Rope shovels are the most common large-capacity excavator used by the Australian coal industry. There are some 50 rope shovels operating in Australian mines and 300 machines worldwide. Maintenance accounts for 30 to 50% of machinery operating costs, and every 1% improvement in availability or productivity has the potential to increase profits by up to 3% [Moore 1998].
This thesis argues that shovel productivity can be increased and maintenance costs reduced by establishing a best operator practice. This work divides naturally into two parts.
The first part identifies the elements of such a 'best practice' by analysing data collected over a ten-day period from a P&H5700 shovel operating in the Hunter Valley, NSW. The purpose of these trials was to identify factors that affect shovel performance. Inter alia they revealed:
(i) There are significant differences in operator style.
(ii) Operator style affects productivity and duty.
(iii) Greater productivity may be realized without increasing shovel duty.
(iv) Several conditions which have a significant impact on machine duty.
The conditions at point (iv) include 'swing during dig', which arises when the operator commands the swing system while the dipper is locked in the overburden, and 'hoist stall', caused by the operator commanding excessive crowd force while digging.
The second part of this thesis develops a framework for monitoring rope shovel performance in real-time using basic electrical signals in the system drives (such as armature currents and armature voltages). This framework exploits natural clustering tendencies in these signals by employing discriminant functions to perform hypothesis testing. It is demonstrated that it is possible to:
(i) Track shovel activity in real time, thereby providing a capability for basic shovel monitoring including cycle time analysis.
(ii) Detect events that result in significant duty loading on the machinery, so providing a basic tool for performance and duty monitoring.
(iii) Predict these events with sufficient warning time to allow corrective feedback to be provided to the operator.
The swing-during-dig and hoist-stall conditions identified in Part I are used as exemplars to illustrate this approach. It is argued that this framework extends naturally to other duty causing events, and as such, could provide the basis for a commercial shovel performance monitoring system.
To demonstrate the method, and provide infrastructure for further systematic investigation of rope shovel performance, a 1/7th scale rope shovel was designed and constructed. An overview of the design and build is included for completeness.