METALLURGICAL ACCOUNTING

Metallurgical Accounting - Systems & Procedures For Modern Day Mineral Processing Plants

 Author: D.E. G. CONNELLY - Director/Principal Consulting Metallurgist

Mineral Engineering Technical Services Pty Ltd (METS)

PO Box 1728, West Perth WA 6872

Australia

 

 For a copy of the full paper: [email protected]

Abstract

 Metallurgical accounting can be defined as the monitoring of the valuable metals from the time the ore is broken at the primary crusher until the time that saleable products and residues are obtained. The overall purpose of metallurgical accounting is to provide management with up-to-date information that can be used to control mining and metallurgical operations. The decisions that are taken on the basis of the information that the metallurgical accounting system provides should result in improved control and, ultimately, increased revenue. The current state of accounting is a result of an extended evolutionary process and will continue as process plants become more sophisticated. Differences in mine “call factors” and grade control reconciliation can lead to frustration between the site’s geoscientists and its metallurgists leading to a level of distrust in the mill grade.

 Metallurgical accounting is not included in undergraduate courses and the acceptable industry standards have developed based on what is considered acceptable to industry personnel and consultants. There is no handbook or reference manual. Computer based databases and spreadsheets have transformed tedious bookkeeping into an important management tool. Over time these standards have improved with developments in analytical procedures, weighing technology and sampling.

 This paper provides anecdotal examples of problems and improvements observed over a number of years within the gold industry in Western Australia.

 INTRODUCTION

The metallurgical accounting standards are based on generally accepted industry standards and while methodologies vary from base metals, gold, mineral sands, iron ore and alumina industries  there is a trend towards accepting more uniform standards as a result of the use of computers and the dissemination of information via the internet.

An SAIMM mineral processing workshop held in Cape Town (SAIMM, 2001) identified the pressing need for an industry-wide standard in metallurgical accounting.

 The reference from the workshop is quoted below:

“AMIRA International research coordinator, Mr Richard Beck said at the conference the problem for many mining and processing operations was to get a balance at the end of each month, particularly in smelters and hydrometallurgical plants where it isn't easy to sample intermediate products. He said many operations measured the amount of metal in feed and waste streams stating recovered metal as the difference: But the metal going out isn't actually measured compared with the metal coming in - the true recovery is metal out against metal in. Fundamentally, the metallurgist wants to know what the exact answer is and if supplied with the right tools to help trace these things, then there would be major advantages. Mr Beck said AMIRA would aim to set up a collaborative project to draw up a standard set of metallurgical accounting practices for the mining industry.”

 The author believes that for mineral processing plants.

 “YOU CANNOT MANAGE WHAT YOU DO NOT MEASURE”

 Metallurgical Testwork

 Mineral processing projects start life is based on metallurgical testwork conducted during a Feasibility Study. This sets the benchmark.

 It is not uncommon for testwork samples to be unrepresentative of the actual plant feed. Testwork results are regularly misinterpreted. Operating efficiencies have a negative impact on the accounting and poor plant design can limit testwork results being achieved. However when testwork results are not achieved, the usual first assumption of management is that there is something wrong with the metallurgical accounting. “Where did all the gold go” because it has been measured in the grade control samples? The original, usually expensive, testwork is often regarded by management as a sacred cow, so they base their judgement on figures that may not have been sampled representatively.

 Other common mistakes include:

• Using leach recoveries while not allowing for solution and slag losses
• Misinterpreting oxygen demand from a batch test (3 litres)
• Misinterpreting scale up effects, such as where sulphides and pulp density impact the leaching rate
• Performing testwork in high shear agitators when the plant has low shear mixers
• Not realising that batch grinding in a laboratory is not the same as a dynamic grinding circuit (although this can be a positive effect)

 

Usually there is a learning curve, particularly for base metals flotation plants. Quite often the metallurgical accounting is blamed, or at least not believed, because of performance results being different to the original testwork.

 It is a very common problem where “mine call factors” do not reconcile with mill reported head grades. Management is invariably suspicious of mill explanations preferring to think that inefficient mill practises have resulted in more gold being lost to tails than that being reported or the characteristics of the ore are different. It is very disappointing for grade control geologists to have their mine grade reduced based on the metallurgical accounts which, invariably, they distrust and do not understand. The basic metallurgical accounting principles are not well understood outside of the metallurgical profession.

 The Metallurgical Balance

 A metal balance in a processing plant equates the material going into the plant with the products leaving the plant. In a gold plant, there are usually two final products, gold bullion and tailings. For a base metals concentrator, it is a concentrate for shipment to a smelter and tailings which are stored on site.

 The basic method used in calculating the metal recovery for a CIP plant follows but it  applies equally well to base metals concentrators, mineral sands plants, acid leach plants etc. The basic flows and analyses are determined by measuring:

• The weight of the ore for a given period (as measured by a weightometer)
• The percentage moisture content of the plant feed
• Representative samples of solids and liquids from the plant feed and tailings
• The pulp density of the feed to the process circuit
• Assay values of the solids and liquids

 The contained metal in the ore should be calculated from the plant feed or, preferably, cyclone overflow, because of the fineness of the solid fraction. Sufficient sample can be collected to obtain a representative sample of the feed. At coarser sizes, the sample size must satisfy statistical requirements (Pitard 1993). At a size of P80 75 microns, two kilograms is an acceptable minimum for ease of drying and preparation.

 Plant Losses

 Plant losses are normally measured in the tailings stream. However there are also other causes for losses that need to be considered and can influence the metallurgical balance in a negative way. Metallurgists need to be diligent in order to achieve a true metal balance.

 Stream losses These can occur as physical losses of gold bearing material leaving the plant. At one plant, a drain was discovered allowing metal in solution to exit the plant, to tailings, at irregular intervals. Pipes need to be walked to ensure the flowsheet is correct and all losses are accounted for.

 Theft This is not as rare as one would expect and at one operation, the poor reconciliation with the metallurgical accounting was sufficient cause to call in the Gold Stealing Squad. The result was the discovery of gold stealing on a grand scale by a number of people and prosecutions followed with the metallurgical accounting used to determine the quantity of gold stolen.

 Preg robbing Carbonaceous or graphitic ore results in preg robbing from solution. Some active clays have been reported as preg robbers and where high copper ores are treated, gold cyanide can precipitate on sulphides.

 Disposal of gold smeared steel from gravity concentrators Steel is removed from the gravity concentrators and discarded. Attached gold is difficult to remove from steel.

 Gold on old smelting pots Old pots contain gold prills and handling these is difficult because the pots are made from clay graphite, which should not be returned to the mill.

 Lock up of gold on mill liners, sumps, gold traps etc. This is common. One operation sold their old liners for scrap, until someone suggested they be cleaned - the result was 10 kg of gold recovered.

 Gold lost in slag If copper or silver levels are high in the gold room slag (matte phase) the gold may not necessarily be recovered in the CIP circuit if returned to the grinding mill. The gold is in solid solution within the matte/slag phase and therefore it is not recovered by cyanide leaching.

 Re-precipitation Gold can re-precipitate from solution if cyanide levels are too low.

 Gold lost on fine carbon Fine carbon can carry gold to tailings. Fine carbon must be removed after stripping. Often the fine carbon screen blocks and these carbon fines are returned to the CIP tanks and bypass the inter-tank screens to tailings resulting in gold losses. Fine carbon is normally collected and sold to a smelter for gold recovery.

 Mill upsets and coarse grinds Floods and plant disruptions can occur, which may not be picked up with infrequent sampling.

 Gold settling in tanks, gold room sumps or locked up in roasters This can be difficult to detect unless one goes looking for it or there is a policy of regular clean up. Similar problems occur in hydrometallurgical plants.

“Dry” Feed Rate Determination

 All the methods for mass measurement determine the total mass of “wet” ore into the plant. These measurements are usually made by a weightometer on the feed belt and correction is required for moisture content to obtain the “dry” mass flow for reporting and calculations. The standard convention is always to use dry tonnes of ore milled.

 All weightometers have a limit of accuracy depending on the make, installation and attention given to calibration. Modern electronic weightometers should achieve an error within plus or minus half a percent if properly maintained.

 Regular calibration of the weightometer is recommended, preferably monthly, using a "chain" along with good maintenance, checking and housekeeping. Checking the zero setting when the feed is off is a simple procedure, which should be carried out regularly.

 Wind deflectors should be used where the wind introduces an artificial side tension on the conveyor belt. Where conveyors do not have gravity take up devices, changes in temperature between day and night will introduce a small error.

 Shovelling off three metres of load on a belt, weighing the contents and measuring the speed is only useful for determining gross errors and on fluctuating feed rate belts is totally useless. Most weightometers can be electronically calibrated and this should be checked on a monthly basis and checked with a chain on a twelve monthly basis. It is an advantage if the conveyor has a belt plough or can be run into a truck and checked with a weighbridge.

 While methods of moisture determination such as microwave measurement, neutron absorption or electrical conductivity are available, none of these are in use on typical gold mines. Microwave technology is used on line at QNI, SLN Nickel and a number of other sites using MOISTSCANTM.

 A more common practise at mine sites is to take samples from the mill feed belt every two hours and to determine the moisture content by calculating the weight lost in drying. For slurry density measurement, usually as percent solids, the most accurate method is, again, wet and dry weights. Indirect measurements, such as Marcy densities, are not considered accurate enough for inventories even though it is acceptable for daily mill control.

 Other Measurements

 SCADA stands for Supervisory Control and Data Acquisition. As the name indicates, it is not a full control system, but rather focuses on the supervisory level. As such, it is a purely software package that is positioned on top of hardware to which it is interfaced, in general via Programmable Logic Controllers (PLCs), or other commercial hardware modules.

SCADA systems are used not only in industrial processes: e.g. steel making, power generation (conventional and nuclear) and distribution, chemistry, but also in some experimental facilities such as nuclear fusion. The size of such plants range from a few 1000 to several 10 thousands input/output (I/O) channels.

 SCADA type control systems are now more common for mineral processing plants. Mineral processing plant control has traditionally relied on the skill of the plant metallurgists and operators. Frequently, vital data such as head, concentrate and tailings grades have been unavailable until the following shift or even the following day. As a result, plant control has been highly variable from plant to plant. It is the complexity of mineral processes, however, that makes the effective manual control of plants so difficult, and therefore, some means of automatic control must be introduced if optimum plant performance is to be approached.

 In the 1960s and 1970s, the introduction of appropriate instrumentation (including computers) provided operators with vastly increased information on plant operation.  This in turn has led to the development and installation of automatic control systems, stimulating further demand and improvements in the instrumentation available.

 In general terms, automatic control in a concentrator is aimed at one or more of the following:

• Increased throughput
• Improved recovery of the valuable minerals
• Improved concentrate grade (resulting in cost savings in subsequent processing)
• Reduced operating costs (e.g., by savings in reagents, more productive use of mill personnel)

 The following instruments have allowed real time data collection:

• Density gauges
• Weightometers
• Tank level gauges
• Mass flow meters
• Feed, concentrate and tailings grade analysers

?  Flowmeters

 SCADA systems allow direct input into the metallurgical accounting and shift reporting.

 Sampling

 Credit for developing an advanced theory of sampling technology and its application is given to Pierre Gy (Pitard 1993).  Proper application of accepted principles of sampling takes into account methods to correctly employ sampling equipment and to determine the quantity of sample to collect.  Sampling is a necessary adjunct to metallurgical accounting to assure correctly evaluated results. Great responsibility rests on a very small sample; therefore it is essential that samples are truly representative of the bulk of the material. Proper collection of a representative sample requires an understanding of the physical characteristics and mineralogy of the material to be sampled as well as determination of the minimum number and size of increments to be taken in order to produce the overall sampling precision required.

 An overview of common industry practises reveals very few operations have, and maintain, automatic samplers. Manual sampling is commonplace and often a cause for concern, unless the importance of the sampling is explained and reinforced with operators and management. The reality is such that, within twelve months, even new plants let the samplers fall into a state of disrepair.

 Gold Plant Case Study

 An excellent sampler system was installed at Marvel Loch in Western Australia where ore was to be toll milled, on a blended basis, for the Cornishman Joint Venture. A sophisticated sampling system was installed to determine the ore feed grade prior to mixing with other ores. The sampling station was properly engineered and commissioned. Sizings carried out on samples collected indicated that there was no sampling bias in the system. This system consisted of a PLC controlled, primary cross cut sampler into a Wescone crusher and then a rotary Vezin sampler to split out two samples into duplicate canisters.

  Sample Preparation

The sample collected from the canisters was further processed to produce samples for assay and composite leach tests.  The sample preparation flowsheet is shown in Figure 1.

 Recovery Determination

A composite was prepared and two leach tests conducted on both the Cornishman ore and the combined mill feed. Recovery was then assigned to the Cornishman ore in the blend based on the results shown in Figure 2.

 The recovery was factored to:

 

R   =   R(test) * Rplant(actual)

Rplant(test)

 This adjustment takes into account a plant recovery, which could be higher or lower than the laboratory test recovery. Tests were carried out at a fixed residence time, fixed grind size and fixed reagent conditions.

 The procedure provides an excellent example of good sampling systems and management. In the past, operators were loath to mix ores for fear of not being able to account for each ore.

 Assaying

 Accurate analyses of routine plant samples and end of month inventory stocks are two of the most important aspects of metallurgical accounting. In terms of cost, the aqua regia digestion method is cheaper than fire assay but, generally, the results are not as accurate because a well performed fire assay determination provides total gold. There are many round robin survey results which confirm the above findings. Aqua regia digestion is better suited to oxide ores because it is faster turnaround and lower cost than the classic method, fire assaying.

 The presence of sulphides or micro fine gold may force a change of procedures rather than a complete changeover to the fire assay method. It is common for grade control assays to be carried out by PAL (pulverise and cyanide leach) or aqua regia while plant samples are processed by aqua regia or fire assay methods, with the aqua regia results cross referenced with the fire assay results.

 It is well accepted that where fine gold is present, the assay handling procedures do not need to be as rigorous as those used where coarse gold is present. Likewise, a change in the ore mineralogy from oxide to sulphide may necessitate a change in procedures.

 The PAL technique is particularly beneficial where the “nugget” effect is a serious problem.

 PAL advantages:

• High gold recovery (95-97%)
• Simple recovery testing (Fire Analysis "tails")
• Suitable for oxide or fresh material
• Large sample charge size (up to 1 kg)
• Ability to enlarge machine for 3 or 5 kg assay charges
• Excellent repeatability of assay results, as shown in Figure 3
• Excellent grind  (95% passing 75μ)
• Good accuracy compared to standards
• Time (Standard <21 hours turnaround, minimum 3 hours)
• Partial batches possible (run pots with water and grinding medium)
• Good mechanical reliability (robust)

 Inventory Measurement

 The gold inventory, that is the gold present in the plant at any instant in time, will vary due to factors such as feed grade, equipment availability, feed rate etc. Inventory sampling is extremely important because metallurgical accounting is based on the fundamental premise of measuring real, physical quantities and the differences in them is over the reporting period which is usually monthly. The inventory sampling must be as accurate as possible and include all sources of gold (or other valuable metal) within the plant. The following outlines the more detailed facets of the month end gold-in-circuit determination (commonly referred to as the GIC).

 Solids

The solids are recovered from the leach tanks by using a 1 mm screen so as not to include any carbon in the pulp sample. The percent solids should be determined by wet and dry weighings and one litre of the pulp weighed to check for specific gravity variations.

 The solids should be washed with raw water to ensure the removal of gold present in solution. Insufficient washing is a common cause of errors. The solids should be assayed in duplicate as a minimum, especially where the gold assay is variable.

 Solutions

Samples of the solutions are obtained by filtering a sample of slurry from each tank, as above, and storing it in a sealed plastic bottle. The solution can be assayed singularly and does not present any special assaying problems.

 Some plants check the specific gravity of the solution because of the high salt content, but this is not considered necessary unless the process water is hyper saline.

Common sources of accounting error does not include solution and solids where feed grades vary significantly and measuring only the gold on carbon.

Carbon

The carbon determination should be based on the concentration of carbon in the tanks.  Poor mixing characteristics in adsorption tanks may lead to swaging of the carbon (variable concentration) and this makes it nearly impossible to determine the carbon concentration. It is well known that at densities above 50% solids by weight, in conventional ore, the carbon can float giving erroneous concentrations and conversely at densities below 40% solids, the carbon may sink to the bottom of tanks.

 A typical sampling regime for carbon content and gold content determination involves collecting at least 100 grams of carbon from each tank, which is then washed and dried so that the grams per litre carbon concentration can be determined. Any grit in the carbon should be removed. Then, the carbon is ashed and assayed in duplicate. Some laboratories pulverise the carbon but this is not considered necessary for good results.

 Taking five litres where the concentration is 5.0 grams/litre can give highly erroneous results and in such a case, 20 litres is more acceptable.

 It is very common for grit to be present in CIP circuit because of problems with the trash screen and tramp oversize emanating from the cyclones. This grit must be separated out of the carbon before assaying so that the true concentration of carbon and the correct assay is assigned to the carbon.

 The carbon assays are carried out after pulverising the carbon and ashing five grams as a single assay. Single assays on carbon are not acceptable and as a minimum standard as the carbons should be done in duplicate to check for reproducibility.

 It is extremely important that the carbon is ashed for sufficient time (at least six hours) and with a small ingress of air to complete the ashing. If this is not carried out then the values reported will be lower than actual.

 The flowsheet shown in Figure 4 describes a typical process.

  FROM EACH TANK SAMPLE

 The gold on carbon is simply the concentration of carbon in each tank multiplied by the average of the carbon assays for each tank.

 It is very important not to forget the carbon in the column, stripped carbon in the pressure vessel or any miscellaneous carbon in the gold room.

 At month end, it is too difficult to measure the gold on the cathodes, thus the standard procedure is to strip the cathodes and convert all the gold to bullion. Even if only one strip has gone onto the cathodes, this must be done. A clean cut off is essential and this gold is usually called “gold shipped” even if it is actually shipped after month end. A typical GIC calculation is shown in Table 1.

 If gold is in the safe at month end then it can be included in the inventory under the gold room section. Some people prefer to call all gold as “gold shipped” and this is fine provided that the logic is consistent from month to month. One should be wary of double counting particularly with gold shipped in the early part of one month which has already been counted as gold shipped last month from gold on wool.

 Thickeners

Where thickeners are involved, inventory measurement is often difficult and the following procedures are suggested. The depth of interface is measured using an air lance and a sample of the solution is taken for assay. The tonnes of solids and solution are calculated and content calculated based on solids and solution assays.

It is also imperative that the procedures are documented and the laboratory staff receive sufficient training in carrying out inventories.

 Gravity Concentrates

The most common form of gravity concentration is a Knelson concentrator treating a bleed stream from the mill discharge or cyclone underflow. The Knelson concentrator is usually located inside a locked cage and is cleaned out on a daily or less frequent basis, as required. For clean out, the locks are removed and the unit hosed out into a plastic bin with the concentrate transferred to the gold room. More modern units have auto dump facilities.

 The gravity concentrates are usually intensively leached using a strong cyanide and caustic solution with an accelerant (IL). The IL pregnant liquor is assayed and the volume measured. Previously the concentrate was tabled and cleaned up by removing iron, magnetite and other foreign matter. The table concentrate was weighed, dried and weighed again until sufficient concentrate was available for a smelt.

 Based on these results, a weekly gravity factor is calculated and this allows a daily head grade to be calculated, based on gold recovered by gravity and the leach feed assay. This system is, in many ways deficient, but is the only means available to arrive at a daily mill feed grade in such a situation.

 Gold Room Records

The gold poured should be drilled or, where the bar assay fineness is less than 80%, vacuum sampled. This is because, on cooling, separate phases in the lower grade metal precipitate out of solution leading to a non-uniform bar that is difficult to sample accurately by drilling.

 The gold shipped must be clearly understood as the gold shipped between inventories to be theoretically and stringently correct from month to month. It is a good practise to include gold on wool at month end as gold shipped after smelting and if necessary uses provisional assays until such time as final figures are available. This procedure must be consistently adhered to each month for the figures to be accurate.

 The drilling returns from the assay office should be stored in the gold room safe and added to the next smelt after three months have expired.

 A mint and site reconciliation must be carried out as these assays determine the mine revenue and small errors have very large effects.

 After a monthly end clean up, the slag should be assayed and returned to the mill feed conveyor. It is also important to note if gold prills are present in the slag. The consistency of the slag should be as a brittle glass; otherwise flux mix needs to be changed. If gold prills are present, this can result in significant gold assays. Gold that has been smelted from cathodes containing copper results in very high gold-in-slag losses. The usual practise is to crush the slag, grind it in an amalgam drum and table to remove the gold prills. At one location, where copper was a problem in the feed, a significant gold loss was noted in the slag.

 A permanent hardcover book should be maintained in the gold room in a safe place.

It is a good practice to record all persons entering the gold room and make them sign the book as a permanent record.

 If the accounting records point to a discrepancy in gold accounted for, the matter should be discussed with the mill superintendent and certainly no one else. Be mindful of the fact that one needs to be extremely careful and use discretion in these matters. If necessary, the mill superintendent will make the decision to inform the Gold Squad. Each operation should have written procedures for gold room security.

 Gold room records, particularly conversion figures, are very important to provide historical records and highlight theft or gold room irregularities.

Flotation Concentrates

 Most oxide gold deposits become sulphidic at depth and in some cases CIP treatment is not suitable. Therefore, flotation must be used to recover the gold. In some cases a simple pyrite/gold concentrate or a copper/gold concentrate is produced while in other situations, two separate products may be produced such as a pyrite/gold and an arsenopyrite/gold concentrate.

 The production of concentrates requires a concentrate reconciliation at month end and resolution by way of physical stock take every three months. The same basic principles apply to copper, lead, zinc, nickel and tin and processors.

 The three product formulae give equations that are normally used, particularly based on sulphur and gold assays. This depends on the accurate weight of the ore milled during the period and reasonably accurate assays of the feed, concentrate and tailings. In the case of pyrite gold, the concentrate may be assayed for sulphur and gold in order to average the weight of concentrate derived.

 F, C, T are the tonnages of feed, concentrate and tailings.

f, c, t are the respective assays.

R is the metal recovery.

 

R = 100* c(f-t)/f(c-t)                                                                                                   (1)

 

C = F * (f-t)/(c-t)                                                                                                        (2)

 The fact that errors in assaying of this magnitude are usually compensating, makes the net error for a shift or a day quite small. The gold content of any process stream is the tonnes times the assay.

 Relative Accuracy

 It helps to place the relative accuracy of measurements in some order.

 

                                                                           Acceptable Relative Error %

Weightometers                                                                       0.5

Stockpile surveys                                                                   5

Moisture measurement                                                          0.5

Inventory reproducibility                                                       5-10

Carbon determination inventory                                            5

Assays             Au                                                                   10

Cu, Pb, Zn                                                       3-5

Iron ores                                                          1-2

Mineral sands                                                 2-5

Calculated vs. Assay Head (not coarse gold)                                    3

Flow meters                                                                            3

Density gauges                                                                       1

Weighbridges                                                                         0.1

Calculated Versus Assay Head

 The mill calculated grade is the official mill grade because it is based on physical gold poured and the result of physical inventories. The calculated assay in grams per tonne should be +/- 0.1  Au g/t to conform to acceptable industry standards, based on a 3 Au g/t mill feed grade. There are some situations where this level of accuracy is never achieved, principally due to a coarse gold problem - the “nugget” effect. This accuracy should be the aim until it is established that for a particular ore it is not achievable. Some mills, where the gold is fine, consistently achieve much better than this.

 Mine vs. Mill

 It is very easy to prove using Pierre Gy theory (Pitard, 1993) that the mine grade can never be as accurate as the mill calculated grade. The sampling of stockpiles and trucks is ‘sheer nonsense when one looks at the amount of sample required to achieve a high confidence limit with the resultant assay’. The 95% confidence level stipulates a sample weight of 250 tonnes for -12.5 mm material but only several kilograms for the same material at a P80 of 75 microns size.

 The calculated mill grade, as determined by inventory change, tailings loss and gold bullion poured, is the only true grade that can be reported with confidence, provided all losses are accounted for and the procedures and assays are validated. If this agrees with the assay grade as sampled, then it confirms the accounting records balance.

 Mine vs. mill discrepancies can be a very serious matter where underground and open pit operations feed to a central mill and ore grades must be reconciled to each mine. This is particularly true when underground mines require higher head grades to remain viable.

 Modern Methods Used

 Manual Systems

In previous years, accounting was carried out manually. However, the advantages of computerised systems are so great that they are the only systems that will be utilised within any modern company. Such advantages include:

• A common database for the company group, with no duplication of data
• Simultaneous availability of data, e.g. at regional and head offices
• Labour productivity is higher with a computerised system
• The computer system can cope with changes more easily than a manual system
• The computer system is quicker and more accurate
• Evaluation and statistical correlation of the data is possible
• Graphs and reports can be produced that are beyond the scope of a manual system

 Excel Spreadsheets

Spreadsheets have been the workhorse for the industry replacing folio and volume of books of accounts. They consist of a number of worksheets and templates that are easily constructed by site based personnel. Some metallurgists are loath to move to databases because they like to see the values in the cells and use cross-referencing and linking cells exclusively. Graphs and reporting is also fairly simple.

 Access Databases

A significant number of mines have changed over to databases because they are very powerful; allow query, search, reporting and graphics while requiring less maintenance than, for e.g. Excel spreadsheets. The downside is that construction of a database requires a higher skill level, testing and development. Some metallurgists distrust databases because they find it difficult to validate the software code. The development of site based systems currently cost between AUD$15,000 and AUD$25,000, depending on complexity.

 JK Tech Met Account

This package (now on sold) was specifically developed for metallurgical accounting and includes reconciliation and mass balancing functions. It costs approximately AUD$70,000 and requires training of personnel to operate effectively. It has been taken up by larger operations where accurate accounting figures are mandatory.

 The package has the following attributes:

• The JKMRC developed mass balancing, model fitting and simulation algorithms
• A new mass-balancing algorithm was released in 2000. The distributors claim that it:
  • Harnesses the power and flexibility of the model fitting and simulation algorithms
  • Uses statistical methodology, redundant data and simultaneous balance for improved confidence in the reconciled data
  • Is ideal for corporate reconciliation and production accounting
• The package produces balanced results across a circuit, site or company
  • It is NOT just data collection, calculations and reporting
  • It balances circuits other programs cannot handle
• It is logical that numbers that agree with each other from one end of the process to the other are credible
• Credible numbers can point to operational problems and opportunities
• Credible numbers encourage decisions to be made

 The benefits of JK Met Account are as follows:

• Reconcile tonnes and quality through the value chain
• Single source for all data
• Decrease audit costs
• Improve corporate governance
• Consistent protocols corporate wide
  • Transparent and credible
  • Auditable

 ABB Knowledge Managers Metallurgical Account

ABB's IndustrialIT solution for metallurgical accounting offers a true and integrated solution for day-to-day and monthly production accounting. For twenty years, they have accumulated a diversified experience and acquired a deep knowledge of processes and data analysis techniques. The methodology they propose transforms data, acquired in real time, into information that is coherent, reliable and concise for production monitoring and accounting for all levels of plant personnel, up to the top management.

 Metallurgical Accountant?, their most recent product, is claimed to be a complete solution for metallurgical accounting. The package includes a configurable database, a database server, data reconciliation software and reporting system as well as on-site installation and training and a maintenance service.

 Metallurgical Accountant? transforms a tremendous amount of process data into a resource of valuable information distributed in concise reports to anyone included in the decision making process. It can:

• Transform process data into a resource of valuable information
• Improve the reliability of the plant performance indices
• Speed-up the generation of coherent production reports
• Facilitate the decision making process and maximize the benefits

 This is a complete product that collects, organises and reconciles production data, then generates and distributes concise reports to acknowledged personnel.

 Metallurgical Accounting and KPI’s

 The purpose of key performance indicators (KPI’s) is to provide a unified reporting tool for all performance measurements. These indicators include production, reagent usage, tails grade, concentrate grade, recovery, safety, health and environment, personnel, operating costs, continuous improvement initiatives and opportunity losses. The new metallurgical accounting systems have databases with web page and Excel interfaces. The database is structured around daily inputs of all measurements and calculations. The data is stored and adjustments and reconciliations made to this data as required. The focus is on accounting rigour, with all adjustments recorded in the database to allow for full auditing. These KPI’s are reported on in the mine monthly report.

 The KPI systems were developed using web-based technologies and are available to any internet-enabled device with Internet Explorer installed. This removes the need for software roll outs and user training.

 Toll Milling

 Toll milling has been used for some time now in the Kalgoorlie, Leonora, Southern Cross and Murchison regions. This regime allows small prospectors to exploit deposits using another party’s mill, for an agreed treatment rate. This toll milling is usually covered by an agreement covering the method of payment, accounting methods to be used and legal considerations.

 There are benefits to the treating company, such as additional revenue, good relations with local prospectors, incremental tonnage etc, and the area has been a minefield of legal battles with some operators who did not do their homework prior to treatment.

 To reduce the main area of disagreement it should be agreed that the leach feed grade should be used for comparison with the calculated head. Samples taken from the mill feed belt should never be used for payment purposes. All samples should be retained for a period of at least six months after the campaign for dispute resolution.

 Preparation and planning, as well as open and honest communication with the other party, are the keys to avoiding the pitfalls so commonly made with toll milling. It is not in the long-term interest of one party to take advantage of the other. Common sense and goodwill needs to prevail when trying to reconcile any problems and this is best done between the technical persons from each side coming to an understanding and recommending acceptance to both parties. The resolution of differences by legal means should only be considered as a means of last resort. It is best to make the agreement work or recommend changes should that be necessary.

 Laboratory testwork be carried on representative samples of ore to determine the suitability of the plant to process the ore before treatment commences. The key areas to look at are the feed grade, gravity gold extraction rate, grind size and treatment rate, viscosity problems, leach kinetics, reagent consumption and adsorption characteristics. If coarse gold is present, then it needs to be agreed that either cyanidation in the mill will be used or bottle roll assays will be used on the leach feed samples.

 A properly signed legal agreement is required prior to the start of ore processing. It is essential that a strong metallurgical input is included in any agreement and all aspects are covered.

 Small tonnages of less than ten days treatment are better bought outright by the treating company, as the errors in inventory determination can be large when handling small tonnages. In fact, the inventory errors due to sampling and assaying could be larger than the campaign gold production.

 Troubleshooting

 Problems with metallurgical accounting are far more common than normally understood and accepted, within the industry. The main problems causing these errors, which are difficult to minimise, are as follows:

• Large sample weights of unground ore, while statistically required, are virtually impractical to achieve.
• Greatly differing specific gravities of ore and gangue lead to serious segregation in the ground ore.
• The commonly referred to "nugget" occurrence of gold in ore introduces assay errors. Coarse gold introduces a whole range of problems into accounting for gold, particularly sampling and assaying. The pulverise and leach (PAL) method has been used where standard methods fail to give reproducible assays. The PAL method involves pulverising 500-2,000 grams of ore and carrying out a cyanide leach bottle roll on the sample in order to calculate a head assay. This technique is very successful on these ores because the leaching "irons out the spottiness" in the sample caused by the coarse gold.
• Errors in weightometer calibration or drift, in addition to errors in moisture measurement, are not uncommon. All weightometers have a limit of accuracy depending on the make, installation and attention given to calibration. Modern electronic weightometers should achieve an error within plus or minus half a percent, if properly maintained.
• Improperly executed inventories introduce errors. This is a common area for sources of errors creeping into the accounting system and as a pre-requisite, regular duplicate inventories should be carried out and an assessment made of the error variance as a baseline case for a particular plant.

 If there is a problem with the metallurgical accounting, rather than hiding the problem or ”adjusting” the figures, one needs to seek outside help from a specialist metallurgical group after checking all the likely causes.

 In the case of small amounts, it is unlikely the metallurgical accounting could prove that gold was missing, however where larger amounts are involved, the accounting should pick up the discrepancy.

 In the case of suspected gold theft, the matter must be kept confidential with the mine manager and the figures scrupulously checked for errors and other sources of gold loss.

 Anecdotal Examples

 The following examples are given to highlight the problems that can occur in different situations. In all but a few cases, no names are used for obvious reasons and any similarities with known companies are, probably, purely coincidental.

 Spotty Mill Feeds

Repeat assays of the same samples of Riverina underground ore showed the following variability:

8.6; 19.6; 3.7; 84.4; 10.6; 9.8; 46.9; 3.4; 18.5 Au g/t

 In order to get some sensible grade on the samples, 500 gram samples were taken, pulverised and bottle rolled with cyanide for twenty four hours. The solids and solutions were assayed giving reasonable reproducibility of assays on the sample.

This was necessary for toll milling purposes to get some idea of the grade on a -12.5 mm ore feed to the plant.

 Use of Leach Tank 1

At a mine site, the calculated head and period cyclone feed assays were starting to show a negative bias. The plant metallurgist started taking leach tank 1 samples and comparing the result. At the period end, this showed a much better agreement with the calculated head and the 0.4 g/tonne difference was eliminated.

 In fact the cyclone overflow assays tended to be far more variable and spikey than the leach tank samples.

 Salted Mill Feeds

One company found a large gold shortfall at the end of their own period campaign.

 During the period, they had toll treated a parcel of ore on behalf of a local prospector, who had taken the feed samples from the mill feed belt on a thirty minute basis and sent the samples to an independent laboratory.

 After re-assaying the inventory samples and bullion, the only conclusion that could be drawn, but not proven, was that the prospector had salted the feed samples. As a result, the company had its own personnel sample all future parcels of ore and used the cyclone overflow as a cross check.

 Short Campaigns

Several companies have found that short campaigns can create problems in achieving a balance mainly because the errors in inventory can be greater than the gold produced during the campaign. This means that difficulties are encountered in reconciling the total gold produced and the individual entitlements.

 Coarse Gold

In this situation, the presence of coarse gold was recognised. However, when the agreement between calculated and assay head was a problem, the toll milling company stopped reporting the assay grade but kept reporting the calculated grade which they paid on (some of the differences were up to 100%).

 The plant metallurgist recognised that there was a problem and the gold was being locked up in the mill. When he did eventually add cyanide to the grinding circuit, excess gold was recovered for the next two periods.

 The company supplying the ore was not aware of the problem and had no understanding of the metallurgical accounting but assumed everything was fair and reasonable.

 Two Stage Sampling

At one operation, where toll milling was undertaken on a campaign basis, the cyclone overflow sample was used for payment purposes but did not agree with the calculated head. As a result, the agreement had to be changed and a two stage automatic sampler was installed on the cyclone overflow which resulted in good agreement thereafter.

 Contaminated Acid

At one mine site a shortfall in gold at the end of the period was traced back to an assaying problem. This was further pinned down to contaminated acid, containing approximately 2 g/tonne of gold, being used in the aqua regia.

 Missing Nickel Concentrate

One mill superintendent, not happy with the tails assays being reported by the laboratory, decided to use figures that made his recovery look much better. This was not detected until many months down the track when, though the concentrate stockpiles were empty, the marketing people sold the reported concentrate that had supposedly been produced. The mill superintendent was asked to explain and was subsequently fired on the spot by the mine manager.

 Missing Gold

At this particular mine site, two ores were being milled and both parties were missing a considerable portion of gold.

 The ore contained lead and copper, and although all of the gold was never fully accounted for, the following areas did yield about half the missing gold; the mill liners, the gold room slag, some drums of gravity concentrate that had been cyanide leached and gold settled out in the leach tanks.

 Gold in Eluate Tank

One company after treating an ore high in copper was short some 100 ounces of gold after a treatment campaign. The gold room operator reported that the wool in the cells was breaking up due to the copper and suggested the sludge in the electrowinning tanks and the eluate tank should be smelted. This resulted in some 200 ounces of gold being recovered.

 Calculated Head Changed to Mine Grade

At one operation the calculated assay was significantly below the predicted mine grade and the manager directed that the calculated grade for the period be adjusted to the mine grade. This was done by reducing the mill tonnage for the period, even though no problem existed with the weightometer.

 This was done to placate head office management, who were not happy with the low grades at the time. The mine manager argued that the effort and detail that went into the grade control gave him more confidence that this was the correct figure rather than the mill grades.

 Gold in Thickeners

KMA sampled different zones in their concentrate thickener to calculate the total inventory in the thickener. This pyrite concentrate assayed approximately 60 g/tonne.

At Harbour Lights, when toll treating oxide ore, clients were not happy about putting ore through the thickener as they  feared a lock up of gold. As a result, the thickener was bypassed and cyanide was added to the mill for oxide ore campaigns.

 Gold Lock Up

Gabanintha always high pressure cleaned the old mill liners after a liner change and typically recovered 10 kg of gold from the clean up. Similarly clean up in the gold room of material in the gold trap recovered a significant amount of gold.

 (Phillips, 1988) reported a discrepancy between feed belt and cyclone overflow sampling of some 280 kg of gold which was treated with some scepticism. The mill was 4.27 m by 10 m and had grid liners which yielded a total of 261 kg of gold when the mill liners were changed out and the concentrators scraped.

 White Range

The geological mine resource was stated at 4.7 g Au/tonne while the actual mill feed grade was 2.8 gAu/tonne. This was caused by down-hole drills smearing off coarse gold during the initial drilling. Once this was recognised, the ore resource was drastically cut.

 Within the mill, the metallurgical balance was poor until a gravity circuit was installed. Thereafter, the balance was not a problem.

 KMA Mill Feeds

For years the gold call at KMA resulted in more gold being produced than was predicted from the metallurgical accounting.

 A long-standing project was initiated with fine milling and splitting of samples each being assayed in duplicate. At the end of the project, no suitable procedure was developed which would give a reliable feed grade.

 Aqua Regia Reports Low Assays

At this operation, the mill assays were carried out by aqua regia while the grade control assays were carried out by fire assay. The transition from oxide to sulphide ore resulted in a larger discrepancy between the calculated and assay head reconciliation.

 Gravity Gold

At one operation, where toll milling was being undertaken, a consistent shortfall in gold production was established as gold lock up in the mill. The cyclone underflow, when assayed, gave very high gold assays and as no gravity circuit was installed, it was decided that cyanide be added to the mill.

 This saw most of the shortfall recovered and a much-improved balance thereafter. The client was particularly relieved because they were confident of the mine grade and were somewhat relieved when the situation improved.

 Toll Milling Dispute

At one toll milling operation, the calculated head and daily assay matched, but the grade was approximately half the anticipated grade.

 The mining company concerned would not accept this and employed a consultant metallurgist. They even resorted to taking their own samples during campaigns.

 Eventually the mining company was forced to admit the problem was in the mine. As a result, they sacked the mine manager and sold off the leases to another party.

 Carbon Assay Issues

At one site, there was a lot of coarse grit and laterite in the carbon which was being dried and reported as grams per litre of carbon.

 However, the assay laboratory personnel were cleaning up the carbon and reporting assays that were used to determine the total gold content. This resulted in a larger inventory of carbon-based gold than reality.

 Bullion Assays Incorrect

In this instance the bars were high in silver and copper and several check assays were found to vary for the same bar by up to 30%. This was explained by phase segregation when cooling, resulting in the assay being subject to the position the bar was drilled.

 As a result vacuum sampling was introduced and one sample was sent with the bar to the refiner, another sample to a local laboratory and a third kept in the safe for cross checking.

 Grade Problems

This operator had several, small open pits, which were to be toll milled, had stored the ore in, essentially, three stockpiles. Namely the high, medium and low grade with the first two to be toll milled and the third to be heap leached.

 Prior to the toll milling campaign, he sampled and had assays done on the high grade stockpiles which came back with very scattered assays. The net result was a ‘break even situation’ from a company that could ill afford the experience.

 Roaster Lock Up

The KMA roasters in Kalgoorlie and North Kalgoorlie both recovered gold from the bubble caps in the roaster during shutdowns. This gold accumulation was reportedly caused by chlorides in the hyper saline water and in the case of KMA amounted to some 9000 ounces for a typical six monthly shutdown. The present Gidgee roaster uses scheme water to wash the concentrate, on belt filters, prior to roasting.

 No Written Procedures

At one operation, the metallurgist had been doing the job for some five years when he resigned and went to another company. The new metallurgist found that he could not achieve agreement between the calculated head and the assay head, whereas previously there had been no problems.

One problem was that there were no written procedures and the original metallurgist had done the sampling himself, with very little documentation.

 After three months of poor agreement, an investigation was called for and this revealed that the past records were incomplete and using the current procedures, repeatable carbon quantities could not be achieved due to mixing characteristics in the tank.

 There is a possibility that the previous metallurgist adjusted the carbon quantities and since nothing was documented, new procedures had to be implemented to obtain reasonable agreement.

 Assay Differences

At one operation, the assay head had agreed with the calculated head but over time the differences started to become significant. The difference was finally resolved after a detailed investigation shows that it is being due to the following:

• Weightometer reading low
• Assay head low due to sulphides in the feed and the aqua regia method
• Errors in the carbon assays
• Errors in carbon measurement

 Fixed Bias

At one operation, a consistent 6% negative bias could not be resolved despite detailed investigation. It was eventually put down to the “nugget” effect and a skewed gold distribution.

Conclusions

Metallurgical accounting is the basis for sound decision making and together with KPI’s form the basis of management reporting in mineral processing plants. Overall, there have been significant improvements within the mineral processing industry.

 The software available for metallurgical accounting is now very powerful and better choices than simple spreadsheets are available. The development of specialist metallurgical accounting packages has provided an effective tool for practitioners to implement systems which transform the data into information that provides the basis for decision making.

 Instrumentation accuracy has improved greatly and SCADA type control systems are more common, allowing better data capture, reporting input into the metallurgical accounts and trending.

 Assaying with PAL procedures has helped enormously while systems and procedures associated with quality control continue to improve.

 The sampling systems have improved, especially for tolling treatment plants, but generally it is poor due to maintenance issues and a low priority placed on installation and support.

 There is still no recognised textbook or standard on metallurgical accounting however conformity of standards across companies has improved greatly with the free exchange of methodology via email and the internet.

 Whilst there are suitable textbooks for sampling and assaying, there is no such “bible” for metallurgical accounting that students and metallurgists can use as a guide. Operators will say they understand when clearly they do not and the terminology gets wrongly used even in company accounts.

 Conflicts between the accounting results are common and can be reduced by:

• Ensuring that the original testwork is performed on representative samples and critically reviewed before plant design is undertaken
• The metallurgical accounting procedures should be clearly documented and subjected to a peer review before being adopted
• An appreciation of relative errors is important in assessing whether a balance can be achieved, Short run balances can be meaningless
• Metal losses should be clearly identified, Regular control checks should be undertaken on mill heads and tailings

 The basis of the methodology has been historical or “home grown”. The introduction of databases and statistical packages has revolutionised the reporting function.

 Acknowledgements

 The author would like to thank all colleagues, metallurgists at various sites, METS staff and other consultants for their contribution and the management of METS for their permission to publish this paper and constructive criticism of various drafts.                                       

  APPENDICES

 DEFINITIONS

Checks must be carried out, on a regular basis, to compare the gold produced, both predicted and actual. If these balances do not occur, then the matter needs to be investigated further for possible causes.

 Calculated Head

The following should be true for any month.

 Calculated head assay  =  Assay head over a month.

 Calc head = (Change in inventory) + Gold poured + Tails loss

                                                 Dry tonnes milled

 Gold Produced

The gold produced is based on the gold poured plus the change in inventory. Many people confuse the terms but a clear distinction must always be made between the two terms, e.g. gold produced is not equal to gold poured.

 Gold produced                       =   Fine gold poured as bullion + Inventory change(close-open)

 Bullion + inventory change    =   Tonnes x assay head x recovery

 Gold Poured

The gold poured should reconcile with the gold on loaded carbon minus the barren carbon. This is sometimes difficult to achieve without measuring the actual carbon density.

 Strip summary predicted    =   Actual gold poured

 Balance

At any point in time and over a month the gold into the circuit minus the gold shipped should be equal to the physical GIC.

 Gold in - gold shipped   =  Gold-in-circuit (GIC)

 Calculated Recovery

The calculated recovery should be within an absolute 3% of the assay recovery, for good agreement.

 Calculated recovery        =  Assay recovery

 Calc recovery =           (Change in inventory) + gold poured            * 100

                            (Change in inventory) + gold poured + tails loss

 The assay recovery is based on the daily assay results.

 Calculated vs. Assay Head

The calculated grade is the official mill grade because it is based on physical gold poured and the result of physical inventories. The calculated assay in grams per tonne should be within plus or minus 0.1 g/tonne to conform to acceptable industry standards.

 There are some situations where this is never achieved, principally due to a coarse gold problem. This, however, should be the aim until it is established that for a particular ore that this is not possible. Some mills, usually where the gold is fine, consistently achieve much better than this.

 Mine vs. Mill

It is very easy to prove that the mine grade can never be as accurate as the mill calculated grade.  The calculated mill grade, as determined by inventory change, tailings loss and gold bullion poured, is the only true grade. If this agrees with the assay grade as sampled, then it confirms the accounting records balance.

 Internal Unit Checks

While the plant metallurgist has a responsibility to be diligent in the execution of the metallurgical accounting, the figures are not to be determined by internal or external influences. This means that all possible sources of error are addressed when there is a discrepancy between the calculated head grade and any other grade that may be in dispute.