Open Cut » Overburden Removal
This project had the following aims:
- Monitoring of bulldozer operations and development of bulldozer productivity algorithms.
- Development of computer modelling software to allow the design, optimisation and graphical depiction of bulldozer operations. This software will use three-dimensional digital terrain modelling together with the bulldozer productivity model developed in the first stage.
- Validation of the software and algorithms.
The bulldozer monitoring system developed for this project performed as expected and about 1500 bulldozer cycles were monitored over three different operations. These operations included dragline spoil reshaping (rehabilitation), cross pit stripping of overburden and dragline bench preparation. The software developed to reduce the monitored data and produce cycle-by-cycle information was used to reproduce about 1000 bulldozer cycles.
The cycle-by-cycle data were used to produce 3d-dig design files of the operation and graphs of cycle time vs. distance with points grouped according to average operating gradient. These graphs displayed considerable scatter but showed a distinct speed upper bound and a less distinct lower bound. The variability in speeds appeared to be due to the following factors:
Functional variation between cycles. Some cycles performed the function of spoil transport and some were used to re grade the operating surface.
Non-slope geometric variation between cycles.. The smoothness of the operating surface appeared to produce variation in maximum speed, especially for the reverse cycle component.
Operator variability. . The above two sources of variability appeared to be related to operator technique. Some operators were able to conduct the reshaping operation with a minimum of preparatory type cycles and the majority of their time was spent on productive slot dozing. These operators also tended to produce and maintain a smooth operating surface allowing higher speeds to be achieved.
Cycle payload.. Payload varied considerably, especially in the preparatory cycles and when pushing at low gradients. Payloads appeared to greatly influence maximum speeds.
The cross pit stripping operation had a much lower range in the above factors and hence the data for this operation had much less scatter. The spoil reshaping operation had considerable variability in the above factors and hence had the greatest scatter, data for this operation were grouped according to cycle function and scatter was greatly reduced.
The monitoring system used did not allow precise determination of blade loads for individual cycles, although desirable this was not an intended outcome of this project. Precise blade loads could not be determined due to the very high level of terrain mapping precision required to establish volumes. Blade load determination is also difficult due to the nature of material transport in dozing operations. Productive bulldozer cycles aim to excavate a volume of material at a discrete point, transport it and deposit it at a discrete point. In practice considerable material is transported sideways and deposited in a windrow. It is recommended that the issue of blade load be further investigated. This would required a monitoring and data processing system which has the following capabilities:
- A very high level of terrain mapping accuracy thus allowing volumes to be resolved with sufficient resolution.
- A capacity to distinguish between windowed and productively transported material.
In the field program and in the predictions of the model suggest selection of optimal blade load is essential to maximizing productivity. Selection of the optimal load is not always intuitively obvious, especially in cycles that involve downhill, horizontal and uphill push. In such cases an optimal load for the initial downhill push may produce very low speeds or loss of traction at horizontal and up hill grades. In practice this may result in partial shedding of the load part way through the cycle. If this process is repeated the material deposited during the load shedding will distort the bulldozer path. The productivity model has been developed beyond the original specification to include blade load optimization for each cycle. It is hoped that these optimal loads can be used to assist operators in the field.
The bulldozer productivity model was developed and tested by re playing monitored cycles using the 3d-Dig files produced during the field program. The cycle time vs. distance points from the simulated cycles were superimposed on the monitored data. For a similar range of cycles, it was found that the simulated cycle times we generally in good agreement with the observed times.
The optimization system is 90% complete and currently capable of subdividing a complex operation into a set of contiguous dozing regions and simulating a single lift, over all regions. An optimization system capable of running all lifts and searching for the optimal strategy is expected to be ready for testing in about two months. Using the existing system a cross pit dozing operation was modeled with four different strategies simulated. These strategies varied in the gradient of push. Details of this study are contained in this report.