Table of Contents
Mineralogy unique to the Lone Tree deposit allowed for the design of a partial oxidation process that takes advantage of the heterogeneous nature of the pyrite morphology. Partial oxidation has resulted in an autoclave plant that is able to achieve low operating unit costs.
Process Description
The oxidation circuit consists of one autoclave, two stages of slurry heating, and two stages of flash cooling. The autoclave was designed to operate at 180°C (356°F) with a slurry residence time of 48 minutes. Design criteria calls for 75% sulfide oxidation through the circuit.
Gold is leached and recovered from the oxidized slurry in a six stage CIL circuit. Each stage incorporates a 10.2 m x 11.2 m (34 ft. x 36 ft.) air sparged tank fitted with a vibrating in-tank screen. Retention time is twenty four hours. Spent slurry is treated with peroxymonosulfuric acid to reduce the weak acid dissociable cyanide concentration and is placed in the tailings impoundment using thin layer deposition techniques.
Loaded carbon is periodically removed from the CIL circuit for gold elution. Carbon is acid washed in 7.0 ton batches with a dilute nitric acid solution and transferred to the strip vessel for gold stripping. The barren eluate is heated to 140 °C (284°F) and advanced through the carbon at 1.5 bed volumes per hour. A 1000 kw electric solution heater has been replaced with a steam heat exchange system. Pregnant solution is pumped to an electrowinning cell for gold removal. Stripped carbon is regenerated in a 400 kw electric kiln prior to transfer back to the adsorption circuit.
Partial Oxidation
The gold recoveries are generally lower than the theoretical recoveries, this is due to cyanide soluble losses rather than refractory losses. One significant difference between the design test work and the actual plant operating conditions is that higher temperatures and pressures are required in the plant to produce the oxidation necessary to reach the 90% recovery level. In September 1994, the autoclave was being operated at a pressure and temperature of 1862 kPa (270 psi) and 196° C (385° F) respectively.
The mineralogy also plays an important role in the use of partial oxidation for the Lone Tree ore body. To date no gold in the deposit has been associated with silica veins or random dissemination in the silica based gangue minerals. The gold is associated with the pyrite, other sulfide minerals and goethite. Often the micron sized gold particles are encapsulated within the sulfide minerals, particularly within the pyrite and arsenopyrite particles.
Since the control of the sulfide level in the autoclave feed is so important to the success of the oxidation circuit at Lone Tree, weekly meeting are held with the Geology and Process departments to discuss the new ore being stockpiled during the week. The ore is actually stockpiled by sulfur grade rather than gold grade. The table below shows the present practice being used:
Oxygen Utilization
The oxygen plant at Lone Tree is owned by Santa Fe Pacific Gold. Air Products and Chemicals, Inc. have been contracted to operate and maintain the plant. The oxygen plant capacity is 204 mt/d (225 st/d) of oxygen.
The operating temperature in the autoclave was increased from 180°C (356°F) to 196 °C (385 °F) at the same time. These efforts targeted a higher oxidation level. The combination of these changes has; lowered the oxygen costs by increasing utilization and lowered steam costs by improving the oxidation reaction. One side benefit which was not anticipated was the ability to process lower than design sulfide ores.
Autoclave Operating Conditions
To maintain the desired temperature in the first compartment required a substantial addition of steam. Addition of cooling water in the subsequent compartments was then required to maintain the targeted temperature. The volume of steam and water combined to displace ore that could have been fed to the autoclave. Operating parameters were changed to maintain a minimum temperature of 168°C (335°F) in the first compartment and to allow the temperature to increase to a maximum.
Lime/Neutralization
In a milling circuit the reagent costs comprise a significant amount of the total operating costs. This is even more important in autoclave oxidation circuits where the lime usage can approach the 195 Kg/ton (100 lb/ton) range. At this consumption level significant savings in operating costs can be made by cutting back the amount of lime added to the circuit. A direct molar relationship exists between the tons of sulfur oxidized and the tons of lime required to neutralize the acid in the neutralization circuit.
By operating at an oxidation level of 50 to 60%, the lime usage is controlled at approximately 134 Kg/ton (55 lb/ton) and the gold recovery is maintained at 88 to 90%. At 90% oxidation or greater, where most other oxidation circuits are operated, the lime consumption would be at least 207 Kg/ton (85 lb/ton). The ability to operate at 50 to 60 % oxidation allows Lone Tree to have significantly lower lime costs. Therefore; the overall reagent costs are lower.
Cyanide Consumption
The cyanide consumption at Lone Tree can be quite variable. Some factors affecting consumption are: whether oxide or sulfide ore is being processed, the water chemistry of the reclaim solution, the amount of ferrous iron in the CIL feed, the amount of copper in the ore and the percent of sulfide sulfur oxidized.
The amount of ferrous iron in the CIL feed in the pilot plant (Hazen, 1992) had a dramatic effect on the cyanide consumption therefore intensive aeration in the neutralization circuit is required to convert ferrous to ferric iron thereby reducing cyanide consumption. Ferrous iron levels are dictated by the percent oxidation in the autoclave.
Dissolved Oxygen in CIL
While operating the plant on oxide ore, dissolved oxygen levels in the CIL was not a concern since random checks showed that levels were at saturation. During the first three months of autoclave operation, random checks also indicated dissolved oxygen levels were at saturation. The mill started experiencing higher gold tails value than expected. Checks of the dissolved oxygen showed measurable levels of oxygen were only present in the last three CIL tanks. Initial speculation was that partial oxidation of the sulfides was the cause of the oxygen depletion. Further investigations showed that the lack of oxygen was due more to the specific mineralogy of the sulfide ores than to the partial oxidation. Different ore types would have very different levels of dissolved oxygen in CIL at the same oxidation levels.
Electric Solution Heater
One problem that did arise in the mill shortly after start-up was the inadequate capacity of the strip solution heater. The heater was designed to raise the temperature of the strip solution from 49 °C (120°F) to 138°C (280°F) at a flowrate of 4.73 l/sec (75 gpm) in approximately two hours. In actual practice it took much longer at a lower flowrate to achieve the increase in temperature. The flowrate was partially limited by the fact that the electrowinning cell could only process about 3.15 l/sec (50 gpm), which meant that strips had to run longer than the designed 20 hours.
Gold Pressure Oxidation Plant Control System
FirstMiss Gold installed a supervisory control system based on Pavilion Technology’s Process Insights neural network software and Oil System’s PI data historian. The primary goal of the system was to control the acid preconditioning circuit. However, our first successful control loop was on the limestone feed rate to the neutralization circuit.
Plant Flow sheet
The plant flowsheet has a conventional grinding circuit feeding an acid preconditioning circuit that destroys the carbonate minerals in the ore before feeding three parallel autoclaves. These hold the ore at 210 C (400 F) and 2.8 MPa (410 psi) for 90 minutes while oxygen from a 290 t/d (320 stpd) Air Products cryogenic plant reacts with the sulfide minerals. The reaction converts the sulfides into sulfuric acid, ferric iron, and ortho-arsenic acid (H3ASO4).
The neutralization circuit is next. Here a staged addition of limestone and milk of lime raises the pH, precipitating the arsenic as a stable blend of ferric and calcium arsenates. The higher pH is also required before cyanidation in a conventional Carbon in Leach (CIL) circuit. Following a standard Zadra strip, the gold is cemented from the pregnant solution with zinc dust. The resulting precipitate is retorted to remove mercury, then fluxed and smelted to oxidize the impurities, produce gold dore.
Supervisory Control System
The supervisory control system consists of three parts. The PI System from Oil Systems, Inc. acts as the data historian, collecting data from the Bailey Network 90 DCS via its own Computer Interface Unit. Process Insights generates the neural network models, and the Run-time Controller runs current plant data from PI through the model and sends control signals back to PI, which sends them back to the DCS.
Besides acting as a data historian transform module (via the performance equation package) for the neural network software, we have set up a terminal in the autoclave control room for the operator’s use. This has proven very popular, as the operators can build their own customized trend and graphic displays.
Application to Neutralization Circuit
The autoclaves discharge slurry at 38% solids and a pH of 0.5 to 1 with 15 to 45 grams sulfuric acid per liter. However, safe cyanidation requires a pH of over 9.5, preferably about 10.5. The neutralization circuit, which consists of four, 570 m³ (150,000 gallon) agitated tanks in series, uses ground limestone to raise the pH from 1 or less to about 5.7 in the first tank. After allowing the reaction to stabilize in the second tank, we then add milk of lime to the third tank to bring the pH to 10.3. Cyanide is added in the fourth tank, and the resulting slurry is pumped to the CIL tanks.
Overall, our goal is to raise the pH to 10.3 at the minimum cost. This means that we add limestone until we reach the point of diminishing returns, then add lime to bring the pH up the rest of the way. Operating experience seemed to show that a pH of 5.5 was the correct setpoint for the first tank, and the operator manually controlled to this value by calling the grinding operator and telling him to adjust the speed of the limestone mill’s belt feeder up or down depending on where the pH was. This control system resulted in frequent out of specification pH values, rapid swings in pH, and was strictly reactive except in the case of planned shutdowns.
Several weeks of data for these inputs was collected in PI and transferred to Process Insights. After cleaning up the data in the preprocessor, several models were run to identify the process delay times. These matched well with our expectations. At this point, several transforms looked useful, (i.e. The model insisted that the free acid content of one autoclave was 5 times more significant than the free acid of another unit, so we created a transform of “average free acid” to use in subsequent models.) These new inputs were then fed into the next run of the model.
The run-time controller had interesting effects. The model used somewhat more limestone than had been the case previously. However, it used substantially less lime. This tells us that we were not as far up the diminishing returns curve as we should have been.
