Underground » Strata Control and Windblasts
The main objective of this project was to automatically assess the risk of experiencing a windblast in a longwall operation. The logical extension of the risk assessment was the automatic generation of an alarm should the risk exceed pre-determined levels.
Certain conditions need to be present to experience windblasts of an intensity that threatens the safety of the underground operator. These conditions include the geological configuration and the dimensions of the mining excavation (mine layout). Mine planning and operation constraints provide limited flexibility in preventing or controlling the occurrence of large goafs and subsequent windblasts.
Windblast early warning is being achieved at Moonee Colliery. The early warning system has proved to be an effective operational tool. This project endeavoured to provide the methodology to automatically generate windblast early warning alarms. The windblast problem at Moonee, the management and effectiveness of the seismic system and the recorded data were used as the basis for developing and evaluating the automatic early warning methodology.
The report describes the installation of the seismic system and the operator-based real time application of the ISS-stability concept. This concept was first coded into a threshold-based decision tree in an attempt to duplicate the decisions made by the on-site operator. Individual early warning parameters were tested. These parameters were: the precursive event rate; changes in the volume of coseismic inelastic strain (Apparent Volume); and the comparison of radiated energies of seismic events of similar moments. It was found that the precursive event rate was the most reliable parameter.
In order to show clearly that the seismic response to mining, as reflected by these seismic parameters, exhibits an early warning to a pending large instability in the roof, a probabilistic risk-assessment approach was developed and tested. An off-the-shelf expert system was used as the basis for applying a probabilistic philosophy in risk assessment. The result was an early warning system more effective than the threshold-based decision tree concept and as effective as the full-time operator.
The report also describes the attempt to normalise the seismic parameters to the production in a way other than through a temporal reference. The use of the automatic risk assessment in parallel with control measures, such as the hydrofracturing initiation of goafing, is also described.
The report concludes with recommendations on further development, testing and implementation of the expert system philosophy in automatic assessment and recognition of windblast risk.
The CD also contains the software developed at Moonee to undertake the automatic event forecasting function.