Strategies For Optimisation of Gold Processing Plants

by Chris Campbell-Hicks

ABSTRACT

Optimisation of many gold oxide processing facilities, in most cases, either does not really take place or if it does take place it happens too late or takes too long. This is unfortunate as extra profits and early cash flows are lost, in new projects, and/or may not be realised at all in others. The situation is easily avoided by ensuring that the plant optimisation happens immediately following commissioning or for an established operation simply ‘biting the bullet’ and making it happen immediately as a ‘better late than never’ decision. This type of optimisation requires intense undiluted focus on each and every problem or restriction that contributes to reduced plant performance or bottlenecks, without letting go, until a predetermined outcome is produced. Thus a dedicated person with specific quantitative goals is required for these tasks. Methodology and strategies for optimisation are discussed and numerous examples presented. The secret of successful heap leaching and the role of mineralogy is discussed.

INTRODUCTION

From the late 80’s and continuing through the 1990’s and the new millennium Australia has demonstrated expertise in gold processing technology that places it in the forefront of capability for development of gold projects both on and off shore.

With more than 300 gold processing plants/projects, designed, built and commissioned within Australia, since the early 80’s, coupled to the current large number and diverse range of off shore developments Australia is probably without rival for fast tracking Fixed Lump Sum or Cost Plus(EPCM) and Turnkey projects for gold.

For instance some of the countries where Australia is currently designing, building or operating gold processing plants include Argentina, Bolivia, Borneo, Botswana, Brazil, Bulgaria, Burkina Faso, China, Chile, Cuba, Egypt, Fiji, France, Ghana, Guinea, Indonesia, India, Ivory Coast, Kazakhstan, Mali, Mauritania, Mexico, Mozambique, New Zealand, Pakistan, Papua New Guinea, Russia, Solomon Is, Sumatra, Tajikistan, Tanzania, Turkey, Uzbekistan, Venezuela, Vietnam, Zambia and Zimbabwe.

This has generated and continues to generate a great deal of metallurgical expertise, particularly among those people fortunate enough to experience ‘hands on’ in a large number of different plants. Much of this expertise is, unfortunately, perishable (Menne 1990) as people take it with them as they retire or otherwise leave the system. In an effort to stop this waste of valuable information, access to which can if nothing else, at least prevent us from constantly reinventing the wheel, an Expert System Shell called Electromagnetic Books has been developed by Menne (1989) and others. This is presented in more detail later.

For others who stay in the system for the medium term, growth in experience leads to a career path through metallurgy via Senior Metallurgist to Mill Superintendent or Mill Manager and occasionally to Site Manager as a logical progression. The scope of work outlined below that is demanded by these positions, for an operating plant, is considerable and it is at this point the opportunity to quickly and efficiently optimise plants is lost.

For optimisation of a recently commissioned plant the options are:

Hold the contract commissioning metallurgists back for several months after the completion of the performance testing.
Employ extra metallurgists.
Engage an experienced short term contract metallurgist/consultant.

The first option will prove difficult to arrange with the contracted engineering group once the plant performance verification is complete. The engineer will have requirements for the metallurgists services on new projects. After the long hours and sustained efforts through commissioning the contract metallurgists are ready for a break and/or a complete change and are seldom keen to continue past the contract completion departure date.

Extra metallurgists can be employed but :

They are usually young and therefore relatively inexperienced.
They will initially need a deal of time consuming supervision and/or direction.
They will take longer to produce desired outcomes.
As extra staff they will have to be engaged as permanent employees.

Engaging an experienced consulting metallurgist or specialist for the short term clearly becomes the preferred option. This is not, however, suggesting engaging the traditional consultant who comes on site, conducts a detailed survey, makes a number of recommendations and departs. This approach leaves the permanent site staff little better off since they are left to implement the recommendations, most of which they were already aware of anyway.

What is required, at this post commissioning stage, is a person who after reviewing and auditing the operation not only generates a report but remains on site and with undiluted focus pursues those areas bottlenecking the operation until the desired outcome is produced to the satisfaction of the Plant Manager. Valuable intensive one-on-one ‘hands on’ training will also automatically be imparted to relevant operating personnel as part of this work and, obviously, at no extra cost.

For an already established operation the same situation applies except that virtually 100% of all that has to be done is already clearly recognised by the incumbent mill staff but who lack the dedicated resources to produce this desired outcome.

NEW PROJECTS

It is desirable to bring the Manager Metallurgy/Mill Superintendent on board during the design phase or commencement of detailed engineering at the latest. If a SAG mill is being considered, which is increasingly likely in most new operations, then he must be employed as early as the comprehensive testwork program prior to establishment of plant design criteria. Failing this a recognised consultant should be used to represent the client. This is often overlooked and is a situation where the Operations Manager, usually a mining engineer, feels sufficiently metallurgically informed to cover these requirements well enough with the Engineering house and the Manager Metallurgy is not brought on board until the engineering is complete and often as late as construction. This is poor practice and in the long term saves nothing. The penalties for his/her late inclusion are:

Insufficient testwork is carried out resulting in flawed design criteria and therefore less appropriate equipment selection.

The metallurgist is not afforded the opportunity to take early ‘ownership’ of the design and therefore may lack the passion to make ‘his/her’ plant stand up quickly.

Minor design changes that he knows, from previous experience, need to be made, now become variations on the construction contract and attract hefty penalties.

The goldfields of W.A., which has the largest number of successful gold mines in Australia, can provide numerous examples of the less successful consequences of such decisions.

During commissioning of a new operation there is always an intense metallurgical input, by experienced metallurgists, lasting from several days to several weeks. For a new operation this commissioning period is normally considered finished at completion of the performance trial. Throughout this period the commissioning engineers are working closely with the owners mill staff, many of whom will be new to the industry, but the engineers themselves actually operate and run the plant.

This is normal for the purposes of carrying out the performance trial where, if the engineer is accountable for performance then, it is only fair that the engineer is given maximum latitude to operate the plant their way. The fastest and most efficient way to do this is to essentially operate the plant themselves for this period.

Although a degree of training is imparted and experience passed on to the owners staff by the engineers during this period, most of the plant and equipment is new and what breakdowns and commissioning hiccups do occur are quickly resolved by the experienced commissioning crew.

The sudden departure of this experienced pool of people, after hand over, in most cases leaves a large vacuum. The new owner has a new plant being run and operated by new people. For a short time all goes well, during this post commissioning honeymoon period, and although the plant is operating well short of optimum it is still early days. The first gold bar has been poured, the project is now a producer and the Board celebrates with champagne.

The gold price falls. Forward sales are not in place. Optimum efficiencies are suddenly required. The hard yards start.

A Mill Superintendent or Manager, as part of his/her scope of work will normally have accountability for and must deal with all of the following:

Plant production and process control.
Production scheduling and forecasts.
Metallurgical accounting and reporting.
Metallurgical balance, daily, weekly, monthly.
Operating budgets.
Capital budgets.
Cost control and supply.
Safety, training and personnel development.
Routine metallurgical investigations.
Laboratory operations.
Maintenance planning and scheduling, shutdowns etc.
Borefield/water supply.
Tailings Dam Management
New projects.

For an average sized gold operation in 2007, say a 1,000,000tpa plant, and typically only one plant metallurgist and perhaps a recent graduate, in addition to the plant manager, this is a very healthy workload.

With only 24 hours in the day this very busy Mill Manager simply won’t make time to dedicate himself, and/or his plant metallurgist, to a full and complete optimisation exercise, without distraction, until results/outcomes are produced in all those areas requiring investigation and rectification.

For existing projects, which may have been operating for several years, exactly the same situation can arise from the same drop in gold price and/or because of a sudden loss of experienced staff which is a common occurrence after several years. The same formula applies. Much time is saved and a faster improvement in profit is realised by contracting an experienced consultant for a short term.

THE ROLE OF MINERALOGY

It is quite extraordinary that Mineralogy, which can provide a greatly enhanced understanding of ore body chemistry, is only given cursory attention or worse still in some cases completely omitted during the testwork schedule upon which plant design criteria are based from the mid 80’s through into the 1990’s. Understanding the mineralogy and the mineral species/forms in which the metal value is present provides a crucial first step guide to the structure of the subsequent metallurgical test work program.

Claudia Gasparini (1986) suggests that microanalytical studies at a very early stage, and before any recovery is attempted, assist in decisions concerning grind sizes and sources of problems such as:

The presence on electrum – particularly coatings over the surface of electrum
The presence of tellurides.
The presence of very small grain sizes too fine to be conventionally leached.
The presence of cyanicides.
The presence of clay/sericite.
The presence of graphite/carbon.
The presence of mercury.
The presence of stibnite and/or arsenopyrite and pyrite and oxygen scavengers such as pyrrhotite.

It is interesting that electrum is accorded a high profile by Gasparini as it is often more common that supposed. Silver tarnishes quite quickly in air forming coatings on the electrum surface slowing down leach kinetics or even completely stopping the reaction.

Gasparini gives an excellent mineralogy testwork flow chart illustrating the importance of mineralogy testwork during the early stages of the metallurgical testwork and which shows the direction this testwork should take.

(After Gasparrini 1983)

A plan for the study of gold ores combining test work, chemical analysis and mineralogy studies. The dashed lines enclose the boxes indicating where the mineralogy studies are more effective.

As an example, the interference by Realgar (α-As₄S₄) and Orpiment (As₂S₃) in gold extraction by forming complex reducing compounds that retard or prevent gold dissolution is well documented. The active inhibitors here are thioarsenite and sulphide ions that adsorb and coat the surface of the gold. A mineralogical study would have revealed this possibility allowing a mitigation strategy to be developed. This also serves as a good example of the use of lead nitrate to improve leach kinetics. In this case, lead ions remove the sulphide as lead sulphide and also accelerate the rate of oxidation of the thioarsenite to thioarsenate thus freeing the surface of the gold (Gasparrini 1993).

If the offending mineral had been Antimony, however, the action is more interesting – particularly at elevated pH as discussed below. (Hedley & Tabachnick 1968)

Impact of Stibnite on gold recovery

pH	% Au rec
12	1.6
11	34.11
10	66.67
9	98.4

Effect of pH on Au Recovery in the Presence of 0.25% Stibnite

(After Hedley & Tabachnick 1968)

Extrapolating the above trend back to the ordinate axis we note that 100% recovery is once again achieved at pH 8.9 but with the reservation that gold recovery will reduce somewhat, and cyanide consumption increase, as the free CN- ion is replaced by dissolved HCN at the lower pH.

Hedley and Tabachnick also document the effect of lead nitrate on the same synthetic ore at 2500ppm stibnite at pH 10 as below:

Pb(NO3)2	% Au rec
(g/t)	% Au rec
0	40.48
150	98.60
750	98.18

Effect of Lead Nitrate on Au Recovery in the Presence of 0.25% Stibnite and pH 10

(After Hedley & Tabachnick 1968)

For Orpiment (As₂S₃) the addition of lead nitrate can provide similar benefits as seen below:

Pb(NO3)2	% Au rec
(g/t)	% Au rec
0	59.35
150	60.71
750	96.16
1500	98.89

What is immediately noted, in the case of arsenic sulphide, is that there is virtually no improvement at 150g/t and significant improvement was not measured until the lead nitrate addition rate increased 5 fold to 750g/t and 99% recovery not achieved until a 1500g/t addition rate was employed.

Effect of Lead Nitrate on Au Recovery in the Presence of 0.25% Orpiment (As₃S₃) and pH 10

(After Hedley & Tabachnick 1968)

In summary the sulphide compounds formed by the decomposition of arsenic or antimony sulphides minerals in (caustic) cyanide solutions, like pyrrhotite, consume the oxygen necessary for gold dissolution.

Also Fink and Putnam investigated the effect of sodium sulphide on the dissolution of gold leaf in cyanide solution. They found that minute amounts of sulphide had definite retarding effects. This could not be accounted for by oxygen depletion and was attributed to an insoluble aurous sulphide film on the surface of the gold preventing attack. This is consistent with the use of lead nitrate for precipitating the sulphide and thus restoring the gold surface and rendering it amenable again for attack by oxygenated cyanide.

STRATEGIES FOR OPTIMISATION

Committing to Outcomes

Before commencing on any optimisation exercise a clear intended outcome should be stated. This must be a quantitative rather than a qualitative written statement that commits to a target that must be achieved. This is important as ‘the best that I can’ is an unacceptable qualitative notion and in reality is nothing more than a cop out for accepting whatever outcome results? Clearly and quantitatively state the intended goal as an increase in tonnage or % increase in recovery such that the desired outcome becomes a committed result and therefore must be produced.

A good technique to use as a starting point, for plant optimisation, is to start at the Goldroom and work backwards through elution, CIL/CIP, milling, crushing, materials handling and finally ending up back in the mine.

GOLDROOM

This is seldom a bottleneck as there is usually very little to prevent more pours/day or week other than more hours being spent in the goldroom. A high grade gold operation, in PNG, increased the gold outturn by 240% from March to June 1996 with the addition of one extra (part time) person in the goldroom.

Most gold rooms easily allow for increased gold production. If some sort of expansion, of this facility, were required it would be unlikely that the Directors of any company would hesitate to spend a few extra dollars to allow more gold bars to be poured each week!

ELUTION AND ELECTROWINNING

As a useful guide for optimisation of this facility a gold industry sponsored research team produced the following from multiple plant case studies :

The loading of gold on to carbon is favoured by low temperature, low free cyanide concentrations, low pH, high ionic strength and the presence of calcium. (AMIRA report July 1989, pp29).

The affinity of carbon for Au(CN)₂^– is such that it may still load quite adequately with one or more of these variables substantially outside optimum recommended levels when other parameters are optimum. Sustained operation outside these guidelines must, however, inevitably have an adverse effect on recovery when (these) other less easily controlled conditions vary.

For efficient stripping of gold, from activated carbon, reversing the above is logical. Thus successful elution should be favoured by high temperature, high free cyanide, high pH, low ionic strength and the absence of calcium.

This is an excellent starting point to commence elution optimisation. Optimisation of the 125°C AARL elution circuit at a small high grade gold mine in PNG in 1996 increased the weekly stripping rate from 6 to 21 strips/week. This facility had been badly bottlenecking the rest of the plant as it regularly campaigned up to 30gpt Au and 200gpt Ag. The increased stripping frequency, at the more elevated temperature of 140°C, coupled to several excursions to higher grades and increased milling rate, lifted the weekly gold outturn by 240%. This was achieved by following the procedures below.

High Temperature

This has been well understood since the work of Salisbury at USBM and Homestake in 1952 and Campbell-Hicks & Beer 1991 which followed directly from the excellent work of Zadra. Figure 1 clearly shows the importance of temperature in shortening elution times using Salisbury’s apparatus.

Translating this to a typical 2 tonne column in pressure Zadra configuration Salisbury’s x-axis scale can be read directly as hours. After 6 hours at 125°C elution is only 80% complete. At 135°C it is 92% complete. At 140°C it is 98% complete when use in conjunction with a closed circuit pressurised electrowinning cell (Baxter and Siddall 1988).

The physisorption of gold on to activated carbon is exothermic releasing ~42 J/mole/⁰C. Where as this is not high, compared to the adsorption of, say, acetone at 560J/mole/^oC, and where a beaker of carbon in acetone becomes quite warm to the touch. It then follows, in agreement Le Chatelier’s principal, where addition of heat forces the equilibrium backwards and desorbs the gold from carbon.

An additional benefit of elevation to 140°C is cyanide free stripping where the higher temperature completely offsets the cyanide requirement while still allowing rapid stripping to barren values of <20ppm Au in 6 hours using either the pressurised cell Zadra or AARL. With the pressurised cell Zadra thermal reactivation requirements are also reduced, with the strong caustic environment, at 140°C and pH13, removing a significant proportion of the organic foulants during the strip.

For AARL circuits operating at 135-145°C it is even more interesting with successful elutions carried out both without cyanide presoak, then without caustic in the presoak and then finally with caustic and cyanide both absent (Tolukuma 1996). This latter result was curious. There was obviously enough cyanide still remaining in the carbon micropores, even after cold acid washing, to allow successful elution. This was confirmed by the evolution of Ammonia gas(NH₃) during subsequent electrowinning after cyanide free stripping and is also consistent with the difficulty that can also be experienced in removing all the HCl when rinsing carbon after acid wash(see anode corrosion below). The pH, of the reagent free eluate, was immediately adjusted to >12, with caustic soda, to provide conductivity during electrowinning. 500ppm cyanide seems to be sufficient to prevent precipitation of gold during electrolyte storage (Menne 1996).

The ability of carbon to hold onto cyanide ions, even after acid washing is evidenced by measured levels of 500ppm, free cyanide, in eluate and/or evolution of ammonia during electrowinning from the cyanide free IPS type stripping plants. This was also observed in the surprising result from the AARL test strip, discussed above.

Lower temperature AARL circuits, operating at a maximum of 110-115°C, although stripping affectively in 4-5 hours are losing ground to 135-145°C units which strip effectively in 2-3 hours. However, many are still in service and the lower temperature 115°C AARL’s occasionally preferred by some engineers. These circuits lack the reagent consumption benefits of elevated temperature and conformance to the AMIRA criteria highlighted above. The AARL circuits, are more expensive, have a high water and low TDS demand but nevertheless are clearly preferred in many cases. Problems with these circuits are normally traditional material transport problems. Zadra circuits operating atmospherically at 90°C are now days seldom built at all.

Optimisation Recommendations

Set a goal of reduced barren values, reduced elution time and reduced cost/strip.
Always aim for maximum possible elution temperature where the higher temperatures will tend to dominate other factors/variables.
Temperatures of 140°C , or above, will enable cyanide free stripping.
AARL plants at 140°C, or above, will allow cyanide and/or caustic free stripping.
For low temperature AARL plants watch for water contamination ie low ionic strength water is essential and check reagent strengths. 0.5% cyanide and 2% caustic will be more than adequate.
For low temperature Zadra plants(ie <115°C) at least 0.5% cyanide is mandatory or high barren carbon values will result.
For low temperature Zadra plants electrowinning efficiency is essential. > 60% cell pass efficiency is mandatory with 75-80% highly desirable. See comments below on electrowinning. Don’t hang about on this one. If all checks out AOK then either enlarge existing cell or purchase/add a new one.
If all else fails go back to basics. Personally check caustic and cyanide reagent strengths and make-up and follow through each step of the cycle.

Carbon Transport

This is a classic post commissioning bottleneck area and one where excellent gains can be quickly made. It is a typical area where the Owners Metallurgist (Owners Representative) can make a contribution to both the conceptual and detailed design engineering. This involvement is dealt with later under Design Phase.

The most common post commissioning bottleneck in stripping circuits, both AARL and Zadra, is the carbon transport system. Engineers have tidy minds, always think in right angles and have an understandable aversion to untidy lengths of plastic poly pipe. Consequently carbon transport lines, normally used to water educt carbon in and out of the elution circuit, have numerous right angle bends in order to follow the pipe racks in a neat and tidy fashion with each bend producing a potential blockage point. And block they do!

This problem is easily resolved by replacing the ‘torturous path’ pipelines with smoothly sweeping sections of black poly pipe of approximately the same I.D. Although these lines will seldom follow the pipe rack they can still be done in a tidy fashion while maintaining minium angles of curvature.

Linear velocity in the pipes should be kept between 2 and 4m/s with 3m/s being optimum. For a typical 2 tonne column which has around 18m³/h raw eduction water flow rate available this equates to a 46mmID(or 50mm OD) pipeline and a line velocity of 3m/s. In this case the original pipeline will have been 50mm. A 63mm(OD) poly line could have been used and would have reduced the restriction problem. Line velocity would then have been just adequate.

There is often a temptation to overkill, on the pipe line size, when replacing these sections. With 90mm OD or even an occasional 110mm OD poly pipe being commonly fitted in place of the original 50mm steel line. For the 90mm line velocity now drops well below 1m/s and for the 110mm line it falls to 0.6m/s. Both pipe sizes will bog continually for transport water flow rates below 25m³/hr.

Optimisation Recommendations

Set a maximum transfer time of 30-40 minutes. This is achievable.
Don’t waste time on this one either, if restrictions are suspected that are slowing down transfer rates immediately replace the torturous path steel pipes with smoothly sweeping sections of poly pipe of similar ID.
If there is any doubt about the line diameter, do the calculations. Flow rate should be 2-4 m/s
Don’t kneejerk and overkill on pipe diameter

Acid Washing

Pachucca type tanks, ie those with air agitation only, will introduce large quantities of carbon dioxide(CO₂) into the circuit, from the air, and if lime or caustic is already present large quantities of calcium carbonate(CaCO₃) will be formed and report as very high levels of inorganic foulant of carbon. This will require abnormally high levels of acid wash to remove the calcium. At a gold mine in the East Murchison area, in 1997, calcium carbonate levels were measured in excess of 60,000ppm requiring more than 140kg of 32% HCl/2 tonne batch of carbon to remove all the calcium which had built up over several years.

Since the presence of calcium favours adsorption of gold, onto carbon, and the absence of calcium favours elution, stripping times were being greatly extended and still resulted in high barren levels. Typical elution times were 18-20 hrs even at 140°C. A special, high strength, acid wash campaign was introduced reducing the calcium levels below 10,000ppm. Stripping times immediately reduced back to <12hrs and continued to fall as the calcium levels were lowered still further.

Higher acid washing strengths and even normal 2-3% strengths require thorough rinsing. This means good air or other agitation and ‘rest periods’ between rinse/flush periods. Acid washing or rinsing without adequate agitation is akin to washing and rinsing our work clothes, in the washing machine, with good water flow but with the agitator turned off! The smell of ammonia after opening the cells in cyanide free IPS systems bears testimony to the retention of significant cyanide in the carbon micropores even after acid washing? Acid can be similarly retained and simple prolonged flushing can take several hours. The acid takes time to diffuse out of the pores and several ‘stand and soak’ periods to allow this diffusion to take place can greatly reduce flush times. Three half hour flush periods with two diffusion soak periods in between can be of great benefit in shortening the rinse cycle.

Optimisation Recommendations

At the design phase select mechanically agitated tanks in preference to pachuccas.
If pachuccas are already installed regularly check levels of inorganic fouling by thermal gravimetric analysis(TGA) to monitor calcium levels.
Set a maximum allowable calcium level on carbon of 5000ppm. Acid wash until this is achieved.
Ensure adequate agitation during both the acid wash and rinse cycles.
Incorporate rest periods into the water rinse cycle to allow time for residual acid to diffuse out of the micropores

Thermal Reactivation

This is one of the simplest areas to optimise and yet is often one of the most poorly managed. All carbon needs to be thermally reactivated, by removal of organic foulants, is the presence of steam at a temperature of 650°C.

The role of steam

Much carbon that was being reactivated during the early to mid eighties showed little difference in activity before and after reactivation. This was a typical result from the early vertical tube kilns where moist carbon was fed down the kiln tubes without further water injection into the hot part of the kiln once temperature had been reached. This moisture evaporated quickly with steam passing back up the tube leaving the carbon essentially dry before it reached the hot part of the kiln where the surface pyrolysis takes place that is needed to remove the organics. At this point, without steam, no pyrolysis reaction means no reactivation. Water, which quickly flashes off to steam, must be injected into this hot regime. Fitting a rotameter as well as a flow controller is highly advisable.

The problem is less apparent with the horizontal type kilns where the steam must pass down the length of the kiln and discharge from the hot end. There is also less chance of the organics redepositing back onto the carbon as they do not have to pass back up the cooler part of the tubes as is the case with the vertical kiln. It is interesting to note, however, that nearly all modern Armstrong Holland or Ansac horizontal kilns can now come fitted with water injection capability at the hot end. These equipment suppliers will still often maintain that this additional water injection is unnecessary. Do not accept this and demand the additional water injection.

The role of temperature

With vertical tube kilns fitted thermocouples measure the carbon temperature, which in turn, via a PLC, controls the high, and low fire on the burners. These are usually mounted vertically upwards from the bottom cone and protrude for a short distance up one or more tubes. This is interesting since any tendency for the tubes to block or become restricted in downward flow, say due to cementing together of dirty carbon, will naturally happen at the most restricted point. In general then it is highly likely that flow through these tubes will be slower than through the rest of the less restricted tubes. This means that the one or two tubes that have the thermocouples in them reach temperature but all the other unrestricted tubes, with the carbon travelling more quickly, do not. Only poor or partial reactivation takes place.

This problem is easily rectified by relocating the thermocouples to a horizontal position 75mm below the bottom end of the tube so that the carbon flows freely over the probe(s) at a velocity representative of all tubes.

Don’t rely on a single point comparative test for in house activity testing. Use a multi-point method and have the activity independently tested by reputable laboratory such as the former Mineral Processing Laboratory (Chemistry Centre W.A.) at Curtin University in Perth (now part of CSIRO at the same facility).

At Lawlers gold mine(W.A.) in 1989/90 both of the above problems were existing in the kiln being used. The probes were relocated and water injection lines, rotameters and flow regulators fitted. After these modifications it was noted that the high fire periods on the burner more than doubled in duration. Activity testing before and after showed an increase in activity from less than 60% to 118% of green(fresh) carbon activity when operating the kiln at 700°C. This result was activity tested and confirmed independently by MPL who had been producing >100% green activity from carbons thermally reactivated through their test kiln. Carbon hardness was also measured before and after and no significant deterioration had occurred. The conditions were considered extreme, however, and the temperature reduced back to 600°C for a returned activity of 98% that of green carbon.

Optimisation Recommendations

All carbon needs is >600°C in the presence of steam. If less than 90% activity is being returned after thermal reactivation there is a problem.
Check that adequate water/steam is being injected into the hot part of the kiln. If unsure fit a rotameter.
If a vertical tube kiln is in service check that the thermocouples are not restricting flow and giving a high bias in the temperature indication. If flow restriction is indicated refit the probes in the horizontal position as recommended above.
If elution times become protracted of barren carbon levels are becoming too high, check the carbon for inorganic metal foulants. This can be done through the CSIRO Mineral testing facility at the Curtin University campus in Perth W.A.

Poor Cell Efficiency

This is almost always due to poor electrical contacts and/or low conductivity and is therefore easily corrected. Because they are only operating at low voltage, typical 2-4 volts, then any additional resistance of a few milliohms will greatly reduce the current. For a typical cell with 12 cathodes, of say 600 x 600mm operating at 3 volts and 600 amps total, an increase in resistance of only 5 milliohms will reduce the current by 50%.

Using Ohm’s law V=IR
Volts (V)	3	3
Current (A)	600	300
Resistance (Ώ)	0.005	0.01

This is a huge decrease in current for such a small increase in resistance.

Increasing voltage to produce greater current will only electrolyse more water and often dislodge gold and produce excessive cell mud. It may also melt the plastic insulation by producing ‘hot spots’. Voltage should never exceed 5 volts. If voltages higher than this are required to maintain current there is another problem.

Cell pass efficiency is given by:

Efficiency(%) = (Solution in(ppm Au) – Solution out)/solution in * 100

and should be between 60-80%.

If less than 70%, all electrical contacts, including bus strips should be checked. This should be initially done with the electrician who will instruct in a thorough cleaning procedure. This should be done at least monthly.

Prior to electrical checks, if the current is low, the solution conductivity should be checked. If caustic soda is being used as the main electrolyte then, in the absence of a conductivity meter, the pH should be a minimum of 12.5. At pH<12.5 insufficient ions are present to effectively carry the current.

Optimisation Recommendations

Set a maximum current/electrode as per suppliers/vendors recommendations to be achieved at a voltage not exceeding 4 volts.
Check solution conductivity and add caustic as required.
Check and thoroughly clean all electrical contacts.

The Effect of Base Metals

Copper, nickel and other base metals have variously had serious effects on electrowinning in a number of operations. For high temperature operations these metals can be removed by carrying out a cold cyanide pre-strip on the carbon first. Rinse and then carrying out the normal high temperature elution.

Anode corrosion

This problem is common to many operations, and costly, with a set of 10 316SS anodes costing approximately $6,000 for a typical industry standard cell. Savage corrosion is usually caused by chlorine gas attack with the chlorine coming from either a high chloride water supply or acid attack from incomplete rinsing of carbon after acid washing in hydrochloric acid coupled to poor attention to eluate solution make up.

High levels of chloride in the electrolyte will report to the anode as chlorine gas (and initially in its monatomic or nascent form!) This is bad news and will eat through 316SS anodes like a hot knife through soft butter.

As mentioned above under acid washing, many operations water wash after acid washing until a pH > 6 or more is achieved. The soak period of an extra 40 minutes to 1 hour, in one WA operation, saw the pH drop back from 5 to 3.5 as the acid slowly diffused out of the micropores. A further rinse after a soak period is essential to remove additional chloride ions.

Caustic and hydrogen embrittlement of steel are particular forms of stress strain corrosion that potentially could occur. To date the author has not observed any direct evidence of this but it is a common problem in the alumina industry where hot and very concentrated caustic is pumped around circuits for sustained periods.

An extremely surprising result emerged from an IPS elution circuit operating in the Southern Cross area of WA. The 316SS anodes were replaced with copper yielded excellent service for more than 7 years without replacement!!! One would have expected that the strong caustic and cyanide environment together with ammonia generation would have quickly dissolved the copper to form copper tetramine? The electrochemistry, to explain this phenomena, remains under investigation. The other obvious advantage of using copper is its lower resistivity (greater conductivity). The resistivity of 316SS is around 95 ohm/cm while that of copper is more than an order of magnitude lower at 6.8 ohm/cm.

Optimisation Recommendations

Set a goal of nil anode corrosion. Check pH of electrolyte and if necessary measure chloride levels.
Introduce a ‘stand and soak’ step into the acid rinse procedure (already mentioned).
Check caustic and cyanide solution make up is happening correctly.
Trial copper anodes?

Oil Heating Circuit

For elution circuits using an oil circulation system and heat exchangers (Hex), the oil systems will often absorb moisture or gas producing vapour locks that cavitate the pump and trip the heaters. This situation is easily relieved by ensuring that the vapour release line to the oil header tank comes off the main oil circulation line at the highest point and reports directly to the header tank without any horizontal sections. This line should be separate from the oil fill line.

The heat exchanger should be regularly descaled. Sulphamic acid is ideal for this purpose but if not available 5% HCl can be used provided it is thoroughly water washed and flushed with caustic immediately afterwards

Optimisation Recommendations

Ensure vapour relief line proceeds straight up to the header tank without ‘engineering right-angles’.
Regularly acid wash the heat exchanger.

CIL/CIP

Some trivial? chemistry

The simple and most commonly accepted equation for the dissolution of gold in caustic cyanide solutions remains Elsner’s equation (Elsner 1846) usually expressed as:

4Au + 8NaCN + O₂ + 2H₂O = 4NaAu(CN)₂ + 4NaOH

Bodlaender suggested that this could proceed via a peroxide intermediate step as:

2Au + 4NaCN + 2H₂O + O₂ = 2NaAu(CN)₂ + 2NaOH + H₂O₂

The hydrogen peroxide formed being used in the reaction:

2Au + 4NaCN + H₂O₂ = 2NaAu(CN)₂ + 2NaOH.

Bodlaender found that he was able to account for more than 70% of the hydrogen peroxide predicted in his equation. Bodlaender’s equations probably more accurately express the true reactions that take place. The overall equation, however is the same as Elsner’s.(Bodlaender 1886)

What is more interesting is that if we consider the oxidation of gold as the sum of the two partial redox equations:

4Au – 4e^–= 4Au⁺ (oxidation)

2H₂O + O₂ + 4e^– = 4OH^–(reduction)

Which added yields:

4Au + 2H₂O + O₂ = 4Au⁺ + 4OH^–

Where then is the cyanide?

Oxygen and cyanide

This may seem trivial but is a key point to maintain focus on when optimising the leach and adsorption areas. Certainly gold would not effectively leach and load onto carbon without cyanide present to complex the aurous ions (Au⁺) immediately they are formed. All this is doing, however, is driving the equation to the RHS and we should not over focus on its importance at the expense of the vital role O₂ plays.

Many old and abandoned copper and gold heap leaches in the USA have been successfully reopened simply by ramming perforated steel rods into the base of the heaps and injecting O₂/air and simply adding water? (Brunysten A 1991)

Cyanide

Oxygen is oxidising the gold not cyanide and substandard performance of leach circuits will 9 out of 10 times be due to low dissolved oxygen levels rather than low cyanide levels. Referring back to the AMIRA conclusions we note that loading of gold onto carbon is favoured by low free cyanide.

Unnecessarily high levels of cyanide in the first adsorption tank will actually inhibit gold from loading. This due to the cyanide ions (CN^–) occupying sites that could more usefully be occupied by the aurous cyanide ions (Au(CN)₂^–).

Cyanide requirements for a circuit should be derived from the change in cyanide value from the first to the last tank. Generally, 70-90ppm cyanide in the last adsorption tank is more than sufficient to prevent gold from re-precipitating and the first cyanide tank should be run at whatever level allows 80-100ppm in the last tank. As a general rule of thumb, unless significant copper or other cyanicides are from time to time present in the ore, levels of cyanide above 250ppm in the first tank are a complete waste of money.

Oxygen

This is the most important single consideration in optimising leach circuits and having a dissolved oxygen (DO) meter in modern gold plants is an absolute necessity. Efficient use of oxygen in most circuits will reduced the required leach circuit capacity(size) by up to 50%. That is, more leach efficiency rather than more tanks. Oxygen levels of 20 – 30ppm are now commonly being achieved, at the top of the leach circuit, in many plants with levels to 30ppm reported at Tolukuma(PNG), Mt Kasi(Fiji) and Ocampo (Mexico).

What now becomes very important is to quickly take advantage of the high gold solution values that result from the high DO levels by immediately contacting the slurry with active carbon. This demands that the number of non-carbon tanks in preleach should be a minimum. The sooner the solution gold can contact carbon after leaching the better. Improved CIL and elution efficiencies will result by reducing any potential for pregrobbing.

Fortunately the last 10 years has seen a dramatic increase in the use of oxygen in gold treatment and to very good effect. This together with shear reactors, such as the Filblast system, combine a continually ‘swept’ fresh gold surface presented in the presence of higher dissolved oxygen levels to greatly enhance leach kinetics.

LeachWELLÔ, a proprietary cyanide leach accelerator product, now available commercially, as one of its functions, produces the same effect by chemically removing the passivating substances from the gold particle surface.

More importantly, for optimisation purposes, LeachWELLÔ will determine the ultimate cyanide leachable gold for a given circuit this providing a yardstick for leach circuit performance. At Boddington gold mine in Australia, which operated at low grade and high tonnage, optimum performance was essential and LeachWELLÔ was used to indicate a final insoluble tail value to be targeted.

If testing for the benefit of oxygen to retrofit to an existing plant is being considered it is cautioned that a less positive result is often produced by laboratory testing with oxygen compared to actual plant performance as these O₂tests tend to understate the benefit.

Impact of Stibnite on Gold Extraction.

This was mentioned in the Mineralogy section above but in summary the following.

Hedley and Tabachnick, in an excellent monograph published by American Cyanamid in 1968, made up a synthetic gold ore using a mixture of very finely divided gold in pure silica (sand) and then added various deleterious minerals to a concentration of 2500 ppm to check their effect on gold dissolution. When stibnite was added to the ore and leached with cyanide at varying pH gold recovery dropped dramatically as the pH increased. These results are presented graphically below and it is immediately noted that the impact on recovery is extreme at higher pH, reducing to 1.65 at pH 12 but increases back to 99% at pH 8.9. pH 8.9 is a very low pH to use in any cyanide circuit and it is cautioned that CIP/CIL circuits that use oxygen injection rather than air for oxidizing the gold are preferred in this case leaving only the mill discharge trommel and the carbon screens as the remaining places where free atmospheric cyanide levels may exceed 10ppm.

Effect of Antimony on Gold Recovery

Optimisation Recommendations

Dismiss the notion of % recovery completely and think purely in terms of an absolute cyanide insoluble final tail value.
Use LeachWELLÔ60X as a post leach bottle roll to determine what the ultimate insoluble tails target should be. Set a final tail solids value of no more than 10% higher than this.
If a dissolved oxygen meter is not available purchase one immediately and determine if the circuit is oxygen deficient.
Almost all plants will benefit from oxygen injection by a greater degree than will be indicated from testwork.
Running high cyanide levels will inhibit adsorption of gold onto carbon. Cyanide should be dosed at a rate that will allow just under 80ppm in the last tank provided Ag/Au ratios are not high.
The case for high silver/gold ratios is dealt with separately.

Carbon Circuit

This circuit generally runs well and without major problems with regular stripping and a good carbon advance rate and high carbon activity. This is most important and a cost/benefit analysis of stripping costs Vs recovery % (in the leach circuit) will need to be performed. This will determine what the cut off point is for the number of strips/week to maintain generally low carbon loadings and lean and mean gold hungry barren carbon being returned to the bottom of the circuit.

The adsorption circuit should not be taken in isolation to the leach circuit, particularly where preg robbing species are or may from time to time be present. When O₂ is being injected to assist leach kinetics it is important to allow this rapidly dissolved gold cyanide to be exposed to carbon as early as possible to take advantage of its high liquor tenor and minimise preg robbing. Additionally, after 24 or 48 hours leaching, the benefit of initially high oxygen levels is lost with the same(or sometimes even less^*) total gold dissolving at high O₂ levels, say 20-30ppm, as will be dissolved using air producing only 6-9ppm O₂if both are left in slurry without carbon contact.

Some people may be familiar with the ‘strange’ testwork leach result that often arises at 24 or 36 or 48 hours when recovery appears to fall again? The leach recovery curve actually dips downwards? At a PNG gold operation with seven adsorption stages in the CIL the lowest tails solids and liquor value were at T5, with T6 and T7 consistently running at higher tails solids values. This preg robbing effect is widely reported and has been given extensive treatment by Menne(1992-1996).

Optimisation Recommendations

Fast carbon turnaround strategies are the answer to excursions into high grade gold campaigns or when significantly higher silver values raise their ugly head. This means more strips/week.
Stripping less frequently at higher loaded carbon values and carrying a higher gold in circuit inventory value(GIC) may be a false economy.

Higher concentrations are sometimes preferred in the first adsorption stage eg to handle high grade excursions where 20gpl will demand faster carbon turnaround strategies. This offsets higher loadings and is a matter of judgement and the practice will vary from site to site. In all cases, however, if the carbon and slurry are reporting directly to a vibrating recovery screen then the slurry return and wash water should be lined out to either the second of third adsorption stage since:

Slurry liquor concentration will be more compatible with the second stage. Wash water will dilute the density of this important tank making carbon recovery more difficult (slower) as the carbon tends to sink to the lower regions of the tank.

GRINDING AND CLASSIFICATION

Milling or ‘metal liberation’ is normally seen as the most important metallurgical step in the mineral extraction process and is a common step for nearly all metals. Because of its wide ranging application more work has been done in this area and it is historically better understood than other steps in gold processing.

Nevertheless grinding is often taken for granted and opportunities to fully optimise grinding and classification circuits, especially using the sophisticated modelling techniques currently available, are not taken. Recent use of such models have produced results than are considerably better than Bond throughput predictions. At a Leinster operation, and a number of other Australian operations, the maximum tonnage predicted by Bond’s equation have been comfortably exceeded. In one case by almost 21% after optimisation had started (Campbell-Hicks 1996). This effect may, at least in part, be related to ‘free grinding’ discussed below.

Most of the following discussion centres around SAG rather than ball or rod milling as the latter two options are more straightforward.

Modelling

To effectively model and optimise comminution circuits at least two circuit surveys are necessary with the plant operating at average throughput and at maximum throughput. This involves comprehensive sampling of all ore and slurry streams at these different plant conditions. The samples are accurately sized and this information, together with hardness testing, is then modelled. Excellent modelling, in Australia, is carried out by the Julius Kruscnitt Institute(JKSimet) in Queensland, Steve Morrell (SMC testing) and by Siddall at Orway Mineral Consultants (OMC) in Perth. OMC are also licensed to use JKSimet in addition to their own proprietary software.

OMC will also provide guidance/supervision for the sampling, particularly on the selection of the larger ore/rock samples where shape and size of the selected specimens is crucial.

If a SAG mill has been installed as part of the circuit, SMC and/or other Advanced Media Competency Testing (AMCT) is absolutely necessary to fully define the ores comminution characteristics. There have been many instances where SMC or AMCT has not been carried out on the ore and a SAG mill has been inappropriately selected for service. The resulting plant performance can be a disaster. Siddall(1989) gives a useful table indicating when selection of SAG or AG mills will be appropriate and when ball milling is more appropriate. Insert Bernie’s table

AMCT can be carried out by AMDEL in Adelaide/Perth or JK and will require up to 400kg of appropriately selected rock. AMDEL will also carry out additional proprietary tests, devised by OMC, under licence. SMC testing, now the almost universally preferred SAG/AG testing technique, can be carried out by almost all mineral testing laboratories of note.

Ore Feed

With the increasing popularity of SAG mills there is a critical factor that is acknowledged but not seriously taken on board during the planning stage of many gold mining operations. AG/SAG mill at all times require a blended feed. Quite often the most difficult job in SAG mill management is to convince the mining department of the mandatory nature of the hard/soft blending requirement of SAG mill feed. Unless this consideration is deliberately accommodated as part of the pit design, both in the initial planning stage and throughout all further developments during the life of the mine, the mill will always under perform, quite often with disastrous results. The West Australian goldfields are littered with the carcasses of such cases.

In summary, the optimisation of SAG mills begins in the mine.

If low aspect ratio SAG mills are in circuit the problem is exacerbated. These mills have always been popular in South Africa and are now quite common in the Australian gold industry. Their advantage is lower lead times during procurement and circuit simplicity. They are also very ore specific and require constant hour-by-hour management. Some of these management strategies are given below.

Ball charge

The response of many SAG operators is to immediately charge more balls or larger balls to their mill at the first hint of poor grind or reduced mill throughput. This is particularly the case where they have come from many years of ball milling. Before this step is taken, however, a number of other variables can be used to optimise grind and throughput (Campbell-Hicks 1989. Siddall 1989. Napier-Munn et al 1996) Some of these variables are:

ball size
ore feed size
ore feed hardness
mill speed
mill density
circulating load
improved classification – cyclones
pebble relief ports and scats crushing

Ball Size & Charge Volume

Often a change in size to a smaller ball can assist with grinding, especially with softer ores. Large ball sizes can be used for harder coarser ores, particularly in high aspect ratio mills, to promote impact breakage, if ore becomes more competent at depth. Larger balls will greatly accelerate wear, however, and often will require a change from rubber to steel liners.

Promoting impact breakage, by increasing ball charge, to achieve additional throughput comes at the expense of a coarser grind. As the coarse grinding media breaks at a more rapid rate it spends less time wearing down by abrasion and producing a fine product. Additionally the middle (critical) size rocks in the 25-50mm range are also broken down so they can no longer act as ‘small media’ suited to the grinding of smaller particles. This will then require charging to larger volumes >15% to compensate for the grinding media removal?

Ore Feed Topsize

An increase in ore feed topsize can occasionally be used to promote impact breakage in preference to increasing the ball size in SAG milling situations where the inlet trunnion diameter permits. The basic rule of thumb here is that the internal diameter of the trunnion must be at least three times the diameter if the largest ore lump that it sees. There is more commonly a temptation to decrease the feed topsize, by closing up the primary crusher aperture, which can contribute to a build up of critical size material in the 50-80mm size range.

Ore Feed Hardness

For SAG mills softer ores will give higher throughput at coarser grind. For AG mills softer ores will be unable to allow survival of grinding media and coarse grind will also result. For very competent ores the situation is worse, with inability of the largest rocks that report to feed being able to break at all, drastically limiting throughput.

Mill Speed

This has more recently become a useful additional variable for SAG milling. As speed increases we see an increase in impact breakage leading to increased throughput but with this destruction of coarse grinding media a coarser product results. Thus variable mill speed can be easily utilised to improve grind but at reduced throughput.

Mill Density

A quick fix to improve grind is mill density. Viscosity permitting an increase in mill density will produce a finer grind. At higher densities the particles are closer together and therefore come into contact more frequently. Additional attrition grinding results.

Circulating Loads, Hydrocyclones and the Free Grinding Phenomenon

Ball mills are commonly in closed circuit with cyclones or other classifier and will not be discussed. SAG/AG circuits were not traditionally run in closed circuit, particularly where high aspect ratio mills are used, except in South Africa where low aspect ratio SAG milling is more common. It has until recently been preferred, in Australia, to run a SAG-Ball combination (SAB) circuit with the SAG in open circuit and closed circuit on the ball mill. More recently, in Australia, the practice is becoming more widespread as the effect of a phenomenon know as ‘free grinding’ is used to advantage albeit in many cases not recognised as such. This effect can be observed when a given mill is changed from open circuit to closed circuit resulting in a finer grind without any apparent increase in power or decrease in throughput. The mechanism is not well understood but the JK Institute have hypothesised that it is due to increased attrition breakage of small particles in the interstices between larger rocks as these interstices fill up with slurry from the cyclone underflow.

Pebble Relief Ports and Scats Crushing

Pebble crushing of critical size material that can be relieved from the mill by pebble relief ports in the 25-50mm size range have become increasing popular. Such crushing is a much more efficient use of power and results in an overall circuit specific energy drop. Pebble crushing is particularly effective because it relieves those critical size rocks that have built up and that exist because the mill is having obvious difficulty in breaking them. It selectively crushes the ‘problem’ rocks into a finer crush that can be returned back to the mill for further size reduction.

It should be noted however that their removal reduces the volume of that material responsible for fine grinding by abrasion and thus will yield a coarser grind by removing this fine grinding media.

Coarsening the Grind

With improvements in leach kinetics by better control of dissolved oxygen levels the effect on recovery of coarsening the grind can be greatly reduced. Many operations at 20ppm DO will deliver the same recovery at 106m as they will at 8ppm DO and 75m. This can be a very useful for increasing throughput as a coarsening of the grind from 75m to 106m can allow up to 20% increase in throughput. These calculated increases come directly from the Bond power equation. This is, however, ore body specific.

Application of PLC’s and Expert Systems

To assist with SAG mill management decision support software that can be driven by inexpensive PC’s and PLC’s are slowly gaining acceptance. One such expert system is HelpSAG developed by Siddall and Menne(1992). Testing, measurement, and development of site specific data for these programs is time consuming and requires input from an appropriate ‘expert’. This is difficult and the experts time is seldom cheap. This may account for the slow development of such systems.

One simple and easy management system for SAG milling, that can be quickly implemented to maximise tonnage can be derived from the mill power curves. The plot of mill power(P) Vs mill weight(L) will generate a series of curves which peak just short of critical volume. Driving the mill beyond this point will result in slurry spill from the mill feed trunnion. The optimum mill feed rate will be at the top of the mill curve where the first derivative of the curve at this point will be just > 0. An algorithm can be programmed and via a simple loop to PLC can control the mill feed rate to maintain the mill at this point (dP/dL³ 0). Inputs will be the mill weight, power and weighometer reading and the output to the variable speed feed to the mill. Thus the mill feed rate decision can be taken out of the hands of the operators and placed in the hands of the PLC. Alternatively the system can be used simply for decision support and only provide recommended tonnage feed rates. The rate is then set by mill operations personnel.

Mill Alignment and Lubrication

In this author’s experience there are two and two only vitally critical parameters, over which we have control, for virtually any type of mill that need to be as close to perfect as possible. This needs to be done from commissioning and maintained with regular input by reputable specialists in this field. These two requirements are:

Transmission alignment and,
Lubrication system.

All mills subside or ‘settle in’ on the plinths and mountings after several months of operation often requiring quite substantial alignment adjustment. Failure to address this issue, if indeed there is a developed misalignment, can result in expensive massive failure and/or accelerated wear of major transmission components and bearings. We seem passionate to re-invent the wheel on this issue, and it is common in a large number of plants familiar to us in W.A., although many mill and maintenance superintendents will deny it

All other (important) factors effecting mill performance and availability are either trivial by comparison or issues of design inheritance and/or issues over which we have less control. Close to perfect lubrication is also essential and it is recommended that an expert in this field is given the opportunity to provide reassurance. The cost could be around $10,000/year and is money extremely well spent.

Available mill power

Mill motors can often be re-rated to provide additional deliverable power to the pinion provided adequate motor protection is in place. A specialist engineer, such as BEC Electrical in Perth, should be used to carry out this investigation and may recommend the fitting of additional protection. Authorisation to increase motor loading 10% or more often results which, mill volume permitting, may allow an immediate proportional increase in tonnage.

Optimisation Recommendations

Set a target of 10% increase in tonnage through the mill!
Carry out a comprehensive mill survey and if data is not available sample ore body and comprehensively test ore for Advanced Media Competency(if a SAG mill is in service). Supply data to OMC or other appropriate expert for optimisation modelling. OMC are ideally suited to supervise such a program, which they have in house, and are also the W.A. agents for JKSimet.
Unlike ball or rod mills SAG mill are grind sensitive and require management on an hour by hour basis, require variable speed drives and the maximum possible amount of instrumentation.
Have the mill alignment and mill lubrication system checked by a recognised expert.
Have the mill motor checked for possible re-rating – this may provide an instant 10% or more increase in tonnage.

CYCLONES

Many operations are presented with the problem of a small quantity of coarse gritty material reporting to the cyclone overflow (COF). In some cases this is quite marked and can contribute to a build up of coarse material in the first leach tank and in extreme cases completely fill this tank reducing leach residence time and volume and requiring the tank to be removed from service and cleaned.

At a PNG operation grits were noted that appeared to coincide with surging of the cyclone feed noted by a constant swing in the cyclone feed pressure gauge. By lowering the mill discharge hopper level it was observed that coarse material was alternately building up on the sloping base of the hopper and then slumping into the mill discharge/cyclone feed pump causing this surge in flow.

The problem was resolved by placing a low friction high density, high molecular weight polypropylene base on the bottom of the hopper allowing smooth passage of material into the pump suction.

Optimisation Recommendations

If the cyclone pressure is fluctuating wildly and the cyclone feed pump hunting try fitting a 25mm sheet of high molecular weight high density polypropylene to the base of the mill discharge/cyclone feed hopper.

DATA ANALYSIS AND CORRELATION COEFFICIENTS

Metallurgy has taken full advantage of the exponential growth of computers in the last 20 years with virtually all gold operations having access to a large data base of information. Metallurgical clerks can spend the best part of a full day entering information into comprehensive spreadsheets that provide, apart from the daily metallurgical balance, an excellent data base for variable comparisons and analysis and plant performance monitoring.

One of the simplest and most powerful tools that can be used for problem analysis and diagnosis and to assist with optimisation is to obtain correlation coefficients, by linear regression, between plant variables. These programs are easily accessible from Lotus or Excel spreadsheets. In Excel, for instance, the CORREL program is found under the function wizard(f_x).

During optimisation of the AARL elution facility at the PNG operation, plant recovery began moving by several percentage points at irregular intervals and dropped from 98% to 79% within a period of 5 days in late April 1996. The operation was, at the time, mining a +10g/t gold and +30g/t silver ore body and the fall in recovery reflected as a tailings loss which peaked at +2g/t cyanide soluble gold. All daily recorded plant variables or levels such as cyanide, pH, carbon concentration, carbon loadings, head grade, cyanide insoluble tails grade(refractory component), grind size, slurry density and dissolved oxygen levels for the previous three months were tabulated and compared to plant recovery %. Correlation coefficients for all these variables were run against recovery with oxygen clearly emerging as the culprit with a correlation coefficient of +0.78. When the oxygen data was ‘conditioned’ ie invalid data points, such as those days when the cyanide was exceptionally low, are removed the correlation was 0.88. The PSA oxygen plant was overhauled and dissolved oxygen restored to previous levels. Recovery responded immediately by moving back to 98%.

An expert system such as HelpSAG, as mentioned above, is a SAG mill specific example of a more general expert shell developed by Menne (1988 – Electromagnetic Books) where we combine qualitative human knowledge with quantitative data into a knowledge base ie the rules.

These decision support systems can then advise people on what action to take and when to take it, assist with problem diagnosis and take action with less human guidance (Pena 1989). Menne further suggests these systems;

“Reduce the cost of human experience by capturing perishable expertise and making it available for shared access. No computer skills are required to store and recall know-how. This allows the basis and authority of decisions to be assigned thus allowing experts to pass routine expertise down the line, freeing their time to generate more. Consistency of judgement and advice is brought to decision making by making the rules, facts and preferences explicit. Access to know-how remains easy no matter how large the knowledge base grows”.

Citect and Honeywell type systems are becoming increasingly popular in the larger plants and, although making some inroads in this area, essentially remain as PID plant control systems for normal operation and data storage and retrieval.

Optimisation Recommendations

If recovery has fallen or is moving around, run the correlations of all plant variables Vs recovery %. This may immediately diagnose the offending variable.
For SAG milling a simple gradient from the mill power Vs load can be used to optimise feed rate where dP/dL is > or = 0.
Software driven decision support (expert) systems are now starting to become available and should be considered, particularly in developing countries, where less skilled human resources are used.

GRAVITY CIRCUITS FOR IMPROVED RECOVERY.

The requirement for a gravity circuit will be dictated by accumulation of coarse gold in the cyclone underflow. If there is no build up in this stream, to values of 15p/t or higher, a gravity circuit is probably not required. Gravity circuits are inefficient with individual steps seldom operating at better that 70%. At these efficiencies, if three steps are incorporated, overall circuit recovery has dropped to less than 35% recovery of gold in the original concentrate. To avoid these inefficient downstream steps, after the original concentrator, Intense Cyanidation(ICN) after a Knelson or Falcon concentrator or an In-Line Pressure Jig is becoming popular. This technique, when used in conjunction with a cyanide leach accelerator such as LeachWELLÔGC, can avoid these down stream steps and allow a very high liquor tenor gold solution to report directly to the goldroom. An additional advantage here is reduced security requirements, normally associated with gravity circuits, particularly in developing countries.

Optimisation Recommendations

Before incorporating a gravity recovery step monitor the cyclone underflow for coarse gold accumulation to ensure that it is a real, rather than apparent, problem.
Consider the use of accelerated cyanide leaching as an alternative to further high security, capital and labour intensive downstream gravity steps.

LeachWELLÔ for ULTIMATE MILL RECOVERY and GRADE CONTROL

Accurate grade control has been correctly identified, by many, as the single most important aspect of mining and perhaps an entire project operation. Failure to get this one right can result in gross mismanagement of ore handling and poor and inconsistent feed to the processing facility resulting in greatly increased mining costs and poor mill recoveries. Both of these factors can ‘make or break’ a mining operation.

Thus it is sensible to know accurately, at all times, feed grade to the processing facility and how much oxygenated cyanide recoverable gold is available from a given plant under optimum operating conditions. Equally as importantly anything that can clearly indicate whether plant leach conditions are in fact optimum, is useful. LeachWELL^TM provides a fast reliable means of determining exactly how much gold can be extracted and what the lowest possible gold tail should be under normal oxidation and cyanide conditions or more simply how much gold can we get and are we recovering all of it.

Optimisation Recommendations

Consider using cyanide (bottle roll) assays for grade control. A useful machine for this is the PAL 1000 unit used in conjunction with LeachWELL 60M that can deliver 50 samples/2 hours with crush, mill and leach in a single step.
LeachWELL can also be used to to post leach the CIP/CIL tail to determine that absolute amount of gold that is cyanide extractable and this result then used a the target for the processing facility. This leach will only need two hours after LeachWELL addition.

HEAP LEACHING

This has been included as many of the same principles apply. Heap leaches however have one very major difference in that they are totally unforgiving of any mistakes made during construction. Once the heap is stacked and the irrigation system in place there is very little that can be done to optimise if the heap doesn’t work.

Optimisation of a heap or vat leach must be done during testwork, design and construction.

It has been observed over a large number of heap leach operations, where the heap leach project is an addition to an already established milling circuit, that the heap project is often considered as an ‘also ran’. Because it is only the low grade stockpile then as long as it breaks even then there is no harm done if it doesn’t make good profit. This is a shame because good leaching at high recovery is possible by adopting appropriate strategies.

Strategy

The key to successful heap leaching is simply to ensure that all testwork, pad construction, agglomeration, stacking and irrigation is done with absolute and unstinting attention to detail. All aspects, over which we do have control, must be done as close to perfectly as possible. A near enough is good enough is poor practice anytime. For a heap leach it is a disaster. A heap leach will always find its own unanticipated and unplanned problems over which we will have no control. These will matter very little if the things over which we do have control are completed with excellence. If this philosophy is pursued, with passion, from project commencement and the testwork has been done well the heap is assured of success.

Testwork

This must be done on each and every ore type with no omissions. For a typical heap percolation rate of 8-12litres/M2/hr the basic rule for percolation tests using columns up to 1 metre diameter is a flow rate of 100litres/M²/hr. The scale down factor to operating heaps is a full order of magnitude. For prediction of outflow rates on the basis of percolation testwork the column result should be reduced by a factor of ten for sizing the heap channel, pipe work and pump capacities.

If agglomeration is required there are numerous tests for decrepitation and agglomerate stability. More importantly when the heap is under construction a simple field test, on the agglomerated fines, such as dropping lumps of cured agglomerate from 1 metre height onto concrete is a useful test that should be done continuously until the amount of cement/lime is optimum. The agglomerated ball should break apart on impact into not less than 3 and not more than 4 pieces. This procedure should be carried out several times/day for the entire agglomeration campaign. It should be assumed that the ore will be variable and of changed characteristics every day unless it is absolutely known otherwise. This is something over which we do have control and can optimise prior to stacking.

Completely overlooked during heap leach testwork is oxygen consumption on the ore. This needs to be done as an agitated or bottle roll test on a slurried sample in order to predict if the heap is likely to suffer from oxygen starvation as column leach testwork, often used to simulate heap characteristics, will understate the requirements since a 500mm column provides far less resistance to air/O₂ saturation than a 9 metre heap.

The Role of Oxygen

As was the case with CIP/CIL the role and importance of oxygen, in heap leaching, is once again frequently overlooked or understated and occasionally appears to be completely ignored. The difference between leach testwork results and the leaching rates from operating heaps, in W.A., has resulted in a significant number of less than successful heaps and a few disasters.

Worse still the cause of failure of the heap to perform, in many of these cases, is not recognised or understood. What is being suggested here is that the cause of failure, as was the case for poor CIP/CIL leach circuit performance, is oxygen starvation. This is why heaps respond so well to ‘rest periods’. During these periods air(oxygen) diffuses down into the heap and re-initiates the leach reaction.

In the U.S.A., the importance of O₂/air has been widely recognised for a number of years and several cases where air/O₂ has been injected, via steel air spears that are hammered into old heaps, have been very successful in reactivating these heaps for both gold and copper.(Bruynesteyn 1991).

Given the success of this new lease of life to old heaps simply by injecting oxygen it is surprising that hard pipe valved plumbing, on the pad base(instead of drain coil), has not been introduced to pump in O₂/air directly to the base of the heap during rest periods.

Pad Preparation

Again this is something over which we have complete control and there is no excuse for not getting this absolutely correct. Close supervision of the grading and careful surveying and pegging to assist this and produce the right fall is necessary. A +/-10mm tolerance is possible on this exercise, although not very popular with grader operators. Do not accept less than +/-25mm.

If plastic welding is required the plastic must be kept totally dry and moisture free. Not a good wet season job.

If a trafficable layer is being used on top of a plastic membrane the same attention to detail is required for surveying and grading. Trafficable material should be well screened to remove fines and unnecessary traffic minimised once it is down. Drain coil in top of the trafficable and prior to stacking the heap is advisable. Any extra help to drain pregnant liquor is recommended. The drain coil should be kept as straight as possible as the heap is stacked and not allowed be pushed into snaking curves by the heap stac

king. A ‘near enough is good enough’ approach is not recommended for any type of operation. It is a disasterous attitude for heap construction.

Agglomeration

This should be straightforward and is right in the plant metallurgists area of expertise. Continual ‘drop tests’ and a minimum 24 hour curing time must be observed. The longest possible trommel should be secured for service. Longer trommels yield more stable agglomerate balls. Ensure that cyanide is added to the agglomeration water or added separately to the agglomerate. This promotes much faster leaching from the heap in the initial stage with up to 70% of the gold being leached into solution during the agglomeration stage.

Stacking

This is the most important aspect of heap construction and where many heaps are made or lost. If a crushing circuit is being used to precrush the oversize then the undersize/fines from this circuit are often agglomerated. The ratio of agglomerated fines to coarse crushed rock may vary and it is important that the agglomerate is blended as homogenously as possible with the/any coarse material. Truck dumping followed by excavator stacking, where the excavator can spread each bucket over a full turret swing, works well.

Irrigation

Irrigation is the most forgiving aspect of heap construction since it can at least be changed or modified as required. Nevertheless spacings should be exactly as designed and then modified according to ‘throw’ after commissioning. Throw of cyanide liquor should include the outside walls of the heap.

Optimisation Recommendation

It is absolutely essential that every single task that is performed in building a heap is done with unstinting attention to detail – as close to perfect as possible.
Make sure testwork is also very thoroughly done including oxygen uptake testwork.
Percolation rates of 8-12litres/M²/hr are acceptable.
Agglomerate lumps, when cured, should break up into 4 pieces when dropped onto concrete (or other hard surface) from a height of 1 metre.
Closely supervise the stacking & blending. Close supervision is the key.

Conclusions

It is hoped that the above will prove useful to operators of gold treatment plants and heap leaches on oxidised gold ore bodies.

References

Extensive use has been made of the following papers, publications and texts.