN

A regular practice used to guide practical blast lengths in relation to Q-classes, which
ensures a sensible balance between the desire to make tunnelling progress where rock is
good, and to prevent stability problems when the rock is bad, is shown in the second
column of Table 2.

Note that the apparent ‘linearity’ is in reality strongly log-linear due to the logarithmic style of
the Q-value classes. The suggested ‘unsupported span’ (the Bieniawski dimension shown
in Figure 7b) is shown in the third column, and involves a simple squaring of the relevant
blasting-round length. Due to a likely mix of qualities – such as Class C following Class B
as the tunnel is advanced, there will be a series of ‘support pending’ jobs to be done
following the face advances.

Table 2. A simple set of easily remembered and relevant criteria concerning blast length. Also
included are suggestions for unsupported longitudinal span. This is the longitudinal along-the-
tunnel-axis dimension between the tunnel face and the first support, as shown in Figure 7b.

Rock class Q Recommended Unsupported span


blasting length (longitudinal)
(m)

A > 40 5m 25m
B 5 - 40 4m 16m
C 0.5 - 5 3m 9m
D 0.05 - 0.5 2m 4m
E 0.005 – 0.05 1m 1m

Definition of SPAN in the Q-system Bieniawski’s UNSUPPORTED ROOF SPAN

Figure 7 a and b. The ‘span’ definitions in the Q-system (left) and the Bieniawski RMR methods.

11
Nick Barton & Associates Rock Engineering

Blasting responsibilities and overbreak

Since it is the contractor/consortium who are responsible for the excavation progress, they
are logically also responsible for blast-hole layout. Of course it is possible that at times an
Owner can feel that over-break ‘as a result of’ blasting is excessive. However, as may be
noted in Figure 8, it is impossible to control over-break when adverse jointing is present. By
chance, both the photographs in Figure 8 are of a cavern excavation in granites. The ratio
of Jn/Jr (number of joint sets/joint roughness) needs to be lower than 6 to avoid marked
over-break, because then block formation and release due to too smooth joints is
compromised. The Jn values of 6, 9, 12 and 15 seen in the inset in Figure 8 mean
respectively two sets of joints plus random, three sets, three sets plus random, and four
joint sets.

Figure 8 Despite careful blasting, it will be difficult to prevent unwanted excessive over-
break if the ratio Jn/Jr is greater than or equal to 6. Barton, 2013.

5 GROUTING ‘TIMING’ FOR MECHANICALLY-ANCHORED BOLTS IN


DIFFERENT ROCK CLASSES

The question of grouting ‘timing’ of mechanically-anchored bolts in different rock classes


is difficult to answer without some doubt remaining, because several factors play a role in
the relevance of the suggestions. For example – can one guarantee the mechanical
anchors hold tension in the poorer rock qualities? The right-hand column in Table 3 is a
suggestion for defensible ‘delayed’ grouting of the bolts.

12
Nick Barton & Associates Rock Engineering

The suggested ‘unsupported span’ (the Bieniawski dimension shown in Figure 7b) is shown
in the second column of Table 3, and involves a simple squaring of the relevant blasting-
round length. Due to a likely mix of qualities – such as Class C following Class B as the
tunnel is advanced, there will be a series of ‘support pending’ jobs to be done following the
face advances.

Table 3. A simple set of easily remembered and relevant criteria concerning unsupported
longitudinal span, and the probably acceptable ‘delay’ in grouting tensioned bolts.

Rock class Q Unsupported ‘Delay’ dimension for


span grouting tensioned
(longitudinal) bolts

A > 40 25m 100 m


B 5 - 40 16m 50 m
C 0.5 - 5 9m 25 m
D 0.05 - 0.5 4m 10 m
E 0.005 – 0.05 1m 5m

Figure 8 A tunnel photo from the Rio Romaro HEP, which can be used for visualizing the
‘delayed’ fully-grouting of end-anchored tensioned 32 mm bolts. There is no problem with
‘delay’ when rock quality is good. Problems arise (with maintained end-anchored tension) if
the rock mass quality is e.g. compromised by some clay in the joints (e.g. classes D and E).

13
Nick Barton & Associates Rock Engineering

Figure 9 The consortiums arrangement of end-anchored bolts, which can be grouted at a


certain distance from the tunnel face, according to the suggested guidelines in Table 3.

6. SHOTCRETE APPLICATION, THICKNESS, CLASS A and B

Table 2 will logically apply to the application of shotcrete in relation to the (Bieniawski)
longitudinal span-without-support. The different possible thicknesses of S(fr) in relation to
rock quality and tunnel / cavern span (the ‘diametral’ span) are shown in Figures 3a and 3b.

It is normal to apply shotcrete in layers of e.g. minimum 4 or 5 cm. Less than this causes
the shotcrete to dry out too quickly before it has had a chance to harden under the intended
moist conditions. The result is a low strength product which may eventually fall off the
tunnel wall, especially if applied on flat joint planes where it has little to adhere to in the way
of rough surfaces.

Figure 10 Note that insistence on a ‘perfect’ circular/ ‘horseshoe’ tunnel profile may involve a lot of
unwanted shear stresses in the shotcrete in the ‘45°’ haunch areas (see arrows). If the rock
structure (especially when bedded/jointed) causes a natural tendency to over-break, the tunnel will
be more stable with the over-break and even an eventual local flat roof, provided the shotcrete is
adequately bolted. Shotcrete, being a thin shell, is not designed for high tangential stresses. Its
primary purpose is to hold jointed rock blocks in place, thereby maintaining the natural cohesion.

14
Nick Barton & Associates Rock Engineering

Figure 11 Note that S(fr) tolerates close placement to the next blast because it is in a semi-plastic
early-strength state. The same applies to the grouting in rock bolts. These comments apply to the
poor quality rock masses requiring only a short delay/or short unsupported roof span.

7. FIBRE-TYPE, DRAMIX OR POLYPROPYLENE

The following comments on shotcrete reinforcement fibre type are of a general nature, but
were stimulated by the next topic concerning control of stress-fracturing and possible rock
bursting in the deepest parts of the headrace tunnel. Here it will be an advantage to use the
longest and most energy-absorbing S(fr), which in fact has proved to be shotcrete
reinforced with long, polypropylene fibres (often ca. 48 x 1.4 x 0.4 mm). These fibres
should have a distinct roughness for best energy/deformation absorbtion. They are used
increasingly in road tunnels, and where significant deformations are expected.

Figure 12 The desireable type of polypropylene fibre with distinct roughness (Barchip Kyodo).

15
Nick Barton & Associates Rock Engineering

Figure 13 A comparison of load-deformation characteristics of shotcretes reinforced with Dramix


steel fibres (25 or 35 kg/m3) shown on the left, and of shotcretes reinforced with Barchip Kyodo
polypropylene fiber with 6 or 8 kg per m3 dosages, shown on the right.

It can be noted that the polypropylene fiber causes a higher post-peak resistance to
deformation, meaning higher energy absorbtion. This of course is positive when used on
the tunnel face in the case of burst or stress-fracturing prone rock, as tunnel workers who
are drilling the next round, or performing other necessary operations, have an increased
measure of protection. It is also easier to muck-out after blasting (if placed on the face).

The writer is aware of longevity concerns of some consultants. Nevertheless, there has
been quite a significant move/change internationally to use PP fibres in the case of road
and rail tunnels (where fire is possible) and where larger deformation or corrosive
environments are expected, such as deep sub-sea tunnels.

8. STRESS FRACTURING –– DESTRESSING PLANS

Figure 14 reproduces the headrace tunnel longitudinal profile. This is for the writer’s benefit.
On the occasion of this first site visit there was no time scheduled to travel to the upstream
end of the project, so progress was not seen. A short section (top-heading only due to
jumbo availability in powerhouse area) was seen at the downstream end. For reference this
is shown in Figure 15.

16
Nick Barton & Associates Rock Engineering

Figure 14 A reproduction of the headrace tunnel longitudinal section for the benefit of the writer.
The maximum cover which reaches about 1,400 m, is certainly within the range where significant
stress-fracturing and rock bursting can be experienced, even in the case of tunnels driven, as here,
predominantly in quite hard granites and gneisses. It is recommended that samples are collected by
the engineering geologists, at e.g. 100 m intervals for point-load testing and occasional uniaxial
compression testing (the latter after coring of cylinders from the recovered blocks). Naturally one
should try to avoid samples with obvious micro-cracking, as these may give false, lower values of
strength.

17
Nick Barton & Associates Rock Engineering

Figure 15 Short downstream section (top-heading) of headrace tunnel. This of course is being
driven full-face from the upstream end, where the largest drill-jumbo does not have ‘power-house
duties’ as is the case at this end.

The reason that the rock (reportedly) already shows signs of possible popping and stress-
fracturing at the upstream end, well before reaching the maximum depth of cover is easily
explained by the combined use of ‘Table 6’ from the Q-system of characterization, and
Figure 16.Table 6 concerns SRF, the ratio of stress and strength. The data in Figure 16 is
independent experiences from mining and deep nuclear-waste research tunnels. Both have
been tested numerous times in the writer’s consulting practice and they appear to be
realistic, which is to be expected since developed from empirical performance-based
experiences.

On the basis of case records of about fifteen deep road tunnels in Norway, where maximum
tangential stresses (σθ) were mostly estimated to be in the range of 50 to 100 MPa, and
from some even higher stress experiences in China, the recommended SRF (stress
reduction factor) shown in Table 3 were developed for excavations in massive, burst-prone
rock masses (Grimstad and Barton, 1993). Note that the maximum tangential stress in the
case of massive rock is of the order of 3 σ 1 + σ3. At e.g. 1,200 m depth the vertical
(maximum) stress might be close to 30 MPa (approx. density: 2.65 x 1,200 / 100 MPa) and
the minimum might be 20 MPa. The maximum tangential stress would then be 3 x 30 + 20
= 110 MPa. This may be close to the UCS of a reduced strength granite, so stress-
fracturing would be expected, since the maximum stress needs only to be about 0.4 to 0.5 x
UCS for stress-fracturing to start. If the granite was 200 MPa it might resist fracturing.

18
Nick Barton & Associates Rock Engineering

Independently from the above SRF update dating from 1993/1994, and this time coming
from the field of mining as opposed to deep transport tunnels, the collection of case records
shown in Figure 16 also shows stress-fracturing initiating when the stress/strength ratio
σθ/σc exceeds 0.4-0.5

There is some controversy concerning the reason for the stress-fracturing starting already
when the maximum stress is ‘only’ 0.4 to 0.5 x the UCS (laboratory-scale uniaxial
compressive strength). A simple-minded explanation is that there is believed to be a strong
Weibull-based scale effect on UCS as sample size increases: the more flaws/microcracks,
the lower the compressive strength as size increases. A value of lab-scale UCS x 0.4-0.5
can be an approximate in situ estimate of large-scale strength.

Table 4 The 6th Q-system table of parameter-ratings concerns SRF, in which the ratio of rock stress
and rock strength (or vice versa) determines whether there is likely to be stress-induced fracturing.
Note the ‘acceleration’ that occurs with the SRF value when the ratio σθ/σc exceeds 0.4-0.5. From
Barton and Grimstad, 1994.

As written beneath the longitudinal section of the headrace tunnel (Figure 15), some rock
block sampling is recommended, in order to measure some tensile (point-load) strengths,
and a smaller number of UCS tests on cored cylinders. Of course if over-coring stress
measurement could be performed (using e.g. a mining consultant for Peru) working some
distance back from the advancing tunnel face (i.e. no delay involved), a better basis for
estimating stress-fracturing would be possible.

19
Nick Barton & Associates Rock Engineering

Figure 16 The depth of stress-fracturing / potential fall-out of fractured rock, as a function of the ratio
of maximum (estimated) tangential stress compared to the uniaxial strength of the rock. A rough
estimate of the potential depth of stress-fracturing is given by the dimension (Df – a). An example of
about 3 m (from Ita HEP in Brazil) is illustrated in Figure 17. If the vertical stress was maximum, the
stress-fracturing in this figure would be at ‘3 o’clock and 9 o’clock’ (not 6 o’clock and 12 o’clock’).

Figure 17 Left: A deep headrace tunnel in marble, in which the maximum stress was vertical,
hence the stress-induced fracturing in the walls. Right: An example of stress-induced fracturing and
significant fall-out / ejection (also in the invert) in a large diversion tunnel in basalts (Ita HEP) where
the largest stresses were horizontal and approximately perpendicular to the tunnels.

20
Nick Barton & Associates Rock Engineering

Figure 18 Some numerical models showing the possible types of stress-fracturing in respectively
massive or jointed rock. These show the effect of a nearly vertical maximum stress on the left, and
the effect of a steep valley-side effected result on the right, with the major (sub-vertical) stress
coming in the ’10 a.m. to 4 p.m.’ direction.

It is understood that there are plans to use de-stress blasting by drilling longer holes ahead
of the tunnel, and charging these in order to pre-fracture the rock, in order to partly ‘de-fuse’
the high stresses. This is considered a wise move. Other measures, such as use of long-
fiber S(fr) on the tunnel face (already discussed) and ‘D-bolts’ (next section) may also be
adviseable.

9. SUPPORT AND REINFORCEMENT MEASURES TO RESIST ROCK BURST

Figure 19 The writer does not have personal experience of the use of ‘D-bolts’ but knows and
respects the Chinese professor inventor, who developed the method in the LKAB mines of Northern
Sweden, and has since become the rock engineering professor in Trondheim, Norway. The
principle is very sound and is explained in the ‘box’ which follows (Figure 20). It has been adopted in
the mining activities of several countries in the last two to three years.

21
Nick Barton & Associates Rock Engineering

Figure 20 The ‘D-bolt’ for safer control of dynamically yielding rock.

Figure 21 Comparisons between the load-deformation characteristics of common rock bolts. The ‘D-
bolt’ shows superior load resistance, and at least 3 to 4 times the capacity of the split set.

22
Nick Barton & Associates Rock Engineering

It is recommended that the ‘D-bolt’ is used for the standard radial bolting, if stress fracturing
and dynamic release of rock becomes common. There is also the possibility of face-control
bolting, using fiber-glass bolts whose yield characteristics are also superior to standard
grouted-rebar bolting, and they have the necessary added advantage of not complicating
blasting and muck removal.

10. COMMENTS ON TUNNELLING METHODS THROUGH THE TRANSITION


ZONE

The inevitable transition between the igneous and metamorphic rock, shown in an
approximate position in Figure 15, is an ‘unknown’ in terms of rock quality and permeability.
The method presently envisaged involving forepoling and steel sets is ‘conservative’, but
may not be optimal in terms of minimizing delay and ensuring deformation control, in case
of severe fracturing and e.g. water. Probe drilling is advised in case of increased fracturing
and water, in the general area where the transition is anticipated. If flow becomes heavy,
pre-injection at high pressure (> 5 MPa, perhaps up to 10 MPa) might be adviseable.

For reasons explained in Figures 22 and 23, it is not ideal to use steel sets as loosening is
unintentionally allowed, and the frail rock mass conditions can be worsened. In Q-system
terminology, SRF inevitably increases due to loosening. It is suggested that such conditions
(assumed here for the rock transition) and other rock class D occurrences along the
headrace or elsewhere, could be tackled by the RRS method illustrated in Figure 25.

Figure 22 The implicit and practical difficulty of making contact between a steel set and the tunnel
profile (which may show overbreak) is illustrated. The spiling bars are however an effective means
of stabilizing the next blast, in case loosened rock, as here, is encountered in the transition zone.

23
Nick Barton & Associates Rock Engineering

Figure 23 Use of steel sets as a ‘standard method’ when conditions get bad is generally
not recommended. An example was seen in a section of the access to the tailrace tunnel
where conditions had worsened. Steel arches are the most deformable method of
(partly) securing a tunnel profile in bad rock, because they invite loosening as illustrated
above. From Barton and Grimstad, 1994.

24
Nick Barton & Associates Rock Engineering

Figure 24 An example of the use of (subsequently sprayed in) steel arches.Compared to


the more robust (because bolted) RRS system shown in Figure 25, steel sets or lattice
girders are very deformable while they are being fixed, and remain deformable ‘from
behind’ due to the difficulty to make effective contact with the rock / tunnel profile.

25
Nick Barton & Associates Rock Engineering

Figure 25 Examples of the formation of bolted sprayed-reinforced-shotcrete (RRS)


arches, as used in the ‘very poor’ rock quality area of the Q-support diagram (blue
arrow). See details in Figure 3b. This would be a generally recommended method for
rock class D and as a preliminary technique for rock class E, to be followed by local cast
concrete, at least in the case of the exceptional rock class E.

RRS or steel-reinforcing-bar reinforced shotcrete arches, for the next-to-worst categories of rock mass,
e.g. 0.01 < Q < 0.1. 1= first layer of general S(fr) – accelerated with non-alkali additive, 2 = build-up local,
smooth but not necessarily circular arch (or arches) of non-alkali accelerated S(fr), 3 = drill bolt holes at e.g.
1m centres round arch, and install end-anchored bolts, and fix pre-fabricated, welded cross-bars. Grout bolts
before spraying in the bolt heads. 4 = attach (wire and weld) 4-6x16mm reinforcing bar ‘steel-arches’ to each
bolt-head cross-bar (pre-fabricate in bundles, for easier attachment. (Note: these bars can be bent into
overbreak zone, therefore requiring less shotcrete volumes than with e.g. stiff lattice girder), 5 = spray over
reinforcing bars with shotcrete, to complete arch and provide foundation for: 6 = bolts and washer, tensioned
(bolt thread pre-protected with plastic caps. 7 = Spray over bolt heads to complete RRS arch.

11. DEFORMATION EXPECTED IN POWERHOUSE

Rock mass conditions seem to be quite good in the powerhouse area, as presently
excavated and observed during the site visit, with Q-values at least in the range 4 to 20,
and perhaps with many (half-round visible) areas with Q exceeding 40 (‘very good’).

With these levels of rock quality, the deformation expected at the modest depth of about
300 m (with possible vertical/sub-vertical stresses of no more than 10 MPa) will cause
rather limited deformation. In fact as shown in Figure 26, the general approximate trend of
∆ (mm) ≈ SPAN(m) / Q, given by the central trend of hundreds of tunnel deformation data,
suggests that a cavern span of 20 m (approx.) may only deform by 1 mm if the Q-value is a
uniform 20, but could be 5 mm if the Q-value was uniformly as low as 4, which is clearly not
the case.

Figure 26 The central trend of tunnel and cavern deformation data is given by the simplest of all
relationships: ∆ (mm) ≈ SPAN(m) / Q.

26
Nick Barton & Associates Rock Engineering

Figure 27 Empirical improvement over the central deformation trend seen in Figure 26. With SPAN
= 20,000 mm, Q = 10, σv = 10 MPa, and σc = 90 MPa (a very conservative estimate made for
simplicity), the first equation suggests ∆v = (20,000/100.10)(10/90)1/2 = 6.7 mm. It is clear that
deformations in practice will be a very few mm (due to higher UCS) and difficult to measure.

12. PERMANENT SLOPES AT THE SITE (PENDING – separate report suggested)

REFERENCES

Barton, N. 1984. Effects of rock mass deformation on tunnel performance in seismic regions. Proc. Caracas
Symp., Adv. Tunnel. Technol. and Subsurf. Use. Vol. 4: 3: 89-99.

Barton, N. 1994. A Q-system case record of cavern design in faulted rock. 5th Int. Rock Mechanics and Rock
Engineering Conf., Tunnelling in difficult conditions, Torino, Italy, pp. 16.1-16.14.

Barton, N. & Grimstad, E. 1994. The Q-system following twenty years of application in NMT support
selection. 43rd Geomechanic Colloquy, Salzburg. Felsbau, 6/94. pp. 428-436.

Barton, N. 2002. Some new Q-value correlations to assist in site characterization and tunnel design. Int. J.
Rock Mech. & Min. Sci. Vol. 39/2:185-216.

Barton, N. 2013. Integrated empirical methods for the design of tunnels, shafts and caverns in rock, based on
the Q-system. 3rd Int. Symp. on Tunnels and Shafts in Soil and Rock, SMIG/Amitos, 17p. Nov. 2013, Mexico
City

Grimstad, E. & Barton, N. 1993. Updating of the Q-System for NMT. Proceedings of the International
Symposium on Sprayed Concrete - Modern Use of Wet Mix Sprayed Concrete for Underground Support,
Fagernes, 1993, (Eds Kompen, Opsahl and Berg. Norwegian Concrete Association, Oslo.

Dr. Nick Barton


NRBarton
27

Das könnte Ihnen auch gefallen