Underground » Strata Control and Windblasts
STAGE TWO
The primary objective of this stage of the project was to develop a simple and reliable method to predict in situ horizontal stress magnitudes from existing borehole breakout data. In situ stresses are always crucial parameters for engineering designs and underground mining activities. Sudden changes in stress magnitudes can have significant adverse impacts on mining operations, such as increasing the risk of violent failures. Identification of high-risk zones associated with in situ stress alteration can guide engineers to implement appropriate controls for potential hazards.
Three areas of study were carried out and discussed in this report, including laboratory experiments, model development, and field data validation.
To investigate the impact of specimen size and borehole size on breakout initiation, two series of normal compression tests (constant borehole-specimen ratio and constant specimen size) were conducted on Gosford sandstone, and borehole radius varies from 6mm-12.5mm. Experimental results revealed that the specimen size does not have an apparent influence on the breakout initiation stress under the same borehole radius where the hole radius specimen ratio is over 10, and constant specimen size and constant borehole-sample ratio specimens are both suitable for borehole size investigation. The results also confirmed that larger borehole size would result in smaller breakout initiation stress given the same rock properties and loading conditions, which is consistent with previous research. Further analysis was conducted to correlate the borehole size and breakout initiation stress using analytical solutions (e.g. stress averaging concept and pressure dependent model) and empirical method, and the results indicated that the empirical equation provided reasonable estimation of experimental data and it was adopted for further model development due to its simplicity and accuracy.
Polyaxial experiments were also conducted on Hydrostone-TB to study the influence of minimum horizontal stress (σh) and vertical stress (σv) on breakout geometries. During the experiment, the samples were loaded under constant maximum horizontal stress (σH) and different σh and σv. Results revealed that both borehole breakout width and depth decreased with increasing σh and σv, although the influence of σh is more significant. This suggested that the effect of intermediate stress should be considered during stress estimation using borehole breakout data.
For model development, as breakout depth can vary over time, breakout width is chosen for horizontal stress estimation. 79 experimental data was collected from literature and this project, and nine failure criteria were assessed on the prediction of σh with given σH, σv, breakout width, and rock properties. Results revealed that σh prediction values are sensitive to σH, such that it is not possible to estimate a reasonable value of σh using σH. Nonetheless, the examinations suggested that the Mogi-Coulomb and Stassi D'Alia criteria are capable of predicting σH with known ��ℎ, with average error rates of 13.37% and 13.74% against the laboratory data, respectively. To overcome this limitation and to estimate both horizontal stress magnitudes from borehole breakout data only, an Artificial Neural Network (ANN) model was introduced to estimate σh. The ANN model utilised σv, breakout width, and rock strength from 79 laboratory data as training inputs, and it was optimised based on a training, validation and testing process. Once the model was constructed, it was independently validated against 23 field data collected from literature and one mine site in New South Wales. The ANN model yielded an acceptable average error rate of 15.88% on σh prediction.
To determine the most reliable method for σH prediction, a comparison analysis was carried out on Mogi-Coulomb, Stassi D'Alia, and ANN models. Based on σh values predicted by the ANN model, Mogi-Coulomb failure criterion gave the most reliable σH estimation on the validation data, with an average error rate of 6.82%. Whereas for Stassi D'Alia failure criterion, the error rates ranged from 13.13% to 13.89% depending on the ratio between rock uniaxial compressive strength and tensile strength. In addition, the ANN model on σH estimation had an average error rate of 13.89%. Based on the comparison, the 'ANN'-'Mogi-Coulomb' model is chosen as the most reliable approach for estimating both horizontal stress magnitudes from borehole breakout data.
In order to further validate the proposed model, borehole logging data were collected from five underground mine sites (mine A-E).
To overcome the inaccurate ��h predictions of the current model, the development of a new model has been commenced. The new model will still adopt 'ANN'-'Mogi-Coulomb' method, but the training dataset for the ANN model will be revised. To accurately estimate the Australian stress conditions, only breakout data that is under the stress regime where vertical stress is the minimum principal stress and horizontal stresses are the intermediate and maximum principal stresses will be selected for model training. The preliminary ��h estimation results from the new model have exhibited a more reliable increasing trend with depth, and the accuracy of ��h predictions has increased significantly. With more experimental and field data, the accuracy of the model is expected to be further improved.
STAGE ONE
The main objective of the first stage of this project was to develop a reliable and simple and cost-effective technique that can be used to estimate the magnitude of horizontal stresses based on borehole breakout data. It is well known that ground stresses have a major impact upon roof and rib behaviour and stability. In underground coal mines, stress magnitudes can vary significantly according to direction. In situ stress orientation and magnitude are also altered when adjacent to major geological structures. These changes in stress can have a substantial adverse impact on mining conditions such as increasing the risk of violent failures, via coal burst or other major roof failures. Knowledge of changes in stress orientation and magnitude will indicate the high risk zones within a mine site which can enable mine operators to implement appropriate controls.
Four areas of studies were carried out and discussed in this report, including literature review, laboratory testing, numerical simulation and back analysis of field data. Based on the review of current underground stress measurement techniques, the advantages and disadvantages of each are summarised.