FAILURE MECHANISMS, SOURCE PARAMETERS and QUANTIFYING THE SIZE OF MINING INDUCED SEISMIC EVENTS Witwatersrand Basin South Africa R. Ebrahim-Trollope*, G. Smith* and R. J. Durrheim^. *University of Cape Town ^University of the Witwatersrand and CSIR

Introduction 1.Elastic waves have the shortest wavelength of any Geophysical wave. Sensitivity is localised spatially and temporally and allows for the highest resolution (Femto scale). 2.Geophysics (seismology) is a remote and observational discipline. Based on inverse theory and simple mathematical models. Most advances still based on empirical analysis. 3.Many unresolved hypotheses Complexities and heterogeneities of failures require resolution and quantification.

Measuring the Size of an Earthquake Inversion Instruments record some distance from the source Time domain Amplitude, duration, etc. Frequency domainSource parameters (Ω 0, f c, f c -γ ) Simple Models Double Couple Shear rupture type failures. Brune, Madariaga, Haskell, Savage, etc. (Moment, Energy ; Stress drops, Radius, etc.)  Common assumptions  Homogeneous (material properties and mechanisms)  Shear rupture and scale invariant or Self-similar  Azimuthal Bias (ignored)  Q, kappa, etc.

Measuring the Size of an Earthquake

Tectonic earthquakes – GeneralM L = Log (A max ) +B log (R) – C – SANSNM L = log (A max ) log (R) R – S – M w = 2⁄3 log M o – 6 – M e (β) = 2⁄3log E β – 2.9M o = µAD – Others (M c, M S, M D, m B, etc.)E β = 4пρV c R 2 E Flux Induced seismic events – M KRSN = 1.45 log T – M L = A Log (E) + B Log (M o ) – C – Carletonville: A , B , C – Klerksdorp: A , B , C and – Welkom: A , B and C – Current: M L = log (E) log (Mo) Measuring the Size of an Earthquake

Available Data

Unresolved Inverse techniques and Hypotheses ? The “Magnitude Scale” This is not a measure of a seismic event but rather a rating. 1.Self Similarity (Scale invariance) Is the physics of failure the same across the size spectrum?. Two camps. 2.Failure models Point source, circular spreading shear rupture. 3.Self organised criticality or characteristic. Size distribution (Gutenburg-Richter “Law”) 4.Fractal sets and fractal dimensions. Quantifying Hazards and Risks.

Results

Fiii Fi Fii

Results

Conclusions Source parameters are noticeably different for different lithologies and failure mechanisms. Self-similar scaling appears invariant for fracture events but not for structural events. These differences significantly affect the “representability” of magnitude as an estimate of size. A single number rating for a multi-dimensional failure is hopelessly inadequate. Multi-model distributions are a mix of mechanism datasets Probabilistic methods based on the G-R become invalid and non representative of hazards and risks i.e. ineffective. New techniques for hazard quantifications need to be developed

Proposal S - EventsC - EventsF- Events MLML Energy Log 10 (J) Length m E/L J/m 0.7 KJ/m 16 KJ/m 180 KJ/m 1.6 MJ/m 80 MJ/m 1.7 KJ/m 6.7 KJ/m 35 KJ/m 223 KJ/m 5.2 KJ/m 23 KJ/m 100 KJ/m 171 KJ/m E/Area J/m 2 30 J/m J/m KJ/m KJ/m KJ/m 2 85 J/m J/m KJ/m KJ/m KJ/m KJ/m 2 12 KJ/m 2 14 KJ/m 2 The various sources / failure mechanisms exhibit different source parameter relationships Significantly different energy radiated per M IMS for the different ensembles and regions I have grouped these into 3 classifications S – Structural Events (faults, fault filled dykes) C – Combination events (dykes, foundations, abutments, facebursts etc) F – Fracture events After: Ebrahim-Trollope et al, 2014

Proposal Size and strength of mining induced seismic failures Represented in physical SI units Area in meters Energy release per square meter (J/m 2 ) Azimuthal bias in degrees and maximum as per maximum spectra

The Magnitude Scale A magnitude rating for quantifying the size of seismic events is inadequate o General comments by seismologists (and consensus) o Crude form of quantification (McGarr, 1982) o Not directly related to any physical parameter of the source, should not be used beyond recognisance purposes (Kanamori, 1983, 2007) o Hopelessly inadequate measure of the size of an earthquake (Dziewonski & Woodhouse, 1983) o It is pervasive and remains widely used (Gibowicz, 1994) o After 80 odd years “remains a dilemma” for tectonic earthquakes o Greater depths and distances to induced failures o Shear rupture models form the basis of source parameter estimates o The biggest limitations are recording frequency bandwidth & azimuthal bias o Multiple wave types o More of a dilemma for induced seismicity o Very close to or in the source o Known differences in mechanisms (shear rupture is only one failure mechanism) o Design does change seismicity o Inhomogeneity amplified due to proximity (Including material properties)

Introduction After: Internet download

Introduction After: Nguuri et al, 2001 and Singh et al, 2011

Introduction After Bohnhoff et al, 2010

Sensors & Event Frequency Range

Going Underground

Damage Underground

Fatalities due to seismicity and FOG Fatalities due to 9 major events Fatalities due to seismicity (ARMgold / ANGLO) 14% of all fatalities –7.7/year (969 – 53.8/year - Injuries) 1998 – Fatalities due to seismicity (ARMgold) Fatalities Fatalities Fatal Average 2.6/year (Since becoming ARMgold) Revenue 5% of revenue (calculated in 80’s) Example M L = 3.3,3.0 events cost R lost revenue SOME EFFECTS OF SEISMICITY

Induced seismic environment southern africa