A System Dynamics Approach to Mine Modelling and Cut-Off Grade Management K Sinding 1 and E R Larsen2 ABSTRACT We discuss the concept of optimal cut-off grade policies and their advantages in tenns of maximisation of net present value of mining operations. There are indications, however, that such optimal policies are not widely used in the mining industry. In order to explore why this is the case we identify a number of stakeholders who may have diverging interests in the operating policies of a given mine. These various interests are mapped and a number of important relationships are explored in a large qualitative systems dyanrnics model. INTRODUCTION The cut-off grade is defined as the grade, or relative content of valuable material, which is used to distinguish between ore and waste in a specific mine or orebody (Dagdelen, 1992). The cut-off value is highly site specific'and is defined by the unique characteristics of a particular mining operation. The modem concept of how an optimal cut-off grade policy can be calculated was formulated by Kenneth Lane in 1964 and later extended 'and generalised to a range of different mining situations (Lane, 1964; Lane, 1988). This approach is based on the idea that calculation of the cut-off grade must take into account the fixed (opportunity) cost of not receiving future cash flows more rapidly as a result of the cut-off grade decision made now (Dagdelen, 1992). The theory is well established and has not been challenged. However, there are signs that the methods advocated by Lane are not widely used. Indeed, the introduction to Lane's 1988 book is not sparing in its criticism of arguments used in defence of cut-off determination not based on optimal cut-off determination taking the opportunity cost element into account. The other major problem influencing determination of cut-off grade is proper accounting for costs in the various stages of the mining process. This comes back to the very site specifi'c nature of many cost elements. For example, there will be significant differences in the cost of ore located close to or far from the shaft in an underground mine. If these cost differences are accounted for there will be a diff~rence in cut-off grade between the two locations. However, if the costs are averaged out in the accounting process, then ore close to the shaft will be subsidising ore of the same grade far away from the shaft. Indications from numerous mine visits, and comments from colleagues in the mining industry, indicate that little or no weight is given to the opportunity cost element in determining cut-off grade. On the other hand, the need for exact and detailed cost accounting is more widely appreciated, perhaps because the ,concept is easier to understand than opportunity cost. Nevertheless, from the point of view taken by Lane, the use of sub-optimal cut-off grades is unsatisfactory and inefficient, especially from a shareholder point of view. It also begs a number of questions about what factors really determine observed cut-off grade policies. ' In this paper we employ a management science technique whereby mental models of reality (eg a mining operation) are transformed first into maps (causal loop diagrams) which show how various factors interact, and second into quantitative computer simulation models. The work we report here is of an exploratory nature in that we concentrate on formulating a qualitative model of mining operations. We begin by outlining the principles involved in determining optimal cut-off grade policies based on precise accounting of cost and inclusion of the opportunity cost element related to cut-off decisions made now. The apparent lack of enthusiasm for this approach indicate that other interests and stakeholders may be able to influence the cut-off grade decision. To analyse this possibility we first introduce the concept of causal loop diagrams as a tool for mapping the interrelationships of the possible stakeholders in a given mine. The stakeholders are then m!lpped and the possible avenues of influence on the cut-off grade decision are identified. In conclusion we discuss what areas are most fruitful for further quantitative modelling. OPT~ALCU~OFFGRADES The seminal work on cut-off grade selection by Lane (1964, 1988) shows that any determination of an optimal cut-off grade must take into account the fact that there is an opportunity cost associated with not receiving cash-flows more rapidly in the future as a result of cut-off policies adopted now. While emphasising this as the key requirement, Lane also stressed the importance of limiting ·factors in a mining system, for example the capacity of a mill to treat ore, or the capacity of a customer (ie a smelter specified in a long tei.1ll contract) to receive concentrate. Using the simplified case presented by Dagdelen (1992), where the milling capacity is the limiting factor, the optimal cut-off grade in a year i between 1 and the end of mine life N is calculated as: gMil1 = 2. Research Associate, The Learning Center, London Business School, Sussex Place, Regents Park, London NW1, UK. APCOM XXV 1995 Conference dNPV i (P-s)y where c is milling cost; fa is annual fixed cost; C is annual mill capacity; d 1. Research Associate, Department of Earth Resources Engineering, Imperial College of Science, Technology and Medicine, Royal School of Mines, Prince Consort Road, London SW? 2BP, UK. fa c+C+-C- i NPV is the discount rate applied; i is the net present value of future cash flows from i to the end of mining; P is the mineral sales price; s is the marketing cost; and y is the fraction of minerals recovered. Brisbane. 9 - .14 July 1995 241 K SINDING and ER LARSEN With the exception of NPyi, all of these variables are exogenously given in the sense that they change only in response to external events or if action is taken to change them. If none of the factors are altered, th~ cut-off grade depends only on the gradual decline in NPY' as mining progresses. From the expression for optimal cut-off grade it is then clear that cut-off will decline as the end of the mine life approaches. In practice, the determination of the optimal cut-off grade depends on the develop1TIent of an iterative model with which the initially unknown NPY' is estimated. An algorithm which solves this problem, and generates an optimal cut-off grade policy has been developed by Dagdelen (1992). Using this algorithm on a hypothetical mine, and comparing the resulting cut-off grade to a traditionally determined cut-off grade policy, Dagdelen obtained radically different results. Most importantly, the optimal cut-off grade policy generated a much higher net present value (a 90 per cent improvement) while undiscounted cash flows were somewhat lower (by 35 per cent) and mine life cut to less than a third (ten years instead of 36). This simplified example did not allow for stockpiling of low grade ore. Had this been possible, an even larger improvement in NPY might have been possible. This brief outline of how optimal cut-off grade poilcies are determined omits many of the refinements described by Lane (1988), and is based on the simplified case where processing capacity is the only limiting factor. There are no published indications of how widely the prescriptions for setting optimal cut-off grade policies are used in the mining industry. A few papers have mentioned the method in passing (Real and Torres Lopes, 1994; Taylor, 1985), but seem to indicate that other considerations are equally or more important than optimisation when the cut-off grade is determined. On the one hand this is consistent with our observations at minesites and with the descriptions from colleagues in the mining industry. On the other hand, this use of non-optimal cut-off grades is, to some extent, predicted by Lane (1988), who sees this as the result of special interest groups motivated by other considerations than maximisation of overall NPY. The example cited by Lane in support of this contention is that staff at a given mine will be interested in setting a lower than optimal cut-off grade so as to extend their period of employment. However, while this may be a valid argument for managers in mines close to the end of their operating lives, it is not necessarily true that managers are in a position to do so when reserves can support a long mine life. For example, if management is replaced by natural turnover every five years and the remaining mine life is twenty years, extending mine life is unlikely to be a primary concern, until late in a mines life. This means that we must look for other causes to explain non-optimal cut-off grade policies. To analyse the possibilities we have encountered we must first digress in order to introduce the concept of causal loop diagrams and their use. increased supply of the metal. The linear cause and effect diagram is shown in Figure 1. But when supply expands, a situation of excess supply may develop, pushing down prices. By taking this effect into account, beginning and end of the linear cause and effect chain is connected to form a closed loop, as shown in Figure 2. The other type of loop is reinforcing, indicating that change leads to more change. As an example, consider the standard prescription for how a mine should adjust its cut-off grade in response to a price increase, ie by lowering the cut-off grade. If each mine follows the same rule, all will still put the same quantity of ore through the mill, but less metal concentrate will be produced, both locally and overall. This restricts supply, and prices are pushed upwards even further. A causal loop with this reinforcing effect is shown in Figure 3. Price '~E & D Investment'~Supply, FIG 1 - Linear cause and effect diagram. Supply FIG 2 - Balancing casual loop diagram for a metals market. 242 Cutoff grade ( Metal Price ~ CAUSAL LOOP MODELLING Causal loop diagrams are a part of what management scientists refer to as 'systems thinking'. This is a way of seeing structures in complex systems. The approach emphasises interrelationships rather than linear cause-and-effect chains, and processes of change rather than snapshots (Senge, 1990). This involves mapping feedback processes, where causal loops are used to represent 'cause' and 'effect' relationships. There are two distinct types of feedback loops: balancing and reinforcing. The following examples serve to illustrate how they work. Consider first the relationship between metal price, exploration and development, and metal supply. A price increase indicates a situation of excess demand. This is an incentive for mining companies to increase investment in exploration and mining capacity development. In turn, and after a long lag, this leads to Exploration and development investment Metal Price FIG Supply \ Concentrate output ~ 3 - A reinforcing feed back loop. MAPPING THE BENCHMARK MODEL The principle of including the opportunity cost element developed by Lane (1988), and the iteration algorithm suggested by Dagdelen (1992) may be described as the benchmark against which alternative solutions to the mine optimisation problem must be measured. The optimal cut-off determination and the iteration involved can, in its simplest form, be mapped as shown Brisbane, 9 - 14 July 1995 APCOM XXV 1995 Conference SYSTEM DYNAMICS APPROACH TO MINE MODELLING Total reserves ~ ~ \ ~ Interestrate ~. ( ~ Oremined~ "" Cuwff(ade ~ . Mme caP'""ty ~ Mill capacity \ costs) ( c)oncentrate NPVofyears remaininig ~ Total NPV Mine Cash Flow . Metal Pnce Fro 4 - Causal loop diagram for the benchmark optimal cut-off grade model. in Figure 4. The main loop in this model represents the iteration needed for determining the optimal sequence of cut-off grades over the life of the mine. The total NPV is derived from the NPV of years remaining as the NPV from now to the end of mining. The small loop connecting ore mined and total reserves reflects that total reserves contributes to ore mined, while ore mined reduces total reserves. This model reflects the mechanics of the process which determines an optimal cut-off grade policy. ne only action needed is to set the process in motion. The model does not, for example, relate cash flow, or remaining NPV, to costs, although an attempt to reduce costs is one of the first and most obvious responses to lower metal prices, or higher supply costs. Lenders Loc~ue"'l own",\ / Parent company management~ 1-----__ Cash Flow RealizedNP A COMPLEX MODEL OF CUT·OFF GRADE DETERMINATION Non-optimal cut-off grades may be employed for a considerable number of interrelated reasons, and as a result of the influences of a range of stakeholders in individual mining operations. The identity of the principal stakeholders is shown in Figure 5, while the interrelationships between these stakeholders is mapped in Figure 6 and also discussed below. The parent company or major shareholder are concerned not only with subsidiaries, but also their own financial performance. While a cut-off grade designed to optimise total NPV at the mine may be in the best long-term interest of the parent company, realities such as parent debt service, performance of parent company shares on the stock market or internal financing needs may indicate a different cut-off grade policy (typically an even higher cut-off grade). Lenders are not concerned with optimal cut-offs, but with debt service. As a result, they will be in favour of stable cash flows over the repayment period. However, many factors influellce the debt required, and the debt is not just incurred at the time when a mine is first developed. The mineral owner, most commonly the state in the form of a regional authority, invariably lays claim to part of the mineral rent derived from the operation by raising tax revenues. The rents are captured using various tax instmments (Gamaut, 1983), most of which are based on either profits, calculated net rent or revenue. APCOM XXV 1995 Conference Neighbours Customers RegIonal authority Fro 5 - Stakeholders in the cut-off grade setting. For these, government will receive the largest discounted revenues if an optimal cut-off grade policy is pursued at the mine. However, some tax instmments are based on the physical mineral output or on fixed annual payments. For these a cut-off grade policy which maximises mine life is preferable. This is not the only problem facing the regional authority. Even if it is both optimal and rational to favour an optimal cut-off grade policy, it may serve to exacerbate economic cycles in the local economy. An optimal cut-off grade policy tends to concentrate cash flows in a much shorter period and tax revenues are correspondingly concentrated. This gives large but short-lived revenue streams to the regional authority, and, since the authority is likely to be politically unable to resist pressures to spend the Brisbane. l;l- 14 July 1995 243 K SINDING and E R LARSEN Concentrate customers Local benefits 1 Environmental pro} Environmental problems Tax revenue Investment cost FIG 6 - Complex causal loop diagram offactors influencing cut-off grade policy. revenue, roughly corresponding bursts in government spending. This creates two problems in the regional economy. First, boom and bust cycles associated with the concentrated exploitation of mines under optimal cut-off grade policies are exacerbated by high levels of g~vernment spending. Second, once government expenditures have increased they are much more difficult to reduce when incomes decline. Local management of the mining operation is the real target of the comments by Lane (1988) noted above. However, adoption of sub-optimal cut-off grades is only a likely scenario when such managers have a rational expectation of extending their employment beyond the closure date indicated by an optimal policy. If the turnover rate for local management is such that management is replaced over a period shorter than the remaining mine life then only the last manager has a real incentive to extend mine life. A better understanding of what motivates local management can be gained by understanding what their pay-off from various courses of action will be. As managers they can influence most or all types of decisions in the operation (and for that reason management is not represented in the 'causal loop diagram) in order to maximise their own pay-off in terms of financial remuneration, corporate advancement opportunities, and external career possibilities. Employees do not figure in the causal loop diagram but they derive benefits from production in the form of local benefits. More than management, employees have an interest in extending mine life. The principal way' of doing this is by reducing the cut-off grade. But the most likely avenue of influence is not management, but rather the regional authority (which must in that case balance its preference for maximum discounted tax revenues with the pressure to retain jobs and with them indirect personal income tax revenues). 244 A further complication is caused by the relationship between employees as workers and employees as neighbours of an activity creating environmental externality effects. Thus, environmental benefits may be moderated by the level of benefits the local residents derive from the mining activity (directly due to employment in production, as well as indirectly through transfers from the regional authority). The customers of a mine are the smelters and refiners which carry out the downstream processing of their output. Relations between mines and their principal customers are frequently governed by long-term contracts (Gentry, 1984), even where the smelting operation is a downstream part of the same company or group of companies. When external events result in a change in price, the optimal response may be to increase cut-off grade, and produce more concentrate. But this extra quantity cannot be sold within existing contracts, leading either to sales at less attractive terms (since smelter payments are not only a function of metal prices) or to sales outside the corporate family. The neighbours as employees or dependents have already been noted as having an interest in the way a mine operates. From an environmental perspective the extent of environmental disturbance clearly depends both on how long a time the mining operations go on, and on how extensive they are. An optimal cut-off grade policy results in a shorter and more concentrated mining operation, followed by closure. The amount of ore mined is lower and as a result the quantity of waste generated is also smaller. While local environmental concerns favour optimal cut-off grade policies, and local employment favours maximisation of mine life through lower cut-off gades, another environmental consideration favours lower than optimal cut.off grade policies. This is what may be called the materials efficiency perspective. By extracting more material from each mine fewer sites will have to be disturbed to produce the same quantity of metal. Brisbane, 9 - 14 July 1995 APCOM XXV 1995 Conference SYSTEM DYNAMICS APPR0ACH TO MINE MODELLING CONCLUSIONS REFERENCES In this paper we have discussed the concept of optimal cut-off grade policies and indicated their advantages in terms of maximi&ation of net present value of mining operations. Observations have suggested, however, that such optimal policies are not widely used in practice. In order to explain why this is the case we have identified a number of stakeholders who have diverging interests in the operating policies of a given mine. These various interests have been mapped and a number of important relationships have been explored in a large qualitative systems dynamics model. The complex model presented here opens a number of avenues for computer based modelling. The initial challenge is to use the algorithm presented by Dagdelen to create a tool which can be used as a point of departure for extensions to the model. These are not only those which involve stockpiling of intermediate material Jor later processing, or for optimisation with several minerals, but also the extensions which takes behavioural or discretionary decisions into account. Dagdelen, K, 1992. Cutoff grade maximization, in Proceedings 23th APCOM Symposium (Ed: Y C Kim) pp 157-165 (SME Inc). Gamaut, R and Ross, A C, 1983. Taxation of Mineral Rents (Oxford: Clarendon Press). Gentry, D W and O'Neil, T. 1984. Mine Financial Analysis (SME: Littleton, CO). Lane, K F, 1964. Choosing the optimum cuttoff grade, Colorado School ofMines Quarterly, 59:811-824. Lane, K F, 1988. The Economic Definition of Ore (Mining Journal Books: London). Real, F and Torres Lopes, A, 1994. Setting of planning objectives for mine management in sUlphides mines, in Mining Planning and Equipment Selection, June, pp 263-267 (Balkema). Senge, PM, 1990. The Fifth Dicipline. The Art and Practice of the Learning OrfJanization (New York: Doubleday). Taylor, H, 1985. Cutoff grades - s,ome further reflections, Trans Inst Min Metall (Section A: Mining industry) October:A204-A216. APCOM XXV 1995 Conference Brisbane, 9 ·'14July 1995 245 246 Brisbane, 9 - 14 July 1995 APCOM XXV 1995 Conference