-
1263
Construction of the desilting chambers for the Nathpa Jhakri
hydroelectric project, India
T.G. Carter & M.J. Telesnicki Golder Associates Ltd., Canada
M.L. Kenny, D.M. Brophy, J.L. Carvalho, D.E. Steels & H.S.
Dhillon Aecon Constructors Ltd., Canada
SYNOPSIS: Construction of the four, closely spaced 525m long,
16m wide, 30m high caverns forming the Desilting Chamber Complex
for the Nathpa Jhakri Hydro-electric Scheme in Himachal Province in
Northern India posed numerous challenges as a consequence of (i)
the difficult rock conditions, (ii) the end use design requirements
and (iii) the physical layout arrangements of the Chambers with
respect to the multiple access tunnels and waterway conduits. This
paper discusses the rock mechanics measures undertaken to safely
excavate and support the complex gneissic and schistose rock mass,
including the reinforcing of several large wedge failure geometries
evident in the curved Chamber sidewalls. The detailed construction
steps taken to develop the staged sequence for excavating the full
Chamber profiles are described, and outlines are presented of the
controlled excavation methods and extensive rock reinforcement
undertaken to create the required curved wall excavation shapes and
preserve the slender pillars between the Chambers. The application
of the fibrereinforced shotcrete lining is discussed. Because of
the difficult rock conditions and relatively high stress state
several detailed 2-D and 3-D numerical modelling analyses were
undertaken to examine the stability of the Chambers as a basis for
reinforcing the Chamber sidewalls, crowns and pillars, taking into
account the numerous inter-connecting access and waterway tunnels
and shafts. The results of the modelling are explored in the light
of observed deformation behaviour of the Chambers during
excavation.
1. INTRODUCTION
Construction of the Desilting Chamber Complex for the Nathpa
Jhakri scheme, which was carried out from 1994 to 2004, involved
excavation of four major Chambers, each of a size similar to that
typical for the main cavern of an underground power station,
(Figure 1). The Chamber layouts and associated tunnel works, which
were principally sized and dimensioned for silt control purposes,
were designed by the Central Water Commission (CWC) for Satluj
Vidyut Nigam Limited (SJVN), a Joint Venture of the Government of
India and Himachal Pradesh, formerly known as the Nathpa Jhakri
Power Corporation (NJPC). As such the layouts of the tunnels and
chambers were optimized principally from a hydraulics viewpoint,
with rock mechanics aspects only considered of secondary
importance. However, in the steep Himalayan topography of the site,
steeply dipping geology
(phyllites, gneisses and schists) dominated rock conditions,
making excavating and reinforcing the Chambers and Intake
structures challenging. From the contractors perspective the main
Chamber constructability issues were wall and crown control, pillar
reinforcement and excavation sequencing, while for the Chamber
Intakes, rockslide stability was of most concern. As shown on
Figure 2, the 1500 MW Nathpa Jhakri Hydroelectric Power Project is
located in a remote northern area of India in the upper reaches of
the River Sutluj in the state of Himachal Pradesh, (HP) almost on
the Chinese border. The project has been implemented at a total
project cost in excess of US$1.2 billion, about a third of which
was funded by the World Bank. Aecon Constructors, through its
wholly owned subsidiary, The Foundation Company of Canada, was
Managing Party of one of the joint ventures constructing the
project, responsible for undertaking two of the main
World Tunnel Congress 2008 - Underground Facilities for Better
Environment and Safety - India
-
1264
Figure 1. Desilting Chamber #3
Figure 2. Project location & scheme layout
-
1265
civil work contracts, totalling $640M. These construction
contracts included the Main Dam, the Intakes and the Desilting
Chambers (Figure 3) and approximately 16km of the Headrace Tunnel.
Throughout the eleven years of the joint ventures contract works,
Golder Associates provided geotechnical and geological engineering
advice, with much of the emphasis on the Chambers.
1.1 Desilting arrangements
As designed by SJVN, the Nathpa Jhakri scheme is basically a
run-of-river power development and was the largest project
identified on the Satluj for harnessing its hydroelectric
potential. A design discharge of 486 cumecs is diverted from the
river into the four gate intake arrangements, by a 61.5 m high
concrete gravity dam, (Figure 4). In an attempt to exclude silt
particles of up to 0.2 mm diameter from the water before it enters
the Headrace Tunnel, the Intakes feed into a complex of four
Underground Desilting Chambers (the largest in the world) through
independent approach tunnels. As is evident from Figure 3, the
Desilting Complex required significant rock engineering input in
order to safely excavate and support the four 525 m long, 30 m
high, 16m wide chambers, which are each lined with steel fibre
reinforced shotcrete, and heavily supported with a surficial
rockbolt anchorage pattern and long cable and bar anchor systems.
Completion of the Chamber complex required removal of more than 1
million m3 of rock. From the Chambers, as per the right hand
diagram on Figure 2, the water runs in a 10.15 m diameter circular
section Headrace Tunnel for 27.3 kms. to terminate in a 21 m
diameter, 225 m deep Surge Shaft. Three Pressure Shafts of 4.9 m
diameter each then take the water from the Surge Shaft to feed the
six Francis Turbine generating units of 250 MW, each set within a
225m long, 49m high and 20m span underground powerhouse, allowing
full utilization of the approximately 425m developed water pressure
head.
1.2 Rock mechanics influences on construction
Numerous technical papers describing various aspects of the
scheme from a geological or geotechnical perspective have already
been published elsewhere, (eg., Kumar and Dhawan, 1999, Dasgupta et
al, 1999, Hoek, 1999, 2000; Bagde, 2000, Mahajan, 2000, Hoek &
Marinos, 2000 and Carter et al., 2005). This paper does not discuss
the overall scheme in any detail, rather it concentrates on
examining the rock conditions encountered during excavation of the
Desilting Chambers by the Continental-Foundation Joint Venture
(CFJV). It is of note that the steeply dipping foliated nature of
the rockmass has been a dominating influence on almost all aspects
of the underground excavation works. As shown in the before and
after cable anchoring photographs of the Desilting Complex Intakes
works (ref. Figures 5a and 5b respectively) the foliation in the
rockmass at the Desilting Complex is pervasive, also controlling
much of the surface topography. Largely because of difficult rock
conditions and access, the Contract works undertaken by Aecon
Constructors as part of Continental-Foundations Joint Venture
(CFJV), which were envisaged to take about 57 months, ended up
taking 131 months to final completion and startup of operation of
all the Generators. The largest delays to progress in the Intake
and Chambers areas occurred because of:
(a) pillar stability issues in the Chambers due to the more
foliated nature of the rockmass than expected and the slenderness
of the inter-Chamber pillars, the rock support designs indicated in
the tender documents needed enhancing. This was accomplished by
installation of three rows of 20m long 60T cable anchors through
massive concrete beams cast on each Chamber wall, together with
over 20,000 supplemental deeper patterned reinforcement elements
(bolts and dowels). . and .
-
1266
Figure 3. Desilting complex
Figure 4. Completed Intakes and dam
Figure 5a. Original rock condition Figure 5b. Desilting area
intakes after cable anchor installation
-
1267
(b) slope stability issues in the Intakes area due to the fact
that the valley side slopes at the intake site were not stable
enough for the required rock cuts needed to reach the structure
foundations, remedial stabilization works were required. To
completion of this works area, eventually 600 cable anchors of 40m
length and 200T capacity were required to be added as additional
support.
2. CHAMBER EXCAVATION
Excavation of the four Desilting Chambers progressed in parallel
as much as possible in order to expedite construction progress and
also to minimize any stress-related rock interaction effects (for
example of having one Chamber excavated significantly ahead of the
adjacent Chamber headings). The initial stage of excavation for
each of the Chambers utilized a crown access, developed by
conventional top heading and benching, with side slashes to develop
the full haunch profile (Segments A and B as shown on Figure 6).
Several of the main adit cross cuts were also constructed at much
the same time as the initial crown drift in order to allow access
to two different elevations within the overall Chamber Complex (ref
layouts shown on Figures 3 and 6). The remainder of the Chamber
profiles were then constructed by benching, again with the
intention of maintaining reasonable sequencing in the benching
operations so as to avoid inter-Chamber interaction problems due to
stress readjustments. Figures 6a and 6b show in section and plan,
respectively, where segments of the top headings and segments of
benching were planned to be underway sequentially. The diagrams
respectively illustrate the staged excavation sequence, as tendered
and one typical stage (time snapshot) during the development (as
modelled for rock mechanics evaluation). Comparison of the tender
layout (Figure 6a) with the inset diagram on the left of Figure 6b,
shows the differences in excavation benching between the tender
proposal and actual (as constructed) staging.
2.1 Crown top headings
Excavation of the Central Pilot Heading of the four Chambers
(i.e., Excavation Sequence 1, Modelling Segment A as shown on
Figure 6a) was completed concurrently over the following
periods:
Pilot Side-Slashes
Chamber 1 4-10-95 to 17-4-97 20-4-97 to 23-9-97
Chamber 2 21-10-95 to 12-12-96 25-3-97 to 17-9-97
Chamber 3 21-10-95 to 10-11-96 20-4-97 to 6-11-97
Chamber 4 21-10-95 to 15-11-96 11-1-97 to 21-7-97
Typically 3.5 - 4.0m rounds were excavated for the top heading
and haunches, and at each stage the rock condition and details of
each face were mapped and classified using standard rock mass
classification methods (i.e., following Barton et al, 1977,
Bieniawski, 1976, Grimstad & Barton, 1995, and/or Marinos &
Hoek, 2000). Such classifications then formed the basis for
definition of blasting charge weights and rock reinforcement.
Figure 7 shows the typical blasting pattern used for the mid grade
rock conditions (Class III, Q = 4-10; RMR76 = GSI = 55-65) with a
typical powder factor in the 1.2 1.6 kg/m3 range. Throughout these
excavations, support in accordance with the Construction Drawing
layouts was installed and regular proof testing of bolts was
carried out by SJVN to ensure the adequacy of the installations.
Due to some secondary grouting difficulties encountered with some
of the crown installations in the driving of the Pilot Headings;
and, with the approval of SJVN, cement cartridges were substituted
for uphole reinforcement, and these agreed procedures were then
utilized throughout development of the side-slashes (Fig. 7).
During the latter stages of excavation of Sequences 1 and 2,
(Modelling segments A & B) SJVN and Geological Survey of India
(GSI) geologists recognized that on a large scale the Chambers were
being crossed by numerous adversely oriented, weak geological
structures (biotite and micaceous schist units, and various shear
zones). The pervasive foliation and multiple shear zones which were
found dipping into the development headings, together with the two
predominant orthogonal joint sets also created a number of local
scale (but still significant) ground control problems during
excavation. In fact, the extent and weakness of the rockmass,
created by these features was deemed so adverse to stability and
operational efficacy that an unprecedented six months period of
shear seam treatment (including additional localized excavation,
cavity filling, grouting and additional bolting) was initiated in
each Chamber, starting in June 1997.
-
1268
The program of shear seam treatment which was started in Chamber
#4 after the main pilot was through and the side slashes were
complete, was then carried out concurrently into each of other four
Chambers over the following periods:
Chamber 1 - 8-01-98 to 9-02-98
Chamber 2 - 6-10-97 to 2-01-98
Chamber 3 - 31-01-98 to 27-04-98
Chamber 4 - 28-09-97 to 21-12-97
Up until initiation of the shear seam treatment, most of the
excavation works had been sprayed with a 50mm layer of plain
shotcrete directly applied onto all agreed final "treated" rock
surfaces. Following successful trials in February, 1998 and
contractual agreement relating to application of the Contractor
proposed Steel Fibre Reinforced Shotcrete (SFRS), the first
production phase of "final" SFRS lining was initiated under SJVN
direction in Chamber #1 in March 1998. Application was conducted
under skilled control of experienced shotcreting crews with regular
quality control panel and spray test cores being routinely checked
and approved by SJVN's testing laboratories. SFRS lining
application, which proved fundamental to maintaining the integrity
of the near-surface zone of the rockmass, was completed within each
of the four Chambers over the period from March to July 1998.
2.2 Benching
As soon as the full SFRS lining had been placed down the Chamber
sidewalls as far as permitted by SJVN (2.5m above the then invert)
for the Sequence 1 and 2 crown excavation, SJVN then allowed
initiation of Sequence 3 benching, (Modelling Segment C). This
bench excavation between RD 0 and RD 490 was carried out within
each of the four Chambers over the following periods:
Chamber 1 - 8-06-98 to 17-12-98
Chamber 2 - 13-08-98 to 18-06-99
Chamber 3 - 25-06-98 to 1-03-99
Chamber 4 - 21-10-98 to 23-10-99
During execution of this sequence over the period January 1999
to May 1999, difficulties were experienced with sidewall profile
control in the downstream portions of several of the Chambers, in
particular Chamber #3, due in large part to the adverse orientation
of the prevailing geological structure. To better control rockmass
behaviour, Sequence 3 and indeed all excavation of the lower parts
of the Chambers, including the hoppers, was developed by
conventional benching, with in the early phases the centre zone
drilled and charged vertically from the bench above, and the more
critical sidewall zones excavated using parallel development
rounds, drilled longitudinally. Later on, after completion of
sidewall cable anchoring, when excavation actually recommenced, all
further bench development was carried out only by horizontal
drilling, typically as indicated in Figure 8. Even with these
measures, significant wedge shaped overbreak zones commonly
developed due to the unfavourably oriented 50 dipping foliation.
Consequently, several methods of benching development were
attempted (including central pilot and slash, with parallel,
longitudinal drilling). Further measures to control wedge and block
slide releases from the sidewalls were also implemented, including
installation of downward inclined reinforcement into the lower part
of the Sequence 2 sidewalls, thereby adding some support to the
walls alongside the next bench zone to be blasted. However, as
these additional reinforcement members could only be installed into
the corner zones of the walls and approval from SJVN was not given
for placement of these elements as countersunk installations
through the next bench blast invert, these measures
-
1269
Figure 6a. Chamber excavation sequence (as tendered)
Figure 6b. Stage 3 of planned development sequence (as modelled
in examine 3D to check stress-interactions)
-
1270
Figure 7. Typical pilot blasting pattern
Figure 8. Hopper portion Drilling pattern & initiation
sequence
-
1271
only achieved marginal wall profile improvements. As the
problems of sidewall control were not completely resolved by the
above methods employed during the benching of Sequence 3 (height
5.0 m), in the next bench (Sequence 4) in addition to maintaining
the central pilot and slash excavation sequence, bench heights were
further reduced to 3.8 m. For both Sequence 4 and Sequence 3, SFRS
placement followed sequentially with the bench blasting and the
regular program of shear seam treatment, as per the procedures
developed and agreed with SJVN during the previous treatment
program. It is of note in this regard that throughout benching,
approval for final application of SFRS was given by SJVN only after
completion of geological mapping, and shear seam identification and
treatment. Final SFRS placement was therefore delayed somewhat,
allowing some further wall loosening before the remainder of the
Sequence 2 and 3 sidewalls were permitted to be sprayed. Sidewall
overbreak issues continued to dominate wall profile problems, until
some improvement could be achieved when the lower limit of SFRS was
revised by SJVN to allow placement to within 1 m of the bench
horizon. SFRS was thereafter applied to all the walls of Sequence 3
excavation in phase with the development progress, but still with
some lag for shear seam treatment behind the excavation face,
during the following periods.
Chamber 1 - 3rd Aug., 1998 - 8th Mar., 1999
Chamber 2 - 29th Sept., 1998 - 13th July, 1999
Chamber 3 - 16th July, 1998 - 7th June 1999
Chamber 4 - 18th Dec., 1998 - 31st Oct. 1999
Although all of the walls were sprayed in each sequence with
SFRS, all as part of the lining operation, any zones of unstable
rock, identified by CFJV or SJVN during initial development were
immediately then sprayed with a fresh layer of 50mm SFRS.
Excavation and support, including application of SFRS, of Bench
Sequences 4 and 5 then continued over the period from June, 1999 to
May, 2000 maintaining the lesser bench height and central gullet
and side slash sequence with regular approval checks from SJVN.
However cracking problems through the lining continued and at this
time, all bench excavation was halted following a
third major wedge failure close to the intersection of Adit 2
and Chamber #4, (ref. Figure 3 and 6b for location). Up until this
halt on excavation, sidewall bolting was being placed in sequence
behind the excavation, with secondary grouting of bolts being
accomplished using pumped in cement in sub-horizontal and downward
inclined holes, with cement cartridges being used in any required
up-holes.
3. GEOLOGICAL RE-EVALUATION
The onset of significant cracking in the SFRS lining in several
of the Chambers and the failure of three almost identically shaped
large rock wedges (ref. Figure 9b for geometry of one of the
failures) highlighted the need for re-examining prevailing
geological conditions within the Desilting Complex. Detailed
geological mapping by SJVN staff and staff of the Contractor showed
that the Chamber area, like the Intake cuts (Figure 5a), was
dominated by north-east striking, north-west dipping foliation,
interacting with several other cross-cutting steeply dipping joint
fabrics, and that these discontinuity fabrics had played a major
role in the wedge slides.
3.1 Foliation
The intensity of foliation in the rockmass forming the sidewalls
of the various Chambers varied markedly depending on rock type,
with zones of the most intense fabric exhibited in the more
schistose sections of the rockmass, as compared with a much more
gneissose texture in the more competent rock units of the Chamber
complex. In the vicinity of the large sidewall wedge failures
foliation was well developed, but not especially schistose. It
however had a controlling influence on wedge geometry, as can be
inferred from the photograph and isometric diagrams in Figure 9b,
(note, for scale, the people standing in the Chamber to the right
of the wedge). Detailed examination of the individual wedges and
plotting of stereonets of the controlling jointing showed that the
configuration of bounding jointing was quite complex. The failure
block in all three cases was found to have been created by the
intersection of no less than 5 joint sets with the Chamber
sidewalls, as follows:
-
1272
Joint Set Dip Dip Direction
J1 (Foliation) 57 352
J2 60 173
J3 82 098
J4
Main Cross Joints
60 253
J5 (Top release) 30 229
As indicated in the above table these five joint planes, which
formed the block margins as illustrated in the isometric diagrams
on the right hand side of Figure 9b, can be subdivided into three
groups:
the foliation, on which the wedge slid,
the three steeply dipping major cross-joints that bound the
subvertical sides of the block, and
the 30 low angle joint fabric that formed the top release plane
of the wedge.
In fact, several families of similar structures were mapped (as
shown on Figure 10), extending across the entire Chamber complex,
potentially defining a whole suite of wedges and combinations of
wedges transecting the inter-Chamber Pillars.
3.2 Pillar conditions
Figure 10 shows in section and plan the typical geological
characteristics of the rockmass surrounding the Chambers and
constituting the core of the pillars separating the Chambers. As is
evident from Figure 10a, the cross-sectional dimensions of the
inter-Chamber pillars varies markedly because of the curved profile
layouts of the Chambers as required for hydraulic reasons. The
pillars are nearly 60m wide at the Chamber crowns and Hopper chutes
but reach minimum widths of only some 30m at mid-height. Because of
the size of the wedge failures that had occurred in the Chambers,
and as it was considered that any further, possibly progressive,
wedge release could compromise the stability of the pillars between
the Chambers, a detailed phase of geological structure mapping was
initiated by GSI geologists on behalf of SJVN and a major program
of numerical modelling and analysis
was undertaken both by the contractor and the designer. Based on
the mapping, the zones highlighted as being of most concern were
where adverse combinations existed of steep cross-jointing
cross-cutting the NE-SW striking foliation. As the foliation,
although typically dipping NW at about 55 was found to swing up to
30 in strike and vary in dip by up to 15 or so degrees, a variety
of different wedge geometries were possible in different segments
of the Chamber sidewalls.
4. NUMERICAL MODELLING
In order to examine these possibilities in some more detail,
several phases of numerical modelling were conducted by Golder on
behalf of the Contractor and by NIRM on behalf of SJVN. Modelling
was directed at answering several different questions at various
stages of the project, namely excavation sequencing,
stress-interaction effects and pillar stability issues. Although
the purposes of the modelling at the various stages was different,
results from all of the modelling phases were found germane to (a)
furthering understanding of the behaviour of the rockmass, and (b)
developing appropriate support layouts to eliminate further wedge
instability issues. For the Chamber sidewalls and pillar zones
initial reinforcement layouts (as per the Tender drawings) required
5m and 6m bolts as the standard pattern of routine surface zone
support, but not specifically providing deep reinforcement into the
pillar cores. With the pervasive nature of the foliation, the
degree of geologically controlled overbreak of the sidewalls due to
the 50 foliation dip and other cross-cutting pervasive jointing,
pillar thickness concerns and sidewall stability issues suggested
that much deeper, heavier reinforcement patterns would be needed to
completely stabilize the pillar core zones. The fact that three
major wedge failures, each up to 15m high and 8m deep had occurred
was further impetus to re-evaluate the need for deeper,
through-pillar reinforcement and more comprehensive integral
sidewall support. The fact that these failures developed deeper
than the surface support, and involved release on foliation and
major cross-joints prompted a further phase of very detailed
numerical modelling to examine the influence of potential wedges on
overall pillar stability. The modelling work undertaken by
Golder
-
1273
on behalf of CFJV (which used the Examine2D or 3D, UDEC, Flac
and/or Phase2 codes) was directed towards checking the magnitude of
deeper, heavier reinforcement necessary to ensure stability for
excavation, while the work undertaken by NIRM on behalf of SJVNs
design team (principally using 3DEC) concentrated on Chamber
stability under operating conditions.
4.1 Initial 2-D modelling assessments
Early phases of Golder Associates modelling studies for CFJV
were targeted at examining optimum construction sequencing that
would minimize excavation difficulties and assist in maximizing
excavation progress throughout the benching sequencing (ref.
Figures 6a and 6b). On the basis of these studies, recommendations
were made regarding preferred sequencing. These modelling studies,
which were carried out either in Examine2D or PHASES concluded
that:
sliding failure problems would be more prevalent in the west
(right) walls of each Chamber, with buckling and delamination
problems more prevalent in the east (left) wall,
zones of the most adverse potential distress would occur in the
west (right) haunch, and in the vicinity of either side of the
desilting drift (Hopper area),
optimum wall control of benches required pre-support of the
sidewalls, as foliation dips were steep enough to allow wedge
sliding and/or fall-outs immediately on excavation, prior to being
able to place sidewall reinforcement, and
excavating Chambers 1 and 4 ahead of Chambers 2 and 3, stress
shielded the inner chambers, thereby minimizing differential
displacements in the pillars between all the interior chambers.
4.2 Initial 3D sequence modelling
The initial 2D modelling analyses concentrated on overall
excavation sequencing and stress-interaction effects, but did not
specifically examine behaviour of the individual Chambers to
benching. These analyses were therefore supplemented by specific 3D
evaluations aimed at checking intersections and benching in order
to optimize pilot bench drift and sidewall slash approaches. To
achieve this, two major series of 3D modelling were undertaken
using the Examine3D code, one looking at intersections, the other
examining sequencing within a single Chamber. Example outputs from
the two series of models are shown in Figure 11. These modelling
studies, which were mainly undertaken using a multi-chamber
configuration for the 1995 analyses and a single chamber,
multi-staged sequence for the 1996 analyses, concluded that:
major interference problems could occur for many of the
intersection areas of the construction access drifts (Adits 1, 2
and 3) with the Chambers, with the Adit 2 configuration,
potentially being the worst,
minimal adjacent Chamber influence during benching was achieved
by keeping benching ahead in Chambers 1 and 4, over those in
Chambers 2 and 3, with most adverse interaction occurring when
bench faces corresponded with the foliation trend, ie, Chamber 4
leading and Chamber 1 trailing,
optimum bench sequencing was achieved by excavating from the
upstream end to the downstream end in order to allow early
attention to reinforcement installation in heavily foliated areas,
(ie, ensuring that, in general, the foliation always would dip out
of the bench face being excavated),
pilot drift development during benching could be taken up to 25m
ahead of the sideslash excavations for a pilot and slash approach,
without adversely affecting sidewall behaviour, and
-
1274
Figure 9a. Normal sidewall chamber conditions in vicinity of
Adit 2
Figure 9b. Wedge Geometry in Chamber #2 Sidewall Zone created by
foliation & major cross-jointing (evident downstream of Adit
2)
-
1275
Figure 10a. Typical Cross-Section of Chambers 3 and 4 showing
complex pattern of cross-jointing and shears in vicinity of Adit
2
Figure 10b. Typical plan detail of geology of chambers &
pillars
-
1276
Figure 11a. Example of detailed chamber intersection zone 3D
modeling
Figure 11b. Example output from Initial 3D modelling for
optimizing chamber bench sequencing
-
1277
potential damage zone depths in weak foliated rock zones
(schists) were double those predicted to occur within the more
competent gneissic materials.
4.3 Supplementary 2D & 3D modelling With ongoing movement
continuing to cause cracking and displacement of the SFRS lining
and with the three major sidewall wedge failures having occurred to
depths greater than the, by then, installed bolting patterns,
concerns regarding wall stability prompted action by all parties to
better understand rock mass behaviour. This was achieved by (a)
increasing the number of installed Chamber convergence arrays to
improve displacement monitoring coverage, (b) installing additional
instrumentation principally sidewall extensometers and load cells)
and (c) initiating a further phase of numerical modelling based on
re-analysis of conditions highlighted by improved geological
understanding of the rockmass. As there appeared to be significant
differences in excavation-related rockmass behaviour along the
Chambers, it was suspected that because of the steep valley
topography (Figure 12a), some of these differences might relate
less to changes in excavation sequencing and Chamber geometry and
more to changes in insitu stress state with distance into the
mountainside. Accordingly the available insitu stress measurement
data for the Chamber area (eg., Kumar et al., 2004) was
re-evaluated in terms of ground surface topography, as summarized
on Figure 12b. Based on topographic surveying of the steep slopes
rising up from the river towards the mountains, the rock crown
cover over the Chambers was found to vary by more than 400m along
their alignment, as per the following Table:
Chamber u/s end (Adit 1) d/s end (Adit 3)
1 West 109m 501m
2 124m 502m
3 141m 506m
4 East 158m 478m
In consequence of these elevation differences, quite significant
changes in insitu stress state across the geometry of the Chamber
complex could be envisaged. Based on Figure 12b, which presents
best estimate fits to available insitu stress
measurement data, more than a doubling of the vertical stress to
the crown of the Chambers occurs at the downstream end of each
Chamber relative to the upstream end, with an almost quadrupling of
the horizontal stress as one moves downstream along the chamber
axes. In addition to these topographically controlled stress
differences across the Chamber complex area, it was also felt that
changes in rockmass stiffness might also exert some influence, due
to some quite significant changes in fracture density and
orientations along the cavern axes, as identified by the geological
mapping (Figure 10). Modelling for this phase of evaluation was
therefore focussed more towards use of discrete element codes that
could replicate the actual mapped fracture patterns rather than
just accounting for changes in fracture intensity by making global
modulus alterations in rockmass properties, as a way to reflect the
mapping information. In order to model the behaviour of the
Chambers to the date of the onset of cracking of the SFRS and the
fall-out of the three major wedge failures, a series of 2D section
and plan models were set up in UDEC by Golder and in parallel a 3D
model of the Chamber Complex was built by NIRM using the 3DEC
program code. As can be appreciated from examination of the
diagrams in Figure 13, these UDEC models were very complex and time
consuming to run. Because of this complexity they also needed very
careful calibration to actual conditions in order that any forward
predictive modelling could be considered realistic. Such
calibration was however also not simple. A two stage approach was
therefore taken for calibrating the discrete element models. First
some overview modelling (in plan and in section) was carried out
using the Phase 2 FEM program code, specifically so that the data
from the convergence arrays and extensometer installations could be
rapidly tracked backwards over the known excavation sequence for
which instrumented response data was available. Then the UDEC
sections were time-stepped through the same stress change sequence
as found from the Phase2 responses and Barton-Bandis shear strength
parameters (Barton & Bandis, 1982) for discontinuity fabrics in
the UDEC models adjusted within realistic ranges until accurate
replication of as-measured convergences was achieved (as shown in
Figures 13 and 14).
-
1278
Typically it was found that the best data source for reliable
calibration was the detailed extensometer records (an example of
which is shown in Figure 14). On this diagram the plot shows the
response for the three anchors referenced to the deepest as datum.
A suite of vertical sections were specifically set up in Phase 2 to
model rockmass behaviour at each extensometer site and input
parameters (stress state, and fracture and rockmass properties)
adjusted in the modelled sections until good replication was
achieved for the complete benching sequence for which
instrumentation data was available. The parameters so defined were
then transferred to the horizontal section models and further
refinement of parameters completed until predicted convergences
matched observed readings. Figure 14 shows one of the extensometers
in the Chamber 3-4 pillar, on the Chamber 4 side in response to
excavation in Chamber 4 of Sequences 5 and 6, and then of the
hopper area (Sequence 7). The multi-point borehole extensometer
anchor responses are colour coded, with the deep zone showing about
2mm, the intermediate zone about 12mm and the surface zone about
another 10mm response, which then rises to almost 25mm with
excavation of the Hopper zone. Replication of these responses by
Golder in UDEC and calibration of the overall trends by NIRM into
3DEC and then carrying out predictive modelling of the excavation
progression to completion of all of the Chambers, including
considering watering up the Chambers for operational conditions,
indicated that quite significant further displacements could
develop, that without additional reinforcement, would lead to
significant distress developing in the pillars. The displacement
trends shown in the upper diagram on Figure 13 give some indication
of the type of fracture-controlled differential displacements that
were seen in all of the UDEC runs. The pattern of displacements
shown indicates that with full Chamber excavation general
relaxation occurs, that if unrestrained would pose a risk for
further
potential unravelling of the blocky rockmass constituting the
pillars, allowing further wedge fall-outs. With the information
generated from this modelling and with continued confirmation from
the convergence arrays and installed extensometers that movements
were still ongoing in the pillars, and that the trends were closely
matching the model predicted behaviour, SJVNs design team began to
formulate detailed construction working drawings to add significant
support to the pillars to enhance their stability.
5. CONSTRUCTION SOLUTIONS
The remedial support arrangements for the inter-Chamber pillars,
which were finally adopted with input from the Panel of Experts and
from various members of the contractors staff, who looked
specifically at constructability issues, are shown in Figure 15.
This diagram shows the various components incorporated into the
final configuration, together with the constructed measures
necessary for their installation. The SJVN design arrangement
basically comprised three rows of overlapping 60T cable anchors
into each pillar, installed through cast-in-place anchor beams, and
supplemented with additional rebar surface support and additional
SFRS to stich together the blocky, loosened near-surface zone of
the rockmass. As excavation had already proceeded down to Sequence
4 (el. 1450m) in most of the Chambers and to Sequence 6 (el. 1445m)
in one part of Chamber 4, and the upper row of anchors was needed
at elevation 1462m, (ie., some 12 to 18m above the then current
Chamber floor levels) a considerable effort was required to get
back up to the elevation required. In order to overcome the
challenge of casting the beams and installing the anchors, CFJV
came up with the hybrid scheme of backfill and scaffolding
arrangements as shown on Figure 15.
-
1279
Figure 12a. Topography at intakes to chambers
Figure 12b. Insitu Stress relationships for Desilting Chamber
area with respect to depth below rock surface
-
1280
Figure 13. UDEC representation of discrete fracture mapping
showing predicted and measured displacements
Figure 14. Typical sidewall MPBX behaviour
Figure 15. Benching & final support
-
1281
5.1 Pillar remedial support
Once SJVNs design staff authorized re-initiation of excavation
and issued formal instructions to start work on the beams and
anchors, forming the bulk of the remedial construction works
required for the pillars, plans were put in place to sequentially
cast the required concrete beams, necessitating, in some of the
Chambers, bringing in backfill and erecting scaffolding according
to the schematic shown on Figure 15. This arrangement was developed
by the contractor as one of many different construction sequences
and execution plans that were explored with the designer, all aimed
at minimizing further schedule delays and maximizing
constructability. The optimized solution required leap-frogging the
fill in the various Chambers in sequence with scaffold erection,
concrete placement and anchor drilling and stressing. By the time
that all works were complete to the state shown in the photograph
in Figure 1, the following measures had been completed:
Works Item Quantity
Refilling of Chambers with backfill
92,000m3 (Fill)
Installation of 60 T Cable Anchors
1,624 each
Placement of 24 @ 525 metre long concrete beams
12,400 metres long
Installation of Additional Rock Bolts (6 m to 7.5 m long)
20,000 each
Installation of Additional grouted anchor bars
6,000 each
Additional Rock Bolts 12.0 m long
4,480 each
Installation of three concrete beams along the entire length of
each side of each Chamber (as pictured in Figure 1) was seen as the
most expedient means for a) providing sufficient bearing area for
anchor stressing for each anchor, whilst also providing a good
measure of surface restraint and integral fixing along the Chamber
axes.
5.2 Anchor arrangements
The anchors for the Chamber pillars, which were generally
planned to be spaced at 7.5m c/c along the beams were selected as
four strand 20m long cable packages with an allowable capacity of
60 tonnes and an ultimate capacity of 105 tonnes. Each had a fixed
length (anchor zone) of 5m, and a debonded free-length of 15m. For
the installations after drilling to full depth water-testing was
required prior to grouting in the fixed length, with on-site agreed
procedures adopted for regrouting based on site conditions and
water test results. Each anchor was then proof tested, and locked
off with a pre-stress of 36 tonnes.
6. CONCLUSIONS
The scheme, the largest in India at 1,500 MW, has been
generating power now for almost five years with the Desilting
Complex Chamber excavations and ancillary tunnels providing their
designed role. In spite of so many rock-related problems requiring
significant engineering input, the Desilting Complex, comprising
four Chambers over 500m long, 16m wide, approximately 30m high and
only at 45m centre to centre has been completed to the original
hydraulically designed shapes. Execution of the final works has not
been without challenges however, and the final arrangements have
really only been accomplished through the cooperation and
integration of important ideas put forward by many parties, and
then engineered to detailed design level only as part of the
construction works. The overall final arrangements now comprise
several support elements not foreseen in the original tender design
drawings. Whether layout re-arrangements (spacing, orientation etc)
of the Chambers or different surface and deep rock support
arrangements incorporated into the tender layouts could have
eliminated or reduced the problems encountered in properly
stabilizing the weak foliated rockmass is a moot point. The design
layout as tendered has been completed, but only by placing three
pairs of concrete beams and hundreds of 60tonne cable anchors and
thousands of square metres of steel fibre reinforcement over the
pillar zone rock mass in each Chamber, suggesting to future
generations of designers that hydraulics should not necessarily
dominate early design decisions, especially in a Himalayan
setting.
-
1282
ACKNOWLEDGEMENTS
The views expressed in this paper reflect the opinions of the
authors and may not reflect those corporately held by the various
organizations involved in the construction of the Nathpa Jhakri
scheme. Acknowledgements are due to many individuals in the various
organizations involved in the project whose views and insight have
helped formulate the thoughts expressed in this paper.
REFERENCES
1. Bagde, M.N., (2000) Finite element analysis of underground
caverns of Nathpa Jhakri Hydel Project. Proc. Int.Conf. Tunnelling
Asia 2000 544 pp
2. Barton, N.R. & Bandis S., (1982). The Shear Behaviour of
Jointed Rock. Issues in Rock Mechanics, 23rd US Symposium on Rock
Mechanics, pp.739-759.
3. Barton, N., Lien, R. and Lunde, J. (1977) Estimating Support
Requirements for Underground Excavations, in Design methods in Rock
Mechanics. Proc. 16th US Sump. On Rock Mechanics, Minneapolis, USA,
pp.163-177.
4. Bieniawski, Z.T. (1976). Rock mass classification in rock
engineering. In Bieniawski (ed.), Proc. of the Symp. Explo-ration
for Rock Engineering, Vol. 1: pp.97-106. Cape Town. Balkema.
5. Carter, T.G., Steels, D., Dhillon, H.S. and Brophy, D.,
(2005). Difficulties of Tunnelling under High Cover in Mountainous
Regions. Proc. Int. AFTES Congress, Tunnelling for a Sustainable
Europe, Chambery, pp.349-358
6. Dasgupta, B., R. Singh and V. M. Sharma. (1999) "Numerical
Modelling of Desilting Chambers for Nathpa Jhakri Hydroelectric
Project," in Proceedings of the 9th ISRM Congress on Rock
Mechanics.
Paris, 1999, Vol. 1, pp. 359-360. Rotterdam: Balkema.
7. Grimstad, E and Barton, N. (1995) Rock Mass Classification
and the Use of NMT in India. Proc. Int. Conf. On Design and
Construction of Underground Structures, New Delhi, India.
8. Hoek, E (1999) Putting numbers to geologyan engineer's
viewpoint. Q. Jnl Eng. Geol. & Hydrogeol. vol. 32, no. 1, pp.
1-19(19)
9. Hoek, E., (2000) Big Tunnels in Bad Rock, The Terzaghi
lecture presented at the ASCE Civil Engineering Conf., Seattle, Oct
18-21, 2000
10. Hoek, E. and Marinos, P. (2000). Predicting Tunnel
Squeezing. Tunnels and Tunnelling International, Part 1 November
2000, Part 2 December.
11. Kumar, R and Dhawan, A. K., (1999). Geotechnical
Investigations of Nathpa Jhakri Hydro Electric Project. Proc.
Workshop on Rock Mechanics & Tunnelling Techniques, Shimla
12. Kumar, N., Varughese, A., Kapoor, V.K., Dhawan A.K. (2004)
In Situ Stress Measurement and its Application for Hydro-Electric
ProjectsAn Indian Experience in The Himalayas , Paper 1b 02
Sinorock-2004 Symposium, Int. J. Rock Mech. Min. Sci. Vol. 41, No.
3
13. Mahajan, S., (2000) Practical application of steel fibre
reinforced shotcrete in Desilting Chambers of Nathpa Jhakri
Hydroelectric Project Proc. Int.Conf. Tunnelling Asia 2000 544
pp
14. Marinos, P. & Hoek, E. (2000). GSI A geologically
friendly tool for rock mass strength estimation. Proc. GeoEng2000
Conference, Melbourne: 1422-1442
-
1283
BIOGRAPHICAL DETAILS OF THE AUTHORS
Dr. Trevor G. Carter is a Principal with Golder Associates in
Canada. He is also a specialist geological engineer with world-wide
experience in unravelling difficult and complex rock conditions for
design of heavy civil or mining projects. He has over 30 years of
experience relating to engineering
geological and rock mechanics aspects of tunnelling and civil
and mining underground design and construction. Dr. Carter has
worked with Golder Associates since 1976 and after spending 4 years
on the construction and design aspects of the 1000MW Drakensberg
Pumped Storage Scheme in South Africa, has mainly been involved
with solving geomechanics problems of large scale surface or
underground excavations for mining or civil engineering
applications. Throughout his career he has remained involved with
hydropower projects worldwide, most recently in India as a
specialist review consultant on rock mechanics aspects of the
underground chambers and tunnel construction works for the 1500MW
Nathpa Jhakri Scheme in India. He is currently serving on a review
board for two similar power schemes in Chile.
Mike Kenny graduated from the University of Manitoba Canada with
a B.Sc. in Civil Engineering in 1968 and is a licensed professional
engineer in the Canadian provinces of Ontario and British Columbia.
He has been employed by the Foundation Company of Canada Division
of Aecon
Constructors for 25 years and is currently Vice President. Mr.
Kenny was involved for 7 years at the jobsite for the companys
Nathpa Jhakri projects with the last 4 years as Project Manager,
where he was responsible for overseeing all aspects of the
construction works for the Dam, the Desilting Chambers and several
km of associated tunnels.
Don Brophy joined Aecon in 1977 after graduating from the
University of Ottawa with a Bachelor of Applied Science degree in
Civil Engineering. He has held the positions of Project Engineer,
Project Manager, Contracts Manager and Estimating Manager. Don has
extensive experience as
Senior Estimator, bidding heavy civil construction projects in
the domestic market as well as internationally. He has worked on
major projects such as the Highway 407 Express toll Route (407ETR)
and the Toronto Airport Terminal Development projects in Canada,
and the Nathpa Jhakri Hydroelectric Power project in India. Don
is
currently responsible for overseeing the execution of all heavy
civil projects under Aecon Constructors.
Dr. Jos L. Carvalho Joe Carvalho graduated in Civil Engineering
from the University of Toronto in 1982 and currently is a Principal
with Golder Associates Ltd. in Mississauga, Ontario. He has over 20
years experience in heavy civil engineering and mining projects
with extensive
capability in detailed rock mechanics analysis and numerical
modelling; with particular experience also in coding
instrumentation and monitoring databases. Dr. Carvalho has been
involved in application of advanced numerical methods for numerous
applications related to open pits, underground mine excavations,
underground hydroelectric caverns, utility and transportation
tunnels and foundation projects. Dr. Carvalho was responsible for
the numerical analyses of several aspects of the Nathpa Jhakri
Hydro-electric project in India.
Doug Steels is President of Aecon Constructors and also manages
Aecon's interests on Joint Venture projects as Board member of the
Executive Management Committee. Mr. Steels' 35-year construction
and engineering career with Aecon started as a Field Engineer on
the massive St. Lawrence Seaway, Welland Canal
Relocation project. His unique combination of practical skills
and management capabilities were developed on Aecon heavy civil
construction projects as diverse as British Columbia's Revelstoke
Dam and Mount MacDonald Railway Tunnel, India's Nathpa Jhakri
Hydroelectric project and the Cross Israel Toll Highway project.
Prior to his appointment as President of Aecon Constructors, Mr.
Steels held roles of increasing management responsibility in
Aecon's corporate office, including the positions of Chief
Estimator and most recently as Senior Vice President. Mr. Steels
holds a Bachelor of Science degree in Civil Engineering from the
University of Windsor.
Harjit Dhillon began his heavy construction career in North
America working for the Perini Corporation out of Framingham,
Massachusetts. He joined Aecon Construction Group Inc in 1963 as
Chief Engineer for The Foundation Company of Canada, Aecons wholly
owned subsidiary. Throughout his years with Foundation,
and eventually in the position of President of The
-
1284
Foundation Company, Harjit was in charge of many of Aecons
largest and most complex joint venture projects, with special
emphasis on hydroelectric developments and their related
underground civil construction works. Some of the more noteworthy
projects included the Revelstoke Dam and Powerhouse for British
Columbia Hydro and the Jenpeg, Longspruce, and Kettle Rapids power
projects for Manitoba Hydro. Most recently, Harjit completed the
Nathpa Jhakri Hydroelectric project in Northern India, a $640
million Aecon sponsored Joint Venture.
Mark Telesnicki graduated in Geological Engineering from the
University of Waterloo in 1987.and has over 20 years experience in
heavy civil engineering and mining projects, particularly focusing
on geotechnical investigations; rock mechanics analysis and design;
instrumentation
and monitoring; and preparation of construction specifications
and drawings. He is currently a Principal with Golder Associates
Ltd. and Manager of the Rock Engineering Group in Mississauga,
Ontario. Throughout his career, he has been involved with
underground hydroelectric schemes, several utility and
transportation tunnels, and numerous rock slope stabilization and
foundation projects. Mr. Telesnicki spent just over a year on site
at the Nathpa Jhakri Hydro-electric project in India where he was
heavily involved in all the early aspects of the construction works
for the Desilting Chambers and appurtenant tunnels.