This action might not be possible to undo. Are you sure you want to continue?
A model for induced seismicity caused by hydrocarbon production in the Western Canada Sedimentary Basin
Valentina Baranova, Azer Mustaqeem, and Sebastian Bell
Abstract: Over the past three decades, a significant number of small-magnitude and shallow earthquakes have occurred in the Western Canada Sedimentary Basin and are located along its western flank near areas of oil and gas production. One of the better documented examples is the swarm of earthquakes associated with the Strachan field, in the Alberta foothills. A model based on Segall’s poroelastic theory is developed to account for the occurrence of earthquakes below the Strachan reservoir. Using this methodology, we show that the earthquake of 19 October 1996, underneath the Strachan field, was most probably triggered by gas extraction. The numerical model also implies that gas extraction would cause subsidence and localized changes in in situ stress magnitudes. There is a strong correlation between rates of production and the number of seismic events, but the onset of major seismic activity postdates the commencement of production by approximately 5 years. Poroelastic modelling can account neatly for this observed delay. The modelled stress changes due to gas extraction point to a regime which favours reverse or thrust faulting that is compatible with stress magnitude measurements in the area. The proposed mechanism involves volume changes which decrease the vertical stress Sv and increase the larger horizontal stress SHmax. The mean stress increase beneath the reservoir appears to be small, but increasing the deviatoric stress permits Mohr–Coulomb failure. As a result, the initially high rate and long history of gas extraction appear likely to be the main trigger for the seismicity beneath the Strachan field. Résumé : Au cours des trois dernières décennies, un nombre important de tremblements de terre de faible magnitude et de faible profondeur sont survenus dans le Bassin sédimentaire de l’Ouest du Canada, et localisés le long de son flanc occidental, à proximité de régions productrices de pétrole et de gaz. L’un des exemples le mieux documenté est représenté par le groupe de tremblements de terre associé au champ de Strachan, dans les Foothills de l’Alberta. Un modèle fondé sur la théorie poroélastique de Segall a été élaboré pour expliquer les événements de tremblements de terre sous le réservoir de Strachan. Nous démontrons par cette approche que la cause la plus probable de déclenchement du tremblement de terre du 19 octobre 1996, sous le champ de Strachan, serait l’extraction du gaz. Le modèle numérique révèle en plus que l’extraction du gaz serait responsable d’un phénomène de subsidence et des changements apparus localement dans la grandeur des contraintes in situ. On observe une forte corrélation entre les taux de production et le nombre d’événements sismiques, mais le début de l’activité sismique significative n’a débuté que cinq années approximativement après le commencement de la production. La modélisation poroélastique rend compte nettement de ce délai observé. La modélisation des changements des contraintes engendrés par l’extraction du gaz pointe un régime qui favorise le développement de failles chevauchantes ou inverses, qui est compatible avec les mesures de la magnitude des contraintes dans la région. Le mécanisme proposé implique des changements de volume dont l’effet est d’affaiblir la contrainte verticale Sv et de renforcer la contrainte horizontale SHmax plus grande. Il semble que l’augmentation de la contrainte moyenne sous le réservoir soit faible, mais l’accroissement de la contrainte déviatorique est suffisante pour provoquer la rupture Mohr–Coulomb. En conclusion, le taux élevé de l’extraction du gaz au début de l’exploitation et sa longue durée nous apparaissent comme la véritable cause du déclenchement de l’activité sismique sous le champ de Strachan. [Traduit par la Rédaction] Baranova et al. 64
Received April 4, 1998. Accepted August 28, 1998. V. Baranova,1 A. Mustaqeem, and S. Bell. Geological Survey of Canada, 3303 33rd Street NW, Calgary, AB T2L 2A7, Canada.
Corresponding author. Present address: 3, 2127 7 Avenue NW, Calgary, AB T2N 0Z8, Canada (e-mail: email@example.com).
In the latter half of this century it has become apparent that certain engineering activities can lead to earthquakes in the upper crust (Gough 1978; Simpson 1986; Knoll 1992; Bell 1996). Generally, these activities are not the sole cause of the earthquakes; they merely induce seismic activity by altering stress regimes, so failure occurs. In effect, these are humanly triggered earthquakes. Triggered seismicity can re-
Can. J. Earth Sci. 36: 47–64 (1999)
© 1999 NRC Canada
48 Fig. © 1999 NRC Canada . Theoretically. The Paleozoic section is dominated by carbonates that accumulated in passive-margin environment (Mossop and Shetsen 1994). are overthrust along the western flank of the basin. 1). The Western Canada Sedimentary Basin extends along a southeast–northwest-trending axis across southern Manitoba. the mechanism involved is similar to that which promotes seismicity during mining activities (McGarr et al. Gendzwill et al. Rebollar et al. British Columbia. Contemporary stress field in the Western Canada Sedimentary Basin. Davis et al. over the past several decades. Gendzwill et al. A much stronger case for induced seismicity was made by Horner et al. Alberta. 1982. For example. Pomeroy et al. 1. 1975. The forelandbasin deposits form a series of west-thickening wedges and. Bell and Nur 1978. 1984) and Wetmiller (1986) in- vestigated earthquake swarms near Rocky Mountain House. Segall 1989. J. (1975) have proposed different models for this type of triggered seismicity. Wetmiller concluded that the earthquakes may have been induced by gas production. Total sediment thickness reaches 5 km along the western margin of the basin. 36. have assembled stress and fluid-pressure data which suggest that the probable triggering mechanism for these earthquakes is water injection into oil-bearing Permian Belloy Formation sandstones in the Eagle and Eagle West fields. southern Saskatchewan. (1994) for the recent earthquake swarm near Fort St. (1982. The area highlighted in this study lies on the western flank of the Western Canada Sedimentary Basin in Alberta. He noted that its location was anomalous and speculated that its occurrence might be related to oil and gas production activities in the area. most of which involve porepressure changes (Kisslinger 1976). 1). near the town of Rocky Mountain House (Fig. Raleigh et al. 1972. Gupta and Rastogi 1979). 1999 YUKO N NORTHWE ST TERRIT ORIES CA N SH ADI IEL AN D SA SK AL BE A RT Rocky Mountain House AT C M H AN EW BRITIS H BIA AN IT OB A COLUM Stress trajectory of SHmin >20 20 18 16 <16 kPa / m 0 500 km S Hmin gradient sult from (i) impounding of reservoirs (Gough 1970. This is not an area of high natural seismicity (Rogers and Horner 1991). but this is unproven because of lack of on-site monitoring. Milne (1970) described a magnitude 5. Can. 1976. Alberta. Davis and Frohlich 1993). Vol. whereas the overlying Mesozoic and Cenozoic rock units are predominantly clastic and accumulated in a foreland-basin setting (Leckie and Smith 1992). 1976. we revisit the Rocky Mountain House earthquakes. Yerkes and Castle (1976). and northeastern British Columbia (Fig. 1968. and Smirnova et al. and propose a mechanism to explain their timing and nature. which has been extensively monitored. 1982. 1982). and (iv) underground explosion (Console and Nikolaev 1995). Each of these activities is characterized by specific mechanisms. Rebollar et al. In this study. It contains Phanerozoic sediments ranging in age from Cambrian to Palaeocene which rest unconformably on metamorphic rocks of the Canadian Shield. numerous smallmagnitude earthquakes have occurred along the western flank of the basin in areas where oil and gas have been extracted (Milne 1970. together with the underlying Paleozoic sediments. Many authors such as Segall (1985. Many are suspected to have been triggered by hydrocarbon production. (ii) quarrying or mining and subsurface extraction of fluids (Pomeroy et al. but. Milne and Berry 1976. Alberta. The present paper deals with seismicity resulting from fluid extraction. (iii) subsurface injection of fluids (Healy et al. Hsien and Bredehoeft 1981.1 earthquake that occurred in 1970 at Snipe Lake. 1984. John. 1995). Earth Sci. Horner et al. evaluate their occurrence in the light of gas production from the Strachan field. 1989). Wetmiller 1986).
Adams and Bell 1991). WALA (1993).5 . Seismicity of Alberta (1957–1997) including the location of seismic network. or underpressured.0 3. is the Lower Devonian halite-rich section of the Prairie Evaporite. where geomechanical detachment might take place. 4. 1) in that SHmax.2. Seismograph stations are as follows: 1.3.3. DOWB (1982).0 .0 . The only potential décollement zone.5 2. MCC (1966). The stress regime is directionally homogeneous (Fig. John Cluster 58o ter es W dim Se 56 o n Ca na dia n 56o Snipe Lake Earthquake di s ed rb tu en Brazeau River Cluster 3 tar y Ba sin 54o be lt 8 5 6 4 54 o Edmonton 52o RMHSZ Study Area 52o c Ro ky Mo 7 Calgary Turner Valley Cluster 50o un n tai s 2 50o 1 9 120o 115 o 110 o Some overpressuring is present at depth along the western edge of the basin. 2.5 3. BAN (1955–1966). SLEB (1977).0 # Seismograph stations o 115 o 110o 60 o n dia na ld Ca hie S 58o Fort St. 5. MNB (1981).Baranova et al.0 Above 4. 7. but most units are hydrostatically pressured. There is persuasive ev© 1999 NRC Canada . Years in parentheses are times of establishment of the stations. 6. the maximum horizontal stress.4. Fig. is regionally oriented northeast–southwest across most of the basin (Bell and Gough 1979. EDM (1963). similar oriented horizontal stresses above and below this unit suggest that the entire sedimentary section is attached to the basement (Bell 1996). PNT (1960). SES (1966). 9. 3.5 . 2. However. 49 120 Magnitude of earthquakes 2.
except for major clusters at Turner Valley. does not extend northwards into the Western Canada Sedimentary Basin.1. except for a few earthquakes like the Snipe Lake event of March 1970 (M = 5. In Alberta. 36. where SHmin magnitudes approach and locally exceed overburden stresses. The first recorded seismic event for the Alberta Basin occurred in 1957. and the felt area. The highest SHmin gradients so far documented come from foothill wells drilled in structural settings comparable to that of Rocky Mountain House. This is twice the total number of events in the catalogue for the same period. but the results are inconclusive. Many more smaller events would be recognised if a denser seismometer grid was employed. Most seismic events are barely felt and. © 1999 NRC Canada 1996 .1). Saskatchewan. Rogers and Horner 1991). These elevated gradients may reflect paleostresses derived from the Laramide orogeny and (or) they may be an artifact of late Tertiary uplift and erosion. 1). Horner and Hasegawa (1978) discuss the spatial and temporal distribution of seismograph stations in western Canada. there are four seismic stations located in Alberta. The catalogue lists the time of a seismic event. The basin is not highly active seismically (Rogers and Horner 1991). Rebollar et al. together with the geomechanical properties of a rock unit. its magnitude. 3. they have not caused alarm in this lightly populated region of Canada. This is important because it is the differential effective stress. Events smaller than magnitude 2 were included only from 1993. Number of events (M > 2) recorded yearly in Alberta. Can. the catalogue lists 419 events with magnitudes up to 5. Earth Sci. Rocky Mountain House. General pattern of seismicity of the Alberta Basin Figure 2 portrays earthquakes that occurred in Alberta prior to March 1996. Currently. and the western Canadian earthquakes are recorded reliably by these stations together with many in British Columbia.50 Fig. and Montana south of the United States border (Smith and Arabasz 1991. which determine its propensity to fracture. 1984) and Wetmiller (1986) attempted to determine the depth of specific earthquakes in the Rocky Mountain House Seismic Zone (RMHSZ). Vol. The depth has not been determined for most events. Montana. Thus. its location. Wyoming. respectively) also increases toward the southwest (Fig. Figure 3 shows the number of seismic events per year for this area. (1984) recorded 220 events between January 1976 and February 1980 with magnitude less than 4 in the Rocky Mountain House area. (1982. Earthquakes of magnitude 4 and greater are rare (Fig. J. The increase in number of earthquakes in recent years cannot be explained fully by the improvement of the seismic network. 1999 50 45 40 35 30 25 20 15 10 5 0 Number of earthquakes recorded 1952 1954 1956 1958 1960 1962 1964 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1966 idence that horizontal stress magnitudes vary laterally within the Western Canada Sedimentary Basin. Seismic events are more or less evenly distributed along the Alberta foreland fold and thrust belt (disturbed belt). The high seismicity of the intermontane belt that passes through Utah. Rebollar et al. and the Northwestern Territories. and Brazeau River. Available seismic data The catalogue of earthquakes used in this study is a subset of the Canadian seismic events database for the last 100 years. the differential stress σ1 – σ 3 (where σ1 and σ 3 are the largest and smallest principal stress. 2). Bell (1996) presented a preliminary interpretation showing a systematic increase in the minimum horizontal stress (SHmin) gradients southwestwards towards the overthrust Rocky Mountain foothills.
5 . The majority of the seismic events are enclosed © 1999 NRC Canada .0 Type of wells gas water injection oil (b) 52.2 115.0 .3 hydrocarbon field boundary Ricinus . The significance of this lag will be discussed later. There is a lag between the commencement of production (1970) and the onset of seismic events (1975–1976). The data on gas production are provided by the Alberta Energy and Utilities Board (1996) for the Strachan.5 115. (b) Approximate location of the hydrocarbon fields of RMHSZ. As can be seen. Horner et al.4 Ferrier A' 52.1 115 115.5 Magnitude of earthquakes 2. high-pressure water injection has been carried out for only two wells at Strachan. 4b portrays all the producing wells (oil. By 1966. It includes Strachan. one in Saskatchewan.5 115. 4.61) between gas-production volumes and number of seismic events for Strachan D3-A gas pool from 1976 to 1995. Between 1976 and 1995 there is a strong correlation between the incidence of seismic events and fluctuations in production. many seismic events were recorded around Ricinus West.5 .1 Caroline 52. Ricinus West. We selected the RMHSZ cluster of events for closer investigation into the possible causes of the seismicity. Ricinus. and an in-house database of the Geological Survey of Canada (Calgary). and Fig. which produces mainly gas (five oil wells out of 36 producing wells). Water injection was carried out at Ferrier. seismic events in the RMHSZ were being registered by more than six seismic stations in Alberta. Phoenix.0 .cross section 126.96.36.199 52 115 For the period 1936–1956 there was one seismic station in Alberta (in Banff). and covers approximately 700 square miles. both of which produce mainly oil. and it is the area with the largest amount of relevant data. but there are no associated seismic events (Fig.2.3 115. Ricinus produces half oil and half gas and is characterized by high seismic activity of variable magnitude in the vicinity both of the gas and oil producing wells. and water injection) in the study area. including focalmechanism solutions and gas production statistics. Figure 5 shows the high correlation coefficient (r = 0. The earthquakes are mostly concentrated within the Strachan field. and Montana. it is the oldest producing zone which is also still active seismically. Statistical analysis Figure 4a shows the seismic events.2 52. No seismic events have been observed around the western part of the Ferrier field or around Caroline. and Ferrier fields.3 Strachan 52.5 52.9 > 4. the western part of Ferrier. British Columbia. the seismic events in the study area are concentrated mainly in the vicinity of gasproducing pools and they are far less abundant near oilproducing areas. 4).4 3. The study encompasses the region between 52° and 52°30′N and 115° and 115°30′W.4 Phoenix 52. As reported by Wetmiller (1986). and Caroline fields (Fig.1 A Ricinus 52 115.4 115. part of Ricinus. 51 Fig. Figure 6 is a contour map of the depths of producing wells which also shows the location and magnitudes of seismic events. (a) 52. showing all the producing wells and cross section A–A′. 4).2 Ricinus West 52.9 3.4 2. which is mainly a gas-producing field.Baranova et al. In addition.3. This choice was made because this is the main area of concentrated seismicity in Alberta. Saskatchewan. gas.A' .name of a hydrocarbon field A .4 115.3 115.2 115. (1994). (a) Seismicity of RMHSZ. and one in Montana.
Ricinus. who employed Biot’s Theory (Rice and Cleary 1976) to show how changes in fluid content within a reservoir can result in modifications to the stress regime in adjacent and underlying rocks and that. There are 646 oil and gas wells within 80 km of Snipe Lake with average depths of 2600 m. and T is the temperature in K. Where possible. It appears that the wells shallower than 2500 m do not play an important role in generating induced seismicity in the Western Canada Sedimentary Basin. unaccompanied by fluid injection. in specific circumstances. The viscosity of gas (η) is taken from McCain (1973) and equated to the viscosity of © 1999 NRC Canada Gas production (mln. 8. of which 67 were located. Analytical approach It is not intuitively obvious why gas production. Slider (1976) derives the gas formation volume factor (Bg) as  Bg = znRT 5615 .5 25 6. 40 Gas production Number of Earthquakes 10 35 8. Over a 3 week period. r = 0. Plotnikova et al. This conundrum has been resolved by Segall (1985). Fluid compressibility for gas (β) is taken from Alberta Energy and Utilities Board (1996) for gas subject to a pressure of 7000 psi. With this relationship.75 10 2. z is a deviation factor. 1989). and Ricinus West (Fig. The Snipe Lake event (Milne 1970) is not shown on Fig. Segall’s approach is followed in this study. The geometry of the field used is shown in Fig. To simulate conditions that might trigger seismicity. the seismic monitoring program run by Wetmiller in 1980 (Wetmiller 1986) demonstrated that the majority of the events were concentrated within a thin horizontal interval approximately 5000 m below the surface (4000 m below sea level) and at an average of 900 m below the Devonian Leduc carbonates from which gas has been extracted at Strachan. 1999 Fig.61 (1976–1995). n is the number of moles (1/379). 8. 6. Relationship between gas production and seismicity for Strachan D3-A pool. we have used measured quantities. The rate of liquid extraction per year for gas is reported (Alberta Energy and Utilities Board 1996) without taking into account the gas formation volume factor. 1975. 36. the virgin reservoir pressure at Strachan (Rose 1990). we have simplified the geometry of the Strachan D-3A pool so as to analyse a linear accumulation. Vol. This is portrayed schematically in Fig. a revised figure for gas extraction that gives the true volume of gas extracted from the reservoir pore spaces was obtained. should trigger earthquakes. Beneath the Lacq gas field in southwest France. these changes can trigger seismicity. m /yr) 3 .25 0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 0 1996 by the 3000 m contour. R is Rhydberg’s constant. Wetmiller (1986) deployed five Sprengnether MEQ-800 smoked-paper seismographs and one Sprengnether DR-100 three-component digital seismograph. In the Gazli and Grozny gas fields in the former Soviet Union which are famous for their devastating induced seismicity (Smirnova et al. J. One hundred and forty six microearthquakes were recorded by the network. Similar well depth – seismicity relationships have been reported elsewhere.75 Number of events (M>2) per year 30 7. 7).52 Can. but it also occurred in an area where the production wells were deeper than 2500 m.5 5 1. induced earthquake epicentres coincide areally with wells drilled deeper than 3200 m (Grasso 1992).0 15 3.895 kPa).25 20 5. but some parameters have had to be estimated. In the study area. Nor does it seem reasonable that these earthquakes should occur beneath the rocks from which fluids were withdrawn. as discussed below. 14 sites were occupied. Epicentres were located in the basal Cambrian sequence and within the Precambrian basement. The material properties used in the calculations are listed in Table 1. the majority of the events occurred in proximity to wells deeper than 3000 m. P where P is the pressure in psi (1 psi = 6. Earth Sci. 5.
9 3. Fig.16 2(r 2 − 1)  where νu is the undrained Poisson’s ratio.40 35 52.4 2. or for any other fields in the Western Canada Sedimentary Basin.30 400 0 0 00 30 00 0 350 450 52. Magnitude of earthquakes 2.0 . The values used are from Rice and Cleary (1976) as estimated for Tennessee marble. Relationship between the seismicity and the depth of producing in RMHSZ.5 .50 00 30 2500 3000 depth of producing wells (m) 0 400 350 0 52.Baranova et al.20 30 00 35 52.2 300 40 00 00 0 115. Diffusivity c is estimated after Segall (1985) as © 1999 NRC Canada . Poisson’s ratios (both drained ν and undrained νu). and density is known. The density of limestone (ρ) is taken from Anderson et al. where the compressional wave velocity Vp = 5500 m/s as reported by Anderson et al. and the reflection coefficient r is derived from νu = (r 2 − 2) = 0. The shear modulus µ is estimated using the equation  µ = ρ Vs2 40 00 115.3. Skempton’s coefficient (B).0 53 52.2.00 115.1 3500 methane at 7000 psi and 124°C.6 115. (1989) and conformed with density log measurements at the 10-31-37-9W5 well. 6.5 115.2.0 .4 115.4 3. Vs is derived from Vp/r.5 . (1989).9 > 4.1 4500 52.3.3 115.0 where the sheer wave velocity Vs = 3200 m/s. and the shear modulus (µ) are not reported for the Strachan field.
J. Ricinus West Strachan 6-25-36-10W5 10-30-37-9W5 6-34-37-9W5 6-14-38-9W5 Ferrier SSW 0 NNE 11-33-35-10W5 9-23-40-9W5 0 500 500 1000 Belly River Gp. Earth Sci. © 1999 NRC Canada .62 × 106 1971 7 0.6 × 10–4 31 × 10–6 0. Average properties of the producing horizon.082 2.100 115 (net pay) 2. ξ at the point of extraction.14 2750 3. Parameter Depth (km) Thickness (m) Average rate of gas extraction (m3/year) Extraction starts (year) Characteristic length along strike (km) Average porosity Permeability (m2) Fluid compressibility (Pa–1) Viscosity (Pa·s) Diffusivity (m2/s) Skempton’s coefficient Poisson’s ratio (undrained) Poisson’s ratio (drained) Density of limestone (kg/m3) Shear modulus (Pa) Symbol D T V — L φ κ β η c B νu ν ρ µ Value 4. and ierfc(x) is the first integral of the complementary error function and is given by the equation  ierfc(x) = ∫ x 0 erfc(ξ)dξ = e− x − x[erfc(x)] π 2 where x. β is the fluid compressibility. Anderson et al. Strachan D-3A pool.46 × 1013 1. The changes in pore-fluid content (M) are calculated as  M (y. Vol.16 0. η is the fluid viscosity. 4).51 0. bmsl.54 Can. T is the thickness. 36. 7. below mean sea level.6 × 10–3 (1. t is the duration in years. y is the distance from the point of extraction. and κ is the permeability. Approximate Location of Earthquake Swarm 4000 Table 1. L is the length along strike. Approximate location of the earthquake swarm is after Wetmiller (1986).y = ζ. 1000 1500 Colorado Group 1500 2000 Depth (m bmsl) Mannville Group 2500 Rundle Group 2000 2500 3000 Wabamun Group 3000 Woodbend Group f Leduc Ree 3500 3500 4000 Cookin g Lake Fm.1989 See text  c= κ ηφβ where φ is the porosity. 1999 Fig.89 × 104 m2/year) 0.367 × 109 References Rose 1990 Alberta Energy and Utilities Board 1996 Alberta Energy and Utilities Board 1996 Alberta Energy and Utilities Board 1996 Rose 1990 Alberta Energy and Utilities Board 1996 Rose 1990 Alberta Energy and Utilities Board 1996 McCain 1973 See text Rice and Cleary 1976 Carmichael 1984 Carmichael 1984 Carmichael 1984. where ζ and ξ are the vertical and horizontal coordinates. Cross section through the RMHSZ (A–A′ in Fig. t) = − y V (t / c) ½ ierfc LTφ 4ct 2 where V is the extracted volume.
55 NW L SE a me e ns al ev el D T y x z Fig. L. length of reservoir along strike.6 -0.Baranova et al. Changes in pore-fluid content in fraction (variable rate of extraction for the Strachan field).5 -0. Fig.9 -1 10 yrs 15 yrs 20 yrs 25 yrs © 1999 NRC Canada .1 -0. 9.7 -0.8 -0. Permeable producing layer of thickness T buried at a depth D in a fluid-infiltrated.t) 5 yrs -0.2 -0.3 -0.1 0 -0. impermeable half-space.4 -1500 -1000 -500 500 1000 1500 2000 pore fluid content m(y. 8. Horizontal distance from the point of extraction (m) -2000 0.
2 t=10 yrs t=15 yrs t=20 yrs -0. 1999 Production rate (×103 m3/year) averaged for 5 years and used in the calculations 2744 (1971–1976) 1930 (1977–1982) 1004 (1983–1988) 640 (1989–1994) 430 (1995–recent) Fig. and the Green functions Gmn are  Gxx = (y − ξ) 2 − (x − ζ) 2 (5x + ξ)(x + ζ) − (y − ξ) 2 + r14 r24 − (x − ζ) 2 − (y − ξ) 2 r14 + ½ ierfc(ξ 2 / 4ct) ½ dξ −∞ D 2 + (y − ξ) 2 ∞ & is the volume of gas where D is the depth in metres. Horizontal distance from the point of extraction (m) -25000 0 -20000 -15000 -10000 -5000 5000 10000 15000 20000 25000 -0. y.3 Subsidence ux is calculated after Segall (1985) as  ux (x = 0. y.05 -0.15 t=5 yrs -0.1 Subsidence (m) -0. The following equations are used to calculate the stress changes σmn below the Strachan field:  σ mn(x. Vol. No. of years (n) 5 10 15 20 25 Production rate (×103 m3/year) averaged from the onset of production 2744 2337 1893 1580 1350 Can. Earth Sci. ξ)ierfc dξ −∞ 4ct . t) = & t 2B(1 + νu )VD c 3πL ×∫ ½ where m. 36. t) = & −µB(1 + νu )V 3π(1 − νu )L t c ½ 16x(x + ζ)(y − ξ) 2 r26  G yy = (x + ζ)(3ζ − x) − 3(y − ξ) 2 r24 − 16x(x + ζ)(y − ξ) 2 r26 © 1999 NRC Canada ∞ ξ2 × ∫ Gmn(x. Subsidence using variable rate of extraction in the Strachan field.56 Table 2. D. n = x. J.25 t=25 yrs -0. y. 10. and V extracted per year. y. Rate of production of Strachan field D-3A pool.
Horizontal distance from the point of extraction (m) -3000 -2000 -1000 -2.4 0.0 1000 0.0 1000 2000 3000 0.0 -0.25 -0.25 -1. 12.25 0. after 5 years (solid lines) and 20 years (broken lines) of gas extraction.0 5000 Depth (m bmsl) 0. -0 2000 -0 .1 8000 © 1999 NRC Canada .2 7000 0.0 -1 0.0 0.5 0.1 0.2 -2000 -1000 2. Incremental changes in magnitude of horizontal stress field (MPa) below Strachan reservoir (D3-A pool) shown on a southwest–northeast vertical plane.3 0. after 5 years (solid lines) and 20 years (broken lines) of gas extraction.0 4000 -0.2 3000 1.4 0. ଙ.0 0. approximate location of the seismic event.1 1.0 0. ଙ. 11.2 0.5 0.1 -0.0 -0.25 5000 Depth (m bmsl) 6000 7000 8000 Fig.5 0.0 -0. 5 . 57 Fig.1 -0.25 4000 .0 0. Incremental changes in magnitude of vertical stress field (MPa) below Strachan reservoir (D3-A pool) shown on a southwest– northeast vertical plane.25 0.1 0.Baranova et al.3 -0. Horizontal distance (m) from the point of extraction -3000 -0.3 -0 .3 6000 0. approximate location of the seismic event.0 0 0.
According to Maury et al. (1992).04 0. the subsidence could reach 25 mm/year. The parameters used in the calculation are the same as those used previously. However. 20. when the rate of extraction was the highest (see Table 2).06 6000 0. As mentioned above. approximate location of the seismic event. 13. J. Incremental changes in magnitude of mean stress field (MPa) below Strachan reservoir (D3-A pool) shown on a southwest– northeast vertical plane. 10. For the first 5 years of gas production. 1999 Fig. after 5 years (solid lines) and 20 years (broken lines) of gas extraction. The above theory is also used to calculate the subsidence which might have occurred above the Strachan field as a result of gas extraction.1 mm/year. 00 0.14 2000 3000 0. . The parameters used for the calculations are given in Table 1. decreasing to 11 mm/year after 25 years of gas production. . Unfortunately. the average rate of subsidence of the Lacq gas field in France for 20 years of extraction is 2. we cannot compare calculated results with observed ones. although the production rate is much lower than that at Strachan.16 0. ଙ. 15. Pore-fluid content changes and subsidence due to gas production Figure 9 shows the changes in pore-fluid content calculated from eq. Earth Sci. Changes of stress fields below Strachan field because of production Changes in vertical (σ xx ) and horizontal (σ yy ) stresses below the Strachan D-3A pool are calculated according to eq. Fluid extraction will be associated with reservoir compaction and surficial subsidence above the reservoir.02 . and 25 years).58 Can.02 7000 8000  Gxy = − 2(y − ξ)(x − ζ) 2(y − ξ)(3x + ζ) − r14 r24 − 16x(x + ζ) 2 (y − ξ) r26 where r12 = (x − ζ) 2 + (y − ξ) 2 and r22 = (x + ζ) 2 + (y − ζ) 2 .04 Depth (m bmsl) 0. Vol. The changes in stresses are plotted in the x–y plane which is perpendicular to the strike of the pool and is ap© 1999 NRC Canada 0. Figure 10 shows the calculated rate of subsidence due to variable rate of extraction (for 5. The most recent map of vertical crustal movements compiled by the Geological Survey of Canada (Vanicek and Nagy 1980) shows negligible movements close to 0 cm/year in the vicinity of RMHSZ due to lack of resolution. The rate of extraction is used averaged for each 5 year increment of production (Table 2):  M total = M n− 5 + (1 − M n− 5)M n where Mtotal is the net change of pore-fluid content after n years of production.08 1000 0. it is worth noting that Segall’s (1985) calculated results coincide well with the observed ones and give 3–6 mm/year due to oil extraction.06 4000 0. Figure 9 shows that nearly 70% of fluid content was extracted for the first 5 years of production.10 5000 0. 36.08 0. It is clear that the subsidence increases linearly with the rate of fluid extraction and depends on hydraulic diffusivity. we are unable to relate the calculated results to observations due to lack of data. Horizontal distance from the point of extraction (m) -3000 -2000 -1000 0.
1 3000 2.2 -0. Calculations were made for 5 and 20 years after extraction started in 1971. approximate location of the seismic event. Lower hemisphere equal area projection of the best fitting double-couple solution of the focus for a seismic event (M = 3. 13) and the deviatoric stress changes [(σ xx – σ yy )/2] (Fig.1 MPa) as compared with the deviatoric stresses. This yielded two solutions. The vertical and horizontal stresses are resolved into normal. Horner. Horner. 15).1 6000 0.2 0.B. N SHmax © 1999 NRC Canada .5 0.3 0.0 km below mean sea level (R. In this paper the positive stresses are always treated as compressions. and driving stresses on the two planes and are plotted in Figs. Horizontal distance (m) from the point of extraction -3000 -2000 -0.3 0. The above stress changes are then applied to the focalmechanism solution of the 3. which are of the order of 1 MPa. It is interesting to note that the strike of the D3-A pool is almost the same as the strike of the fault planes suggested by the focal-mechanism solution (northwest–southeast 302 and 336).0 1000 2000 -0. The mean stress [(σ xx + σ yy )/2] (Fig. These stresses represent the changes of magnitude in the existing stress state due to gas production from the Strachan D3-A pool. 14) are plotted on the same x–y plane. Fig. 115°13′W).9 magnitude RMHSZ earthquake that occurred on 19 October 1996 (R.0 1.2 -0. shear. after 5 years (solid lines) and 20 years (broken lines) of gas extraction. 59 Fig. 16 and 17. 11. -0. personal communication. 1997).0 1. 2 -1000 2.Baranova et al.9). Incremental changes in magnitude of deviatoric stress field (MPa) below Strachan reservoir (D3-A pool) shown on a southwest–northeast vertical plane. 12) are positive (compression). 14.5 0. which occurred at an estimated depth of 4. at an approximate depth of 4 km below sea level (Fig.1 1 -0. which is 1 km below the Strachan D3-A pool.2 0. 0 0. dated 19 October 1996 at Rocky Mountain House.4 0.4 0. NP1 302/51° northeast and NP2 156/44° southwest. 0 0.0 4000 -0. This phenomenon can be explained by the fact that due to extraction the changes in vertical stress are negative (dilatation).1 0. It is notable that the changes in mean stress are very small (0.0 0 Depth (m bmsl) 5000 0. whereas the changes in horizontal stress (see Fig.1 7000 8000 proximately parallel to the direction of the SHmax (Figs.2 -0. 1997). 15. ଙ. 12). personal communication. 0. Alberta (52°13′N.
normal (b). 1999 Fig.5 Normal stress (MPa) 0. respectively. 16c and 17c.6) is used. Change in pore pressure is calculated as  ∆p = (1 + νu B) ∆σ nn 3 where ∆σ nn is the change in mean stress. Incremental changes in magnitude of shear (a). Ricinus. and driving (c) stresses with depth at a plane dipping 51° northeast below the Strachan field after 5 years (solid lines) and 20 years (broken lines) of gas production. 17.and northeast-dipping planes are shown in Figs. it decreases from 4. Earth Sci. 16. 17a).5 MPa right below the reservoir to approximately 1 MPa at 4 km below sea level. f is the coefficient of friction. The positive sign of shear stress on both planes favours reverse faulting up to 8 km below the Strachan field (Figs. Figure 19 shows changes of SHmin with depth based on hydraulic fractures run in Mesozoic rocks in the Caroline. A value of 120 MPa was assigned for SHmin at the depth of earthquake occurrence (4000 m below mean sea level). as shown in Figs. normal (b). Since we do not know the exact value of the coefficient of friction for the rocks in which failure occurred. Changes in magnitude of normal stress at both planes are relatively low. The changes in magnitudes of normal stress on both planes decrease with depth.60 Fig. (a) Shear stress (MPa) Can. 36.0 0. Vol. The changes in driving stress on the southwest. respectively. and Strachan fields (McLellan 1989). J. On the southwest-dipping plane it represents compression. (a) 6 6 5 4 3 2 1 0 3000 Shear stress (MPa) 5 4 3 2 1 0 4000 5000 6000 Depth (m bmsl) 7000 8000 3000 4000 5000 6000 7000 8000 Depth (m bmsl) (b) (b) 0. and ∆p is the change in pore pressure. the depth of the seismic event. The proposed mechanism of failure suggests that the overburden pressure (vertical stress) is decreasing while © 1999 NRC Canada . and driving (c) stresses with depth at a plane dipping 44° southwest below the Strachan field after 5 years (solid lines) and 20 years (broken lines) of gas production. the most common value (0. Note that the changes in the magnitudes of shear stress are quite high.1 -0.0 3000 0 3000 4000 5000 6000 Depth (m bmsl) 7000 8000 4000 5000 6000 7000 8000 (c) (c) Depth (m bmsl) 6 5 6 5 4 3 2 1 0 3000 4000 5000 6000 7000 8000 Driving stress (MPa) Driving stress (MPa) 4 3 2 1 0 3000 4000 5000 6000 7000 8000 Depth (m bmsl) Depth (m bmsl) The change in driving stress (∆σ d ) acting to cause displacement is calculated following Segall (1985) as  ∆σ d = ∆σ s + f (∆σ n + ∆p) where ∆σ s and ∆σ n are the changes in shear and normal stresses. Incremental changes in magnitude of shear (a). 16b and 17b. Note that a change of only 1 MPa in stress is apparently sufficient to produce the failure under the recent rate of extraction (Fig. Changes in driving stresses after 5 and 20 years of gas production favour reverse faulting and show the sharp decrease in magnitudes from 4 MPa right below the reservoir up to almost nothing at a depth of 6 km. 16a. 18). whereas on northeast-dipping plane it represents dilatation. Figure 18 depicts the inferred Mohr–Coulomb circle and failure line before and after the gas extraction.2 Normal stress (MPa) 0.5 -1.3 0.
We believe that this period was required to accumulate the stresses needed to initiate movements on the faults in a “disturbed zone” (in our case) below the reservoir. most seismic events at Lacq occur above or within the reservoir (Grasso and Wittlinger 1990). the failure line in Mohr–Coulomb criteria comes closer to the circle because of lowering of the coefficient of friction of the faults. (1994). vertical stress and effective vertical stress. it can explain the lack of correlation between the rate of production and seismicity of the Strachan field during the 5–7 years of extraction. There is no similarity in the relative depths of seismic events in the different gas production areas. as well as in Turner Valley. respectively. It is interesting to note that. the onset of earthquake activity is very similar and started 5–10 years after gas extraction began. the proximity to the disturbed belt and the instability of the state of stress are two factors that play important roles in generating the seismicity in the area. although there are examples of seismic events caused by water injection in western Canada. 1975). Once the faults are activated. 1989) gas production areas. Since the stress regime and focal mechanism favour thrust faulting. In the case of RMHSZ. We have shown that the rate of extraction (especially during the first 5 years for the Strachan field) can give rise to changes of stress magnitudes great enough to trigger earthquakes on reactivated faults. τ. We suspect that seismic events in Snipe lake. SHmax′. However. At Strachan.Baranova et al. Lacq (Grasso and Wittlinger 1990). sheer stress. There is no correlation between water injection and occurrence of earthquakes in the study area. earthquakes in the Fort St. were also caused by water injection. There are several observations which strongly suggest that the earthquakes that occurred in and around the Strachan field were triggered by gas extraction. © 1999 NRC Canada . the correlation increases dramatically after fault movements begin to occur below the reservoir. John oil production area have been caused by massive water injection and occurred immediately after commencement of injection. However. reported seismic events occurred beneath the reservoir (Wetmiller 1986). and Grozny (Plotnikova et al. The regional state of stress in the RMHSZ is close to the condition at which earthquakes could occur (Bell 1996). (1) These seismic events occurred in an area of naturally low seismicity. According to Horner et al. Sv and Sv′. Schematic representation of the Mohr–Coulomb failure criterion applied to the stress regime in the Strachan area before extraction (circles formed using solid lines) and after extraction (circles formed using broken lines). So. The significance of this simulation is explained in the text. so small stress and strain changes could trigger seismic instabilities. maximun horizontal effective stress. and Strachan fields.6 σn 20 10 0 0 20 Sv' Sv 40 SHmin 60 80 100 Normal stress (MPa) the normal stress is increasing due to extraction. but high tectonic stresses. Gazli (Smirnova et al. 18. 61 Fig. 50 40 Shear stress (MPa) 30 τ = 0. The lack of seismic events associated with it allows us to conclude that the volume of water injected simply was not enough to cause stress release. in the case of Strachan (this paper). This can explain the modest changes in mean stress which have the effect of increasing the radius of the Mohr’s circle until it crosses the failure threshold. Ricinus West. Water injection in the study area was carried out in several wells in the Ferrier. we assume that failure occurred within a preexisting low-angle fault zone.
Blackwell. J. The proposed mechanism suggests decreasing pore pressure (vertical stress) and increasing normal stress so that changes of mean stresses are small and failure on preexisting faults can be initiated by increasing deviatoric stresses. Slemmons. (7) Changes of stress due to extraction indicate a regime that favours reverse faulting and coincides well with the natural state of stress for this area. Ricinus. Alberta Energy and Utilities Board. 1999 Fig. We are most grateful for the helpful comments and suggestions of the reviewers.R. natural gas: statistics 1965–1996. Calgary. 54: 1410–1419. J. Zoback. In Neotectonics of North America. 367–385. The Geology of North America. Roest and G..C. Geological Society of America. Alta. Stress magnitude (MPa) 0 0 500 1000 True vertical depth bmsl (m) 20 40 60 80 100 120 140 1500 2000 ap r app pr 2500 3000 3500 4000 4500 . 19. J.D. Decade Map Volume 1.S. gas.J. Horner for his helpful comments and suggestions. Alberta reserves of crude oil. Low. 1996. and D.62 Can. SHmin. Vol. N. Earth Sci. Alberta Energy and Utilities Board. oil sands.S ox (2) There is spatial and temporal correlation between earthquake occurrence and the rate of gas production. and Sv gradients estimated from the hydraulic fractures in the Caroline. and for providing the focalmechanism solution and seismic events database. Edited by D. Engdahl. Drs.D. R.and high-relief Leduc formation reefs: a seismic analysis. p ore su res re causes of triggering of the recent seismicity beneath the Strachan field. pp.L. (5) There is a linear correlation between depth of production and the number of seismic events. Atkinson. and Strachan fields. (6) Application of the model developed by Segall (1985) shows that the 19 October 1996 earthquake below the D3-A pool of the Strachan field could have been triggered by gas extraction. Adams. 1989. This research was supported by the Geological Survey of Canada. 36. Anderson. the high rate and long history of gas extraction can be considered as probable p ox. R. Calgary.B. (3) There is no apparent correlation noted between extraction of oil and the number of seismic events. and Bell. which also would have caused subsidence and stress changes. Geophysics. Brown. E. We speculate that the difference in density between gas and oil can account for this.. 1991. Approximate pore-pressure. Crustal stresses in Canada.. J. We thank R. (8) As a consequence of the above.B. © 1999 NRC Canada p ap x ro . and Hinds. M. (4) There is no relationship between water injection and earthquakes for the study area.S v Hm in .
Healy.Baranova et al.. D. M. A review of theories of mechanisms of induced seismicity.B.. Davis. Davis. Jr. Tulsa. Triggered earthquakes. J. Mossop. Horner.R. Nyffenegger. An overview of western Canadian seismicity. R. 1981. Engineering Geology.E. 1989. Pomeroy. Earthquakes and water flooding in Rangely oil field.. The Canadian Society of Petroleum Geologists. Rose. Milne. 1994. J. Geological atlas of the Western Canada Sedimentary Basin. G. Smirnova. Geological Survey of Canada..W. Strength changes due to reservoirinduced pore pressure and stresses and application to Lake Oroville. Induced seismicity in gas field region. 1975.. Load-induced earthquakes at Lake Kariba — II. The Snipe Lake. sequence of June 1974. 9–46. 1991. Annual Review of Earth and Planetary Sciences. W. Relationship of mine tremors to induced stresses and to rock properties in the focal region. M. D. Gupta. E.. 1993. 1976. In Neotectonics of North America. and Sbar. 1984. 1989. Journal of Geophysical Research.. Stress and subsidence resulting from subsurface fluid withdrawal in the epicentral region of the 1983 Coalinga earthquake. M. The seismotectonics of southern Saskatchewan. Handbook of physical properties of rocks. 1986.H. Earthquakes and hydrocarbon production in the Fort St. Brookfield. J.J. and Machmudova. 1984. Segall. Canada. and Gay. Ten years of seismic monitoring over a gas field. P. Gough.D. M. E..C. The properties of petroleum fluids. 21: 1105–1113. The Denver earthquakes. 14: 21–42.. Bell.. Moscow.B.G.B. J.R. 1979. Boca Raton.. Rogers.. 1979. E. Rice. A. B9: 6801–6816. The Canadian Society of Petroleum Geologists oil and gas pools of Canada. Induced seismicity. R. E.R. W.C. Canadian Journal Earth Sciences. Segall. B9: 4469–4483. J. 30: 39–50.R. Open File 1947. evolution.B. 1995. I.R. and depositional cycles of the Western Canada Foreland Basin. Canadian Journal of Earth Sciences. 1995.R.).. 6: 1564–1567. 86(B2): 903–920. Kanasewich. G. S.G. Kisslinger. Grasso. 17: 942–946. 91–117. Induced seismicity in Canada.. McLellan.J.A. L.D.J. 1973. P. Chapt..H. D. Grasso. 69–76. Northeast–southwest compressive stress in Alberta: evidence from oil wells. 19: 466–475. 1978.L. Seismological Research Letters.. V. and Hasegawa. McCain. and Wittlinger. Engineering Geology. D. D. 1970. Plotnikova. J. 64: 207–224.A. Geoscience Canada. Rubey. C. 1982.S.D. Earthquake swarm caused by intensive exploitation of an oil field. 1990..B. 1972. 1982. pp.. In Proceedings of the 8th International Association of © 1999 NRC Canada . and MacRae. P. Spottiswoode. 41: 20–29.I. CRC Press. and Shetsen. A reservoir analysis of the Denver earthquake: a case of induced seismicity. N.M. Edmonton.P. earthquake of March 8. M. Alberta. Knoll. and Berry. Journal of Geophysical Research. Geology. and Nur. J. G. G. 15: 1341–1355. Source parameters from shallow events in the Rocky Mountain House earthquake swarm. 1976. 80: 450–473. and Bredehoeft.. Okla.L. 45: 475–482. 4. pp. 1990. H. 1970. C. In Proceedings of the 15th International Association of Seismology and Physics of the Earth Interior General Assembly. H. Series 1. 1985. 161: 1301–1310. Alta. M. 66: 685–700. N. C. J. and Nikolaev. R. Gendzwill.D. Slider. Partnership Sub-Series 2.M. 1989. John area of northeastern British Columbis. and Raleigh. P. Bell. Earthquakes triggered by surface quarrying — the Wappinger Falls.. Canadian Journal of Earth Sciences. NATO ASI Series. Did (or will) fluid injection cause earthquakes? — criteria for a rational assessment.D.A.S. D. N. Dams and earthquakes. Monitoring of subsidence and induced seismicity in the Lacq gas field 63 (France): the consequences on gas production and field operation.R. Engineering Geology. 1976. Old Post Road. McGarr. and Hasegawa. (Editors). Bulletin of the Seismological Society of America. 31– 32.J.. Console.S. and Frohlich. Horner. Carmichael. I–III. Earth and Planetary Science Letters. 1992. Rebollar. Calgary. J. American Association of Petroleum Geologists (AAPG). Bulletin of the Seismological Society of America. Horner. 1978. 1976. Proceedings of intergovernmental conference on the assessment of earthquake risk. 1992. Paris. P. 1970.) Healy.I.. In Assessment and mitigation of earthquake risk. Balkema Publishers. (Editor). R.N. D. Zoback. Canadian Journal of Earth Sciences. (Editor). 1994. D.. Petroleum Publishing Company. and Frohlich. Petroleum Publication Co. E.B.. Induced earthquakes at a potash mine near Saskatoon. 85: 1888–1895. 1968. M. Environment. B.J. W. and Rastogi.B.W. 139: 507–534. Bulletin of the Seismological Society of America.W. and Nyland.S. and D. Memoir 55.A. 1. (Compilers). The 9 April 1993 earthquake in South-Central Texas: was it induced by fluid withdrawal? Bulletin of the Seismological Society of America.. Fla. Focal depth and source parameters of the Rocky Mountain House earthquake swarm from digital data at Edmonton.. G. Alta. and Cleary. Griggs. (In Russian. and Gough.. Grasso. Canadian Journal of Exploration Geophysics. H.. 14: 227–241.S. Induced seismicity. Petroleum reservoir engineering methods. Hsien. 1996. Edited by D... Earthquakes triggered by fluid extraction. C. R.L. A. Geophysical Journal of the Royal Astronomical Society. 23: 135– 153.R. Blackwell.A.C. and Bredehoeft. Vols. Slemmons.. Bell... Science (Washington. S. Leckie. J. Simpson. J. 65: 981–993.I..T.W. Canadian Journal of Earth Sciences. Geological Society of America. Kanasewich. Mechanics of seismic instabilities induced by the recovery of hydrocarbons.. Gough.C. 1992. A. In Foreland basins and fold belts. In-situ stress magnitudes from hydraulic fracturing treatment records: a feasibility study. 1975. 1976. Canadian Society of Petroleum Geologists and Alberta Research Council. In situ stresses in sedimentary rocks (part 2): applications of stress measurements. MIR. Istanbul.S. pp. Vt. D. Navitskaya.. C. Raleigh. New York. Earthquake Notes. (Editor). The Geology of North America. P. A.G. and Horner. Okla.G. C. Flynova. Smith. 10: 85–98. Simpson. pp. 32: 123–135. M. Decade Map Volume 1. Reviews of Geophysics and Space Physics. and Shebalin. August 1989. Journal of Geophysical Research. 19: 907– 918. and Nyland. 1978.M. Engdahl. 21: 79– 101..M. Earthquakes induced by underground nuclear explosions.. C. Rebollar. Pure and Applied Geophysics. W. and Wittlinger. Maury. D.V.I. 1992. Tulsa. V. J. Some basic stress diffusion solution for fluid saturated elastic porous media with compressible constituents. 10: 219–226. P.. Barclay. H. UNESCO. Milne. Bohn. Regional setting. J... R.
Earth Sci. and Nagy. In Neotectonics of North America.J. 1999 Service of Canada. Slemmons. 1991. M. Earthquakes near Rocky Mountain House. Yerkes. Seismicity and faulting attributable to fluid extraction. D.D. The Geology of North America. Zoback. p. Geological Society of America. International Union of Geodesy and Geophysics. Blackwell. and Castle. R.B. Seismicity of the Intermountain Seismic Belt. 39. Vol.. J.. Alberta.. W.O.64 Seismology and Physics of the Earth Interior General Assembly.F.J. Decade Map Volume 1. 1986. Canadian Journal of Earth Sciences. 23: 172–181. 10: 151–167. Wetmiller. Geodynamics Can. 36. and Arabasz. P. © 1999 NRC Canada . 1976.D. Report on the compilation of the map of vertical crustal movements in Canada. R. Moscow. R. Vanicek. Earth Physics Branch. Abstracts. and their relationship to gas production facilities. R. Open File Report 802.B.R. Engineering Geology. Edited by D. Engdahl. E. Smith. and D. 1980.