Automation of Load Haul Dump machines : comparative performance analysis and maintenance modeling

Sammanfattning: The Load Haul Dump (LHD) machine and its operating environment create a complex system. Mine productivity depends on the operation of the LHDs and on the mining environment, including fragmentation, size of boulders, navigation techniques etc. Traditional navigation techniques require a lot of infrastructure to accommodate automatic operation. From fully automated fleets of vehicles, the focus of automation has gradually widened to include more flexible solutions, such as semi-automatic LHD machines, with safety as a main goal. The automatic system used for semi-automation is different from that used for fleet automation in that less infrastructure is needed, and the operator controls only one vehicle at a time. A semi-automatic LHD machine can operate in either manual or automatic mode depending on the need and situation. Several issues must be resolved to maximise the benefits of automation. One is to improve the maintenance strategy, especially preventive maintenance, as it is crucial to avoid the loss of time incurred by unplanned breakdowns. Another issue is the complexity of the mining environment; external disturbances such as oversized boulders, road maintenance etc. can throw the entire investment in automation into question.The purpose of this thesis is to explore the maintenance actions connected to automated LHDs as well as the factors influencing the operation of the machine. Research methods include a literature review, unstructured interviews, and data collection and integration. Reliability analysis, fault tree analysis and Markov modeling were performed to comparatively analyse manual and automatic LHDs.This thesis presents an approach to evaluate the performance of manual and semi-automatic LHD machines. It describes the maintenance procedures of automatic LHD machines. It includes a study of the reliability of LHD machines with special attention to automatic operation. It studies the operating environment’s effect on automatically operated LHDs compared to manually operated LHD machines, identifies the external disturbances affecting the automatic operation of LHD machines, and introduces a new way of modeling the maintenance and environmental disturbances to determine the best operation mode for the LHD machine.The analysis shows that the production performance of manual and semi-automatic LHD machines is similar. When it comes to the maintenance performance, hydraulic and electric systems are still the biggest reason for machine downtime but the stops are usually short, which means that LHD machines can start producing relatively soon after failure. However, the automatic LHD machine has more time spent in the workshop for the transmission and engine than the manual LHD machine. The difference in reliability between the machines regarding the engine is not significant. But for the transmission there is a verified difference. One possible reason for the difference in transmission reliability could be engaging/disengaging gears when the machine is in automatic mode.The analysis of the operating environment shows that LHDs suffer from mining related, machine related and/or automatic system related disturbances. Seventy-five percent of the stops causing idle time for LHD machines are related to the operating environment. Better fragmentation of rock to avoid big boulders, better constructed roads to minimise the need for road maintenance etc. are keys to the successful operation of automated LHDs.Fault tree models and reliability block diagrams are effective tools for evaluating the reliability of a system but it can be difficult to include mining related disturbances. Therefore, in this thesis, Markov models are introduced to describe disturbances affecting LHD machines and to identify possible differences between manual and semi-automatic LHDs. A fault tree model can classify and analyse failures but cannot show changes between states; this is something a Markov model can handle. The proposed Markov model built for the application shows that the best mode, from an operational point of view, is semi-automatic operation due to its flexibility handling disturbances of different kinds, especially those that are mining related.