Smart mining complexes: large-scale stochastic optimization, high-order simulation and self-learning decision support systems for sustainable development of mineral resources

点击翻译按钮获取中文摘要

Modelling and optimization technologies have evolved to become an essential tool for optimal mine designs and production schedules, aiming to maximize the economic value of mining assets from their initial assessment and development to closure [1-11]. Recently, including our CRDPJ Grant 411270-10, new methodologies and algorithms have been developed for complex mining systems that enable the explicit and joint management of orebody, mining, processing and market uncertainties [11-23], and show substantial increases in both NPV and metal production. It is evident in all past developments, that the following areas are still main open areas for research: (a) computational complexity of large scale mining optimization problems, (b) endogenous and exogenous uncertainties in the key parameters (geological/mining, financial), and (c) absence of 'holistic' optimization of the elements of a mining complex that intelligently adapts to unveiling information. Our work contributes substantial developments to strategic stochastic mine planning optimization methods, specifically, by integrating and simultaneously optimizing critical elements of mining complexes under uncertainty using our newly developed, computationally efficient simulation and optimization approaches. Through these contributions, we have changed the mine planning paradigm in several ways, particularly by shifting the focus of the optimizers from the assessment of economic values of extracted materials to the economic values of the products generated and sold to customers, while accounting for major sources of uncertainty (geological/supply). This contribution opens new possibilities to develop an advanced, all-inclusive framework to deal with several critical issues that could not be previously addressed.