Professional Documents
Culture Documents
Mike Higgins Julius Kruttschnitt Mineral Research Center Brisbane, Australia Tom BoBo, Kirstin Girdner and Dr. John Kemeny Split Engineering Tucson, Arizona Vicki Seppala Cyprus Sierrita Corporation Green Valley, Arizona
ABSTRACT Optimal blast fragmentation is fundamental to all phases of comminution. Changes in blast design may affect efficiency and productivity of downstream processes such as crushing, milling and leaching, resulting in cost savings or losses. The basic steps in blast engineering are to design, implement and observe the results of a blast. Recently developed technology can enhance this process at every step to provide a structure for the optimization of blast fragmentation. A set of integrated software tools now exists for this procedure covering blast design, analysis and quantification of resulting fragmentation. A first pass approach is to characterize the rock mass to establish blasting parameters. The Kuz-Ram model, an accepted tool for fragmentation prediction, can be used to determine these parameters, but has its limitations, primarily in the area of fines prediction. Given these initial parameters, a blast design is implemented and the outcome observed. However, the implementation of the design (drilling and loading) often does not coincide with the original design due to unanticipated field conditions. Using records of the actual drilling and loading, analysis tools are available to contrast the differences between design and actual in terms of energy distribution, rate of maximum instantaneous charge and instantaneous burden relief, among others. Digital images of the broken rock are used to quantify blast fragmentation with image processing software, to produce size distribution data. This data can be back correlated to the Kuz-Ram model, closing the loop to derive optimal blast design parameters for a single rock mass zone. Results of a case study using a modified Kuz-Ram model at a copper mine are presented in this paper. The goal is to produce zone specific parameters that yield a consistent correlation of the Kuz-Ram size distribution with the measured results. These results show that a structured process for optimization of blast fragmentation is possible. International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Geology of the Sierrita-Esperanza-Ocotillo Deposit The Cyprus Sierrita-Esperanza-Ocotillo complex is located 25 miles south-southwest of Tucson, Arizona, USA on the southeast side of the Sierrita Mountain Range. West and Aiken (1981) reported that the Sierrita-Esperanza copper-molybdenum deposit is associated with a Laramide-aged, porphyritic facies of a large north-northwest-trending granodiorite batholith, which comprises the eastern side of the Sierrita Mountains, with extensive, high-angle faulting and folding" occurring throughout the range. The Sierrita-Esperanza deposit is composed principally of Mesozoic and Tertiary igneous rock. The Triassic, Ox Frame Volcanics are the oldest unit in the mine area. This unit was intruded by the Harris Ranch Quartz Monzonite and overlaid later by the Late Cretaceous Demetrie Volcanics. Late Cretaceous to Early Tertiary intrusive rocksincluding biotite quartz diorite, Ruby Star Granodiorite, and Ruby Star Quartz Monzonite Porphyryinvaded the preexisting volcanic and intrusive rocks to complete the premineral profile. Rehrig and Heidrick (1972, 1976) indicate that faults, joints and mineralized fractures are the notable structural features within the Sierrita-Esperanza-Ocotillo complex.
Local Ocotillo Pit Geologic and Structural Conditions The Ocotillo Pit was chosen for this study because of its relatively homogeneous geologic and structural zone characteristics. The Ocotillo Pit is comprised of the Ruby Star Granodiorite a relatively massive rock unit that is vertically jointed. Joint spacing is between 0.5-feet and approximately 3-feet, with the joint planes dipping out of the bench face. For examples of wall structure see Figure 1 in the Appendix (Typical Structure #1 and #2). The rock mass has a Uniaxial Compressive Strength (UCS) averaging 35,000 psi (240 MPa) and an average Specific Gravity (SG) of 2.65.
Testing Methodology It is widely experienced that in the mining industry things do not always go as planned or designed. With this in mind a method was devised to optimize blast fragmentation in a specific ore zone in the mine. Simulations were run on the designed patterns and compared to the actual patterns that produced the fragmentation of the muck. By comparing the fragmentation that was measured at the muck face to that which was predicted, "zone specific" coefficients can be developed from the modified Kuz-Ram prediction model to better predict fragmentation.
Blasting Parameters Four test blasts have been completed to date in the Ocotillo Pit, numbered 2162 to 2165, to start the optimization process. The same blast parameters were used in all four blasts: square pattern 24 feet by 24 feet, bench height 50 feet, subdrill 6 to 8 feet, hole diameter 10 5/8 inches. Explosives were a 1000 pound bottom charge of heavy ANFO, topped with ANFO to a stemming length of 26 feet. Drill cuttings were used for stemming. Wet holes that could be pumped and sleeved were loaded with both
Page 2
ANFO and heavy ANFO. Holes that could not be dewatered were loaded with slurry explosive. Details of the explosives used are listed in Table 1. Each test blast contained a number of holes that could not be dewatered, therefore the blasts could not be comprised of only ANFO and Heavy ANFO as had been the intended design. Timing was achieved with both downhole and surface delays. Downhole delays were 450 ms in the main body of the blast, with 600 ms in the perimeter holes, except in blast 2165, where 375 and 500 ms delays were used respectively. Surface timing in 2163, 2164 and 2165 was arranged in an echelon pattern, on a half V2-V3, with 17ms between holes and 42 ms between rows. Blast 2162 used the same downhole timing, with 17 ms between holes on the eastern side of the blast, 42 ms on the western side, and both 17 and 42 ms in series between rows. The different layout of blast 2162 was used to create a free face for the remaining blasts in this strip of the bench. The hole layouts and surface timing patterns are shown in Figure 2. Note that blast 2165 shows five holes not fired in the blast, and one hole in blast 2162. The area, and thus the tonnage, for blast 2165 has been adjusted to compensate for this.
Digital Image Analysis Technique For quantification of blast fragmentation, digital image analysis techniques were used. For specific post blast fragmentation analysis, digital images were taken using a Kodak DC 260 digital still camera from the muck face when the shovel had reached the center of the muck pile. The digital images were taken from the center of the shot pattern, to ensure that the image samples truly represented the fragmentation of each test shot. An average of fifteen images were taken of each testing location to quantify the fragmentation for the test. Digital image acquisition techniques were strictly followed to ensure against sampling bias. The images were taken as representative samples from each testing location. Once the images were taken from each testing location they were processed using the user-assisted (desktop) version of the Split image-processing program. The desktop version of the Split program has five major parts. The first part of the program concerns the scaling of images taken in the field. The second part of the program concerns the automatic delineation of the fragments in each of the images that are processed. The third part of the program allows editing of the delineated fragments to ensure high quality results. The fourth part of the program involves the calculation of the size distribution based on information from the delineated fragments. Finally, the fifth part of the program concerns the plotting or export of the size distribution results. Details of these steps are described in Kemeny (1994) and Girdner et al. (1996). The result of the automatic delineation is a binary image (2 graylevels, black and white) that contains white particles and a black background. Figure 3 shows a grayscale image of a muck pile and Figure 4 shows the binary image resulting from the delineation. (Some editing has also been performed in this step). The black areas in these images contain fine material too small to delineate in addition to the unfilled air space between particles. The black pixels are used to estimate the amount of fines.
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 3
Once the image processing is completed, the size distribution is then calculated from the delineated image. The results for the size distribution of the particles identified in Figure 4 are shown in a semi-log plot and equivalent sieve series in Figure 5. The images from each test blast were processed together to compute one size distribution curve for each. These results are shown together in Figure 6. Analysis of these results show little difference between the fragmentation of each shot, indicating that the single rock mass zone responded as expected when using consistent blasting parameters for each test shot . The P80 for these blasts ranges between 12.71 and 13.41 inches. This measurement provides a baseline to develop the zone specific parameters for the Kuz-Ram prediction model. Once the parameters are modified to generate the target P80 size of 10 inches a marked difference in blast fragmentation should be observed.
Blast Characterization The first step in developing site-specific factors for the Kuz-Ram model is to characterize the blasts to ensure that they are similar. This is done using the JKMRC JKSimBlast design and analysis software to describe the actual blasts and perform a detonation simulation to determine timing information. In this case, two characterizations have been performed. The first is a chart of cumulative explosive mass versus detonation timing; the second is a contour of explosive energy distribution. In addition, average powder and energy factors have been calculated for each blast, as shown in Table 2. These give an immediate, although generalized, indication that the blasts are essentially equivalent. The chart in Figure 7 shows the cumulative explosive mass detonated for each blast. The slope of each line gives the rate of consumption, or rate of application, of explosive in the rock mass. Since fragmentation can be affected by both the amount of explosive energy and its rate of application to the rock mass, this analysis can give a quick indication of differences or similarities between blasts that may not be directly obvious from the downhole and surface timing. In this case, it can be seen that blasts 2163, 2164 and 2165 are very similar, both in the slope of the lines and also in the number of holes detonating at any one time. There is some deviation in the latter part of the blasts, mainly due to the different number of rows in each blast and the resultant staggered timing of the perimeter holes. Blast 2162 shows a much higher rate of detonation in the early part of the blast, due to the extra timing between rows and the blast being split into two sides, producing a doubling-up of detonating holes at various times in the blast. As was shown in Figure 6, there was little variation in the fragmentation results for all four blasts. This could imply that there is a minimum slope of the cumulative chart above which consistent fragmentation is achieved, as is the case here. It could then be expected that there is also a maximum desirable slope, above which there could be an unacceptable increase in fines, damage or stemming ejection. Only an extensive history of blast results could show such a condition. The energy distribution is a calculation similar to that for a magnetic or electrical field energy around a wire (Scott et al, 1986). The result shows the concentration of energy experienced by the rock at any point at elevation 3625 feet, at about the mid-level of the bench. In this case, the analysis is static, i.e. timing is not taken into consideration. Figure 8 shows the contours for blast 2162 note the low concentrations around the lost hole, and around those holes with less explosive. Figure 9 gives the International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999 Page 4
relative percentage area occupied by each range of the explosive energy distribution for all four blasts. It can be seen that, despite local differences in loading, the blasts have a similar distribution of energy. Therefore, these two analyses, cumulative detonated explosive mass and explosive distribution, can be used as an indicator of consistency of blast design for a specific rock zone, particularly where variations in blast geometry require modifications to hole or timing layout.
Fragmentation Prediction A modified Kuz-Ram equation (Cunningham, 1983, 1987) was used to predict the size range for the actual blasts, and then to compare that with the standard design parameters and the actual fragmentation image analysis results. The modified Kuz-Ram equation includes a calculation for the rock factor based on the blast index developed by Lilly (1986). The equations and values used in the Kuz-Ram model are (all calculation units are metric, from Cunningham, 1983, 1987): A = 0.06 x (RMD + SGI + HF) (modified Lilly Blastability Index) where A = Rock Factor RMD = rock mass description = >0.1m joint spacing (20) + dip out of face (20) = 40 SGI = specific gravity influence = 25 x SG 50 = 16.25 HF = hardness factor = UCS 5 = 48 (UCS = 240 MPA / 35,000 psi) X50 = A x K-0.8 x Q0.167 x (115 E)0.633 (modified Kuznetsov equation) where X = mean fragment size K = powder factor Q = mass of explosive E = relative weight strength of explosive n = (2.2 14 B/D) x (1 W/B) x ((1 + S/B) / 2)0.5 x (ABS[(BCL CCL) / L] + 0.1)0.1 x L/H where n = uniformity exponent (used in Rosin-Rammler size distribution) B = burden (24 ft) S = spacing (24 ft) W = standard deviation of hole accuracy (0.1) D = hole diameter (10 5/8 in) L = charge length above grade (~24 ft) H = bench height (50 ft) BCL = bottom charge length (~21 ft) CCL = column charge length (~10 ft)
R = 1 e-0.693 (X X50)
where R = mass fraction smaller than X X = diameter of rock fragment n = uniformity exponent
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 5
Figure 10 shows a size distribution from a Kuz-Ram analysis for the standard blasting parameters compared to the measured values from the actual blasts using the Split program. The P80 target of 10 inches is indicated by the dot on the chart. In order to compare with the actual blasts, an average hole was determined for the two types of loading (ANFO & H-ANFO and Slurry) for each blast and a Kuz-Ram calculation performed on these. The results for blast 2163 are shown in Figure 11. This shows that there is little variation in the predicted fragmentation for the two types of explosive loading. The next step is to adjust the Kuz-Ram calculation so that the prediction matches the measured results from the Split program. The slope of the curve is affected by the uniformity exponent; decreasing the Rock Factor will shift the curve to the left. For the blast parameters used in the Ocotillo Pit, a factor of 1.65 on the uniformity exponent and a new coefficient for the Rock Factor of 0.043 will shift the size distribution to give closer agreement with the measured results, as shown in Figure 12. This factor can then be used in future blast design for this geologic zone, in conjunction with the design analysis indicators above, to guide the engineer in making variations to hole layout or loading for modified blast designs.
Optimization Having now determined a zone-specific characteristic indicator for blast design, and new coefficients for fragmentation prediction, it is now possible to make justifiable decisions for altering blast parameters to achieve a P80 of 10 inches. Several of the blast parameters can be altered and their affect on fragmentation as well as drill and blast costs can be calculated. Some proposed alternatives are: Reduce burden and spacing from 24 by 24 to 20 by 20 feet. This would increase the number of holes in a blast (and thus the amount of drilling and explosives) by about 40%. The predicted size distribution from this modification is shown in Figure 13. Increase hole diameter from 10 5/8 inches to 12 , and maintain deck lengths, which would increase the amount of explosive by one third. This would depend on available drill power and time to drill the larger diameter. Use a higher energy explosive. An increase of about 30% is required, but the split between shock and gas energy would need to be investigated first in comparison to the current explosives. Increase the length of the explosive decks by about 15%. However, this would reduce the amount of stemming, which would require a change of stemming material in order to properly confine the energy of the explosive. At present, there is very little or no stemming ejection, so it is unlikely that such a large reduction could be justified.
Blasting and excavation within the Ocotillo Pit has been temporarily suspended at Cyprus Sierrita. The recommendations for blast optimization shown above will be analyzed using a cost benefit type approach before operations resume in this area. The fragmentation from the blasts using the selected option will be measured and compared to the target fragmentation to ensure that the modified parameters give consistent results.
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 6
Conclusion This paper presents a simplistic approach to optimize fragmentation. The measurements described here are a cumulative size distribution from images taken at the center each blast, although in reality there will be significant variations between different regions, such as the top and toe of the blast. To fully model fragmentation it is first necessary to determine the pre-blast distribution and then measure the fragmentation throughout the muckpile. Both of these should then be compared with each other and with the energy and timing distribution in the blast. Obviously, this would be a more complex and time consuming process in both measurement and analysis. The aim here was to gain an initial understanding of blast performance so that first pass options for blast parameters could be assessed. This then allows the mine engineers to make a better informed decision on the next step to take, based on the potential improvement: either perform more detailed measurements and analysis under the same conditions; conduct trial blasts of the proposed options for comparison with the base case presented here, or extend the study to investigate blasting under other conditions and practices at the mine.
Acknowledgements We would like to thank Cyprus Sierrita for their cooperation in providing us with the site and allowing us to perform this study and our continued work in blast fragmentation optimization. We would specifically like to thank the powder crew who makes it all happen, from the planning of the shots to the ultimate initiation of the shot. We would like to thank Southwest Energy for their cooperation with the blasting agents and information that was required for this study.
References Cunningham, C.V.B., 1983 The Kuz-Ram model for prediction of fragmentation from blasting: Proc. 1st Int. Symp. on Rock Fragmentation by Blasting, Lulea, Sweden, 2: 439-453 Cunningham, C.V.B., 1987 Fragmentation estimations and the Kuz-Ram model four years on: Proc. 2nd Int. Symp. on Rock Fragmentation by Blasting, Keystone, Colorado, USA, August, 449-461 Girdner, K., Kemeny, J., Srikant, A. and McGill, R., 1996, The Split System for Analyzing the Size Distribution of Fragmented Rock, Proceedings of the FRAGBLAST-5 Workshop on Measurement of Blast Fragmentation, ed. Franklin, J. and Katsabanis, T., Montreal, Quebec, Canada, pp. 101-108. Lilly P.A., 1986 An empirical method of assessing rock mass blastability: Proc. AusIMM/I.E. Aust. Newman Combined Group Large Open Pit Mining Conference, 41-44, 89-92 Rehrig, W. A. and Heidrick, T. L., 1972, Regional Tectonic Stress during the Laramide and Late Tertiary Intrusive Periods, Basing and Range Province, Arizona: Arizona Geological Society Digest, Vol. X, p. 205-228. International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999 Page 7
Rehrig, W. A. and Heidrick, T. L., 1976, Regional Fracturing in Laramide Stocks of Arizona and its Relationship to Porphyry Copper Mineralization: Economic. Geology, Vol. 67, p.198-213. Scott, A., et al, 1986, Open Pit Blast Design Analysis and Optimization: JKMRC Monograph Series in Mining and Mineral Processing, JKMRC, Queensland, 1986. West, R. J. and Aiken, D. M., 1981, Geology of the SierritaEsperanza Deposit, Pima Mining District, Pima county, Arizona, Advances in Geology of the Porphyry Copper Deposits, Southwestern North America, Ch. 21, ed. S. R. Titley, University of Arizona Press, Tucson, Arizona, 1981.
APPENDIX 1
Table 1 - Explosives Products Density (g/cc) Energy (cal/g) Velocity Of Detonation (ft/sec) Weight Strength Relative to ANFO Volume Strength Relative to ANFO ANFO 0.82 831 12,800 1 1 H-ANFO 1.25 781 13,200 0.94 1.44 Slurry 1.2 722 17,700 0.87 1.27
Table 2 Average Actual Powder and Energy Factors blast short tons explosive (lbs) energy (MJ) powder factor (lbs/ton) energy factor (MJ/ton) 2162 80,420 40,166 57,814 0.50 0.72 2163 72,415 37,271 51,460 0.51 0.71 2164 60,406 28,235 40,455 0.47 0.67 2165 60,031 28,815 42,665 0.48 0.71
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 8
APPENDIX 2
Figure 1
Wall Structure #1
Wall Structure #2
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 9
Figure 5 Split-Desktop size distribution results from the muckpile image shown in Figure 3.
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 10
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 11
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 12
Figure 11 Blast 2163 ANFO & H-ANFO (A) and Slurry (B) Calculated Size Distributions
International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999
Page 13
Figure 12 Standard Blast Parameters with modified Uniformity Exponent and Rock Factor Coefficient
Figure 13 Burden and Spacing reduced to 20 ft x 20 ft International Society of Explosive Engineers 25th Annual Conference on Explosives and Blasting Technique Nashville, Tennessee, February 7-10,1999 Page 14