Statistics enables comparison of rock bolt pull-test results across variable ground conditions. Pull-test data provides geotechnical engineers with the load bearing capacity of the full length of the bolt. To compare the effectiveness of different types of bolts, it is important to consider whether apparent differences, are actually due to other factors, such as ground conditions and hole diameter.
Ideally, this would be done using a factorial ANOVA analysis based upon properly designed experiment (DoE). But, in th real world, pull testing means halting heading development. This means pull tests are usually conducted on available headings, rather one designated by experimental design. Under these circumstances, the best way to compare bolts is to compare like with like ground conditions using simple comparative statistics like t-tests. At this mine, ground conditions in a single heading can vary from highly friable quartz breccias with cavities, to highly altered andesite breccias and clay. This means that ground support needed to perform across a range of conditions. Whilst the trial results demonstrated that inflatable bolts were capable of achieving very high pull-test results in very weak rock (RMR =15-20), there were insufficient case studies to achieve a statistically significant t-test result. In cases where there are insufficient tests for a particular rock type, it may be possible to use a factorial ANOVA to statistically “remove” the effect of ground conditions, and/or borehole diameter from the test result. Sometimes, the mine site’s database is sufficiently large, to enable simulation of a Design of Experiments approach using existing data.
