• This session explores how real-world scheduling problems often involve nonlinear elements that introduce significant complexity — especially when convexity, typically a desirable trait, becomes a bottleneck.
• Drawing from recent research, Hoa Bui shares how Mixed Integer Nonlinear Programming (MINLP) can be adapted to navigate these challenges by strategically de-convexifying models to enable faster convergence and practical performance.
• The session highlights a mining-based case study where nonlinearity was turned from a hindrance into a modelling advantage — demonstrating how optimisation thinking, algorithm design, and careful decomposition can transform intricate scheduling problems into tractable solutions.

Presented by: Hoa Bui | Research Fellow | Curtin Centre for Optimisation and Decision Science & ARC Training Centre for Transforming Maintenance through Data Science, Curtin University.

• Open pit mining involves some of the most complex optimisation challenges in the resources sector — from long-term production scheduling to balancing ore quality, haulage, and operating costs.

• In this session, Karsten will explore what strategic open pit mining optimisation really involves, why it’s such a challenge, and the practical approaches used to address it.

Presented by: Karsten Lehmann | Principal Optimisation Specialist | Fortescue 

• While optimisation methods like stochastic programming are often used to manage uncertainty in decision-making, their practical application can be restricted by system complexity, problem scale, and the difficulty of clearly defining uncertainty.

• This presentation showcases an alternative strategy developed for a complex scheduling challenge at Rio Tinto. Rather than relying on traditional uncertainty-modelling frameworks, robustness was embedded directly into a mixed-integer programming (MIP) model using custom constraints designed to align with specific robustness objectives.

• This approach empowers end users to actively explore and balance trade-offs between operational performance and robustness, enabling more transparent and adaptable decision-making in dynamic, real-world environments.

Presented by: Irene Pivotto | Principal Advisor Optimisation | Rio Tinto

• Recent research applying Primal-Dual Hybrid Gradient (PDHG) methods to linear programming has shown that first-order methods can be a useful part of the mathematical programming toolbox. These algorithms also lend themselves to GPU-based implementations, raising the prospect of exploiting the current surge in hardware development for optimisation applications.

• This session will focus on the current state of Gurobi’s GPU capabilities and explore future directions for leveraging GPUs in optimisation. Attendees will gain insights into how GPU technology can enhance the performance of Gurobi’s solver and what advancements are on the horizon.

Presented by: Steven Edwards | Manager of Technical Account Managers | Gurobi

Panellists include:​

• Ryan Loxton | Director | Curtin Centre for Optimisation and Decision Science​
• Dr. Christina Burt | Principal Optimisation Specialist | Fortescue​
• Gaurav Singh | Managing Director | Decisions 360