Browsing by Author "Nkambule, Sthembile"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Open pit optimisation using monte carlo simulation: a case study of kolomela mine
    (University of the Witwatersrand, Johannesburg, 2024) Nkambule, Sthembile; Tholana, Tinashe
    Iron ore grade and price are among the input variables in the open pit optimisation process. Kumba Iron Ore (KIO) uses the widely used deterministic approach whereby average input variables are used to determine open pit limits. This approach assumes that the estimated variables are known with certainty. However, the input variables are associated with uncertainty. For example, the grade of a mineral deposit is estimated by interpolating relatively limited data from exploration drilling results. This means that the estimated grade is not entirely a true representation of the entire mineral deposit. This can be a significant source of uncertainty. Also, iron ore prices have proven to be highly variable in the past years which can also be a significant source of uncertainty. If uncertainty of these two variables is not well understood and quantified, this can result in sub optimal or overly estimated pit limits. The effect of not understanding uncertainty can lead to poor decisions that can result in loss of revenues or additional cost and thus can impact the net present value (NPV) and life of a mine (LOM) of a mineral project. KIO, like all mining companies, is increasingly concerned with the effects of risk on NPV because of the chosen final pit shells. The aim of this research was to model the uncertainty of iron ore price and grade associated with open pit optimisation. Isatis software was used to produce ten realisations of the case study orebody and Monte Carlo Simulation was used to model iron ore prices. The economic block values of the ten block model realisations were calculated using four prices which were P75, P50, P25 and P5. The ‘P’ in P75, P50, P25 and P5 refers to probability of exceeding a certain iron ore price point. A P50 value is a median value, which means that it is expected that 50% of the time, the iron ore price will be above the P50 value, and 50% of the time, it will be below the P50 value as simulated from @Risk software. Forty pit shells and high-level schedules were then generated using Deswik Pseudoflow. NPVs of the forty pit shells were determined and were compared to the NPV of the deterministic pit shell. The study showed that there are benefits in doing probabilistic iron ore grade estimation as opposed to deterministic estimation. The highest difference in ore tonnages between the deterministic block model and the simulated block models was 15%. The probabilistic pit shell with the highest ore tonnages had lower strip iv ratios and an additional year of mining. This resulted in the pit shell producing the highest NPV. Results of the study also showed that iron ore price has a huge impact on NPV. The P5 price ($170.89/t), which was the highest price with a low probability, produced bigger pit shells and higher NPVs. The P75 price ($70.84/t) which was the lowest price with higher probability, produced smaller pit shells and lower NPVs when compared to NPV of the deterministic pit shell produced at $78.5/t. The study interpolated NPVs of the probabilistic pit shells and showed that KIO deterministic pit shell was planned at P71. This shows that KIO is very conservative in their mine planning. It is recommended that probabilistic pit optimisation be done for all pits at Kolomela Mine. In addition to iron ore grade and price, it is recommended that uncertainty of all pit optimisation input parameters be modelled. It is also recommended that the methodology demonstrated in this study be used also at Sishen Mine. Finally, it is recommended that KIO should implement this methodology annually before their medium-term planning process to assess uncertainty during pit optimisation.