Sensitivities in rock mass propertiesA DEM insight

Cédric Lambert(*) & John Read(**)

(*) University of Canterbury, New Zealand

(**) CSIRO - Earth Science and Resources Engineering, Australia

• Application to a particular geotechnical domain of an Australian mine

• Synthetic rock mass approach, based on real mine data

• Sensitivity of rock mass properties:

• To lithology

• To joint size

• To fracture frequency

Sensitivities of the properties

• Generate a discrete rock mass specimen representative of field conditions

• Based on measurable parameters

The synthetic rock mass (SRM) framework

SRM - Bonded particle assembly intersected with

fractures

Intact rock - DEMJoint fabric – 3D Discrete Fracture Network (DFN)

• Stochastic generation of a DFN model based on structural information

• Available information: window mapping and core logging

• Persistence and density estimated following method proposed by Mauldon (1998) & extended by Lyman (2003)

• Trace length distribution determines the persistence parameters (or diameter distribution)

• Number of traces or spacing controls the joint density

Structural modelling & DFN generation

• Intact rock modelled with PFC3D

• Calibration of microproperties on 2m x 2m x 4m specimens (UCS & elastic properties)

• Resolution of 4 particles across the specimen

Intact rock representation

6m*6m*6m

Basalt

Mafic volcanicsMafic intrusives

6m*6m*6m

Basalt

Mafic volcanicsMafic intrusives

6m*6m*6m

Basalt

Mafic volcanicsMafic intrusives

6m*6m*6m

Basalt

Mafic volcanicsMafic intrusives

UCS [MPa] E [GPa]Lab. PFC3D Lab. PFC3D

Mafic intrusive 113.7 107.9 96.6 98.0Mafic volcanic 74.3 70 28.8 29.5Basalt 170 166.3 30 30Multi rock 112.6 70.9

• Actual geology had to be simplified:

• 3 dominant lithologies considered

• Calibration for each lithology

• Randomly distributed with same relative proportion

• Multi rock model tested under unconfined compression conditions

Synthetic rock mass specimen

smooth-joint contact model

Intact rock

DFN

24m rock mass specimen

( PFC3D v4.0 manual)

• Characterisation of the stress strain behaviour of the rock mass

• Extract rock mass mechanical properties (i.e. UCS, Young’s modulus)

UCS test performed in N-S direction

Unconfined compression test on 24m rock mass specimens

• Mine management wanted to know if the methodology could be applied to their mining environment

• Simplified intact rock representation

Sensitivity to lithology ?

• No censoring information available on trace length distribution

Accuracy of the size distribution ?

Sensitivity to joint size ?

Sensitivities of the rock mass properties

• Compression tests have been performed on specimens considering a homogeneous lithology

• Same 24m DFNs have been used for each lithology

Average compression strength Average deformation modulus

Sensitivity to lithology

• 24m rock mass specimen have been generated applying a multiplication factor s to the size of the joints of the reference sample (0.5 0.7 0.9 1.0 1.1 1.2 1.5)

- Same orientation, same position and same spatial density

- Cover a wide range of interlocking

Sensitivity to joint size

s=0.5 s=0.7 s=0.9 s=1.0

s=1.5s=1.1 s=1.2

• Unconfined compression test in N-S direction

Sensitivity to joint size

Intact rock UCS112.6 MPa

Bilinear relationship two distinct mechanisms ?

s ≤ 1.1 : more than 30% of the rock mass is continuous failure involves brittle failure through

intact rock

s > 1.1: towards blocky rock mass controlled by interlocking?

1

2

1

2

• Fracture frequency (in loading direction) varies with joint size and joint spacing (or density)

• Specimens have been tested varying the spacing keeping other properties of the DFN identical

Sensitivity to fracture frequency

from Ramamurthy et al. (2001)

fcicj Jexp

n) (r / J J nf

Fracture frequency is control parameter in the direction of loading fracture intensity probably more suitable

parameter

• Sensitivity analysis exhibited a linear relationship between intact rock properties and rock mass properties

- Uncertainty could easily be extrapolated from intact rock to rock mass

• Random distribution of lithologies is certainly not a realistic representation of the intact rock condition

- Recent work shows that spatial variation of properties in slopes reduces factor of safety (Jefferies et al. 2008)

- Length of spatial variation is critical

• Enhanced the significant variation of strength with fracture intensity

- Highlighted the importance of getting and collecting structural data right!

Conclusion

Thank you for your attention

Corresponding author : cedric.lambert@canterbury.ac.nz

Acknowledgement to the sponsors of the Large Open Pit (LOP) project (www.lop.csiro.au)