Overview
The end goal in open pit mining is to achieve reliable mine slopes that, if they fail, do not cause loss of life, equipment damage, sustained losses of production, or the inability to achieve published reserves.  Over the years these requirements have been hampered by critical gaps in our knowledge and understanding of the relationships between the strength and deformability of jointed rock masses and the likely mechanisms of failure, with divided opinions on how to characterise the rock mass.  Commonly, there has been an uncritical derivation of physical rock properties from empirical classification systems and an equally uncritical use of such physical properties in design analyses without a clear understanding of the geological framework.  These inadequacies have been exposed by a number of spectacular large open pit slope failures, resulting in multiple fatalities, production losses and unfavourable world-wide publicity.  Collectively, they have demonstrated a need to step outside the box and reassess the fundamentals of rock mass strength and slope failure mechanisms from first principles.

The LOP Project slope design research tasks have been and continue to be directed at addressing these inadequacies by enabling the effective use of geological and geotechnical data in assessing rock mass characteristics, 3D modelling and simulation of slope failure mechanisms, design analysis, and uncertainty analysis.

Originally the principal objectives of the project were to:

These objectives were set to be progressed within the framework of the following seven tasks:

Task 1 - convert a prototype 3D modeller (Siromodel) to software that creates polyhedral blocks from structural distribution data characterised by position, orientation and persistence.

Task 2 – use Siromodel to visualise and evaluate likely pathways for candidate failure surfaces along or through a combination of incipient structures, rock bridges and/or weak rock in the structured 3D rock mass.

Task 3 - evaluate how the structured rock mass deforms in the stress environment of an open pit and how candidate failure surfaces may propagate through the dilating rock mass along the path of least shear and/or tension resistance.  Particle flow codes and/or other numerical codes will be used together with Siromodel to complete this research task.

Task 4 - evaluate the rock mass characterisation data and recognise/define the attributes that control the strength of the structured rock mass as it deforms and the candidate failure surfaces propagate, recognising the possibility that the strength of a dilating rock mass and its constituent pieces of mobile rock is a function of particle interlock or some other energy function rather than particle friction (Ø and c) as dictated by the Mohr Coulomb failure criterion.

Task 5 - evaluate the potential for decay and loss of rock mass strength with time, which may vary significantly from rock mass to rock mass.

Task 6 - describe and set criteria for the different failure mechanisms that may be associated with or attached to the predicted candidate surfaces.

Task 7 - recognising that the rock mass strength may be defined in a different way than before, methods of analysis will be developed to examine the reliability of the described candidate failure surface.

Research Outcomes

During Year 2 of the project, Tasks 1 and 2 were modified as follows.

Beta Siromodel software was developed and distributed for trialling at sponsor mine sites, a process which continued during Year 3 and was completed in Year 4 of the project.   It was decided at the 4th SMC meeting in October 2006 however, that the application of Siromodel would be restricted to bench design at the bench and inter-ramp scale, with the discrete fracture network (DFN) modelling linkages with particle flow codes or other numerical codes (Task 3) being performed by DFN modelling codes such as JointStats and FracMan.

The original JointStats software was produced by the Julius Kruttschnitt Mineral Research Centre (JKMRC), University of Queensland, as part of the International Caving Study research and technology transfer program.  The software accepted standard structural data from a face mapping or borehole scan line and organised the data hierarchically according to its orientation, length, spacing and persistence (trace length) attributes.  A (DFN) model of the fracture network could be exported to a 2D or 3D numerical modelling code using an XML text file.

For the LOP project the software was enhanced jointly by the JKMRC and CSIRO by:

• expanding the existing JointStats data base to include;
• quantitative measures of rock mass parameters, and
• data exported from Sirovision in a form suited to statistical analysis,
• providing a means of analysis of rock mass parameter data that leads to a measure of the reliability of that data from within a single geotechnical domain;
• providing a means of statistical joint analysis of;
• joint orientation that is equipped with measures of the reliability of the calculated mean set direction and Fisher dispersion constant, and
• joint persistence data that is equipped with measures of the reliability of the calculated joint size distribution and joint spatial density of scan line data, with joints modelled as full disks,
• providing a means of determining the goodness of fit of the disk model of joints to a set of data that can also be used to test for a significant difference between orientation or persistence data sets; and
• providing a means of combining the measure of uncertainties arising from orientation, persistence and rock mass parameter analysis to arrive at overall measures of uncertainty.

The enhancements were completed during Year 3 and the software was released to the Sponsors in October 2008.

In parallel with the above, Tasks 3 through 7 were also modified to incorporate new developments that originated from within the project.   In particular, it was recognized that when considering the propagation of candidate failure surfaces through the jointed rock mass there is a need to:

• construct an “equivalent material” that honours the strength of the intact rock and joint fabric within the rock bridges that may occur along a candidate failure surface in a closely jointed rock mass; and
• be able to simulate the brittle fracture that can and does propagate across the joint fabric within the rock bridges as the rock mass deforms.

Accordingly, a number of approaches and numerical codes with the potential to construct an “equivalent material” and model brittle fracture across the rock bridges were considered.

After trial analyses the Itasca PFC code was selected as the vehicle for further study.   The principle reason for this decision was that PFC uses a micro mechanics based criterion (the BPM or Bonded Particle Method) to determine if a crack is to be created within a continuum element, which offered the potential for stepping away from the Mohr-Coulomb and/or Hoek-Brown criteria, a feature that was consistent with the objectives of the research (Read, 2007).

Extensive tests on simulated laboratory samples have shown that the synthetic bonded particle material can be calibrated to produce quantitative fits to almost all measured physical parameters, including moduli, strength and fracture toughness.   It has also been shown that it is possible to use the BPM method to represent the strength of the intact rock and joint fabric within the rock bridges with an “equivalent material” or Synthetic Rock Mass (SRM) material.   In this model the intact rock is represented by an assemblage of bonded particles numerically calibrated using UCS, Modulus and/or Poisson’s Ratio values to those measured for an intact sample.   The joints are represented by a sliding joint model that allows associated particles to slide through, rather than over, one another and so represent joints that slide and open in the normal way (Pierce et al, 2007).

Ongoing work during Year 3 of the project showed that, given reliable intact strength and jointing data, by using the SRM methodology it is possible to construct an "equivalent material" envelope that honours the strength of the intact rock and joint fabric within the rock bridges that may occur along a candidate failure surface in a closely jointed rock mass.  Additionally, because the characterisation tests are done for different sample sizes (ranging between 5 m and 80 m cubes in the test cases) and orientations, when using the SRM methodology it is possible to demonstrate the effect of scale and defect orientation on the strength of the rock mass.  These are the groundbreaking outcomes that were sought by the project.  Quite clearly, considerably more testing and comparisons with empirically derived Hoek-Brown parameters will be necessary to confirm the methodology.  As with development of the Hoek-Brown criterion itself, this will take time and experience, but it is the required next step.

A practical difficulty encountered during the SRM studies was that current hardware limitations prevented 3D SRM simulations for a complete slope.  To overcome this difficulty, the study leader, Dr Peter Cundall, has developed a special purpose code, Slope Model, that is based on lattice mechanics, will run 10 times faster than the current PFC3D code, and will be able to handle much larger models.  It embodies the SRM concept so that a DFN may be imported, with failure involving both movement on joints and breakage through intact material (rock bridges).  The extra capacity of Slope Model will enable the direct modelling of significant portions of a real slope (e.g. a potentially unstable region in one part of the slope).  The new code will also be able to model fully coupled transient flow within a closely jointed rock mass environment, including variation in pore pressures as the slope is excavated, topics which are now being studied as part of the Year 4 hydrogeological research studies outlined below.

Although the understanding of hard rock mining hydrogeology has increased significantly over the last 10 to 15 years, there remains a tendency to apply over-simplified approaches for assessing pore pressure distributions in a closely jointed rock mass:  over-simplification often results in underestimating the pore pressure distribution.

Generally, in a jointed rock mass the distribution of water pressures in the discontinuities behind the slope at any particular time are poorly defined.  The water pressure distribution is usually addressed by considering the rock mass as an equivalent continuum whereby the flow analyses employ some form of an equivalent porous media approach and the resultant water pressures are distributed to all of the discontinuities in the mechanical model.  These analyses typically ignore the scenario where slope deformation is changing the water pressures within the rock mass and the resulting changes in the permeability of the fracture network.  Conventional analytical approaches are also inappropriate if the intact rock is relatively impermeable and there is little connectivity between the discontinuities.  Likewise, for transient conditions (e.g. heavy rainfall), the analyses need to account for variable flow/pressure conditions.

The effects of these simplifications in our pore pressure estimates can and have led to production losses, either from conservatively over-designed slopes (i.e. slopes that are flatter than necessary) or under-designed slopes that fail.  To overcome these shortcomings, in Year 4 a new task (Task 4) was added to the project’s research plan.  The research will be ongoing over the next two years and is designed to achieve the following objectives.

• Develop an understanding of the flow process in rock masses at different scales, particularly those with poor or limited connectivity.
• Develop a numerical model that realistically couples fluid flow, pressure distribution and rock deformation.
• Extend and apply the understanding to an assessment of the effects of pore pressure on the stability of fractured rock slopes.
• Develop and document a methodology that will allow LOP sponsors to assess the effects of groundwater on the stability of their slopes.
• Validate that methodology against existing conditions at sponsor sites.

top
References
Pierce, M., Cundall, P., and Potyondy, D., 2007:  A synthetic rock mass model for jointed rock.  1st Canada-U.S. Rock Mechanics Symposium, Vancouver, B.C., May 27-31, 2007, pp. 341-349].

Read, J.R.L., 2007:  Rock Slope Stability Research.  Slope Stability 2007, Proceedings 2007 International Symposium on Rock Slope Stability in Open Pit Mining and Civil Engineering, Perth, Y Potvin (ed), pp 355-359.  Australian Centre for Geomechanics: Perth.