ELSEVIER GEO-ENGINEERING BOOK SERIES VOLUME 5 Part 7 doc

Rock burst in tunnels 339

22.4 SEISMIC ENERGY RELEASED IN A ROCK BURST

Evidently the center of seismic event leading to rock burst is the region of highest stress

concentration in the elastic zone. Seismic studies of Cook (1962) indicated that such

events occur generally not more than 30 meters from the face of an excavation (Jaeger &

Cook, 1969). Seismic events that end up in rock burst were only 5 percent of all events

recorded and the seismic energy of the order of 105

to 108

ft 1b. was released in bursts.

Otherwise in the remaining 95 percent of the cases, the energy released at the epicenter

of the violent failure and propagating towards the excavation is most probably absorbed

in the deformation of the previously fractured zone of rock mass. This zone in this manner

provides adequate cushion between the epicenter and the face of excavation.

Experience shows that rock masses which are fractured either naturally or artificially

are not prone to rock burst. This is explained by the relatively ductile behavior of jointed

rock masses. It is only the massive hard and brittle rocks (Q perhaps greater than 2) that

pose problem because of low value of E/Ef

. Further, since a fault will render the masses

more flexible as if it has reduced the elastic modulus, the chances of rock burst at the

intersection between the fault and the tunnel or roadway are increased.

Another important factor is the rate of excavation which cannot however be accounted

in the theory. Laboratory tests show that the ratio E/Ef

increases with decreasing rate of

deformation. Thus a slower rate of excavation may cut down the frequency and severity

of rock bursts.

22.5 SEMI-EMPIRICAL CRITERION OF PREDICTING ROCK BURST

It is obvious that failure of rock mass will occur where tangential stress exceeds its biaxial

(plain strain) compressive strength. Singh et al. (1998) have suggested that the effective

confining stress is nearly the average of minimum and intermediate principal stresses.

Thus the biaxial strength is given by equation (19.3) in Chapter 19.

In situ stresses should be measured in drifts in areas of high tectonic stresses to know

Po and σθ realistically. It will help in predicting rock burst conditions in massive rock

masses.

Kumar (2002) has studied the rock burst and squeezing rock conditions at NJPC head

race tunnel in Himalaya, India. The field data is compiled in Table 22.1 for 15 tunnel

sections of 10 m diameter where overburden is more than 1000 m. No rock burst occurred

at lesser overburden. According to Barton et al. (1974), heavy rock burst was predicted

as σθ/qc was more than 1.0, where qc is the uniaxial compressive strength of rock mate￾rial (gneiss). Fortunately, values of σθ/q

′

cmass are between 0.55 and 1.14, which predict

very mild rock burst conditions. Actually there were no heavy or moderate rock burst

conditions along the entire tunnel. Slabbing with cracking noise was observed after more

than one hour of blasting. According to site geologists, Pundhir et al. (2000), initially

cracking noise was heard which was followed by the spalling of 5–25 cm thick rock

Table 22.1 Comparison of Mohr’s and Singh’s criteria of strength of rock mass (Kumar, 2002).

Rock Predicted Rock

cover UCS Q Parameters φp Po σθ qcmass q

′

cmass σθ/ σθ/ rock behavior

S.No. Chainage, m (m) (MPa) RQD Jn Jr Ja Jw SRF Q (deg) (MPa) (MPa) (MPa) (MPa) q′

cmass q

′

cmass behavior (observed)

1. 11435–11446 1430 50 70 6 2 2 1 2.5 4.7 45 38.6 77.2 31.6 124.8 2.4 0.62 Heavy burst Mod. slabbing

with noise

2. 11446–11459 1420 32 60 6 2 2 1 2.5 4.0 37 38.3 76.7 30.0 87.9 2.6 0.87 Heavy burst Mod. slabbing

with noise

3. 11459–11525 1420 50 67 6 2 2 1 2.5 4.5 45 38.3 76.7 31.1 123.7 2.5 0.62 Heavy burst Mod. slabbing

with noise

4. 11621–11631 1320 32 55 9 1.5 2 1 2.5 1.8 37 35.6 71.3 23.1 77.0 3.1 0.93 Heavy burst Mod. slabbing

with noise

5. 11634–11643 1300 50 70 6 1.5 2 1 2.5 3.5 45 35.1 70.2 28.7 113.4 2.4 0.62 Heavy burst Mod. slabbing

with noise

6. 11643–11650 1300 60 60 6 1.5 3 1 2.5 2.0 45 35.1 70.2 23.8 108.6 2.9 0.65 Heavy burst Mod. slabbing

with noise

7. 11656–11662 1300 55 55 6 1.5 3 1 2.5 1.8 45 35.1 70.2 23.1 107.9 3.0 0.65 Heavy burst Mod. slabbing

with noise

8. 11662–11796 1300 50 65 6 1.5 2 1 2.5 3.3 45 35.1 70.2 28.0 112.7 2.5 0.62 Heavy burst Mod. slabbing

with noise

9. 11860–11917 1230 50 67 6 1.5 3 1 2.5 2.2 45 33.2 66.4 24.7 104.9 2.7 0.63 Heavy burst Mod. slabbing

with noise

10. 12044–12070 1180 42 70 62212.5 4.7 55 31.9 63.7 31.6 175.9 2.0 0.36 Heavy burst Mod. slabbing

with noise

11. 12070–12077 1180 34 60 6 1.5 3 1 2.5 2.0 30 31.9 63.7 23.8 55.7 2.7 1.14 Heavy burst Mod. slabbing

with noise

12. 12087–12223 1180 42 67 6 1.5 2 1 2.5 3.4 45 31.9 63.7 28.3 105.2 2.3 0.61 Heavy burst Mod. slabbing

with noise

13. 12223–12267 1100 42 75 42212.5 7.5 45 29.7 59.4 37.0 108.7 1.6 0.55 Heavy burst Mod. slabbing

with noise

14. 12273–12322 1090 50 70 43312.5 7.0 45 29.4 58.9 36.2 107.2 1.6 0.55 Heavy burst Mod. slabbing

with noise

15. 12359–12428 1060 50 75 6 1.5 2 1 2.5 3.8 45 28.6 57.2 29.4 98.5 1.9 0.58 Heavy burst Mod. slabbing

with noise

Notations: Po = γH; σθ = 2γH; qcmass = 7γQ

1/3 MPa; q

′

cmass = biaxial compressive strength from equation (19.3); Q = post-construction rock mass quality; φp = peak angle

of internal friction in degrees and H = height of overburden in meters.

342 Tunnelling in weak rocks

columns or slabs and rock falls. This is very mild rock burst condition. Another cause

of rock burst is the class II behavior of gneiss according to tests at IIT, Delhi, India

(i.e. axial strain tends to reduce in comparison to peak strain after failure, although lat￾eral strain keeps on increasing due to slabbing). Further, only the light supports have

been installed in the rock burst prone tunnel even under very high overburden of 1400 m.

These light supports are stable. It may also be noted from Table 22.1 that according to

Mohr’s criterion, σθ/qcmass is estimated to be in the range of 1.6 to 3.1 which implies that

moderate rock burst conditions should have occurred. Kumar (2002), therefore, made an

observation that Singh et al.’s (1998) criterion (equation 19.3) considering σθ/q

′

cmass is a

better criterion than Mohr’s criterion for predicting the rock burst conditions in tunnels.

It is interesting to note that q

′

cmass is much greater than uniaxial compressive strength

(UCS) of rock materials. However, q

′

cmass would be less than biaxial strength of rock

material. Hence equation (19.3) appears to be valid. It is important to note that q

′

cmass

(biaxial strength) is as high as four times or more of uniaxial rock mass strength (qcmass).

The peak angle of internal friction (φp) in Table 22.1 is found from the triaxial tests

on the rock cores. It is assumed to be nearly same for moderately jointed and unweath￾ered rock mass. This appears to be a valid hypothesis approximately for qc > 10 MPa as

micro reflects the macro. There is difference in the scale only. The φp is not affected

by the size effect. Table 29.1 offers more explanation considering non-linear effect in

Chapter 29.

It is important to know in advance, if possible, the location of rock burst or squeezing

conditions, as the strategy of support system are different in the two types of conditions.

Kumar (2002) could fortunately classify mode of failures according to values of joint

roughness number (Jr) and joint alteration number (Ja) as shown in Fig. 22.3. It is observed

0 2 4 6 810 12 14

High

Squeezing

Mild Squ. Moderate squ.

Moderate Slabbing

with Noise (Rock

Burst)

σθ

/q′

cmass = 0.6 - 1.0 2 - 3 3 - 4 >4

0

1.0

2.0

3.0

4.0

Joint Alteration Number (Ja

)

Joint Roughness Number (Jr

)

Fig. 22.3 Prediction of ground condition (Kumar, 2002).

Rock burst in tunnels 343

that mild rock burst occurred only where Jr

/Ja exceeds 0.5. This observation confirmed

the study of Singh and Goel (2002). If Jr

/Ja is significantly less than 0.50, squeezing

phenomenon was encountered in many tunnels in the Himalaya. Thus, a semi-empirical

criterion for mild rock burst in the tunnels is suggested as follows:

σθ

q

′

cmass

= 0.60 − 1.0 (22.2)

and

Jr

Ja

> 0.50 (22.3)

The support pressure may be assessed from modified Barton’s criterion which is found

to be valid upto an overburden of 1430 m by Kumar (2002),

proof ∼=

0.2(Q)−1/3

Jr

f MPa (22.4)

where

f = correction factor for overburden,

= 1+(H−320)/800 ≥ 1,

H = overburden above crown of tunnel in meters and

Q = post-construction rock mass quality.

The dynamic support pressure may be αvproof like equation (21.1) where αv · g is the

observed maximum acceleration of rock pieces. The αv may be as high as 0.35.

22.6 SUGGESTION FOR REDUCING SEVERITY OF ROCK BURSTS

Suppose a tunnel opening is supported by very stiff supports so that support pressure

develops to the extent of cover pressure, no rock burst will occur. But, this is a very costly

way of solving the problem.

Another way of reducing chances of rock burst is to make opening of small size. This

is because amount of strain energy released per unit area of excavation will be reduced

considerably.

Since stress concentration is responsible for initiation of cracking, it may help to

select a shape of excavation which gives minimum stress concentration. For example, an

elliptical opening is best suited in non-hydrostatic stress field. Its ratio of span to height

should be equal to ratio of horizontal stress to vertical stress. In hydrostatic stress field,

circular openings are better than square openings. As mentioned earlier, it may also help

to slow down the rate of excavation in the zone of stress concentration, as rocks will be

able to absorb more strain energy due to creep.

It may be recalled that the de-stressing technique has been used with some success

in mines. In tunnel opening, if rock is broken intentionally by blasting or drilling, etc. to