Abstract
The current asset management approach for continuous miners is based on a generic-static lifecycle strategy that is universal to all continuous miners in the fleet. It employs reactive maintenance of the asset, which is primarily time based. This approach results in uneconomical maintenance and lifecycle intervention activities, unplanned breakdowns, failures that lead to damage to the asset or expensive components beyond a repairable state, and a reduction in the life of the asset.
It is essential for a transition in the asset management of continuous miners from a reactive universal solution approach, to a predictive approach that is based on the individual asset throughout its operational life. Digital twin technology provides a proven feasible tool utilised by multiple industries to manage complex dynamic systems according to a predictive maintenance approach.
This study answered the research question:
How can digital twin technology be utilised in the asset management of continuous miners, and how should a framework be developed for the creation of such a digital twin?
The research question was answered by applying a pragmatic philosophy and inductive approach to fulfil five research objectives following the design science research approach and utilising mixed research methods.
The first research objective established a definition of a digital twin within the context of this study. The second research objective identified the roles and applications of digital twins in other industries for the asset management of complex dynamic systems. Third, the current asset management practices for continuous miners were documented, and four applications were derived for digital twins in the asset management of continuous miners. As the fifth and final research objective, the derived applications and development framework were validated as viable answers to the research question.
The study derived and provided four applications of digital twins in the asset management of continuous miners:
1. Condition monitoring and forecasting (incl. fault detection and diagnosis) of continuous miners to detect developing deviations/faults and act accordingly to prevent failure of the asset.
2. Implementing a predictive maintenance-based strategy for continuous miners and planning maintenance as well as lifecycle milestones such as interventions and overhauls.
3. Management of the continuous miner’s lifecycle according to a dynamic strategy founded on the digital twin of the continuous miner; and
4. Real-time operational control and management of the continuous miners as well as the larger mining operation based on the information obtained from the digital twins of the continuous miner fleet.
Further, a deep long short-term memory neural network utilising dropout layers was derived as the development framework to produce a continuous miner digital twin. A deep LSTM development framework is well supported by proven applications in the development of digital twins used for asset management of complex dynamic systems in multiple industries.
This study proved that digital twins can fulfil multiple roles within the asset management of continuous miners and provided a development framework through which a digital twin of a continuous miner can be developed for deployment in asset management applications.
Keywords: Continuous miners, digital twins, asset management, digital twin development framework