Grinding Circuit Expansion

Grinding Circuit Expansion

Increasing ore hardness led to a gradual reduction of mill throughput rates. This, combined with declining head grades and low gold price, prompted a review of milling practice at Kidston. Plant trials and computer simulation of the grinding circuit indicated the potential to increase mill throughput by up to 45% by reducing SAG mill feed size.

Brief history and description of the comminution circuit

Early in 1986, difficulty in maintaining design throughput was experienced on processing particularly hard ore. The throughput rate

grinding circuit original kidston

dropped as low as 280 t/h due to a critical size build up in the SAG mill charge.

The chosen solution was conversion to SABC. Plant testwork was carried out in 1986 to determine the design parameters for the proposed modifications.

grinding circuit kidston sabc circuit

Grinding Circuit Survey 1

A full grinding circuit survey was carried out to obtain sufficient data to develop an initial computer model. The survey was carried out at maximum achievable throughput on the ore type tested. The results were first mass balanced using the JKMBal program to smooth the data, a necessary preliminary step to setting up a circuit model, as the difficulties in taking a representative sample from a circuit of this size are extreme.

grinding circuit milling rate vs work index

The simulations of primary interest are as follows:


The resultant SAG mill load was 120 % of the surveyed case, indicating an overload situation, as expected.

  • Simulation of 660 t/h feed to the SAG mill, however open circuiting the SAG mill by rejecting recycle crusher feed from circuit, and adding sufficient ball milling capacity to achieve final grindsize requirements.
    The resultant SAG mill load was 116 % of the surveyed case, also indicating an overload situation.
  • Simulation of 660 t/h using the existing SABC circuit but crushing the SAG mill feed to a (nominally chosen) F80 of 16 mm, and adding the required ball milling capacity.
    The resultant SAG mill load was 55 % of the survey circuit load. This result was considered to be indicative only, as the SAG mill breakage and discharge mechanisms in the surveyed circuit model were not likely to be the same in a crushed feed situation, i.e. without coarse ore contributing to breakage of the finer ore particles.

Two SAG Mills – Simulations carried out using another identical SAG mill in circuit, and evenly splitting 660 t/h of feed between the two mills, showed that the existing ball mill would accommodate approximately 580 t/h to produce a cyclone overflow size of 80 % passing 0.2 mm. Additional ball milling capacity would therefore be required to increase circuit throughput to 660 t/h.

Final circuit configuration

When designing the expansion, consideration had to be given to the limited space available.

The secondary crushing circuit had to fit into an area bounded on four sides by the coarse ore stockpile, the stockpile feed conveyor, the primary crusher and waste dumps.

Crusher product is returned to the stockpile feed conveyor via a return conveyor. A flop gate was provided at the transfer point so that the stockpile feed conveyor may be bypassed. This allows the system to be used for producing road base for use in the open pit.

SAG milling: The only change required in the SAG mill circuit was replacement of the 75 mm pebble port with a 35 mm grate. This change had to be scheduled to coincide with commissioning of the secondary crushing circuit.

Ball milling: The original cyclone cluster, containing six 0.66 m diameter cyclones, was replaced with an eight cyclone cluster to accommodate the increased throughput. Underflow is split between the two ball mills and may be diverted to either mill singly if required.

SAG mill control

Full load motor current is 5 600 A. The current set point was dynamic and was calculated by another loop which monitored motor current compared with full load current. Operation above or below full load current was recorded as an integration of amp minutes. Current set point for the tonnage controller was calculated from the value held in this counter and was designed to hold the count at a set point of 1 000. An alarm would activate at a count of 4 000 whilst at 6 000 the mill motor would trip.

The advantages of this approach were two-fold. Firstly, heat generation in the motor during periods of operation above full load current was countered by an equal time of operation below full load current which allowed cooling of the motor windings, thus protecting the motor. Secondly, the average power draw was maintained close to the full load value, maximizing power utilization. In practice, power utilization approached 100 %.

It became apparent very quickly that, with the finer feed size, the SAG mill had to be operated with a much lower muck load. If the muck load exceeds 4 to 5 % by volume (with a ball load of 11 to 12 %) grinding efficiency drops away rapidly and the mill overloads.

The increase in muck load which results in overloading is too small to be detected by a change in the bearing pressure. If such a change is detected the mill is already overloading.

As a consequence of the low volumetric load in the mill, the motor power draw is now typically in the range of 90 to 95 % of installed power. Attempts to increase power draw by increasing ball load have only resulted in an increase in ball breakage, with no concomitant benefit in milling rate. This means that maximizing power draw is now inappropriate. In addition, lack of resolution in measurement precludes the use of bearing pressure for control purposes.

grinding circuit expansion at kidston gold mine