Introduction
Misfires or any other type of malfunction in a surface blast, be it construction, quarrying, or large open-pit operation, is easily detected. Once the source has been identified, remedial action can be taken to ensure productivity and safety (Farnfield, 1966; Segarra et al, 2009). However, the situation in an underground operation is very different, especially in stope blasting or any other mass mining operations. In such cases, fan drilling and vertical retreat mining (VRM) are the usual methods employed. The boreholes in this case could have a single column of explosives or incorporate several explosive decks with various delays.

Any missing or misfired holes in this case are extremely difficult to detect after the blast because it is not possible to uniquely identify the source of a poor blast. In such cases, the presence of oversize fragments cannot be ascribed to the blast in question.

Standard blast monitoring through commercially available vibration monitoring is sometimes employed to monitor some of these blasts, but for reasons explained in this article, it is not possible to carry out a truly diagnostic analysis of such blasts and identify any particular malfunctioning blast hole.

Experimental Program
The study was carried out at a deep underground mine. A typical cross section of a stope blast design at the mine is shown in Figure 1. The borehole diameter was 162 mm, with varying depths as shown. A booster-sensitive packaged water-gel slurry explosive with a density of 1.15 g/cc was used throughout. The quoted effective RWS (i.e. gas expansion up to 100 MPa) by the manufacturer of this product was 96, with a VOD of up to 4,900 m/s. Each explosive column is bottom initiated with a 450 g Pentolite booster. Blast monitoring at the mine is usually carried out with a commercially available monitor (Instantel Minimate) employing three-component geophones.

The first step in this study was to ascertain the adequacy of this vibration monitor in assessing the production blasts at the mine. For this purpose, a comparative study was carried out with high-frequency accelerometers placed side by side with Minimate. The usual precautions were taken in mounting the geophones and the accelerometers (Farnfield, 1996) on to the surface of the rock underground. The accelerometers had an upper frequency range in excess of 10 kHz and the usual acceleration limit was 100 g. The placement of the two types of monitors (i.e., geophones and accelerometers) is shown in Figure 2. The data for the accelerometer recording was carried out with both an analog data acquisition system with a bandwidth of DC to 45 kHz, and a digital data acquisition system with a sampling rate of 1MHz.

Analysis
A typical vibration record obtained with the geophones (i.e., Minimate) along longitudinal, transverse and vertical directions is shown in Figure 3. Individual delay rounds are clearly seen in the vibration record, despite the usual scatter in firing times. The highly unequal particle velocity amplitudes among the delay rounds are clearly evident. The data highlights the need for multi-axial recording of vibrations, but the pronounced variations along each axis can be attributed only partly to the varying charge weights in the holes.

A more diagnostic comparison between high-frequency accelerometer recordings and those by geophones is illustrated in Figure 4 for the same blast. Closer examination of the two records, i.e., particle velocity records from the geophones and integrated accelerometer data, from the same blast shows only a qualitative agreement between the two types of recordings. The peak particle velocity values obtained by integration of the accelerometer data show consistently higher amplitudes than those obtained with direct geophone recordings, despite both geophones and accelerometers placed in the same location.

The difference between the two types of recordings is further amplified when one examines the frequency content of the individual wavelets corresponding to each delay round. The respective frequency spectra of particle velocity from the geophone and accelerometer recordings are shown in Figure 5, for a single delay round. Whereas, the energy content in the geophone recording is seen to be confined below 300 Hz, the same particle velocity values obtained from integration of accelerometer data shows the maximum amplitude closer to 1 kHz or higher. In most cases the peak particle velocity derived from accelerometer data greatly exceeds those obtained by geophone recordings.

These differences have serious implications when one compares the resultant particle velocity records obtained by the two modes of recording. This is illustrated in Figure 6, which shows the comparison between the resultant particle velocity for the geophone and accelerometer recordings for a typical production blast with pyrotecnic delays. The scatter in firing times between the delay rounds is obvious but to be expected. Similarly, the varying particle velocity amplitudes observed with both geophone recording and accelerometer recording among the events is to be expected, but due only partly to varying charge weights and the travel paths for the seismic waves, and related geological factors. The more serious observation is the fact that for the same charge weight per delay and distance, the peak particle velocity obtained from accelerometer recording is often greater than by a factor of two compared to peak particle velocity obtained with geophone recording.

For example, the event at approximately 300 ms shows the resultant particle velocity to be 80 mm/s, whereas the data derived from the accelerometer station at the same location shows it be 190 mm/s. Similarly, the event arriving at 500 ms, the geophone data yields a resultant particle velocity of 90 mm/s compared to 220 mm/s for the corresponding accelerometer recording. This difference cannot be explained away simply because of charge weight differences or the geological factors involved, as both accelerometer recording and geophone recordings correspond to identical charge weights, geological conditions and seismic travel paths.

Particle Velocity Amplitudes and Explosive Energy
The recorded vibration amplitudes can also be related to the explosive energy yield at the source, as the radiated energy can be shown to be proportional to the square of the amplitude of the particle velocity. Throughout this analysis it is assumed that the seismic energy (i.e., blast vibration energy) is directly related to the total explosive energy released in the borehole. This is a valid argument so long as the source function remains unchanged, i.e., there is no change in the type of explosive used, or the decoupling condition in the borehole is changed, or there is a significant change in the initiation mode employed in the blast that would result in a drastic change in the energy partitioning between shock and gas energy from the explosive (Mohanty, 2009). On this basis, the energy yield from the various explosive loads for the blast in question with accelerometer data is shown against designed delay times is shown in Figure 7.

For a true comparison of energy levels, the derived energy values have been linearly scaled (i.e., normalized) with the corresponding charge weight for each delay round, as in earlier works (Mohanty et al., 1997). The effect of varying distances to the corresponding explosive charges on the amplitude of the particle velocity is considered minor in this case because of the relatively large distances involved between the stope blast and the monitoring stations. Therefore, the energy yield values shown in Figure 7 would represent the actual specific energy estimates from each of the explosive columns in the blast, i.e., specific energy per kg of charge.

The radius of each circle in Figure 7 represents the relative specific energy/kg with respect to the maximum value obtained for selected blasts, irrespective of the delay time recorded. If each explosive charge in the blast were to yield the same specific energy (i.e., MJ/kg), as expected, all of the circles shown in the figure should have had the same radius, since there were no variations in explosive type or its diameter or its mode of intiation. However, as the figure shows there is considerable variation in specific energy release from the various explosive charges, with values ranging from 100% to total failure.

For this study, a very conservative criterion of specific energy release is applied, i.e., 0% to 20% of the expected energy release (with respect to the normalized maximum particle velocity amplitude) considered a failure, >20% to <40% considered a partial energy release, and >40% to 100% considered full energy release. On that basis, only three holes in the blast appeared to release full energy, six only partial energy, and seven holes with negligible energy. Another way of illustrating the specific energy release figures for a number of regular stope blasts is shown in Figure 8. It shows 28 blast holes to release less than 20% of the expected explosive energy, and only 22 blast holes yield nearly the full energy (i.e., 40% to 100% of the expected energy).

Discussion and Conclusion
There is considerable data available in open literature on measurement and analysis of blasting vibrations from mining and quarrying operation. Much of this information, however, relates to estimation of damage potential to civil structures and to a lesser extent, to pit slope stability and blast-induced damage to rock mass, and detecting obvious misfires or firing time deviations in a blasting round. In contrast, near-field vibration measurements and their diagnostic use in examining explosive performance and blast design has received much less attention (Mohanty, 1997; Fleetwood et al, 2009).

The results of this study show there is significant discrepancy between the expected vibration amplitudes in terms of specific vibration energy released and that actually measured in the stope blasts under investigation, all other conditions remaining the same. In this case, aside from the usual firing time deviations with pyrotechnic detonators, typically more than one-third of the blast holes release very little energy in the form of transmitted seismic energy over multiple accelerometer stations. As is well known, low vibration amplitudes can be caused by a variety of causes such as varying geology (Fleetwood and Villaescusa, 2011), poor loading practice, hole deviation resulting in dislodgement of or damage to explosive charges in adjacent blast holes, or the quality of the explosive products itself. However, these apparent blast malfunctions cannot be attributed to the above causes in the great majority of the production blasts in question, especially since the blasts were monitored at multiple accelerometer stations.

Although none of the above factors can be ruled out in some individual cases, the widespread malfunctions exemplified in this study point to a more systemic cause for these apparent failures such as lithologic factors. The effect of the latter is evident from wide-scale blast hole deformation due to high ambient stresses that are characteristic in some of the stopes at this deep mine, and is part of the continuing investigation. The study also aims to provide not only superior blast diagnostics but also improved blast designs to counter these deficiencies.

Acknowledgement
The authors are grateful for the financial assistance provided by the National Science and Engineering Research Council of Canada and Vale Canada during the course of this research. In addition, the extensive assistance provided by Vale for field monitoring of vibrations in one of their underground mines is also gratefully acknowledged.

Mohanty and Zwaan are professors at the Lassonde Institute of Mining at the University of Toronto, and Malek is a rock mechanics engineer working with Vale’s Copper Cliff mine in Ontario, Canada. This article was adapted from a paper Mohanty presented at the 2013 International Society of Explosives Engineers conference, which took place during January in Nashville. The next ISEE conference will be held in Denver, Colorado, February 9-12, 2014 www.isee.org).

References
Farnfield, R., 1996,”So you think you are monitoring peak particle velocity”; Proc. 12th Symp. on Ann. Symp. on Explosives and Blasting Res.; Int. Soc. of Explosives Engrs.; p. 13-20.

Fleetwood, K.G., Villaescusa,E., Li, J. and Varden, R.; 2009,“Comparison of traditional near-field vibration prediction models with three-dimensional vibration scaling and blast wave energy”; Proc. 9th Int. Symp. on Rock Fragmentation by Blasting (FRAGBLAST 9), Sanchidrian, J.A.;ed.; CRC Press, p. 579-588.

Fleetwood, K.G. and Villaescusa, 2011, “Measured results of the influence of of a large- scale fault on blasting vibrations in sub-level open stoping”; Proc. 37th Ann. Conf. on Explosives and Blasting Tech.; Int. Soc. Explosives Engrs., ISEE, p. 1-13.

Mohanty, B. and Yang, R.,1997, “Blasting vibrations and explosives performance”; Proc. 13th Ann. Symp. on Explosives and Blasting Res., Int. Soc. Explosives Engrs.,ISEE, p.15-28. Mohanty, B.; 2009, “Intra-hole and inter-hole effects in typical blast designs and their implications on explosives energy release and detonator delay time—A critical review”; Proc. 9th Int. Symp. on Rock Fragmentation by Blasting (FRAGBLAST 9); Sanchidrian, J.A; ed., CRC Press, p.23-31.

Segarra, P, Sanchidrian, J.A., Lopez, J.M.; Querol, E. and Guiterrez, J., 2009, “Assessment of the error of blast vibration measurements”; Proc. 9th Int. Symp. on Rock Fragmentation by Blasting (FRAGBLAST 9); Sanchidrian, J.A; ed., CRC Press, p.551-560.