structures under shock & impact vi, c.a. brebbia & n ... · damage experienced in weak rock...

Damage to stope walls from underground

blasting

E. Villaescusa* & L.B. Neindorf** Western Australian School of Mines, Kalgoorlie, Western Australia^ Mount Isa Mines Ltd, Queensland, Australia

Abstract

This paper presents the results of a monitoring program to characterize the blastdamage experienced in weak rock masses extracted by underground sublevelstoping methods. Triaxial geophone arrays were used to monitor blastattenuation and extensometers were used to measure hangingwall deformationsduring stope extraction. The blasting events were monitored and the vibrationdata were analyzed using the Holmberg-Persson criteria for blast attenuation. Acomprehensive analysis of the measured vibrations was completed. This wasfollowed by the development of predictive tools that can be used to reduce blastdamage during future underground extractions within weak rock.

1 Introduction

Rock mass damage in underground mining usually occurs due to a change inthe induced stresses and also due to blast damage. As excavations are created,the in-situ stresses redistribute around the boundary of the openings. As aresult, high stresses may be experienced on the backs and corners of theexcavations, while low stresses may be experienced within some of the exposedstope walls. The de-stressing of the walls will tend to open existing cracks, as aresult of movement of the rock mass into the excavation. Extensometers areused to measure the amount of wall relaxation and borehole camera surveysdetermine the exact location and nature of damage. Blast damage is defined asthe creation, extension and opening of pre-existing geological discontinuities inthe rock. In some cases, blast damage may occur through intact rock. Damage

Structures under Shock & Impact VI, C.A. Brebbia & N. Jones (Editors) © 2000 WIT Press, www.witpress.com, ISBN 1-85312-820-1

130 Structures Under Shock and Impact VI

weakens a rock mass, leading to stability problems when the excavation size isincreased.

Empirical evidence from underground stoping suggests that the measuredvibration levels from blasting can be linked to rock mass damage [1]. Themagnitude of the vibrations depends upon the nature of the rock mass, theblasting agent, the hole diameter used, the drilled pattern (burden, spacing, holeangle and distance of the holes to the exposed walls) and the hole deviation.Back analysis at Mount Isa Mines suggests that blasting may control up to 15%of the overall hangingwall behaviour [2]. A relationship between the criticalpeak vibration velocity and rock mass damage in the nearfield (within a chargelength) can be determined by correlating the measured blast vibrations and thedamage observed with the borehole camera surveys.

2 Design of experiment

The instrumentation was designed to compare results from a number ofmonitoring programs carried out on a number of stopes extracted using twodifferent blasthole patterns. The placement of monitoring locations was eithernear the mid-span of a designed stope wall, to determine maximum deflection,or at a pillar to determine the effectiveness of that pillar. In order to measureground vibrations across the stope hangingwalls, it was required to drillmonitoring holes from hangingwall accesses. A typical monitoring set-up isshown in Figure 1, where three monitoring holes, an extensometer, a geophoneand observation hole (TV borehole camera) are shown. Monitoring holes drilledperpendicular to a stope wall can be used to determine opening and shearingalong geological discontinuities such as bedding or foliation. If the geophonesare perpendicular to the blastholes, then the first stress wave arrival at eachgeophone will be from approximately the same section of an explosive column.This allows a more accurate determination of vibration attenuation.

Figure 1: Cross-section of a typical monitoring set-up.


Structures Under Shock and Impact VI 131

2.1 Geophones

A number of triaxial arrays of geophones were used to generate vector sums,which were used to determine peak particle velocity for each monitored event.This peak particle velocity (PPV) can be related to distance from a blastingevent and can be used in the prediction of PPV for future blasting events. Eachgeophone hole contained three triaxial geophones, which were placed at variabledistances from the blasthole being monitored. Omni-directional geophones (14Hz OYO^were used to monitor the blasts in the very near field, while 8 HzOYO geophones were used at the furthest monitoring point in each hole.

2.2 Extensometers

Multiple anchor point rod extensometers using mechanical read outs were usedto measure stope wall deformations. Each extensometer consisted of 6 anchorsat varying distances into an exposed stope wall. Measurements were takenregularly before and after the blasting events throughout the entire excavationperiod, so that a complete record of stope wall movement was obtained.

2.3 TV Borehole camera

Bore hole observation holes were drilled parallel to and approximately 1m awayfrom the geophone holes. The holes were designed to intersect the exposedstope boundary and the last 8m (immediately adjacent to the stopes) werediamond drilled. This diamond-drilled section of the hole provided core that waslogged for rock mass characterization purposes. The observation holes wereplaced close to the geophone holes in order to establish a correlation betweenthe measured Peak Particle Velocity (PPV) and rock mass damage [1].

The monitoring program consisted of surveying the hole before and aftereach blasting event. The borehole camera used in this study was a Pearpoint"flexiprobe type" solid state video unit. Surveys were made using both afrontview and sideview lens attachments, with the most reliable data comingfrom the sideview lens. This gave a clearer picture of the borehole wall,(perpendicular to the borehole axis). All surveys were recorded on video tapeand analyzed on surface for the frequency of cracking through intact rock beforeand after a blast, as well as the opening and displacement of existing geologicaldiscontinuities.

2.4 Production Blasting

A number of individual stopes which were designed to be stable atapproximately 21m were monitored (average width of the orebody wasapproximately 5 to 7m). The production strategy consisted of extracting eachstope, filling it (using uncemented rockfill) and then moving to extract theadjacent stope in the sequence. The extraction strategy was aimed at recovering100% of the ore. In order to achieve this, rib pillars left between stopes were tobe blasted following the completion of rockfilling (See Figure 2).



I Temporary pillar (drillied)

New Slot I Production recovered pillarBlasting | Rockfill

Figure 2. A longitudinal view of a typical stoping geometry.

The production rings were drilled with 89mm holes using top hammertechnology (Tamrock SOLO rig). The typical pattern used was approximately3m between rings, in conjunction with an intermediate easer (1.5m burden). Thetoe spacing of the holes ranged from 5 to 6m and the length of the holes wasapproximately 30m. As a standard practice all hanging wall holes were chargedwith ISANOL and the footwall holes with ANFO. In each ring, the footwallhole was detonated ahead of the hangingwall hole. Three or four ICI Powergelpackages were placed at the bottom of each charge, and delay detonators of thesame number were placed approximately 8 and 16m above the bottom ofcharge. Production holes were unstemmed.

3 Rock mass damage

3.1 Extensometers

Extensometers were placed in the hangingwall of the excavation in order tomeasure the displacement of the exposed unsupported walls (pre and post-blast).A plot of displacement versus time for a typical extensometer is given in Figure3. It should be noted that stope 1 was only extracted to production Ring 4 due toa hangingwall failure. Therefore this extensometer was actually located at thevery end of the open hangingwall. Figure 3 indicates that the failure at thisposition was between 1.5 and 3.5m deep into the hangingwall. The wall of stope1 continued to creep during the extraction of the adjacent stope 2, and had afurther jump in displacement with the firing of the pillar toes, shown in Figure 3as main rings 5&6, which caused failure in stope 2. Time dependant creep (notshown in Figure 3) continued after the panel has been rock filled for almost amonth.


Structures Under Shock and Impact } 7 133

1cc

11c

) c RR1N

CCs

ccs

"1

c c0 (s ),

c0s

C C R0 0 |S S N

Dist. into H/W

t_ 13:Oct 16-Oct 19-Oct _22-Oct 25-Oct 28-Qct 31-Oct. 3-NpvTime

6-Nov _9-Nov

Figure 3 : A typical extensometer record.

3.2 Vibration analysis

The vibration from a blasting event experienced at an observation point is afunction of the distance from the charge, explosive type, blasting geometry andthe attenuation characteristics of the rock. Ground vibrations can be predictedby generating a graph of peak particle velocity (PPV) versus distance from thecharge [1]. The method is described by Holmberg and Persson [1] as follows:

1) Calculation of the material constants, K and a is based on the followingrelationship where (a is the symbol given to the Holmberg term):

K(0)° (1)

=> Log (PPV) = a log(o) + log (K) (2)

2) The calculation of a can be be simplified for a coordinate system base [3] to;

a = —[arctan(—) - arctan(—)]/vo /Vo /Vo

(3)

Where L,, and 1̂ are distances (in metres) along the line of charge relative tothe intersection of R^. The sign of LI and ̂ are positive in the direction fromthe bottom to the top of the charge (See Figure 4). To predict PPV, the values ofK and a are placed back into Equation (2), assuming that for a generalgeometric case, the point of observation is perpendicular to the centre of charge,ie. the values of L^ and ̂ are of equal length. Also, to determine the K and a


134 Structures Under Shock and Impact 17

values, it is preferable to eliminate as many of the blast variables as possible,therefore only a limited set of material constants are produced.

dx

Observation point

Figure 4: Configuration of explosive charge.

3.2.1 Analysis for closed wallsThe Holmberg equation was developed for fully confined holes. In the presentstudy two different confinements were experienced due to the 1.5 and 3.0mburdens used in the production blasts. Figures 5 and 6 present the results acrossbedding for both burdens where no stope wall exposure existed in front of thehangingwall geophones when the charge detonated.

y= 1.0868X +2.6742= 0.7289

log (a)

Figure 5. Vibrations across bedding for closed walls with 1.5m burdens.

The material constants for both burdens can be calculated from the fittedEquations. For the 1.5m burdens K is 470 and a is 1.09 while for the 3.0mburdens K is 550 and a is 0.75. This indicates that 3m burdens are likely toproduce higher levels of vibration that attenuate at a slower rate. Consequently,more rock mass damage can be expected if easers are not used within theblasting patterns.


Structures Under Shock and Impact 17 135

y = 0.7488X + 2.7408= 0.6555

-1.5 -1 -0.5 0 0.5 1

log (a)

Figure 6. Vibrations across bedding for closed walls with 3m burdens.

In determining the value of a, the explosive types and the charge lengths aretaken into account in the Holmberg analysis. However, it is possible for bothvariables to have an effect on the calculation of final values of K and a. In orderto determine this, all data collected across bedding (from closed walls) with 1.5and 3.0 metre burdens and having deck lengths of 24 and 30m were combinedand analysed.

Firstly, the data were separated into ANFO holes and ISANOL holes forsimilar burdens, and K and a values were determined for each. Any differencein the calculated constants would indicate that explosive type is a significantvariable likely to have an effect on the final K and a values. The results areshown in Table 1:

Table 1. The effect of the explosive type on measured vibrations.Burden(m)1.51.53.03.0

ExplosivetypeANFOISANOLANFOISANOL

K

456528515633

a

1.121.00.69.72

No. data points

78341812

rvalue0.710.710.61033

The calculated values of K and a for ANFO and ISANOL were very similar forboth burdens. This suggests that the explosive type is accounted for correctly inthe calculation of the a term in the Holmberg equation.

The effect of charge length was also considered, ie. all other variables wereeliminated to examine the effect of charge length on the calculation of the K anda values. The results are presented in Table 2:


Structures Under Shock and Impact VI

Table 2. The effect of the charge length on measured vibrations.Burden(m)1.51.53.03.0

Chargelength (m)

23302330

K

388480779575

a

1.040.980.690.60

No. data points

5358524

r'value

0.630.570.230.59

The results indicate a similar value of a and a slightly different value of K forthe 3.0m burdens. However there were limited data for 30m charge lengths with3.0m burdens, decreasing the confidence in the calculated material constants. Ifthe results obtained for the 1.5m burden data are used, then it can be concludedthat charge length is accounted for correctly in the calculation of a.

3.2.2 Analysis for open wallsThe previous analysis was based on data measured across bedding for closedwalls. A similar study was carried out for open walls, in which the geophoneswere located close to the centre of the mid-span of the exposed stopehanging walls. Similarly to the closed wall situation, only the effect ofconfinement was considered.In order to estimate the peak velocity in an open wall, the variations in the

experimental set-up must be considered. For example, the geophones were notlocated exactly within the outer skin of the exposed hangingwall, but wereplaced at varying distances into the hangingwall. Also, different unsupportedstrike lengths were exposed during data collection. The effect of the differentgeophone distances into the hangingwall can be analysed using a graph of log(a)versus log(PPV) as shown below:

-1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2log (a)

Figure 7: Measured vibrations for varying distances into the hanging wall.

The results suggest that for distances into the hangingwall ranging from 3 to8.5m a relationship can be established. However, the data from llm into the


Structures Under Shock and Impact 17 137

hangingwall appears to be behaving quite differently and similar to a closedwall situation. Consequently, these data were not used in the open wall analysis.

The unsupported span along strike exposed during each monitoring eventwas also considered. Figure 8 is a graph of log (a) versus log (PPV) acrossbedding with varying stope span lengths.

I0.» •

a

2

•

##

1 -0

A#.

#*•

— *

%0

&

8 -0.6 -0

log (a)

*

.4 -0

•

2 (

\

|

) 0.

Open span+ 6.0m• 9.0mO10.5mXl2.0mAl3.5m• 15.0m

Figure 8. Measured vibrations for different stope strike lengths.

The data show no pattern and therefore, strike length is not considered in thecalculation of the material constants for open wall situations. The materialconstants K and a for open wall spans were determined in a similar manner tothose for the closed wall situation and given in Table 3. Very poor correlationwas determined for the 3m burdens.

Table 3. Material constants K and a for open wall situationsBurden (m)

1.53.0

K

329237

a

0.810.46

No. data points

2419

r̂value

0.85037

The predictions of peak particle vibration (calculated using Equation 1 andbased on the distance from the charge to an observation point) for closed wallsituations analysed are presented in Figure 9. The constants K and a weredetermined for each case as described above. The damage predictions aresummarised and presented in Table 4, where the damage threshold is set at 700mm/s. This value has been suggested by Holmberg and Persson [1] and similarresults were reported by Andrieux, et al [4].



Distance (m)

Figure 9: Predicted vibration levels for closed wall situations.

Table 4. Extent of predicted damage into the stope hangingwall.Situation

1.5m burden, Isanol1. 5m burden, ANFO

3m burden, Isanol3m burden, ANFO

Extent of damagefrom charge

5m7.8m

L 7m10.8m

Extent of damage into thehangingwall4m (H/W hole)3m (Baser)

None (F/W hole)6m (H/W hole)3m (F/W hole)

A similar analysis was carried out for open wall situations, the results of whichare shown in Figure 10. There were too little data for 3m burdens to produce aresult, therefore the analysis is carried out for 1.5m burden data only. Theresults suggest that, for similar distances from a detonating charge, a decreasedlevel of vibration is expected for an open wall situation compared with a closedwall. Most of the pre-conditioning (which leads to damage) may be occurring inthe very near field (in an closed wall), when the holes are detonating adjacent tothe walls that will be exposed. Nevertheless, the levels of vibration predicted forthe first 2m across an open hangingwall are still very high and are likely tocontribute to additional rock mass deterioration.

Distance (m)

Figure 10: Predicted vibration levels for open wall situations.


Structures Under Shock and Impact VI 139

3.3 Borehole camera

Damage caused by any particular blasting event can be determined bycomparing pre-blast and post-blast conditions of the walls of the observationholes based on borehole camera surveys. In some cases negligible damage wasdetected when longhole winzes were being fired. For example, a small amountof water was present in the bottom of some observation holes (followingdrilling) and this water remained in the hole until the firing of the last lift of thecut-off slot. The bottom of the hole was placed approximately 2.5 to 3m fromthe long hole winze and the presence of water suggests that no opening ofcracks was taking place as the first four lifts of the winze were fired. Blasting ofthe last lift in the winze was done in conjunction with the remaining holes in theslot (usually very confined with 2.5m burdens) causing significant damage, asshown in Table 5. Subsequent firing of the main rings in a typical stope causedshearing across the observation hole, at the depths listed below:

Table 5. ActualBlasting eveiCOS1L5RING1RING2RING3&4

damage observed iit Depth

n a typicalof damage

;

obseinto5m5m>.4m8m

rvatthe

ionhan

hole.igingwall

As a comparison, the estimated depth of damage into the hangingwall from blastvibrations (Table 4, for hangingwall holes charged with ANFO having a 1.5mburden) is approximately 6.3m.

4 Conclusions

A comprehensive analysis of blast vibration data using the Holmberg-Perssoncriteria for blast attenuation has been used to predict damage boundaries aroundopen stopes in weak rock. The predicted results have been calibrated usinginstrumentation consisting of extensometers and observation holes. Theestimates of damage depth made through vibration analysis compare well withthe measured results using a TV borehole camera.

Acknowledgements

The authors wish to thank the management of Mount Isa Mines Limited for thepermission to publish the data shown in this paper.



References

[1] Holmberg, R. & Persson, P.A. Design of tunnel perimeter blasthole patternsto prvent rock damage. Proc IMM Tunnelling Conference, London,1979.

[2] Villaescusa, E. Tyler, D. & Scott, C. Predicting underground stability usinga Hangingwall Stability Rating, Procc. F' Asian Rock MechanicsSymposium, (H.K. Lee, Yang H.S & Chung S.K. Eds), Seoul, BalkemaRotterdam, pp. 171-176, 1997.

[3] Riihioja, K. Peak particle velocity prediction notes. Julius KruttschnittMineral Research Centre, Internal Memorandum, The University ofQueensland, Brisbane, 1995.

[4] Andrieux, P., McKenzie, C., Heilig J. & Drolet, A. The impact of blasting onexcavation design - A geomechanics approach. Proc of the 10th Sympon Explosives and Blasting Research, ISEE Meeting, Austin, 1994.


structures under shock & impact vi, c.a. brebbia & n ... · damage experienced in weak rock...

Documents