Process Integration and Optimisation A Case Study in Productivity Improvement

A Dance1, W Valery2, A Jankovic3, D La Rosa4 and J Colacioppo5

ABSTRACT
Metso Minerals Process Technology Asia-Pacific and South America (MMPT) has been working with operations around the world to increase their efficiency through Mine-to-Mill or Process Integration and Optimisation (PIO) projects. These projects have delivered significant improvements in mine efficiency, mill throughput and reduced operating costs at mines around the world. The MMPT methodology involves rock characterisation, site auditing, data collection, modelling/simulation and implementation of integrated operating and control strategies on site. This results in significant benefits to the operations for example, five to 20 per cent increases in concentrator throughput have been measured. MMPTs rock characterisation utilises simple and inexpensive measurements that can be performed by trained site personnel. The advantage of these measurements is that a large amount of data can be collected in a very short timeframe, as the samples do not require shipping to an outside laboratory. Typically, a PIO project starts with a site visit to perform blasting and process audits, collect high quality data including measurements of run-of-mine fragmentation and survey data around all crushing and grinding circuits. These measurements are combined with rock characterisation and the definition of strength and structure domains to model the complete production chain. Site-specific models of the blast fragmentation, crushing, grinding and flotation processes can then be developed. This proven methodology has applications ranging from greenfield projects to long-standing operations with AG/SAG or conventional grinding circuits. Process improvements can be higher mining productivity, higher mill throughput, decreased overall operating costs and higher flotation or leach recovery. This paper discusses a recent application of PIO conducted at a large open pit operation in South America and emphasises the potential that every operation has to do more with what they have.

consists of the development of integrated operating and control strategies from the mine to the concentrator that minimise the processing cost per tonne and maximise profitability. It involves rock characterisation, site auditing, data collection, modelling/simulation and implementation of integrated strategies in the field. This results in significant benefits to the operations for example, five to 20 per cent increases in concentrator throughput.

PIO METHODOLOGY
The PIO methodology is divided into a number of components: benchmarking, material tracking, rock characterisation, high quality survey data collection in the mine and concentrator, historical operating data collection, modelling/simulation and implementation of the integrated operating strategies. Experience has found that ignoring any of these elements will result in the project being less successful. A PIO project is normally initiated with a site visit to collect data of the current operating practices and benchmark the performance of both the mine and concentrator. Particular attention is made to the rock conditions and variability in rock properties to define strength and structure domains. Once the different rock strength and structure domains are defined, data has been collected, modelling and simulation techniques are used to determine the best operating strategy and avoid costly (and often inconclusive) trial-and-error exercises. The ability to accurately model the blast fragmentation, crushing and grinding processes allows the entire production chain to be simulated, with the impact of upstream changes being able to be predicted downstream. The conclusions and recommendations of the simulation study reveal a clear direction forward for the operation. The PIO study then enters the site implementation, support and monitoring phase. Once the changes in the mine and concentrator have been implemented and the benefits clearly demonstrated, outside assistance is only periodically required thereafter to ensure that the gains are maintained in the longer term.

INTRODUCTION
Metso Minerals Process Technology Asia-Pacific and South America (MMPT) has been working with operations around the world to increase their efficiency through Mine-to-Mill or Process Integration and Optimisation (PIO) projects. These projects have delivered significant improvements in mine efficiency, mill throughput and reduced operating costs at mines around the world (Dance et al, 2006). The MMPT methodology

Benchmarking and process audits

1. Manager Process Integration and Optimisation, Metso Minerals Process Technology Asia-Pacific and South America (MMPT), Unit 1, 8-10 Chapman Place, Eagle Farm Qld 4009. Email: adrian.dance@metso.com FAusIMM, General Manager, Metso Minerals Process Technology Asia-Pacific and South America (MMPT), Unit 1, 8-10 Chapman Place, Eagle Farm Qld 4009. Email: walter.valery@metso.com MAusIMM, Manager Development and Process Engineering, Metso Minerals Process Technology Asia-Pacific and South America (MMPT), Unit 1, 8-10 Chapman Place, Eagle Farm Qld 4009. Email: alex.jankovic@metso.com Manager Process Control and Information Engineering, Metso Minerals Process Technology Asia-Pacific and South America (MMPT), Unit 1, 8-10 Chapman Place, Eagle Farm Qld 4009. Email: david.larosa@metso.com Process Technology Engineer, Metso Minerals Process Technology Asia-Pacific and South America (MMPT) South American representative, Avenida Independncia, 2500 Bairro Iporanga, Sorocaba, So Paulo 18087-101, Brazil.

2.

3.

4.

5.

A clear understanding of the current operating practices needs to be gained before any changes can be recommended, this must be backed up with measurements and data collected at site. This involves auditing the blast design, implementation and initiation sequence, followed by material tracking (that will be described in more detail below) and fragmentation measurements along the process. Blast design implementation is closely observed, as an operation must be able to implement accurately and repeatedly before it can expect to gain from blast design changes. From the perspective of the concentrator or processing plant, sampling surveys of the crushing and grinding circuits provide the necessary data to model each stage of comminution. Operating practices such as stockpiling and blending of different ore types ahead of the concentrator are also reviewed with consideration of the expected variability in rock properties. Finally, the level of instrumentation and process control strategies are reviewed to ensure that the mill operating conditions are best suited to changing rock conditions.

Large Open Pit Mining Conference

Perth, WA, 10 - 11 September 2007

A DANCE et al

All of these measurements allow mathematical models to be developed for the complete process chain. These models are later used to simulate the impact of operational changes in the mine or concentrator on the entire process.

Material tracking
In some cases, where material is blended before the concentrator or when sophisticated mine monitoring and dispatch systems are not in place, there is a need to monitor material movement from the mine to the concentrator. MMPT routinely use passive Radio Frequency ID (RFID) tags to mark material and track its movement over time. Initially developed to assist in the PIO studies, MMPT have commercialised this system under the name of SmartTag. A schematic of the system is shown below in Figure 1. The small, epoxy-encased tags can be placed in the stemming column of blastholes or post-blast on the muckpile surface. Picked up by the excavator, the tags pass through the primary crusher and are detected by antenna placed over conveyor belts at various points in the process; for example, over the primary crusher and SAG feed conveyor belts. The use of such technology allows the spatially-based mine information to be linked with the time-based concentrator information. That is, linking where the material came from to when it was processed in the mill. In conjunction with online image analysis systems, the fragmentation from different rock domains can be measured from the blast down to the grinding circuit.

Rock characterisation
The MMPT methodology for rock characterisation utilises simple and inexpensive measurements of rock strength and structure that can be performed by trained site personnel, and quite often these measurements are already being collected by the operation. The advantage of simple measurements is that a large amount of data can be collected in a very short timeframe, as samples do not need to be shipped to an outside laboratory. When attempting to characterise and map rock strength and structure of an entire orebody, density of data and their statistical significance are very important.

For rock characterisation, measurements of rock strength (Point Load Index, PLI and/or UCS) and rock structure (Rock Quality Designation, RQD and/or fracture frequency) are normally used. RQD measurements can be taken on drill core while Point Load tests can be done on both drill core and irregularly shaped samples of material collected from around a muckpile face or stockpiles. The PLI value can be correlated to Unconfined Compressive Strength (UCS) as well as the JKMRC Drop Weight test parameters A and B and the SMC Drop Weight Index. The latter tests parameters are necessary in order to model the crushing and grinding circuits. Therefore, the use of the Point Load Index allows sites to characterise their rock properties quickly and easily, while still making use of the available sophisticated models of blasting, crushing and grinding. The rock structure is represented by the RQD value that indicates the fracture frequency present in the drill core. This measurement is routinely taken at operations for geotechnical purposes but has been shown to be very useful in blast fragmentation modelling in the absence of detailed rock mass structure mapping. Once the PLI and RQD data are available, the range of rock properties is mapped out to define possible blast domains, which are zones of ore with similar rock strength and structure that when submitted to blast generate similar fragmentation. In Figure 2, three ore types as originally classified by the operation based on lithology and grade have been mapped to identify their rock strength and structure domains: ore type A shows variable rock strength with a limited range in rock structure. Ore type B is the reverse: variable structure with more consistent strength. Ore type C is the most consistent overall with medium strength and very blocky in structure. Under the same blasting conditions, all three ore types will result in different fragmentation distributions. Rock characterisation involves understanding the reason behind this variability and predicting the impact on fragmentation and downstream process performance. Originally, the same blast design and energy were utilised for ore types A, B and C. The relatively simple rock strength and structure measurements, existing operating data, fragmentation

FIG 1 - Schematic of RFID tag-based material tracking system.

Perth, WA, 10 - 11 September 2007

Large Open Pit Mining Conference

PROCESS INTEGRATION AND OPTIMISATION A CASE STUDY IN PRODUCTIVITY IMPROVEMENT

measurements followed by blast modelling and simulations undertaken in the PIO project indicated that ore types A, B and C result in a very different fragmentation distribution (as far as the predicted concentrator performance is concerned). The variability in rock properties requires ore types to be broken according to their strength and structure domains. Once the domains have been defined, different blasting practices, crushing and milling operational strategies are established. Through modelling and simulation studies, the impact of blending different domains can be reviewed. Most importantly, as the rock properties have now been well characterised and the processes modelled, the variable nature of the material can now be compensated for.

an ore type classification based on assay and not rock

properties;

underlying rock conditions and lithologies resulting in

variable concentrator performance;

a single blast design was being used for all ore types; and a significant change in blasting energy was required to
demonstrate the effect on mill throughput. As a result, this project provides a good case study in how many operations could benefit from Process Integration and Optimisation.

Process conditions Modelling and simulation

The measurements collected while at site are combined with the rock characterisation domains to model the complete blasting and comminution process. MMPT use these data to develop site-specific models of blast fragmentation, crushing, grinding and flotation. This allows customised blast patterns to be developed that optimise both crushing and grinding performance. For each domain, blast designs are defined to generate the optimal fragmentation size for downstream processes. This may involve an increase or decrease in energy (powder factor in the blast), depending on the rock characteristics of each domain. The objective of any modified blast design is to minimise the overall cost for the entire process by distributing the energy required to fragment the rock mass, sensibly and effectively. Near-field vibration measurements and models are used to confirm that pit wall stability, damage and dilution issues are considered in the blast designs. In addition, the crushing and grinding models allow the impact of operational and control strategies to be investigated and optimised. As part of the site visit phase of the PIO study, a review of the current blasting practices and processing conditions was made. A typical ore blast was monitored and this involved an audit of the blast design implementation (hole location, hole length, loading practices, tie-in, etc) and tracking of the ore from the muckpile to stockpiles and concentrator feed using the SmartTagTM system. This operation was characterised by ore types exhibiting very different blast fragmentation distributions and as a result, very different concentrator tonnages through the SAG mill circuit. The range of fragmentation generated by the same blast design is illustrated in Figure 3 (one metre scales shown at bottom of photos). The material on the left fragmented very well with an abundance of fines and little oversize to be crushed by the primary gyratory. This material was tracked using the SmartTag system and processed through the concentrator at between 4000 and 5000 t/h. The material on the right of Figure 3 was more difficult to fragment and despite the primary crusher operating with a tighter gap or setting (125 - 150 mm), resulted in concentrator throughputs between 2000 and 3000 t/h or almost half the rate of the finer material. It is interesting to point out that despite the primary crusher operating with a tighter gap, the amount of fines generated in the blast still had a dramatic effect on mill performance. That is, despite controlling the SAG mill feed topsize, the distribution of feed in the finer fractions (especially in the size smaller than the mill discharge grate size) has a significant impact on mill performance. The fragmentation shown in Figure 3 illustrates the effect of adopting a single blast pattern (or relatively similar blast patterns) that does not take into account varying rock properties.

CASE STUDY LARGE OPEN PIT OPERATION IN SOUTH AMERICA

MMPT was recently involved in a PIO study at a large open pit operation in South America. Recommendations on blast design, crushing and grinding changes are currently being implemented and preliminary results indicate a significant increase in concentrator throughput. A review of the current operating practices revealed a number of conditions common to most operations of this size:
10

High

Ore Type C Ore Type A

Is50 (MPa)

Rock Strength

Ore Type B

Low

Fractured
0 0 10 20 30 40

Rock Structure
50 RQD (%) 60 70

Blocky
80 90 100

FIG 2 - Rock strength and structure map for blasting domains.

Large Open Pit Mining Conference

Perth, WA, 10 - 11 September 2007

A DANCE et al

FIG 3 - Examples of fragmentation for two ore types.

100

90

80

70

% of Rock Type

60

Exoskarn

50

40

30

Endoskarn
20 10 Finer Fragmentation, Higher Throughput 0 A B C D E F G H Coarser Fragmentation, Lower Throughput

FIG 4 - Association of rock lithology with ore types.

The first step in optimising blast fragmentation for mill performance is to understand the material properties and how they affect the run-of-mine size distribution. The second step is understanding how feed size affects mill throughput and how to modify blast design, crushing and mill operating practices to achieve optimal process performance. The third step is to evaluate how well an operation can implement integrated and optimal operating strategies so that the true effect of blast design, crushing and grinding changes are revealed in the concentrator performance.

Rock characterisation
The operation reviewed here defined material types or ore types based on assay differences alone. In Figure 3, the material on the left had a different metal assay compared with the material on the right. Although the metal assay dictated that material processing had to be done differently, the mining stage of comminution (blasting) was done in a similar fashion. The key issue here was the inability to predict how the material would fragment based on ore type or metal assays prior to actually drilling the pattern. It was realised that designing the blast pattern based on ore type rather than rock properties resulted in a highly variable fragmentation size and mill throughput. A review of the different ore types and their associated rock types revealed an interesting pattern represented in Figure 4. In

this figure, the eight defined ore types were ranked (in general) from left to right from finer to coarser fragmentation (and higher to lower mill throughput). Of the eight ore types, A and F represented almost three-quarters of the expected mill feed, with the remaining six types making up the other quarter. (In Figure 3 above, the left photo was of ore type A and the right photo was of ore type F.) Of the main lithological groups, Endoskarn and Exoskarn made up the majority of mill feed material. It was found that the finer fragmenting ore types (like A) were found predominantly Endoskarn while coarser fragmenting ore types (like F) were found in Exoskarn. It is likely that the presence of these two lithological groups (and subcategories within each group) is a better indicator or predictor of blast fragmentation than ore type (or metal grade) as they relate with the rock strength and structure that can be measured. To investigate this further, drill core measurements of rock strength (Is50 or Point Load Index) and structure (RQD) were reviewed. Unfortunately, there is not an extensive set of Is50 and RQD measurements along with rock type/lithology and ore type at the present time. Figure 5 summarises the currently available data. It can be seen that the Exoskarn lithology exhibited (in general) both higher Is50 and larger RQD values. The collection of new rock strength and structure measurements is currently underway at site to refine the definition of rock domains.

Perth, WA, 10 - 11 September 2007

Large Open Pit Mining Conference

PROCESS INTEGRATION AND OPTIMISATION A CASE STUDY IN PRODUCTIVITY IMPROVEMENT

With this new database of PLI, RQD, rock type and ore type, domains of expected similar blast fragmentation can be defined. Due to the complexity of the orebody, it is likely that each blast will contain a number of blast domains. When this occurs, there are three options to base the blast design on:

the average rock strength and structure, the highest rock strength and structure, and adjust the pattern with localised rock strength and structure.
While adjusting the pattern blast to suit local conditions is optimal for minimising cost and fragmentation purposes, it requires an accurate prediction of rock conditions. If the rock conditions are known prior to drilling, then a variable burden and spacing can be designed to account for harder/softer zones within the blast. If harder zones are identified during the drilling process using either a drill-monitoring system output or drill operators logs, then the formulation and amount of explosive can be varied. In most cases, this is not very practical and instead, it is recommended to design the blast for the worst conditions present. Following definition of these domains, specific blast patterns can be recommended to optimise both fragmentation size and overall cost for the entire operation. The set of standard blasting patterns are designed to maintain a consistent fragmentation size and therefore a consistent mill throughput. Some modified blast designs may involve an increase in powder factor while others may warrant a reduction in energy.

recommended to improve fragmentation. With a tightening of the pattern, the powder factor increased significantly by 41 per cent and raised the drill and blast cost per tonne by 40 per cent. Fragmentation modelling and simulations indicated a finer 80 per cent passing size (P80) of 175 mm compared with 401 mm and an increase in the -25 mm fraction of 14 per cent. This represents the step-change in blasting energy needed to clearly demonstrate the positive impact on fragmentation on mill throughput. Historical results indicated that this difficult material processed at a significantly lower rate and needed higher energy levels to overcome the greater rock strength and blockier nature of the material. Further work will be done to refine the blast designs for the other rock types and potentially lower the blasting energy requirements.

Preliminary blast trial results

Following MMPTs recommendations, the operation is currently trialling the modified blast design in the recognised difficult zones. Figure 6 shows two images of the modified blast muckpile (with one metre scales) that can be compared with historical results shown in the right photo of Figure 3. The photos clearly demonstrate the effect of the higher powder factor along with some changes in stemming length and explosive formulation. Measurements of the fragmentation using image analysis confirmed the model-predicted P80 of 175 mm and the mine reported a significant increase in shovel productivity. This material was stockpiled, tracked and processed separately through the concentrator with impressive results. Figure 7 shows a trend of SAG mill tonnage over time before, during and after this modified blast material was processed. For the entire period shown, the ore type was the same and was mined from a similar area of the pit. The values in Figure 7 show the mill tonnage increased from 3500 to 4000 t/h before to around 5000 t/h for the modified blast material. With the stockpile depleted and normally blasted material sent to the concentrator, tonnage returned to below 4000 t/h.

Modified blast designs

At this stage of the project, blast domains are still to be defined for this operation. However, using the modelling and simulations tools available, MMPT recommended a modified blast design for the expected difficult to blast areas of the pit as illustrated by the right photo of Figure 3. Table 1 summarises the current (or historical) blast pattern along with the modified pattern
7

Exoskarn
Is50 (MPa) 4

Endoskarn

0 40 45 50 55 60 RQD (%) 65 70 75 80

FIG 5 - Rock strength and structure map for main lithological groups.

TABLE 1
Summary of blast designs.
Design Current Modified Hole diameter (mm) 311 311 Burden spacing (m) 7 8 6 7.5 Powder factor (kg/m3) 1.15 1.62 Predicted % -25 mm 30.7 44.4 Predicted P80 (mm) 401 175 Cost ($/tonne) 0.20 0.28

Large Open Pit Mining Conference

Perth, WA, 10 - 11 September 2007

A DANCE et al

FIG 6 - Resulting fragmentation from modified blast.

5,500

5,000

4,500 Mill Tonnage (tph)

4,000

3,500

3,000

current blast design 2,500

modified blast design

current blast design

26-Apr 06:00 26-Apr 12:00 26-Apr 18:00 27-Apr 00:00 27-Apr 06:00 27-Apr 12:00 27-Apr 18:00 28-Apr 00:00 28-Apr 06:00 28-Apr 12:00 28-Apr 18:00 29-Apr 00:00

FIG 7 - Trend of concentrator tonnage during modified blast trial.

It was unfortunate that the amount of high-grade ore in the modified blast accounted for only 120 000 t and therefore ran out after 24 hours of processing. Additional blasting trials with the modified design are underway and the operation has stockpiled this material to be processed separately at a later date. The increase in mill throughput was 25 to 40 per cent, exceeding all expectations and more than compensated for the 8/tonne higher blasting costs. Blasting trials continue and it is hoped that supplementary results will be reported during the conference presentation in September. A simple economic analysis can be done using some assumed values to demonstrate the benefit to the overall operation: at a stripping ratio of 3:1 and 5000 t/h concentrator throughput, the additional 8/tonne blasting costs represents close to $40 000 per day. However, the 1500 t/h gained through the finer fragmentation from the modified blasts (assuming 1 per cent copper head grade and 90 per cent recovery) represents almost

$2 500 000 in increased revenue (based on copper alone at current metal prices). In other words, a factor of 62.5:1 for increased revenue to added cost. While drilling and blasting costs represent a significant portion of the overall mining costs, they are minor when compared to the potential gains made through Process Integration and Optimisation.

SUMMARY
The application of MMPTs PIO methodology at a large open pit operation in South America was presented. Recommendations on blast design, crushing and grinding changes are currently being trialled and implemented. Preliminary results indicate a significant increase in concentrator throughput as a product of these changes. This case study is of importance as a number of conditions common to many large mining operations were present:

Perth, WA, 10 - 11 September 2007

Large Open Pit Mining Conference

PROCESS INTEGRATION AND OPTIMISATION A CASE STUDY IN PRODUCTIVITY IMPROVEMENT

an ore type classification based on assay and not rock

properties;

underlying rock conditions and lithologies resulting in

variable concentrator performance;

a single blast design was being used for all ore types; and a significant change in blasting energy was required to
demonstrate the effect on mill throughput. This PIO study revealed that rock lithology was closely associated with rock strength and structure, therefore with blast fragmentation and consequently, concentrator throughput. A characterisation study is ongoing to collect drill core measurements of rock strength and structure for different rock types and further refinement of rock domains. The results of this study will allow a series of standard blast patterns to be designed for each strength/structure or blasting domain. The set of standard blasting patterns were designed to maintain a consistent

fragmentation size and therefore a consistent mill throughput. Some modified blast designs may involve an increase in powder factor while others may warrant a reduction in powder factor. Implementation of the recommendations and blasting trial results gathered to date indicate a 25 to 40 per cent increase in mill throughput for the material found to be more difficult to fragment and process. This tonnage increase more than compensated for the higher blasting costs.

REFERENCES
Dance, A, Valery, W, Jankovic, A, La Rosa, D and Esen, S, 2006. Higher productivity through cooperative effort: A method of revealing and correcting hidden operating inefficiencies, in Proceedings SAG 2006, pp IV-375.

Large Open Pit Mining Conference

Perth, WA, 10 - 11 September 2007

Perth, WA, 10 - 11 September 2007

Large Open Pit Mining Conference