The allocation of gold production from multiple shafts feeding a common treatment plant using run-of-mine sampling of ore deliveries
HE Bartlett, L Korff, RCA Minnitt

Previously, the grade of ore at the shaft head was taken to be equivalent to the grade measured in the faces by means of chip sampling. The tonnage mined from the stopes is determined from survey volumetric measurements. Estimates of shaft head grade and tons include the grades and tons of waste, development ore, sweepings from old areas, and other sources, not all of which are sampled. The tonnage of ore delivered at the shaft is often estimated using skip factors. These factors are subject to manipulation and not all the skips are filled to the same level. Consequently, both the grade and tonnage of ore delivered at shaft head are subject to a measure of uncertainty. For many years, these uncertainties have led to disputes between shaft managers and metallurgists over the reconciliation between shaft tons and grade and plant tons and grade. These problems are compounded when several shafts feed a central processing plant. Shaft bonuses and a shaft’s profitability are affected by poor gold allocation methods. Without measurement of the grade and tonnage from all shafts, a poorly performing shaft could benefit at the expense of the other shafts.
In order to improve the measurement of grades and tonnage at the shaft head, which is the custody transfer point between the mine and surface operations, South African mining companies have developed a methodology to measure the mass of ore from each shaft using rail weighing equipment, and the run-of-mine grades delivered from each shaft using cross-stream (go-belt) sampling.
Harmony Gold Mining Company has implemented the rail weighing and go-belt sampling methods for their shafts.
This paper deals with the statistical tests on the procedures and control measures that are in place to establish confidence in the gold allocations. The metal accounting system for examining the allocations follows the first principle of the AMIRA P754 Metal Accounting Code of Practice namely ‘The metal accounting system must be based on accurate measurements of mass and metal content. It must be based on a full check-in check-out system’. The check-in is mass and grade of run-of-mine ore and the check-outs are the plant input grades and the recovery and residue from the plant. The balances between check-in and check-out are the subject of the statistical analyses, which have been conducted using the classical statistical tests and procedures that are generally used in all fields of endeavour.
It is concluded that the gold allocation procedures that rely on go-belt sampling and rail weighing of ore from shafts has been successfully implemented, and that the system, which has been in operation at Harmony since 1993, has won general acceptance from stakeholders, with the gold allocations being perceived as being fair to all. Apart from the on-mine issues of bonuses and fair allocations, accurate metal accounting assists in the broader issues of generating accurate production figures for management. An added benefit of weighing and sampling ore at the shaft head has been that shafts now have daily moving average grades, which are used for grade control. The go-belt sample at the shaft head is the first accurate sample of production in the value chain from face sampling to gold bullion.
Keywords: metal accounting, go-belt sampling, run-of-mine sampling, codes of practice, AMIRA P754, train weighing, statistical analysis, gold split, allocation of gold, metal balance, grade control.