Thomas, RDH 2013, 'A statistical approach to account for elevated levels of uncertainty during geotechnical design', in PM Dight (ed.), Slope Stability 2013: Proceedings of the 2013 International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, Australian Centre for Geomechanics, Perth, pp. 324-335, https://doi.org/10.36487/ACG_rep/1308_19_Thomas
(https://papers.acg.uwa.edu.au/p/1308_19_Thomas/)
Abstract: Uncertainty in geotechnical engineering results from the inherent variability of natural materials, and the challenges engineers and geologists have in correctly assessing them. Geotechnical design at the definitive feasibility study (DFS) stage should be based on a geotechnical model with target levels of confidence of ideally between 50 and 75%. Due to various reasons such as budgetary constraints for ground investigation or changes to the resource model affecting mine design, a higher degree of data uncertainty can exist in one or many of the model inputs. This can be accounted for by selection of conservative design values, but where probabilistic design is desired further uncertainty can be induced by estimating population characteristics, for instance by adopting coefficients of variation from published literature, e.g. Harr (1987) and Kim (2005).
This paper presents a case study for a DFS level open pit slope design for a gold project in West Africa. The geological, structural and hydrogeological models were suitably defined, however limited geotechnical drillhole data was available for some units. The limited data sets hindered rock mass characterisation and derivation of design values (and distributions) for subsequent slope stability modelling, resulting in elevated levels of uncertainty. The author was faced with the prospect of either accepting the uncertainty and accounting for the small populations by choosing lower bound values and assumed distributions, or relying on regional data from experience of working with the encountered units elswhere. In an effort to ensure the most relevant, deposit specific data was used, the author sought to supplement the limited data sets with additional drillcore data from elsewhere within the project area.
The challenge the author faced was to justify combining data that would conventionally be considered separately. A number of statistical tests were used to demonstrate the validity of combining the different populations of drillcore data. A suite of logged and derived parameters were tested. The results of the statistical analyses and the effect of combining drillcore data populations on the resultant level of uncertainty and ultimate pit slope design are presented.