HAL Id: hal-02461099https://hal-mines-paristech.archives-ouvertes.fr/hal-02461099

Submitted on 30 Jan 2020

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Identifying critically stressed faults without triggeringslip : A discussion

David Eaton, Mason Mackay, Anton Biryukov, Murad S. Abuaisha

To cite this version:David Eaton, Mason Mackay, Anton Biryukov, Murad S. Abuaisha. Identifying critically stressedfaults without triggering slip : A discussion. Microseismic Industry Consortium: Annual ResearchReport, Volume 6, 2016. �hal-02461099�

Chapter 35

Identifying critically stressed faults withouttriggering slip : A discussion

David W. Eatona, Mason MacKaya, Anton Biryukova and MuradAbuAishaaa Department of Geoscience, University of Calgary, Calgary, AB, T2N 1N4, Canada.

E: eatond@ucalgary.ca

Summary

Identifying and mapping critically stressedfaults is a key first step for assessing hazardsfrom induced seismicity. The classical paradigmof mapping faults with seismic or well log datamay not be universally successful, however, asfaults are sometimes cryptic and not easily iden-tified. This is especially true for near-verticalstrike-slip faults, for which vertical reflectionoffsets and/or diffractions may not be conspicu-ous, or in the case of low-angle thrust faults suchas an intracutaneous wedge within a thick shalesequence. We propose alternative approachesfor identifying critically stressed faults that in-volve: 1) modeling and mapping of stress per-turbations caused by the presence of a criti-cally stressed fault system, 2) mapping localizedzones of anomalous pore pressure in a targetshale formation, and 3) dynamic triggering ofmicroseismicity due to high-amplitude surfacewaves resulting from large teleseismic earth-quakes. The goal of this work is to promote fur-ther discussion.

35.1 Introduction

The identification and mapping of potentialseismogenic faults without activating them isan important element for risk assessment forinjection-induced seismicity (Walters et al.,2015). Traditionally, the identification of faultsis carried out using well logs (Rafiq et al.,2016) or by mapping reflection offsets in seis-mic profiles (Figure 1). However, this approachis not always viable, especially in the case ofnear-vertical strike-slip faults or sub-horizontalfaults. Strike-slip faults commonly have a near-vertical fault orientation and may not resultin any obvious vertical offset across the fault.Moreover, thrust-fault systems commonly con-tain low-angle (flat) segments. If such segmentsoccur within a thick shale sequence they may bedifficult to identify.

455

Eaton Microseismic Industry Consortium Vol. 6 – Chapter 35

Figure 1: Seismic line in the Fox Creek region showing inter-preted faulting. From Green and Mountjoy, 2005.

In this paper, we propose alternative ap-proaches to identification of critically stressedfaults. These alternative approaches consist of:

1. Mapping perturbations in stress conditionsnear a fault, including changes in stressmagnitude and direction (Figure 2). Suchvariations in stress could be identified us-ing DFIT, wellbore breakouts (Bell et al.,1992) and cross-dipole sonic logs. Underfavourable conditions it may be possible tomap stress variations in detail using seis-mic data that is calibrated using these in-dependent sources.

2. Mapping pore pressure compartmentswithin a reservoir. We propose that faultsmay act as barriers to stress equalizationwithin a reservoir.

3. Identifying critically stressed faults bymapping microearthquakes that are dy-namically triggered by the passage oflarge-amplitude surface waves from dis-tant earthquakes. Implementation of thisapproach would require deployment of aseismograph array in advance of hydraulicfracturing.

These three proposed methods are examined inthe following sections.

35.2 Stress Modeling

In order to calculate how the stress field arounda fault is perturbed both in magnitude and di-rection, the finite-difference code FLAC3D wasused (Itasca, 2012). The finite-difference ap-proach allows for materials to be represented ascubic polyhedral elements that are free to de-form following Hooke’s elastic constitutive law(Itasca, 2012),

∆σij = 2G∆εij+

(K − 2

3G

)∆εkkδij, (35.1)

where Einstein notation is used implying thechange in stress is a function of the Kroeneckerdelta δij , bulk modulus, K, and shear modu-lus G. An elastic material formulation withouta failure criterion was used in order to repre-sent the complete mechanics of stress distribu-tion around a fault without the complexity offailure mechanisms that one would expect to beassociated with dynamic processes.

The model approximates a stress condi-tion that is characterized by both strike slipand thrust fault regimes based on the criticallystressed crust theorem, wherein the ratio of max-imum and minimum principal stresses, σ1 to σ3

is dependent on a coefficient of friction, µ, of0.6 (Zoback 2010):

σ1

σ3

=(√

µ2 + 1 + µ2)2

. (35.2)

The convention used by the numerical codeis that negative stresses are compressive. Thestress state is at a modelled depth of 3 km as-suming an average crustal density of 2700 kg/m3

and a hydrostatic pore pressure gradient. Thefault is approximated as a circle with a diame-ter of 3 km and the model size is 12 × 12 ×12 km . The 25 m computational grid is gra-dationally meshed in order to balance computa-tion time and numerical accuracy with mesh el-ements at the fault face. The material propertiesand model parameters are summarized in Table1.

456

Microseismic Industry Consortium Vol. 6 – Chapter 35 Identifying Faults

Figure 2: Left panel: deflection of maximum horizontal stress orientation in the neighbourhood of vertical open fractures that arenot oriented perpendicular to the smallest principal stress. Right panel: deflection of mean borehole breakout azimuths from 29wells in the Scotian Shelf, offshore eastern Canada. From Bell et al., 1992.

The effects of gravity were neglected andthe model was initialized with an in situ stressstate. Boundary conditions were applied withfixed corners and constrained to only allow forshear displacement at the model boundary. Themodel was then stepped to equilibrium and thefault region material was changed to a weakermaterial by assigning a fault material with halfthe bulk and shear modulus of the surroundingrockmass. The model was then stepped to a newequilibrium and analysis of the stress perturba-tion was made. The results of stress modellingare summarized in Figures 3-5.

35.3 Pore pressure compartments

In sedimentary basins, regions of pore overpres-sure can be compartmentalized by permeabilitybarriers (Hunt, 1990). Breaches in integrity ofseals can occur due to fracturing, induced bygeneration of hydrocarbons within the compart-ments and/or the thermal expansion of pore flu-ids. An episodic process of resealing and break-out cycles in intervals of thousands of years hasbeen proposed by Hunt (1990). Figure 6 showsan example of compartmentalized pore over-pressure within the Montney region of Albertaand British Columbia and its possible relation-ship to anomalous induced seismicity (Eaton etal., this volume).

Figure 5: Maximum principal stress rotation. Points representtrends and plunge of the primary principal stress orientation.Deviation from a perfect North to south alignment indicates arotation of the stress field by up to 25 degrees.

Table 1: FLAC3D fault zone modelling parameters. The follow-ing parameters are chosen to represent the rockmass and faultproperties of an assumed fault zone that may be a candidate forinduced seismicity.

457

Eaton Microseismic Industry Consortium Vol. 6 – Chapter 35

Figure 3: Maximum principal stress magnitudes contours around a 3 km fault. Stress perturbations indicate a lowering of stressmagnitudes in the core of the fault with increase stress magnitude lobes distally from the fault edges.

458

Microseismic Industry Consortium Vol. 6 – Chapter 35 Identifying Faults

Figure 4: Minimum principal stress magnitude contours. Increased compressive stresses (orange) are observed distally to the fault,while zones of relaxation (green) occur in other regions.

459

Eaton Microseismic Industry Consortium Vol. 6 – Chapter 35

Model Parameters Value Units

Density 2700 kg/m3

Principal stress magnitudes -150, -50, -50 MPaShmax (trend/plunge) 180/0 degreesShmin (trend/plunge) 90/0 degreesBulk modulus K 20 GPaShear modulus G 12 GPaFault diameter 3 kmFault zone width 60 m

35.4 Dynamic Triggering

Dynamic triggering happens as surface seis-mic waves from an initial earthquake propa-gate through the Earth’s crust and perturb thestresses at local critically stressed faults, trig-gering secondary earthquakes. Once the seis-mic wave train has passed and ground shakingends, the crust in the local area returns to itsprevious stress state modified according to thecumulative stress drops associated with locallytriggered earthquakes.

Dynamic triggering of secondary earth-quakes was widely accepted in the scientificcommunity following the 1992 M7.3 Landersearthquake in southern California. Earthquakerates increased dramatically across the westernUnited States at distances well beyond the after-shock zone in a few days after the earthquake(Hill et al., 1993). Furthermore, dynamic trig-gering has been observed across the globe invarious geologic and tectonic environments. Ithas been shown to take place at distances fromthe initial rupture varying from meters (Kilbet al., 2000) to over thousands of kilometers(West et al., 2005). The main ruptures havebeen noticed to trigger other earthquakes at dif-ferent time scales. In many cases secondaryearthquakes occurred within minutes to hoursfollowing the primary seismic surface wave ar-rival (West et al., 2005; Gomberg et al., 2004).

In other cases, earthquakes occurring weeks tomonths after the initial earthquake have been in-terpreted as a delayed response to dynamic trig-gering.

Recently, van der Elst et al. (2013) andWang et al. (2015) noticed that seismic activ-ity on a fault in the areas of wastewater injec-tion may be related to the stress perturbationscaused by strong distant earthquakes. By usingthe novel match-filtering approach, they havemanaged to extract low-magnitude events fromthe continuous waveform data and establish thestatistical causality between the major distantearthquakes and local increase in the seismic ac-tivity in the hydraulically stimulated areas.

Having been located and shown on the map,the series of triggered event locations may bespatially clustered, thus providing an estimateof the dimensions of the fault that has under-gone dynamic triggering. This suggests thatsuch remote triggering could potentially serve asa probe for outlining the areas where criticallystressed faults are present, prior to any hydro-carbon field development. This procedure couldserve as a preventive manner of accidental faultactivation, causing unnecessary slip, and thusreduce the risk of a production disruption.

35.5 Conclusions and Outlook

Recognition and mapping of critically stressedfaults is a key component of hazard assess-ment for anomalous induced seismicity. Undervarious common circumstances, fault imagingand mapping may be problematic using tradi-tional methods, including 3-D seismic imagingor mapping with densely spaced well log data.We propose three novel methods for indirectlyinference of the existence and location of crit-ically stressed faults: mapping perturbations toregional stress, mapping compartments in porepressure, and clustering of naturally occurringmicroearthquakes that are dynamically triggeredby teleseismic surface waves. Ongoing researchis focused on evaluation of the potential utility

460

Microseismic Industry Consortium Vol. 6 – Chapter 35 Identifying Faults

Figure 6: Heterogeneous (compartmentalized) pore-pressure gradient in the Montney trend, courtesy Canadian Discovery, Ltd.Black dots show M ≥ 2.5 earthquakes since 2008, suggesting that pore-pressure compartments may be linked to faults.

Figure 7: Seismicity rate increases associated with Hector Mine and Landers mainshocks, showing that dynamically triggeredevents tend to align along faults. From Gomberg et al. (2001).

461

Eaton Microseismic Industry Consortium Vol. 6 – Chapter 35

of these proposed methods.

35.6 Acknowledgments

Sponsors of the Microseismic Industry Consor-tium are thanked for their long-standing supportof this research topic. We are particularly grate-ful to Canadian Discovery Ltd. for permission toshow the pore-pressure data, as well as the Nat-ural Sciences and Engineering Research Coun-cil of Canada (NSERC) and Chevron Canada fortheir support of an Industrial Research Chair inMicroseismic System Dynamics at the Univer-sity of Calgary. In addition, Aquaterra WaterManagement, ConocoPhillips Canada, Micro-seismic Canada and Nanometrics are thankedfor sponsorship of a collaborative research pro-gram on multi-scale monitoring of hydraulicfracturing and wastewater injection, Duvernayregion.

35.7 References

Bell, J.S., G. Caillet and J. Adams, 1992.Attempts to detect open fractures and non-sealing faults with dipmeter logs. GeologicalSociety Special Publication, 65, 211-220.

Eaton, D., B. Cheadle and A. Fox, 2016. Acausal link between overpressured hydrocar-bon source rocks and seismicity induced byhydraulic fracturing. This volume.

Green, D.G. and E.W. Mountjoy, 2005. Faultand conduit controlled burial dolomitizationof the Devonian west-central Alberta DeepBasin. Bulletin of Canadian Petroleum Geol-ogy, 53, 101-129.

Gomberg, J., P. A. Reasenberg, P. Bodin andR. A. Harris, 2001. Earthquake triggeringby seismic waves following the Landers andHector Mine earthquakes Nature, 411, 462-466.

Gomberg, J., Bodin, P., Larson, K., and Dragert,H., 2004. Earthquake nucleation by transientdeformations caused by the M = 7.9 Denali,Alaska, earthquake. Nature, 427, 621-624.

Hill, D. P., Reasenberg, P. A., Michael, A.,Arabaz, W. J., Beroza, G., Brumbaugh, D.,Brune, J. N., Castro, R., Davis, S., and De-polo, D., 1993. Seismicity remotely triggeredby the magnitude 7.3 Landers, California,earthquake. Science, 260, 1617-1623.

Hunt, J.M., 1990. Generation and Migration ofPetroleum from Abnormally Pressured FluidCompartments. AAPG Bulletin, 74, 1-12.

Itasca Consulting Group, Inc., 2012. FLAC3D- Fast Lagrangian Analysis of Continua inThree-Dimensions. Ver. 5.0. Minneapolis:Itasca.

Kilb, D., Gomberg, J., and Bodin, P., 2000. Trig-gering of earthquake aftershocks by dynamicstresses. Nature, 408, 570-574.

Rafiq, A., D.W. Eaton, A. McDougall andP.K. Pederson, 2016. Reservoir Characteriza-tion using Microseismic Facies Analysis Inte-grated with Surface Seismic Attributes, Inter-pretation, in press.

van der Elst, N. J., Savage, H. M., Keranen,K. M., and Abers, G. A., 2013. EnhancedRemote Earthquake Triggering at Fluid-Injection Sites in the Midwestern UnitedStates. Science, 341, 164-167.

Wang, B., Harrington, R. M., Liu, Y., Yu, H.,Carey, A., and van der Elst, N. J., 2015.Isolated cases of remote dynamic triggeringin Canada detected using cataloged earth-quakes combined with a matched-filter ap-proach. Geophysical Research Letters, 42,2015GL064377.

Walters, R.J., Zoback, M.D., Baker, J.W. andBeroza, G.C., 2015. Characterizing and Re-sponding to Seismic Risk Associated withEarthquakes Potentially Triggered by FluidDisposal and Hydraulic Fracturing. Seismo-logical Research Letters, 86, 1110-1118.

West, M., Sanchez, J. J., and McNutt, S. R.,2005. Periodically Triggered Seismicity atMount Wrangell, Alaska, after the SumatraEarthquake. Science, 308, 1144-1146.

Zoback, M. D., 2010. Reservoir geomechanics.Cambridge University Press.

462