August 11th, 2014, Published in Articles: PositionIT
by Monica Tetteh and Fred Cawood, University of the Witwatersrand
The theory of mine call factor (MCF) compares the sum of metal produced in recovery plus residue to the metal called for by the mines evaluation methods expressed as a percentage. The concept is well known in underground mining scenarios and therefore this article highlights some MCF issues from a surface mine perspective using AngloGold Ashanti Iduapriem mine as a case study.
The mine call factor (MCF) at AngloGold Ashanti Iduapriem Mine from 2006 to 2012 averaged 100% . Despite the perceived efficiency, the difference of “gold called for” and “gold produced” was 5890 ounces at an average price of US$1098 per ounce. The loss in monetary values was equivalent to US$6 469 432. The revenue obtained over this period was US$1230-billion resulting in a revenue loss of 0,7%. Since 2010 there has been a declining trend to a minimum of 97% in 2012. Fig. 1 shows the positive relationship between recovery % and time, while there is a drop in MCF. This suggests a potential problem with the grade of the ore mined.
This MCF investigation established the relationship between actual measurements and reporting against measurement protocols. Such measurements include tonnage measurements based on surveyors calculated tonnage and a system of truck count tonnage (referred as spot tonnage), as well as sampling and assay standards practiced on the mine. This article highlights how these measurements are conducted on the mine as part of a contribution towards best practice.
The theory of MCF was defined by Storrar  as a “ratio expressed as a percentage, which the specific product accounted for in recovery plus residue bears to the corresponding product called for by the mine’s measuring method”. This means that if a mine’s evaluation method is perfect, and sampling, assaying, and tonnage measurements in the mine and plant are flawless and there are no losses anywhere, then the MCF should be equal to 100%. A MCF can therefore be viewed as a measure of the efficiency of all the processes in the mine value chain.
A MCF investigation begins by establishing the relationship between actual measurements and reporting against measurement protocols. In this article the measurements considered are tonnage determination based on survey volume and a truck count tonnage. The assay analysis is based on the mine’s quality control (QC) procedures and analysis.
Significance of MCF
A MCF of 100% theoretically means the measurement systems in the mine and the plant are perfect. It can therefore be inferred in this background that a discrepancy is a result of a gain or a loss of metal produced. Cawood  investigated underground face sampling on narrow reefs and asserted that the MCF is not a reliable indicator when used to identify losses because it covers a wide range of technical disciplines, starting from the work face to the final product accounted for by the plant. This is further emphasised by Storrar [2 p. 265] who stressed that “an abnormally high MCF which persists over a long period is unsatisfactory if the cause is not determined”. He further emphasised irregularities of this nature might undermine the confidence of sampling procedures and therefore sampling methods must be reviewed. It is against this view that Springett , found that the MCF should not just serve the purpose of simply balancing books, but it should identify real problems, which when addressed, can increase profitability.
The potential causes of an abnormally poor MCF ranges from gold loss due to blasting and thefts by syndicates who sometimes operate on several mines. It was therefore declared by De Jager  that the MCF is sometimes faced with myths with regards to the possible causes of gold unaccounted for. The issues vary from mine to mine due to the atypical nature of mining activities and it is therefore appropriate to follow a systematic approach to distinguish between truths and misconceptions.
A MCF investigation begins by analysing the various components which contribute to it. This can be seen on the ore flow diagram from either an underground or surface mine to the smelter or processing plant. Storrar  at the time suggested nine variables for typical underground gold mines namely:
Generally, the reason for an abnormal MCF is usually classified into errors in estimating the quantity and quality of mineral expected from the resource. Cawood  outlined a seven step approach for doing a MCF investigation which is illustrated in Fig. 2. It is worth mentioning that a MCF investigation is likely to be successful when initiated and supported by management. This philosophy is indicated by a triangle turned upside down throughout all the stages in Fig. 2. The triangle widens at the first step and narrows at the bottom because the scale of work reduces as one progresses through the steps.
The first step is to “review existing protocols and establish compliance”. The initial step is crucial and time – consuming because it entails a complete review of all existing protocols (evaluation, drilling and blasting, mining, sweeping and cleaning cycles, grade control, assay protocols etc.) and comparing with best practices. This is because there is no reason in doing any investigation of the MCF if there is non-compliance of mine protocols in the first place.
Step 2 is a literature survey which involves gathering information on previous investigations on the mine or other related mines. The idea here is not to re-invent the wheel but to study solutions which may have been applied previously to similar problems at similar operations.
Step 3 is a thorough understanding of the ore flow process, which begins from the point where the ore is broken at the face to the point where the metal is produced. This process is revealed in a metal content sheet which reflects the estimate of metal content at various points of measurements along the ore flow.
Step 4 requires identifying and quantifying risks. The use of flow sheets is instrumental in identifying various risk parameters. Although a standard flow sheet is illustrated in Storrar , a flow sheet is mine specific. The magnitude of each risk parameter is determined using sensitivity analysis. For instance in South Africa, a study on sampling narrow reefs has revealed that a 1% decrease in sampling value, would result in a 1% increase in MCF .
The 5th step is to rank the risk variables according to their magnitude of risks based on sensitivity analysis in step 4. High and low risk variables will be tested and therefore interruptions with current protocols may be carefully planned, executed and scrutinised, after which results are scientifically analysed.
The 6th step is to distinguish between long- and short term topics. When the tests introduced in the 5th step give justifiable reasons to question existing protocols, then there is a case for further research. It is advisable for such research to be conducted through independent institutions or employees from the mine can do research for a higher degree as part of their career development plan.
The final step is adjusting existing protocols as a consequence of research outcomes. This is a management decision which should be carried out only when outcomes of research results have resulted in a steady and reliable improvement of protocols.
The next section discusses the methods of tonnage measurement at AngloGold Ashanti Iduapriem Limited (AAIL).
Survey and spot tonnage
The survey tonnage is determined based on the surveyor’s calculated volume and the spot tonnage is compiled from a pit tally sheet. A truck tally load system of booking is used to record the movement of materials from the pit to various destinations. Fig. 3 shows the diagram of material movements at Iduapriem Mine. Ore mined from two pits (Ajopa and Block 8) are sampled using the reverse circulation drilling method. Ajoparompad stockpile contains ore from Ajopa pit denoted as Fingers 1, 2 and low grade. Some of this ore is re-handled into a stockpile at rompad close to the crusher while the rest is directly dumped into the crusher. The ore from block 8 is also stockpiled at rompad close to the crusher or directly dumped into the crusher.
A spotter counts the number of trucks for each mining block at the end of every shift and records the total number of trips for each truck after that shift period. The shift geologist ensures all necessary information (such aspit name, bench, block no grade) are correctly entered on the tally sheet before entering into a database management system.
The spot tonnage is computed based on the truck/bucket factor which is usually given by the manufacturer of the truck. The spot tonnage and the surveyor’s tonnage are generated by the mine on monthly basis. The two measurements are not always equal and the difference can be attributed to human inefficiencies in entering correct figures and operator errors. Fig. 4 shows the average of survey and spot tonnage. It can be seen that spot tonnages are consistently higher than the survey tonnage.
Since the survey tonnage is a controlled measurement in the sense that it makes use of specialised instruments with higher accuracies, they are more reliable to use for the purpose of ore and metal accounting in cases where there are high discrepancies.
Correlation between survey and spot tonnage
The survey and spot tonnage show a positive linear relationship in Fig. 5. However, there are some points which have totally deviated from the linear graph which is an indication of over or underestimation. The correlation coefficient (r) measures the intensity of the linear relationship between the two variables (spot tonnage and survey tonnage). Since it (r) is very close to one (1), it means a very strong linear relationship exists between survey and spot tonnages.
From the correlation and margin of error indicated in Fig. 5, it can be said that spot tallies are satisfactory indicators and therefore reasonable checks. However, it is appropriate to question the accuracy of percentage margins that can be considered accurate for the purpose of ore and metal accounting. The average percentage of survey over spot tonnes over this period is 95%. Using statistics, it is established that at a 95% confidence interval the percentage ratio between survey over spot ratio should lie between 92% and 98% (indicated by the dashed lines) in Fig. 5. Therefore any percentage ratio which is outside this range raises questions on whether these measurements should be used for ore and metal accounting. The circled point in Fig. 5 has a survey over spot percentage of 83% which is clearly outside this confidence interval.
The next section discusses the quality of grade estimations based on the mine’s quality control methodsand analysis.
Quality control methods and analysis
Sampling is done by reverse circulation drilling and chip sampling after mining the first 2 to 3 m from the natural surface. The planning of locations for grade control holes is based on geological information such as the ore shapes of the previous lift, ore reserve block outline and the current pit profile. Before drilling begins, the sampler ensures that the splitter, cyclone and the drilling rig (Fig. 6) are clean to prevent contamination of samples.The quality control practice involves the insertion of certified reference materials in a batch of grade control samples that are submitted to the laboratory
For a batch of assay results to be accepted, the standards should have an expected value of ±2 standard deviation (SD) and blanks should be ≤ 0,02 g/t (Shewhart control graph). According to the mine’s quality control protocol, if a batch of samples fails, then it is the responsibility of the laboratory to reanalyse the entire batch at their own cost. Pulps of that batch are retrieved and resubmitted to the same laboratory with different sample names. For the purpose of this article, the performance of the laboratory and quality of sampling is assessed as per standard, blank and duplicate plots.
In Fig. 7, the analysis of a low grade standard with expected value (0,848 g/t), recorded an assay value of 0,88 g/t > +3SD. There is no bias evident in the standard analysis of this batch and therefore results can be accepted. The calculated bias over the period is 0.
|Category||Area of investigation|
|Resource/reserve||How is ore body geometry determined
Study grade estimation algorithm
Truck tonnage factor determination
In situ and soil densities
Procedur of down hole survey
|Survey||Drill hole coordinates
Pit measurements (advance and loss)
Process of markout
Sample identification procedure
Ore geometry interpretation
Selection of cut-off criterion
Quality control methods and analysis
Excessive back break
Excessive flying rocks
Blasting of large boulders
|Material movement||Digging procedures
Temporary stockpile measurements
|Clean up||Grading ore zones
Dozing ramps on ore faces
Running trucks across ore zones
|ROM pad||Ore left on rom pad
Spillages from trucks
Measurements of stockpiles
The blank analysis in Fig. 8 shows the majority of the blank results below the detection limit (0,02 g.t). This means there is no contamination of laboratory equipment from previous sample preparations and therefore the batch of assay results can be accepted.
Variable components of MCF at Iduapriem Mine
The reasons for a MCF being less than perfect can be classified into errors in estimating the quantity and quality of mineral expected from various sources (in the mine and processing plant). In addition to Storrar’s nine variable components discussed earlier, a list of categories outlined in Table 1 is recommended for investigating areas of potential metal losses for surface mines. The table focusses on measurements and processes starting from the broken face to the rompad. Measuring systems within the process plant is not covered in this article. Readers are encouraged to refer to AMIRA P754 code developed for mass measurements, sampling and analysis within the metallurgical plant only.
Conclusion and recommendations
This article assessed the quality of tonnage and grade measurements at Iduapriem mine. The two methods of tonnage measurements (survey and spot tonnage) were compared and it was realised that a strong positive correlation exists between these two measurements. It was further found that grade estimation is generally a potential problem area and can be further investigated; however the quality control analysis in this paper does not show any bias detected
or possible contamination of laboratory equipment.
It is recommended that more reliable methods of ore tracking systems could be investigated. For example, according to JKMRC , the use of passive radio frequency identification tags (RFID) has proven to be useful in effective ore tracking and they are relatively cheap. This system makes it possible for materials to be tagged with markers at the source representing volume and later detected as it flows through the system, or not detected if sent to waste dumps and later detected if sent to a long term stockpile and eventually processed (Fig. 9). Information concerning a particular parcel can be written to the markers and therefore quality and quantity can be monitored at specified measuring points as it flows throughout the value chain. The ore being processed at a particular time can be linked with its geographic coordinate, hence a mill to mine reconciliation can be done.
 Anglogold Ashanti: Iduapriem MR BME BFE FEB 2013, Confidential Mine records, [Accessed on 14 March 2013].
 CD Storrar: South African Mine Valuation, 2nd edition, published by Chamber of Mines of South Africa, Johannesburg, pp. 226, 265,266, 1981.
 FT Cawood: Underground face sampling on narrow gold reefs, Journal of IMSSA, Vol. XXXI, no 7, p. 202,203, 2003.
 M Springett: Solving open pit grade control problems. Innovative Mine Design for the 21st Century, Bawden & Archibald (eds), USA, pp. 225-226, 1993.
 EJ De Jager: The Analysis of the Mine Call Factor In the Gold Mining, with specific reference to Western Holdings Mine. Thesis: PhD (Eng), University of the Witwatersrand, Johannesburg, 1997.
 JKMRC: An introduction to Metal Balancing and Reconciliation, published by Julius Kruttschnitt Mineral Research Centre, University of Queensland, Australia, 2008.
Contact Monica Tetteh, University of the Witwatersrand, firstname.lastname@example.org