The problem the framework addresses
“Digital twin” is used for systems ranging from a geometric visualisation to an autonomous optimisation platform. In mining, this ambiguity makes projects difficult to compare and can obscure the data, modelling and decision-making capability actually required. The paper develops a mining-specific framework that functions both as a shared vocabulary and as a design blueprint for mechanised, human-in-the-loop operations.
Five connected dimensions
The framework describes a twin through five dimensions:
- the physical entity being represented;
- the digital model;
- the update from the physical entity to the digital representation;
- the update or intervention from the digital side back to the physical system; and
- the optimisation process that turns representation into an improved decision.
This makes the flow of evidence and action explicit. A high-fidelity model without a defined update path or intervention is materially different from a twin that continually changes a production or maintenance decision.
Properties and model choices
Seven properties refine the definition: data source, synchronisation frequency, model fidelity, application domain, model capability, mine-life phase and the way an intervention is actioned. The digital model may be geometric, physics based, data driven or hybrid. Digital-to-physical involvement is described through increasing levels—observe, analyse, decide and act—while optimisation can occur online or offline.
These properties allow an engineering team to define what its twin must do before selecting software or instrumentation. The accompanying selection tools help match model form and fidelity to the use case instead of treating maximum complexity as the default.
Case studies
Two examples demonstrate the framework at different scales. A roadheader case focuses on a specific machine and the models needed to understand its interaction with the cutting environment. A mine-operations and maintenance case uses fleet and maintenance decisions to maximise cash flow, showing how the same framework can describe a larger socio-technical system.
Contribution and boundaries
The paper clarifies why two valid mining twins can look very different while still sharing a consistent architecture. It also centres the actionable intervention: the value of a twin lies not only in mirroring a physical entity, but in supporting a better decision or action.
The framework assumes that enabling infrastructure for sensing, transmission, storage, processing and security can be provided. The treatment of knowledge-based models also remains open to refinement. Future applications, including underground drilling and blasting, can test how well the vocabulary transfers to rapidly changing rock and operating conditions.