ARMA 10-122

Rockburst at Jinping II Tunneling Project

Zhu, H. C., Wu, J. Y., and Zhu, Y. S.
Itasca Consulting China Ltd., Wuhan City, Hubei Province, China
Zhang, C. S., Chen, X. R., and Hou, J.
HydroChina Huadong Engineering Corporation, Hangzhou City, Zhejiang Province, China
Zeng, X. H.
ErTan Hydraulic Engineering Development Co.Ltd., Chengdu City,Xichuang Province, China

Copyright 2010 ARMA, American Rock Mechanics Association

th th
This paper was prepared for presentation at the 44 US Rock Mechanics Symposium and 5 U.S.-Canada Rock Mechanics Symposium, held in
Salt Lake City, UT June 27–30, 2010.
This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical review of
the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: Rockburst occurrences are mostly resulted from in-situ stress localization in synclines and near faults at Jinping II
project. All observed rockbursts have been classified into four types as strain-burst, tip-burst, slip-burst, and fault-induced stain-
burst. Stain-burst does not prevail and usually acts as an indication of the increasing risk of severe burst as stress ratio is generally
low. Tip-burst is commonly seen and resulted from the in-situ stress localization in the terminating area of NWW-oriented weak
fractures at Jinping. Slip-burst, widely reported elsewhere in the world, appeared at Jinping along a stiff facture. Severe stain burst
can be triggered when advancing towards a NE-oriented weak fault and thus referred to fault-induced strain burst.

excavation depth of 2525 m from the surface when

1. INTRODUCTION driving headrace tunnels. The mountainous topography
Jinping II is a hydroelectric project located on Yanlong provides no additional aid for excavation, except for the
River in Sichuan Province, southwest of China. As two portals on east and west sides of the mountain.
shown in Fig.1, there are total seven tunnels at this
The tunnels run mostly through massive marbles, except
project: twin transportation tunnels serving as accesses, some 5km weak rocks as slate and schist near west
one drainage tunnel, and four water-conveying or portal. Major geological structures such as large faults
headrace tunnels. Each extends approximately 16.7m
and beddings generally are perpendicular to the tunnel
and makes around 117 km in total.
axis (please refer to Fig. 5 for details). Strata are heavily
folded, with an average dip of 60 degrees (please
referred to Fig.3 A).
By the end of February 2010, the twin transportation
tunnels have been in operation for a few months. Three
Tunnel-Boring-Machines (TBMs) have been heading all
from the east portal for some 6 km, 5 km, and 4.6 km at
the drainage tunnel, #1, and #3 headrace tunnels,
respectively. The other two headrace tunnels have been
driven with traditional Drill and Blasting (D&B) method
for approximately 5 km from the east portal and 1.8km
from the west portal.
Fig. 1. Plan view (left-up) and cross-section view (bottom) This paper summarizes the observations and studies of
showing project and tunnel layout, respectively rockbursts occurred while driving all these tunnels at
Jinping II project. As a start point, investigation and
All the tunnels run beneath Jinping Mountain, which has achievement are firstly introduced of the in-situ stress
a summit elevation of 4455m; this creates a maximum distributions along the tunnel axis inside the Jinping
Mountain with complex geology. Rockburst slight reduction of stress level inside the core area of the
observations are then introduced, followed with the syncline.
analysis of rockbursting mechanisms. Brief study is
From Fig. 2 it can be seen that the actual in-situ stress
lately presented of the influence of opening size in
state on the cross-section of a tunnel depends on its
rockbursting potentials at this project.
location with respect to the core area of a syncline. The
model suggests that high in-situ stress can only be
2. IN-SITU STRESS ALONG TUNNEL AXIS encountered when advancing the tunnel under the core
area of such synclines.
2.1. Investigation of in-situ stress
The attempts to measure the in-situ stresses failed in As it has been well-documented that geological structure
massive marble in the twin access tunnels at depth below such as bedding and fault can significantly alter in-situ
1800m from the surface. Because of the high stress distribution nearby, the FLAC3D model as shown
confinement, the available hydro-fracturing device is Fig. 3 integrates all major folds, beddings, and faults
unable to provide enough water pressure to break the presented along the tunnel axis [2].
rock. Overcoring method becomes much less reliable
due to core damage induced by drilling in marble at such
depth. Lab test indicates that all drilling cores obtained
in the twin access tunnels below the depth of 1800 m
from surface are significantly damaged as the presence
of drilling-induced micro-scale cracks [1].
Numerical model was built based on site observation and
applied as an alternative to investigate the in-situ stress
distribution along tunnel axis, particularly the area below
the depth of 1800 m. Considering the complex geology
along the tunnel axis, the first model was built with
FLAC3D to look into the influence of a typical fold in
the stress and to understand if stresses could locally Fig. 3. FLAC3D model built for stress investigation. A:
build up [2]. topography and geology, B: FLAC3D model, and C: rock
mass quality variation in the model

Fig. 4. Model-referred stress distribution along tunnel axis. A:

Fig. 2. Model-referred stresses within a typical syncline at stress contours color coded by magnitude of the major
Jinping project principal stress; B: magnitudes of three principal stress
components along tunnel axis
As shown in Fig.2, a typical syncline appeared in the
east portal was taken as example for the modeling The above of Figure 4 shows the model-referred stress
exercise. Based on reality, the model considers the distribution along tunnel axis as contours color-coded by
differences of rock mass quality in the folded area, i.e, a magnitude of the major principal stress. The magnitudes
lower GSI of assigned to the core area whereas a value of three principal stress components are given in Fig. 4
of 75 to elsewhere. Selected beddings were included in below. It can be seen that the differences are generally
this simple model. small among the three principal stress components along
tunnel axis, except a few regions. In other words, stress
As can be seen in Fig. 2, the model suggests that stress ratios are generally low, mostly less than 1.30.
can locally build up in some particular areas inside the
syncline. In general stress increases in the underneath of Form rockburst pre-warning point view, the in-situ stress
the core area where rock mass appears competent with distribution along tunnel axis can be further classified
GSI value of 75. So does stress ratio. However there is a into a few domains as the following:
 • Domain #1: highest stress level over the entire drawing, the rockbursts are found to fall into a few
tunnel axis in the core area of Jinping Mountain, ranges as shown in Fig.6. These ranges are classified as:
in the range approximately between 7.8 and 10.2 • Class #1: located in the core area of the mountain.
km from the west portal. Affected by both the very severe rockbursts have occurred frequently
thickness of overburden and the presence of the when access tunnels passing this area;
large-scale syncline. • Class #2: severe rockbursts observed occasionally;
• Domain #2: located in the range approximately • Class #3: rockbursts with less severity appeared
between 5.5 and 7.8 km from the west portal with frequently in access and drainage tunnels.
great depth of overburden. Localized stresses with
relatively high stress ratio may appear
occasionally, corresponding to the presence of a
certain geological structures;
• Domain #3: relatively high stress level resulted
from relatively great depth. Increased stress ratio
as a typical feature to the presence of synclines.
This domain is located generally in the range Fig. 6 Rockburst-prone ranges in which rockburst occurred
between 10.2 and 14.5 km from the west portal; while driving the access and drainage tunnels at Jinping
• Domain #4: the rest areas where stress does not
The rockbursts in class #1 and #3 are associated with the
seem significant for rockburst study.
presence of synclines. The increased frequency of
2.2. Validation of stress localization rockbursting in these areas seem to indicate localized
Stress distribution described in the above can be stress regimes
validated with the ground responses observed while Fig. 7 shows one of the examples of rockbursts occurred
driving the tunnels. Such validation has been further while the drainage tunnel was passing a syncline near
used to understand rockbursting and to help identify the east portal. The ground damage consecutively
rockburst risks for future construction. extends over 20 m in length along the tunnel. Such
The most commonly seen ground response of marble at consecutively extended damage is different from those
Jinping is fracturing, or Excavation Damage Zone (EDZ), triggered by a single geological structure to be addressed
even at great depth below 1800 m from ground surface. later. Syncline seems to result in a localized in-situ stress
Back analysis with PFC2D against field measurement of regime that agrees with what discussed in the above.
EDZ profile indicates that the stress ratio of maximum
over minimum on cross-section is only 1.16, which is
quite low [3]. Fig. 5 presents both the measurements and
modeling results with PFC2D.

Fig. 7. Diagram showing rockburst damage at the section

between 14+800 and 14+780 in the drainage tunnel

As rockburst risk increases when driving tunnel in a

syncline, a pre-warning of rockburst was made when the
TBM in drainage tunnel headed for the syncline located
in the range between 11 and 11.6 km from west portal.
Fig. 5 In-situ stress validation against PFC duplication of However the ground performed fairly well that brought
measured EDZ an unexpected TBM-advancing rate as high as 600 m per
month. Further investigation indicated that the tunnel
However severe rockbursts did occur at some particular passed the syncline at relatively high elevation where
locations while driving the access and drainage tunnels. high stress does not seem to build up. This fact appears
After posting these locations onto the geological
to agree with the model-referred stress distribution in a At Jinping such fracture is typically sub-vertical and
syncline. NWW-oriented. The average strike of this set of
geological structure is generally within 20 degrees off
Ground damages appeared in the range between 5.5 and
tunnel axis.
8 km from west portal extended in a rather short length
usually within 2 m in the direction of tunnel axis. This type of fracture usually appears weak as in
Ground damage appeared at an individual base with a weathering condition or with infillings of weak materials.
certain interval. Both the location and the interval of
Ground responses to the presence of this set of fractures
these damages are quite consistent with the presence and
depend upon how it is stressed, or presented with respect
the spacing of a set of NE-oriented fractures at Jinping.
to the opening. Rockburst can only be triggered when
Note the strike of this set of fracture is sub-perpendicular
such facture terminates in the surrounding rock within a
to tunnel axis. Localized in-situ stress near such fracture
certain distance, say 3m from opening perimeter, when
may be accounting for rockburst.
overburden thickness is less than 1800m.

3. ROCKBURST ONSERVATION AND

ROCKBURSTING MECHANISM
Several types of rockbursts have been observed and
identified at this project. In addition to some well
documented ones such as strain burst and slip-burst,
other types of rockburst have been clearly observed in
the field, contributing to the perfectly circular opening
profile created by TBMs.
3.1. Strain burst Fig. 9. Illustration of ground response to a NWW-oriented
Strain burst has not been commonly seen at Jinping weak fracture
project contributing to favorable stress ratio of σ1 over
σ3 around or lower than 1.3 in general. In most cases Fig. 9 shows an example of tip-burst occurred when a
strain burst appears as an indication of the increase of NWW-oriented facture terminates near an opening. It
rockburst risk. can be seen in the left side photo that the surface of this
facture appears in brown as evidence of weathering. The
facture itself is weak to have a low stiffness. Ground fall
appeared along the fracture trace. However rockburst
occurred in the tip area. As shown in photo on the right-
hand side of Fig.9, burst failure appears in massive and
competent rock right behind the TBM shed.

Fig. 8. Photos showing strain burst occurred in the right-hand

side shoulder of drainage tunnel with V-notch profile

Fig.8 shows a typical strain burst with a V-notch failure

profile. It occurred at the location of some 12 km from
the west portal in drainage tunnel. An interesting point to
Fig. 10. Model-referred in-situ stress along a NWW-oriented
view strain burst at Jinping is its indication to in-situ weak facture at Jinping
stress and severe burst. As having been mentioned in this
paper, small stress ratio and fracturing are dominated at The mechanism of tip-burst appears obvious due to
Jinping. Rockbursts including strain burst, are likely localized in-situ stress distribution in the terminating
linked to localized in-situ stress. Stain burst at Jinping area of a fracture. This presumption can be easily
usually acts as an indication of in-situ stress change confirmed with a numerical model as shown in Fig.10.
associated with the presence of synclines and faults and After taking the consideration of the contrast of rock
thus a warning to the increased risk of severe rockburst. mechanics properties between the fracture and the
3.2. Tip burst adjacent rock, stress localization can be presented with
Tip burst is fracture-associated and most commonly seen the 3DEC model [4]. In other word, the fracture needs to
at Jinping. This burst is usually triggered while work be weak to trigger a tip-burst.
face is approaching to the termination area of a fracture. 3.3. Slip burst
Slip burst refers to a rockburst in fracture-sliding
mechanism. This type of burst will not be addressed in
details in this paper as it has been well documented since
1980s [5]. Generally the fracture needs to be stiff that
allows stress to build up at some interlocked locations.
Only one slip burst has been observed at Jinping. This
burst occurred in the drainage tunnel when TBM hit a
NWW-oriented stiff fault at the location of 9.5 km from
the west portal. Fig. 13: Model-referred stress rotation near weak faults

Fig. 13 presents the stress change near a set of weak

faults, revealed with a conceptual 3DEC model. Hydro-
static stress is assumed as initial condition as shown in
the left-hand side of Fig. 13. At this stage the weak
faults are assumed to have similar geomechanics
properties to the adjacent rock. After weakening the
weak faults and allowing them to move, major principal
stress rotates significantly to be parallel to the fault and
increase stress ratios. Such localized in-situ stress in the
vicinity of a weak fault may be able to trigger rockburst
when a tunnel advances towards it.
Fig. 11: slip-burst occurred in drainage tunnel
Note that how stress is localized near a fracture really
Fig. 11 shows the profile of failure. Damage area depends on how this facture presents with respect to the
extends along a NWW-oriented facture with a length regional stress direction. This issue will not be discussed
over 20m and a maximum failure depth over 6m. Rock in this paper.
adjacent to the fault is severely broken. From site
observation, it seems that seismic events have been
triggered, more or less simultaneously, at various
4. OPENING SIZE EFFECT OF ROCKBURST
locations along this stiff fault. 4.1. Introduction
3.4. Fault-induced strain burst An attempt made during the construction at Jinping was
to predict rockburst potentials for the four large-size
Fault-induced strain burst here refers a strain burst but is
headrace tunnels by using the rockburst occurrences
resulted from localized stress near a fault. This type of
observed in the smaller-scale twin access tunnels and the
burst has been frequently observed in engineering
drainage tunnel. The first warning was made when the
practice worldwide. It is triggered while work face is
large headrace tunnels entered into the syncline some
approaching towards a fault.
12.5 km from the west portal because rockbursts had
occurred in this particular area in both D&B-driving
access tunnels and TBM-advancing drainage tunnel.
However all four headrace tunnels peacefully passed this
area without causing any rockburst problem. Further
investigation did not reveal significant alteration of
geological condition between the large headrace tunnels
and the other ones. It indicates rockburst potential may
be associated with opening size at this project.
The possible “opening size effect” of rockburst can be
investigated from two aspects: scale-effect and
confinement effect of rock mass behaviors. This paper
briefly discusses the latter one only.
Fig. 12. Fault-induced strain bursts in hydroelectric tunnel #2
Rockburst is considered a brittle failure under high stress.
Typical example of fault-induced strain bursts at Jinping In other words, the rock has to fail in a brittle manner
is shown in Fig. 12. The strike of such fault is usually when stress exceeds the strength of surrounding rock.
sub-perpendicular to the tunnel axis. Total four V-notch The brittleness and the strength of marble at Jinping are
profiles representing four bursts were clearly presented significantly affected by the confinement level [1]. At a
in the roof of the headrace tunnel #2. Three of them are given distance from an opening, confinement level is
located on the southeast side of the fault. closely associated with the size of the opening. As a
result, rockburst risk is theoretically linked to the size of On this σ1—σ3 chart, seismicity-prone domain is defined
opening at Jinping. with a stress range within which high energy can be built
It has been well described by all commonly-used and the rock mass intends to fail in a brittle manner. For
strength criteria that the peak strength of rock mass the Jinping case, this range can be defined as 5MPa ≤
increases with confinement level. In other words, higher σ3 ≤ 10MPa because:
energy is allowed to build up to trigger a burst with the z Peak strength of rock mass or maximum tangent
increase of confinement level. It implies that smaller stress reduces to the level less than 60% of UCS
opening intends to have higher rockburst risk at a given when σ3 is under 5MPa. It is unfavorable for a
distance from opening perimeter, as confinement is brittle failure;
higher than that around a larger opening. z Ductile response seems dominated when σ3 is over
10MPa and thus is unlikely to result in rockburst.
On the other hand, studies of rock mechanics behavior of
intact marble has revealed that its post-peak response Confinement stress distribution on radial direction can
transits from brittle to ductile when confinement stress be calculated with numerical models. Taking circle
increases to the level of 20 MPa. To rock mass of marble tunnel as example, Fig. 15 presents the modeling results
with GSI value of 75 at Jinping, this threshold is around when assuming the opening radius varies from 1m to
10MPa [6]. This indicates that rockburst potential 10m. Initial conditions for these model runs refer to the
reduces as the increase of confinement close to, or over reality in Jinping. In-situ stress field is equivalent to
10MPa at Jinping. depth of 1800m from ground surface. The marble is
assumed to have GSI of 80 and UCS of 140 MPa,
4.2. Opening size effect of rockburst respectively.
After creating an opening in massive marble, there will
be a brittle zone with a certain thickness surrounding the
opening, as the brittle response of marble at low
confinement. Rockburst is expected to occur within this
zone only unless confinement changes as failure sub-
sequentially progresses. Beyond this brittle zone to
great depth, ductile behavior intends to present.
Theoretically seismic risk reduces in the ductile zone to
trigger a burst to the ground.
The Hoek-Brown model built in FLAC3D has the
capability to describe the brittle-ductile transiting
response with the increase of confinement, in addition to
the peak-strength change. This model was thus Fig. 15: Confinement level change with radial distance under
employed to investigate the possible “opening-size various circular opening sizes
effect” of rock burst risk.
Taking the seismicity-prone criterion established in the
above, it can be seen in Fig. 15 that there is no rockburst
risk when the opening radius is 1.0 m as the confinement
increases to 14.4MPa, over the brittle-ductile transition
limit, at the depth of 0.25 m from opening boundary.

Fig. 14: Rockburst-prone domain defined by stress altering Fig. 16: Confinement level change with radial distance under
paths at various radial distances to opening boundary various circular opening sizes
Fig.14 presents stress altering paths at various radial Fig.16 shows the maximum different stresses (σ1- σ3) in
distances from an opening with size of 7.2m in diameter. the seismicity-prone domain under various opening sizes
from 2m to 10m in radius. The tendency of maximum The authors would acknowledge ErTan Hydroelectric
different stress variation vs. opening radius can be Development Co. Ltd. for their funding to carry out the
clearly classified into two sections, the one with opening studies and their permission to publish this paper.
radius no more than 4 m and the other over 4 m. As the
index of (σ1- σ3) represents the failure potential it
REFERENCES
suggests, in this particular case, an opening size of 4 m
in radius seems to be the most seismicity-prone. 1. Shan, Z.G., Wu, X.M., and Yang, P. et al. 2010.
Behavior of marble at Jinping II project – part I: intact
It also should be borne in mind that ground damage may rock. Submitted to the 44th US Rock Mechanics
be more significant with large opening sizes when Symposium.
attacked by a same magnitude seismic event as the
2. Itasca Consulting Group Inc. 2007. FLAC3D: Fast
ground is less confined. More efforts are still required to lagrangian analysis code, Version 3.1. Minneapolis,
investigate this aspect but not discussed in this paper. MN, USA
3. Itasca Consulting Group Inc. 2008. PFC2D: Particle
5. CONCLUSION flow code in two dimensions, Version 4.0. Minneapolis,
MN, USA.
The description in this paper can be summarized as the
following: 4. Itasca Consulting Group Inc. 2007. 3DEC: Three
dimensional discontinuity code, Version 4.1.
z In-situ stress varies significantly along the 16.7 km Minneapolis, MN, USA.
long Jinping Tunnel axis. From rockburst study
point view, synclines and faults are the most 5. Ortlepp, W.D. 1983. The mechanism and control of
rockbursts. In Rock Mechanics in Mining Practice, ed.
concerned geological structures that can result in
S. Budacvari, 257-281.
localized in-situ stresses. Stress localization is often
accounted for rockburst at Jinping; 6. Zhang, C.S., Chen, X. R., Hou, J. and Chu, W. J. et al.
2010. Behavior of marble at Jinping II project – part I:
z Generally fracturing is the most commonly seen rock mass. Submitted to the 44th US Rock Mechanics
ground response to excavation due to the low stress Symposium.
ratios around the tunnel, with exception in the core
area of synclines and the vicinity of fault where
rockburst could be a major concern.
z All significant rockbursts are associated with
geological structures such as fracture and fault.
Generally the NWW-oriented weak factures are
mostly favorable to tip-burst. The NWW-oriented
stiff fault has ever induced slip-burst. The NE-
oriented weak fault usually results in severe strain
burst while tunnel face is approaching towards the
fault;
z It seems that rockburst potential is associated with
opening sizes at Jinping. When considering the
influence of confinement in both rock mass strength
and stiffness, it has been found that the large
opening size over 12 m in diameter reduces
rockburst risk. This finding can well explain the
difference of rockburst risks between larger
hydroelectric tunnels and the smaller ones at
Jinping. In addition, the opening size of 4 m in
radius, or 8 m in diameter seems the most favorable
opening size to rockburst risk. This comment seems
to agree with reality as well.

ACKNOWLEDGEMENT