SAG MILL THROUGHPUT

Porgera Joint Venture (PJV) and Dyno Nobel initiated a project to optimize drilling and blasting practices for downstream operations. The SAG mill was identified as a production bottle neck when treating hornblende diorite ore. Consequently, it was agreed that the project would focus on optimizing the fragmentation size distribution for the crushing and SAG milling circuits. The Julius Kruttschnitt Mineral Research Centre (JKMRC) was sub-contracted to investigate the impact of blasting on SAG mill throughput.

Rock characterization, blast and mill surveys were conducted on hornblende diorite. The data was used to build site specific models for blast fragmentation, primary crushing, and SAG and ball milling. The models were linked and simulations performed to investigate the influence of ROM fragmentation on SAG mill throughput. Based on these simulations, modifications were made to blasting practices and primary crusher settings. Two additional surveys were conducted to validate the models. This paper discusses the methodology and results achieved.

The operation employees about 2,000 people. The mine is operated by Placer Dome and produced over 900,000 oz of gold in calendar year 2000.

The Porgera Gold Mine moves over 210,000 tonnes per day of material including about 18,000 tonnes of ore. Mobile equipment consists of 36 Cat 789 haul trucks, 8 Cat 777 haul trucks, 6 OK RH200’s, 2 Cat 992 loaders, 1 OK RH40, 3 Cat 16G graders, 3 Cat 824 wheel dozers and auxiliary contractor equipment. The mine runs 24 hours per day utilizing 3 shifts per day.

After blasting, the ore is loaded and hauled to the primary crusher or runof-mine (ROM) stockpiles. At the primary crushing plant, the ore is reduced to minus 150 mm size material in a single pass through a gyratory crusher to produce a SAG mill feed.

Apron feeders draw the crushed ore from a coarse ore stockpile to feed two parallel primary grinding lines. Each line consists of a variable speed SAG mill measuring 28 feet in diameter by 11 ft long. The mill motors are rated at 5,500 HP. At the discharge of each SAG mill is a pair of vibrating screens. The screen oversize material from both SAG mills is conveyed to a shared pebble crushing circuit consisting of two Omnicone crushers.

The SAG screen undersize from both SAG mills is pumped to a common distributor where it is split to feed three parallel ball milling circuits. The ball mills are in closed circuit with hydrocyclones. The grinding circuit product measures 80% passing 106 microns and feeds the flotation plant.

A bulk flotation process using conventional cells produces a gold-bearing sulphide (mainly pyrite) concentrate. The concentrate is thickened before it is pumped to the pressure oxidation circuit. Oxidized slurry is re-circulated to destroy most of the carbonates prior to the concentrate being fed to four parallel autoclaves. In the autoclaves, the sulphides are decomposed to expose the gold. The autoclave discharge is cooled, thickened and washed. After pH adjustment, the oxidized slurry is cyanide leached in a leach/CIP circuit. The loaded carbon is stripped, and the gold and silver are electrowon from the pregnant solution. Bullion is poured on site.

WHAT DOES A MINE TO MILL PROJECT INVOLVE?

The “Mine to Mill Blasting” approach involves identifying the leverage that blast results have on different downstream processes and then optimizing the blast design to achieve the results that maximize the overall profitability rather than individual operations.

Identification of each opportunity to increase process efficiency is very important to the success of the approach. Potential areas of impact include, but are not limited to:

Increase in comminution efficiency leading to higher mill throughput, Improvement in loader/excavator productivity due to better diggability, and fill factors,

A typical mine to mill blast optimization project involves several phases such as:

• Scoping
• Auditing
• Modeling and Simulations
• Validation
• Implementation

SCOPING

The mine to mill blast optimization project is interdisciplinary in nature and it impacts on several aspects of mining and milling operations. This means a team approach is required with members from the mine and the mill, and a project champion to drive the project. At Porgera, the initial scope of the mine to mill project was limited to maximizing the SAG mill throughput by altering blast design and crusher settings to provide an optimum feed size distribution to the mills. Hornblende diorite was chosen for the study because it is the hardest ore at Porgera.

The impact of blasting results on the efficiency of load and haul operations was investigated, but the details of that study are not covered in this paper. Likewise, the effect of blasting on ore body dilution and damage to the pit walls will be studied separately.

AUDITING

During this phase, an audit of all the processes from drilling and blasting to milling was conducted. A trial blast (#340319) was conducted on hornblende diorite in May 2000 with the standard design and the ore from this blast was processed through the mill without any blending. During this trial all critical input and output parameters from drilling and blasting to SAG milling were monitored, especially the size distribution of ore from in situ to the final product of the grinding circuit.

Rock / Ore Characterization

Generally ore domains are classified based on grade and geology and do not take into account the breakage characteristics. The breakage characteristics of rock /ore are very important for the design and optimization of any mine to mill process. Therefore, it is essential to understand these differences and characterize the rock mass in terms of its blastability as well as comminution properties.

Blastability

Blasting loosens the existing rock mass structure to liberate rock blocks and creates new fractures within the intact material. Fines are produced by the crushing action of the explosion in a zone that surrounds each drill hole. Blastability is a function of mechanical properties of intact rock (eg. stiffness and strength) and the rock mass structure (size of ‘insitu blocks’). Zones within a mine that are of similar strength and structure should blast in a similar way and hence form a blasting domain. It should be recognized that the blasting domains may be different from the current geological domains.

The structural characteristics of ore in the trial blast were estimated from the scan line mapping and face photographs. The mean in situ block size was estimated to be between 0.5m to 0.6m.

Cores from the ore samples collected from the trial blast were subjected to various rock mechanic tests to measure the strength and stiffness properties. Rock samples were also subjected to point load testing to estimate the strength.

Comminution Characteristics

The micro fracture network, grain size and grain characteristics may be important in comminution processes such as crushing and grinding. It is necessary to understand the comminution properties of different ore types to optimize the integrated mine-to-mill operation The JKMRC use a ‘drop weight test’ to determine the breakage characteristics of rocks. A detailed description of the tests for determining the comminution properties is given by Napier-Munn et al (1996).

The actual blast design parameters such as hole depth, burden, spacing and stemming height for each hole was measured and compared with the design parameters.

The results indicate good control in drilling during collar positioning. However in case of maintaining the drill depth, 20% of the holes were either drilled 0.5m deeper or shorter than the design depth. This variation in the drilled depth causes either severe damage to the floor (excessive drilling) or leaves toe (short drilling) therefore producing an uneven floor. The lack of control in maintaining the stemming heights can result in oversize from the collar (excessive stemming) or in stemming ejection (less stemming) both resulting in undesirable blasting results.

Fragmentation Assessment

During this trial, photographs were taken of the muck piles and at the rear of trucks while the ore is dumped at the primary crusher. These photographs were analyzed with the Split image analysis system to estimate the fragmentation size of ROM material. Approximately twenty images from the muck piles and thirty five images from the rear of trucks were analyzed. The images taken at the primary crusher dump pocket provided a better estimate of the ROM fragmentation size. This is because the surface of the muck piles are not true representations of the entire blast.

Mill Survey #1

The first survey was conducted on 28 May 2000 to provide a baseline of current practice and performance. The blast (#340319) was un-modified and the primary crusher gap was set between 120 and 180 mm to maintain crusher throughput.

The ore from the trial blast was campaigned through the mill without any blending and a full mill survey was conducted. SAG mill throughput was maximized at 673 dmtph of fresh feed. When the mill reached steady state, samples were collected from various locations in the circuit every 15 minutes over a period of one hour. The major circuit variables were also monitored and collected from the control system. Figure 2 page I- 278 illustrates the primary crushing and SAG milling circuits, and the sampling locations.

After the steady state portion of the survey, the SAG mill circuit was crash stopped to collect samples from the belt conveyors, to allow inspection of the mill interior and to carry out load measurements. Subsequently, the SAG mills were ground out to determine the ball load. Photographs were taken to estimate the rock and ball size distribution of the muck load using the Split system.

Cameras were also installed to monitor the SAG feed conveyor belts for subsequent feed size distribution determination using the Split imaging system. All samples were processed through the lab to determine % solids, and size distribution. Portions of the sized SAG mill feed were collected and sent to JKMRC for drop weight and Bond Ball mill work index tests. The results of these ore characterization tests are summarized in Table 2.

It can be seen that SAG #1 feed rate was substantially higher than for SAG #2.

This difference in milling rate is primarily due to the fact that the feed size distribution for SAG #1 was finer than the feed of SAG mill #2 (see Figure 3 page I-280). This happens because the feed belt conveyor of SAG #1 is located at the center of the coarse ore stockpile where as for SAG #2 it is offset from the center. Because of segregation on the stockpile, SAG #2 tends to receive coarser feed.

The total charge measurements are consistent with the bearing pressure measurements. However, SAG #2 drew less power than SAG #1 despite having a higher load. This apparent discrepancy is likely explained by the worn out condition of SAG #2 belly lifters which were changed out only one month after survey #1.

MODELING AND SIMULATIONS

Understanding the mechanisms and interaction between different processes in the Mine-to-Mill chain, modeling key processes and integrating them is a key feature of the integrated mine to mill optimization. Based on the information gathered during the benchmarking audit and survey, the blasting, crushing and milling processes have been modeled. These models were linked to investigate the impact of blast fragmentation on SAG mill performance.

Blast Fragmentation Modeling

The ore properties and the blast design parameters measured during the audit for the trial blast, were used in the JKMRC blast fragmentation model to predict the ROM fragmentation. Photographs were taken of the post blast muck piles and at the rear of trucks. These images were processed using the Split system to obtain the size distributions of ROM. A comparison of ROM size distributions predicted by the models and by the Split system indicates that the model predictions are very close to the ROM size distributions estimated by the Split.

Mill Modeling

The JKMRC has been conducting considerable research in this area for the past three decades and developed a number of models to simulate different stages of comminution. These models are encapsulated in a simulator called JKSimMet. A detailed description of these models is given by Napier-Munn et al (1996).

The data collected from the mill survey was mass balanced and then used to calibrate the crushing and grinding circuit model at Porgera. The ball milling circuit was included in the survey to provide a model of the entire grinding circuit.

The breakage rates for the two SAG mills are plotted in Figure 4. For survey #1, the breakage rates are similar for both mills. There is a drop in the breakage rates in the size range of 20 to 75 mm. A noticeable difference is that SAG #2 breakage rates do not rise back up at the coarse end. This may be due to the large amount of coarse material in the feed that could not be broken with the available mill charge, the worn lifters, or poor representation of the mill load size distribution based on which the breakage rate is calculated.

Simulations were run to predict the influence of different blast designs on SAG milling throughput. A tighter drill pattern was selected as the method to increase the amount of fines in the SAG mill feed. The key parameters of the blast designs were shown in Table 1.

The second survey was conducted on 17 October 2000. The powder factor of this blast (#320314) was about 40% greater than for survey #1 (see Table 1) and the primary crusher gap set at 110 mm. The ore was similar for both surveys #1 and #2 as indicated by the results of the ore characterization tests (see Table 2). So the results of the two surveys can be compared.

Unfortunately, a pebble crusher was not operating during this survey. So, it was decided to conduct the mill survey without a pebble crusher under the following two scenarios.

• by re-circulating the uncrushed scats ( the worst case) and,
• by rejecting/dumping the scats to ground (the best case).

A full mill survey was conducted for the first case (re-circulating uncrushed scats). The second case was run at steady state but a full circuit survey was not done. Operating variables were recorded.

The finer SAG feed size distribution achieved in survey #2 (compared to survey #1) is illustrated in Fig 5. Survey #3 feed is also shown.

MILL SIMULATIONS

The purpose of the “Mine-to-Mill” project was to determine the impact of blasting on SAG mill feed size distribution and hence mill throughput. It was also important to establish if the milling rate could be increased by making modifications to the operation of the existing circuit instead of altering the feed size distribution. The effects of changing several major circuit variables were simulated.

Different grate size openings were simulated and the results indicated that an opening smaller than 70 mm would be detrimental to SAG mill throughput. This prediction is not surprising if the breakage rate curve is examined (Figure 4); the 20 to 75 mm material does not break efficiently in the SAG mill. A smaller grate opening will result in lower throughput because critical size material will occupy more mill volume and not break efficiently. It is better to remove this material from the mill and crush it. PJV are currently using grates with 75 mm openings.

Simulations were also run with several different SAG discharge screen aperture sizes. The simulations suggested that SAG mill throughput gains would be marginal by increasing the screen aperture dimensions.

Also, the transfer size to the ball milling circuit would increase.

However, since each SAG mill has two parallel screens on its discharge, one screen has been fitted with 12 mm aperture panels and the other unit with 15 mm aperture panels. This arrangement provides operating flexibility depending on the feed being milled and which part of the grinding circuit is the bottleneck. So far the operating evidence suggest that the ball milling circuit can handle the coarser transfer size from the SAG mill when the screen fitted with 15 mm aperture panels is run.

The choice of ball size is normally dictated by the hardness and size distribution of the feed. The simulator predicts higher SAG throughputs with the 125 mm ball compared to the 150 mm ball. The 125 mm ball can provide the necessary energy level to break the rocks. Since the number of 125 mm balls for a given volume is greater than for 150 mm balls, the number of breakage events is more for the 125 mm ball (other mill variables being equal). This translates into a higher breakage rate with the 125 mm ball. Industrial trials are continuing to resolve this issue.

According to simulations, SAG mill throughput can be increased by running with a higher ball charge. However, the total (ball and rock) charge would have to be reduced so as not to further exceed the motor rating. A higher ball charge and lower rock load would tend to promote impact breakage and reduce the amount of abrasion breakage. These factors would lead to a coarser transfer size. PJV is running the SAG mills at approximately a 12% ball charge and the ball milling circuit is not often constrained.

The effect of the pebble crusher’s closed side setting (CSS) was also simulated. Increasing the CSS does not appear to reduce the SAG feed throughput by much. Nevertheless, the smallest CSS possible is targeted without causing excessive crusher vibration. Bypassing of the crusher does have a significant negative impact on SAG throughput as shown by survey #2 results and predictions.