Printer Friendly

Where did it go?

Matthew Randall and Adam Wheeler discuss the modelling of blast displacement and dilution control at Minas de Rio Tinto

The open pit copper mine at Rio Tinto, situated in the Huelva province of southern Spain (MM, October 1998, pp. 195-199), has been a centre of mining activity for more than 2000. Evidence of extensive mine workings dating back to Roman times can be seen in the final slopes of the current open pit and these along, with other areas of the mine, are being preserved as sites of archaeological interest.

The changing fortunes of the mine during its life have seen production switch from predominately silver, to sulphuric acid and copper from the massive pyritic ores, to gold from the gossan cap, and to low grade copper from the stockwork mineralisation of Cerro Colorado. The advances in technology form milestones in the development of the mine with the use of steam shovels in the 1890s, the introduction of cyanide leaching of the gossan ore, and the use of ANFO for large scale blasting.

Geological setting and mining method

The mineralisation at Rio Tinto, typical of a volcanogenic massive sulphide, is characterised by submarine deposition of sulphide accumulations fed by a number of near vertical breciated feeder pipes. Folding during the Hercynian orogeny and subsequent erosion has left an exposed cap of gossan with a 20-30 m thick layer of rich secondary sulphide overlying an irregular pyritic and chalco-pyritic stockwork. There are two distinct periods of primary mineralisation and the copper minerals appear to have been deposited in the second phase, accompanied by intense chloritisation of the host rocks (Pryor et al, 1972). The red gossan cap, flanked by ancient and modern mine workings, was aptly name Cerro Colorado.

Today, Cerro Colorado continues to be mined on 10 and 12 m benches using a fleet of P&H 2100 BL electric shovels, Caterpillar 994 wheel loaders and 175 t Caterpillar 789 haultrucks. The total installed capacity of the fleet is 95,000 t/d with an ore production of 10 Mt/y.

Drilling is carried out with two Buycrus Eyre 45R and one GD SK75E unit, drilling a 250 mm hole to a depth of 11.2 or 13.7 m depending on bench height. With the introduction of Emulsions (Emumex 8500) in favour of heavy ANFO, the blast patterns have been modified from a 5.5 x 6.5 m or 6.5 x 7.8 m grid to 6.7 x 8.0 m. Recently staggered patterns have been adopted in favour of the conventional square pattern in order to improve the distribution of energy.

The majority of the blasts are four or five rows in depth and typically range from 100 to 200 holes. Due to the sporadic nature of the stockwork mineralisation most production blasts (excluding areas of waste stripping) are made up of a number of ore and waste composites that have to be blasted together. Blasts can rarely be cut according to mineralogical boundaries and controlled blasting is essential to minimise dilution.

The ore is loaded and hauled to the primary crushers where it is reduced to minus 175 mm. The coarse ore is further reduced to minus 19 mm before milling to a final particle size of minus 210 microns. Copper concentrate is produced in a two-stage flotation process and is transported by road to Huelva for smelting.

Grade control and reconciliation

As part of a study of the grade control procedures at Rio Tinto, the effects of blast induced displacement were investigated and a modelling procedure developed to predict dilution. The model does not attempt to simulate the complex thermodynamic or fragmentation processes of a blast - rather, it is aimed purely towards reproducing observed blast displacement characteristics at this particular site.

Whilst there are numerous references in the literature to methods of blast optimisation to minimise overall costs, these studies have generally concentrated on fragmentation and have not considered the effects of ore loss and waste dilution. The authors set out to demonstrate that control of blast displacement and fragmentation need to be considered together in order to reduce costs and maximise profits.

Grade control at Rio Tinto is based on blast-hole samples which are analysed for Cu, Zn, As and Sb. Extensive studies of duplicates has shown that the mean grade variance of such samples is of the order of 0.2% Cu. The blast hole samples are then used to locate ore and waste contacts and define ore and waste composites (zones) within a blast perimeter that will return the maximum profit. This procedure (OPTIMA) is automated within the Datamine mine planning system and uses Datamine's Floating Stope Optimiser to calculate the position of the contacts (Randall and Wheeler, 1998).

The floating stope optimisation method requires that the blast be discretised into blocks of suitable size, and that the attributes for each block are calculated from the blast holes using one of the standard estimation methods, such as Inverse Power of Distance (IPD), Ordinary Kriging (KRG3D) or Polygonal interpolation (POL3D). The optimal contacts are then computed taking into account the likely dilution and the minimum size of Selective Mining Unit (SMU). For the mining equipment at Rio Tinto the SMU was taken to be 12 x 14 m with its orientation dependent on loading direction.

Dilution (normally expressed as a percentage) was estimated from a combination of reconciliation methods:

* Comparing truck load counts with surveyed volume for each blast composite allowance is made for back break and irregularity of the loading platform; and

* Comparing mine and plant head grades over an extended period of time (minimum one month).

By plotting the estimated ore loss and waste dilution per blast against the composite perimeter length or hydraulic radius (perimeter divided by area) a reasonable estimate of dilution can be made. This method is however limited by the following factors:

* Errors may occur in calculated tonnage due to estimates of density and humidity;

* Contacts are assumed to be vertical planes (this is not an unreasonable assumption, in this particular case);

* Dilution is based on relative, rather than absolute, tonnage differences - over-digging in one area may be compensated for by under-digging in another area; and

* Some ore loss and/or waste dilution at the ore/waste interface is unavoidable due to the rill angle of the loading face.

Despite these limitations the value of close grade control can be gauged by the improvement in reconciliation between mine and plant during 1997.

During this time dilution was reduced from around 15% to less than 10% in 1998. The objective is now to reduce this figure to 5% by addressing the limitations listed above through the application of blast modelling.

Blast design software

During the last two decades the modelling of blasts has developed from purely empirical relationships to mechanistic models that attempt to simulate the whole blasting process. The first models were based on static solutions for stress distribution resulting from the pressure generated by a finite length explosive charge in an infinite length cylindrical hole (Jordan, 1962).

Subsequent work on dynamic models (Favreau, 1969 and 1983) were relatively sophisticated in their approach, being based on the laws of physics and chemistry, but were severely limited in their application by the capacity and cost of computing at that time. Subsequent rapid advances in computing have largely removed these limitations, and there has been renewed interest in mechanistic models.

Blasting software can be divided broadly into two types:

* Design and analysis of blast layouts using empirical relationships to predict fragmentation (may also include Monte Carlo simulations of initiation sequences to detect problems of connectors or delays); and

* Blast simulations that incorporate mechanisms of crack generation, heave and rupture to predict fragmentation and muckpile profile.

The first model type is typified by products such as Shotplan (Carter, 1992), Disvol (Jimeno, 1987) and Compu-Blast (Wheeler, 1998). These are generally simple to understand and are used routinely at the mine sites to predict the effects of minor changes to the blast design.

The mechanistic models, such as Sabrex (Jorgenson and Chung, 1987) and 3 x 3 (Grobler et al, 1994), are considerably more complex and are primarily used by the explosive manufacturers or blasting consultants. These models also mainly deal with the prediction of fragmentation and do not claim to accurately predict displacement.

At Rio Tinto it was recognised that the control of dilution was an important consideration and a program of study was undertaken to (a) develop a detailed cost model to monitor mine costs and (b) develop a blast displacement model that could mimic field observations. The displacement model was developed within Datamine using the 3D graphical tools of Guide to visualise the results.

Blast optimisation

Numerous papers have been written over the years describing how blasting can be optimised to minimise overall operating costs (Hunter et al, 1990, Eloranta, 1995). The unit costs for drilling, blasting, loading, hauling and crushing have been expressed in terms of: degree of fragmentation (MacKenzie, 1965); rock compressive strength (Borquez, 1981); powder factor (Nielsen, 1983); and diggability (Gold et al, 1987).

Whilst drilling and blasting costs can be easily calculated for any blast design, the cost associated with dealing with the broken muckpile are complicated by a number of factors that contribute to diggability which are difficult to quantify reliably. More importantly, optimisation of these parameters fails to take into account the effects of dilution associated with displacement during blasting and loading conditions at the face.

The approach at MRT has been similar to that described by Nielsen, whereby the unit costs can be related to powder factor, if all drilling is with 250 mm (9 7/8 inch) holes and drilling patterns are in the range 6 x 6m to 10 x 10m. The problem then becomes one of best using the available energy of the explosive to achieve the required comminution of ore and minimum cost for disposal of waste.

Analysis of the unit costs during the period April 1997 to March 1998 identified that the costs associated with dilution and crushing represents more than 65 % of the total costs (see table overleaf). During this period the powder factor was increased progressively from 0.78 to 0.89 kg/[m.sup.3].

Whilst it is logical that an increase in energy input into blasting will normally increase the amount of fragmentation and reduce in other unit costs, it is evident that at Rio Tinto, there is no discernible correlation, despite correcting for price changes, mining rate and extraordinary costs.

It was concluded, therefore, that an optimal powder factor must be less than 0.78 kg/[m.sup.3]. This was confirmed by:

* Back-analysis with the program DISVOL, which suggested an optimum powder factor of 0.7kg/[m.sup.3];

* Visual inspection of the muckpiles during loading (characteristic size of less than 400 mm); and

* Few problems (costs) with secondary blasting and generally good floor conditions.

A reduction in powder factor is also likely to reduce dilution through blast displacement and therefore decrease overall costs - it can be seen from the table, for example, that a 10% reduction in dilution will reduce costs by almost US$0.05/t. The question is how to quantify the relationship between dilution [TABULAR DATA OMITTED] and blasting parameters so that a true overall cost can be calculated.

Modelling of blast displacement

The philosophy behind the modelling of blast displacement has been to try and match observed behaviour. The program takes into account the following:

* Blast layout (actual location of holes);

* Initiation point and sequence;

* Final muckpile surface of representative blasts;

* Measured face velocities from blast videos; and

* Measured displacement of additional marker holes.

Starting with the surveys of the blasted muckpiles, the three main blast configurations used at MRT are shown in Figures a),b) and c). In each case the pre and post blast surfaces are shown to demonstrate the relative movement. These simulations, and others, can be used as 'templates' to predict the muckpile profile resulting from any given blast configuration.

By selecting a blast that is similar to the proposed layout, the displacement algorithm attempts to fit the estimated profile to the blast template chosen. The template may require to be stretched or shrunk to fit the proposed layout and allowances are made for differences in charging. This approach avoids the need to dynamically model displacement and ensures that the result is close to that actually achieved in the field. Ideally the actual post-blast profile should be used as the control, although this is not always possible due to the need to commence mining immediately after the blast.

To predict dilution the blast is modelled as a 3D matrix of mini-blocks with plan dimensions of half the blast hole spacing (3 to 4 m) and height of less than 1/3 the bench height. The block displacement vector is then calculated according to a set of simple power laws that take into account:

* Burden and spacing;

* Initiation point;

* Charge length and distribution; and

* Delay timing.

Using these relationships the parameters can be adjusted so that the face velocity profile and lateral displacement match the information from video images and marker holes. Examples of the output from the displacement algorithm are shown in Fig 8a and b.

The output block model will have the same attributes as the input model (grade, ore type etc) and can be used as the input to OPTIMA to locate the ore/waste contact lines. The advantages of this system are that a consistent and auditable method of grade control is used which utilises all the available information make the best estimate based on overall profit. In this case the dilution is handled explicitly by modelling displacement and defining SMU. Ore loss and waste dilution are therefore accounted for in the calculation of overall profit.

Currently the rill angle of the loading face is not handled correctly and an adjustment to the contact lines must be made in the field. This problem can be overcome in the floating stope optimiser by using a non-rectangular search volume, which will require sufficient miniblock in the blast model (vertical plane) to provide the required slope accuracy.

Conclusions

It has been shown at MRT that dilution is an important factor in project economics and that it should be considered along with the other unit costs in the optimisation of mining costs. The interrelation between drill and blast, loading, hauling, crushing and blast displacement can be expressed in terms of the specific charge, which can be modified to minimise overall costs. The specific charge is one of a number of parameters that control the heave and rupture of the rock mass. This movement of the rock assemblage is generally referred to as blast displacement.

It blast induced displacement can be predicted then dilution can be controlled (engineered). The solution proposed here provides a means of conditioning the blast simulation to observed data and generating a block model of the muckpile. Using this model, the Floating Stope Optimiser generates contacts that will return the maximum profit for a given SMU size.

It is hoped that the ideas presented will be applicable to other mines, particularly where visual grade control is not possible and/or the ore and waste zones cannot be segregated by blasting. In these situations dilution will become a major factor, which will require special grade control procedures.

References

Avery D. 1973 'Not On Queen Victoria's Birthday - the story of the Rio Tinto Mines' Published by Collins, London 464 pages

Borquez G.V 1981 'Estimating drilling and blasting costs - an analysis and prediction model' Engineering and Mining Journal, January. pp 83-89

Carter C.L 1992 'Blast Design System' Third Large Open Pit Mining Conference, Mackay, Australia, September 1992. pp 247-250

Eloranta J 1995 'Selection of Powder Factor in Large Diameter Blastholes' EXPLO '95, Brisbane, September. pp 25-28

Favreau R.F 1969 'Generation of strain waves in rock by an explosion in a spherical cavity' Journal of Geophysical Research, Vol 74(17) pp 4267-4280

Favreau R.F 1983 'Rock displacement velocity during a bench blast' 1st International Symposium On Rock Fragmentation by Blasting, Lulea, Sweden August. pp 753- 776

Gold R.D, Kennedy D.A and Gray J.H 1987 'A review of drilling and blasting practices at the Fording River Operations' Proceedings of the 11th Canadian Institute of Mining and Metals District 6 meeting. Vancouver Canada.

Grobler H.P, Guest A.R and Chitombo G.P 1994 'The implementation of numerical analysis models in blast design and fragmentation optimisation' in Application of Numerical Modelling in Geotechnical Engineering, Pretoria RSA pp 139-146

Hunter G.C, Sandy D.A and Miles N.J 1990 'Optimisation of Blasting in a Large Open Pit Mine' FRAGBLAST '90, Brisbane, Australia. pp 21-30

Jordan D.W 1962 'The stress wave from a finite, cylindrical explosive source' Journal Mathematical Mechanics 11 pp 503-551

Jorgenson G.K and Chung S.H 1987 'Blast simulation surface and underground with the SABREX model' CIM Bulletin, August. pp 37- 41

Lopez Jimeno, E and Muniz E, 1987 'A new method for the design of bench blasting' 2nd International Symposium On Rock fragmentation by Blasting, Colorado.

MacKenzie A 1966 'Cost of Explosives- Do you evaluate it properly?' Mining Congress Journal Vol 54 May pp 32-41

Nielsen, K 1983 'Optimisation of open pit bench blasting' 1st International Symposium On Rock Fragementation by Blasting, Lulea, Sweden. pp 653-664

Pryor R.N, Rhoden H.N and Villalon M 1972 'Sampling of Cerro Colorado, Rio Tinto, Spain' in IMM Geological, mining and metallurgical sampling, (ed) M.J Jones, pp 156- 172

Randall M.M and Wheeler A.J 1998 'Blancing the Books - Improved mine reconciliation through automated grade control' MINING Magazine, May 1998 pp 337-342

Taylor S.L, Glibride L.J, Daemen J.J.K and Mousset-Jones P 'The impact of blast induced movement on grade dilution in Nevada's precious metal mines' in Rock Fragmentation by Blasting, Mohanty (ed), Balkema, Rotterdam. pp 407-413

Wheeler R.M 1998 'Compu-Blast Software' Software produced by White Industrial Seismology Inc, Joplin, MO, USA.

Matthew Randall, Mining consultant, Rio Tinto Technical Services. Tel: (+44117) 927 6407.

Adam Wheeler, Independent consultant. Tel and Fax: (+441209) 890 733.
COPYRIGHT 1998 Aspermont Media UK
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:modeling of blast displacement and dilution control at the Minas de Rio Tinto copper mine in Huelva, Spain
Author:Randall, Matthew; Wheeler, Adam
Publication:Mining Magazine
Date:Nov 1, 1998
Words:3025
Previous Article:Flotation advances.
Next Article:Raglan - an Arctic achievement.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |