Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

The Effect of Powder Factor on Fragmentation

C. Mick Lownds, PhD


Viking EXPLOSIVES & SUPPLY, Inc.,
12955 Courthouse Blvd., Rosemount, MN55068

Abstract:

This paper examines the hypothesis that the fragment size distribution of blasted rock
contains information about the changes in fragment sizes with distance from the borehole,
and that this information can be interpreted and used to predict the effect of powder factor
on fragmentation. A set of simple relationships between the fragment size distribution
curve and the dependence of fragmentation on powder factor is proposed. The equations
derived are compared with published data on fragmentation in small scale blasting. Two
data sets were analyzed, and the results confirm the hypothesis.

Introduction

Blasting in mining is affected by geology, excavation geometry, hole pattern, timing, the
explosive and stemming. The local geology is outside the control of the mining engineer,
and the excavation geometry (bench height, stope width, etc.) is usually subject to
constraints unrelated to blasting. Hole pattern, explosive and stemming are often
conveniently combined into the parameter called powder factor (or energy factor or
ANFO equivalent powder factor). This is the primary variable of blasting, and is under the
full control of the mining engineer or blaster. No other design variable of blasting has a
greater impact on the cost of blasting or on the results of blasting.

Fragmentation is a vital result of blasting, particularly in hard rock mining, where it is


usually the variable that controls blast design. Fragmentation is often described
qualitatively because of the difficulty of measuring the fragment size distribution. Even
when the distribution is known, blasting is often described in terms of a single parameter or
modulus of the distribution. The most commonly used fragmentation moduli are (a) the
screen size through which 80% of the blasted material will pass (b) the 50% passing size,
and (c) the characteristic size from the Rosin Rammler distribution function.

Regardless of the exact definitions of powder factor and fragmentation, the quantitative
relationship between them is important in blasting; it is especially important in hard rock
mining for two reasons:

First, a quantitative description of the dependence of fragmentation on powder factor is


the basic element in building a model of mining costs, from drilling to crushing (1) or
through crushing to downstream comminution (2). Even when quantitative cost models are
not attempted, the blast designer is well served by knowing even the approximate change
he should expect in fragmentation from a contemplated change in powder factor. Second,

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 101 1 of 9
the relationship between fragmentation and powder factor is so central to blasting that for
a blasting model to be of any value it must predict this property of blasting accurately. As
computer models of blasting become more available and potentially more useful, their
validity should be measured critically against how well they handle the fragmentation vs
powder factor relationship.

The work described here is part of an ongoing effort (3,4) to model fragmentation in
taconite. It has been shown (3) that the relatively scarce data on taconite fragment size
distribution after blasting agrees with the proposal by Stagg, Rholl, Otterness and others
(5,6,7) that there is a zone of fracture around each borehole that is not significantly
affected by preexisting jointing or by interaction of fractures from adjacent boreholes.
Fragmentation in this region is thought to be due primarily to radial fracture, which is the
clearly visible mechanism in most model scale blasting.

Hypothesis

The fragment size distribution and the dependence of fragmentation on powder factor are
intimately related. This hypothesis results from the earlier study (4) of fragment size
distributions. The purpose of this work is to examine the hypothesis.

The intensity of radial fracture around a borehole can safely be assumed to decrease
monotonically with distance from the hole. It can also be safely assumed (at least in
homogeneous rock) that the fine fragments are produced in this region. The details of how
quickly fracture intensity decreases are encompassed in the fragment size distribution
curve for the smaller fragment sizes. In principle, a surface will exist around each hole on
which all fragments eventually formed will be about the same size, and within which all
fragments will be smaller than those on the surface. The surface for the 50% passing size
encompasses a known volume (50% of the blast). If its shape were known, its distance
from the borehole in various directions would be known. Thus the distribution curve (at
least close to the borehole) contains fragment size vs distance information. Decreasing
burden and spacing decreases the volume broken per hole, and the distance of the 50%
surface from the borehole must therefore decrease. Thus the fragment size distribution can
be used to implicitly derive fragment size with distance from the hole, which can in turn
yield the fragment size distribution for a new hole spacing and therefore a new powder
factor.

This logic assumes that the fracture mechanisms are not affected by changes in powder
factor. The models described here, both physical and mathematical, are based on this
assumption. It is also assumed that the larger size fragments are strongly influenced by the
explosive. There are certainly cases in practical blasting where the 80% passing size results
largely from the frequency of pre-existing joints. Such cases are not considered here.
Comparison with experiment is offered as a test of the hypothesis and of the assumptions
about fragmentation mechanisms.

Fragment Size Distributions

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 102 2 of 9
Following Stagg, Otterness & Siskind (7), and Lownds(4), a power law was chosen to
represent the material smaller than about x~so, the screen mesh size through which 50% of
the blasted material will pass. This distribution will be used in the form

P(x) = k.xª for P(x)<~0.5, x <~x50 (eq. 1)

where P(x) = mass fraction passing size x


k,a are constants.

The power 'a' was found from the measured fragmentation (as described below) by linear
regression of the data for 0 < x<~x50. Since such regression fits are least reliable at the
extremes of the range, it was decided to use x 40 for the size modulus of the small fragments
rather than the more commonly used x50.

For fragment sizes larger than x50, the Rosin Rammler (RR) or Weibull distribution is
preferred, since it is now widely used to describe fragmentation by blasting. It is used here
in the form

P(x) = 1 - exp( -(x/xc)ü ) for P(x)>~ 0.5, x >~x50 (2)

where P,x are the mass fraction passing and size, as above
xc = characteristic size
n = exponent; high values of 'n' describe narrow size distributions.

Although the Rosin Rammler (RR) distribution has xc as size modulus, it is more common
to use x80, especially when examining the effects of powder factor.

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 103 3 of 9
There is thus a simple relationship between the exponents 'b' in the powder factor equation
(3) and 'a' in the size distribution equation (1). This model is illustrated in Figure 1a.

A logical extension of this approach for the 80% passing size is now considered. Again, the
analysis is simplified by considering only a plane through the charged portion of the
boreholes. If the fracture patterns radiating from each hole continue to have uniform
intensity in all directions as distance from the borehole increases, the surface on which x80
is formed will be an area within the area formed by the 3 or 4 nearest neighbor holes. This
model is illustrated in Figure lb, and results in a new relationship between the powers 'b'
and the distribution parameters for the larger fragments:

b = 2.2¹´² / ( n.(1-P)¹´².1n(1-P) ) (11)

= 3.93/n for P = 0.8

It will be shown below that measured fragmentation does not support equation (11).

A concept more relevant to practical blasting is that described by Stagg and coworkers
(5-7): In sequential blasting where free faces are important, fragmentation occurs as a
near-field zone of radial fractures which extend predominantly in the direction of the free
face. In that case the 80% surface would be roughly parallel to the free face. This situation

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 104 4 of 9
is illustrated in Figure 1c, and results in a second equation for 'b' as a function of the
distribution parameter for the large size fragments:

b = ( 2.n.(1-P).ln(1-P) / P )¯¹ (12)

= - 1.243/n for P = 0.8

Comparison with Measured Fragmentation

The published results of model scale blasting by Otterness and others (6) and Petrosyan (8)
have been studied. The work reported in (6) involved small scale bench mining at the
University of Missouri-Rolla's experimental mine in a massive Dolomite deposit. A total of
29 shots was fired, of which 23 involved powder factor changes due to changes in spacing,
burden and hole diameter. These 23 shots were analyzed.

In examining 40% passing size here, the mass of each sieve fraction was compared to the
design mass of rock per hole, since the 40% sizes originate immediately around each hole.
The powder factor used was also the design powder factor. For the 80% passing analysis,
the sieved masses were compared to the actual mass broken per hole, and actual rather
than design powder factor was used. A slight improvement in fit resulted from this method
of analysis over others tried.

The experiments in ref (8) were done in concrete blocks 500x500x400 mm. Considerable
attention was paid in the original work to scaling, and the authors satisfied themselves that
the model work accurately represented full scale blasting. Two series of tests at so-called
short and long delays were done, each covering a wide range of powder factors. The
published data do not cover fragments smaller than about x50, so these tests could only be
analyzed for the large (x80) fragment sizes.

For each shot in both references (6) and (8), (P, x) pairs were calculated from the raw data
provided. Linear regression of In(ln(1-P)) vs 1n(x) was applied only to pairs for which
P>~0.5 to derive the RR parameters for the larger sizes. This involved a minimum of 3 but
usually 4 or 5 data pairs. For each of the 23 shots analyzed in (6), linear regression of
1n(P) vs 1n(x) was applied only to pairs for which P<~0.5 to derive the power 'a'. Again
the number of data points was 3 to 5. All the derived data, including correlation
coefficients ( R² ) for the fits to the power law or the RR equation are shown in Table 1.

The best fit fragment size distributions for P<~0.5 and for P>~0.5 were used to calculate x
40 and x80 from the power law and RR distributions respectively. These derived
fragmentation moduli for each shot are plotted against the corresponding powder factors in
Figures 3-5. Also shown in Figure 3 are the lines reflecting the powder factor exponents
derived in equations (10) and (12). Since the derivation above focused on the exponents
and ignored the constants, the vertical positions of the theoretical x 40 vs q and x80 vs q
curves were adjusted to fit the data.

As can be seen from the graphs, the slopes of all the theoretical curves are in good

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 105 5 of 9
agreement with the values of x40 and x80 derived here from the published measurements.
The graphs also show a tendency in all the x80 data sets for the fragment size (x80) to be
constant for powder factors above a certain value. Basic statistical analysis of the data sets
of Tables 1 and 2 leads to the following comparison of the value of the exponent 'b' from
the theory and from data analysis:

Theory, eqns 3 & 10 or 12 Regression, 1n(xp) vs 1n(q)


b +- b +-
Ref(6) X40 -1.449 0.060 -1.409 0.098
Ref(6) X80 -1.128 0.094 -0.970 0.220
Ref(8) X80 Long Delays -1.026 0.062 -1.035 0.159
Ref (8) X80 Short Delays -2.119 0.036 -2.113 0.219

Discussion and Conclusions

There appears to be a change in the mechanism of formation of the 80% passing size as
the powder factor becomes relatively very high. Both data sets examined here show a
roughly constant x80 for powder factors above a critical value, although this value was not
the same for the two sets of data. Since these small scale blasts were done in massive
material, jointing is not the explanation. No other explanation for this observation can be
offered here, but it is potentially of great practical and economic significance in blasting
and deserves further investigation.

All the data sets analyzed here support the hypothesis within the experimental error. The
agreement between theory and experiment for the 40% passing sizes (Figure 2) is
excellent. The agreement for x80 from Ref (6) is less striking, but still within the scatter of
the data. Unfortunately, Petrosyan's experiments (8) did not have many shots at low
powder factor, especially in the long delay series.

Nevertheless, the agreement between the experimental and theoretical values for 'b'
derived from his work is strong support for the theory. In this context it should be noted
that the two series of Ref (8) exhibit significantly different values for the RR exponent 'n',
and significantly different values for the exponent 'b'. The values for the long delay series (
n = 0.59 and b = -2.1 ), are unusually low and unusually high respectively, supporting the
inverse relationship between them derived in equation (12).

Comprehensive data sets are required for the kind of analysis presented here; such data
are scarce, and are generally available only for model scale blasting. In full scale blasting,
the RR exponent 'n' is typically in the range 0.7 to 1.1, commonly about 0.9. This theory
(eq 12) predicts a corresponding 'b' for the 80% passing of -1.38. In comparing values of
exponents, a less negative number will be called higher; measured values of 'b' as high as
-0.8 for the 50% passing size have been reported (9). If the analysis here is valid, such high
values of 'b' must arise from the influence of pre-existing joints on fragmentation. As the
role of such fractures increases and the influence of the explosive in determining
fragmentation decreases, the value of 'b' tends to zero. From this hypothesis, then, the

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 106 6 of 9
value of 'b' in full scale blasting with an RR exponent n = 0.9 will be about equal to -1.38
in massive (unjointed) formations, and greater than about -1.38 in jointed formations.
Further work on the effects of jointing on the value of 'b' is planned.

Figure 2 shows that the modulus of fragmentation becomes much more sensitive to powder
factor (higher negative values of 'b') the closer the modulus is to the largest size fragment.
For the 95% passing, which might correspond to the oversize, an RR exponent of about 1
gives a value of 'b' of about -3. This means that a 10% decrease in powder factor would
increase the 95% passing size by a factor of 1.1³ = 1.33, or 33%.

The model represented by Figure lb and equation (11) predicts very high negative values
for 'b', which are not supported by the data examined here or by any other data known to
this writer.

The hypothesis allows a value of the powder factor power law exponent 'b' to be
calculated from a knowledge of the fragment size distribution. If the 40% passing size is
formed in a volume of the blast where the explosive overwhelms pre-existing jointing, the
value of 'b' appears to be reliable over a wide range of powder factors. For the 80%
passing size, the value of 'b' which is derived from the distribution is, at worst, a lower
limit. Both very high powder factors and pre-existing jointing will increase the value of 'b'.
Values as high as zero are theoretically possible in heavily jointed formations, and have
been observed in small scale tests in unjointed materials.

Acknowledgments

The author is grateful for the support of Viking Explosives & Supply, Inc. and for the
encouragement of Jack Eloranta of Minntac Mine, USX Corporation.

References

1) Nielsen, K. "Sensitivity Analysis for Optimum Open Pit Blasting", Proc 11th Ann. Conf
on Explosives & Blasting Techniques, ISEE, 1985.

2) Eloranta, J. "Efficiency of Blasting vs Crushing and Grinding", Proc 23rd Ann. Conf on
Explosives & Blasting Techniques, ISEE, 1997.

3) Lownds, C.M. "Prediction of Fragmentation Based on Distribution of Explosives


Energy", Proc 11th Symp on Explosives and Blasting Research, ISEE, pg 286, 1995.

4) Lownds, C.M. "Derivation of Fracture Intensity from Measured Fragmentation", Proc


12th Symp on Explosives and Blasting Research, ISEE, 1996.

5) Stagg, M.S., Otterness, R.E., Siskind, D.E., "Effects of Blasting Practices on


Fragmentation", Tillerson & Wawersik, Editors. 1992, Balkema Rotterdam, ISBN90 5410
045 1. pg 313.

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 107 7 of 9
6) Otterness, R. E., Stagg, M. S., Rholl, S. A., Smith N. S., "Correlation of Shot Design
Parameters to Fragmentation", Proc 7th Symp. on Explosives and Blasting Research,
ISEE, pg 179, 1991.

7) Stagg, M. S., Rholl, S. A., "Effects of Accurate Delays on Fragmentation for Single Row
Blasting in a 6.7m Bench", 2nd Int Symp on Fragmentation by Blasting, ISEE, Keystone,
CO, pg 179, 1987.

8) Petrosyan, M. I., Rock Fragmentation by Blasting, A. A. Balkema, Rotterdam, ISBN 90


6191 902 9, 1994.

(9) Kuznetsov, V.M., "The Mean Diameter of the Fragments Formed by Blasting Rock",
Soviet Mining Science, Vol 9, pp 144-148.

Copyright © 2000 International Society of Explosives Engineers


1997R - The Effect of Powder Factor on Fragmentation - P 108 8 of 9
Copyright © 2000 International Society of Explosives Engineers
1997R - The Effect of Powder Factor on Fragmentation - P 109 9 of 9

You might also like