In Situ Stress State of the Ruhr Region (Germany) and Its Implications for Permeability Anisotropy

In this study, we carried out reactivation potential analysis of discontinuities revealed from four exploration boreholes penetrating heavily faulted and folded Upper Carboniferous rock strata of the Ruhr region. We performed this study based on the notion that slip is controlled by the ratio of shear to effective normal stresses acting on a pre-existing plane of weakness in the prevailing stress field configuration. The results of this analysis were supported by indicators of localized fluid flow, both on micro- and macro-scales, which confirm relationship between secondary permeability and in situ stress state in the Ruhr region. Findings from this study, in conjunction with results of destructive laboratory testing, indicate that the steep NW–SE- and NNE–SSW-striking planar discontinuities are likely to be either close to the critical state or critically stressed in the in situ stress configuration in the Ruhr region. These planar structures, as evidenced by indicators of localized permeability, are the main fluid pathways in the studied region. The NE–SW-striking discontinuities, on the other hand, are most likely to be closed and hydraulically inactive in the prevailing stress state. Based on results gained from this study, implications for utilization of deep geothermal energy in the region were discussed.


Introduction
It is generally accepted that fractures and faults are the main fluid conduits in the Earth's crust (Hickman et al. 1995). The in situ stress tensor is one of the major controls on the reservoir permeability anisotropy, where fractures and faults (regardless of scale) with high shear to normal stress ratios (i.e., ones having high reactivation potential) are considered as primary fluid pathways in both crystalline and sedimentary formations (Barton et al. 1995;Ito and Zoback 2000;Moeck et al. 2009;Zoback 2007). The presence of these, socalled, critically stressed planes makes the permeability of crust about four times larger than for a case of an intact rock subjected to appropriate confining pressures (Townend and Zoback 2000). An understanding of reactivation potential of fractures and faults is a critical issue in the development of man-made, and especially conduction-dominated, geothermal systems. First, the success of such a system relies on the number of intersected permeable fractures and/or faults oriented favorably to the in situ stress tensor (Zoback 2007). Second, if a fault cutting a reservoir has high effective normal stresses acting on its plane (i.e., sealing fault), it can isolate fluid flow and compartmentalize a reservoir. On the contrary, fault with high dilational stresses may induce fluid leakage during drilling operations. Third, knowledge about reactivation potential of fractures and faults is important in cases where rock matrix permeability is insufficiently low and stimulation treatments are necessary to enhance it. Fracture initiation, in case of hydrofracturing, or reactivation, 1 3 in case of hydro-shearing, may be needed to generate flow paths and enhance reservoir productivity. Significant pore pressure disturbances during fluid injections, not adjusted to the prevailing in situ stress tensor and rock frictional properties, may reduce the frictional strength of faults and induce their either aseismic or, if the stress conditions will be sufficiently altered over a large area, coseismic shear slip. As a result, before an establishment of any deep geothermal system and carrying out subsequent stimulation efforts, in situ stress tensor and its effect on the fracture and fault network shall be well understood.
The Ruhr region is located in the heart of the North-Rhine Westphalia state in western Germany. It is one of the largest metropolitan areas of Europe with more than 5 million inhabitants extending over 130 km from west to east and 35 km from south to north, covering an area of ∼ 4600 km 2 between the valleys of the Ruhr, Emscher, Lippe, and Rhine rivers. The Ruhr district heating network supplies 6500 GWhyear −1 with an installed thermal capacity of 2310 MWth (Bartelt et al. 2013). The wider metropolitan region of Rhine-Ruhr is one of the largest district heating networks in the world, fed primarily by fossil fuels, i.e., coal and natural gas (Klaus et al. 2010). To meet national and international climate protection targets, this network is to be converted into a CO 2 -free energy network in the forthcoming decades (Wegener et al. 2019). The Ruhr region is defined by the rich coal-bearing layers of the Upper Carboniferous period, belonging to the sub-Variscan Trough, with hard coal extraction starting in the thirteenth century. In 2018, after more than 700 years of extensive hard coal mining, the last mine was decommissioned in the region. The conversion of the district heating system in the Ruhr region could rely on the use of its vast geothermal potential. Favorable conditions for establishing deep geothermal reservoirs may exist in the fractured sandstones of the Carboniferous age (Hahne and Schmidt 1982) and karstified and/or dolomitized carbonate rocks of the Devonian age (Przybycin et al. 2017;Balcewicz et al. 2021). Devonian carbonates were intersected at depths between ∼ 4500 and ∼ 6000 m in several deep exploration boreholes, north from the Ruhr region (Eder et al. 1983;Drozdzewski 1993;Hesemann 1965), and are outcropping south of the Ruhr region in the vicinity of Wuppertal, Schwelm, Hagen, and Iserlohn (Balcewicz et al. 2021;Meschede and Warr 2019) (Fig. 1). Based on the temperature profiles from exploration boreholes in the northern part of the Ruhr region gathered in this study (Fig. 2), the deepest borehole in the region (i.e., Münsterland 1) (Hesemann 1965), and temperature measurements from local coal mines (Wedewardt 1995), the average geothermal gradient amounted to ∼ 36 • C km −1 . As a consequence, for reservoir depths between ∼ 4500 and ∼ 6000 m , temperatures of ∼ 170 up to ∼ 230 • C can be expected (assuming temperature of 10 • C at the surface). Such thermal conditions may enable not only heat but also electricity production (Hettiarachchi et al. 2007) if sufficient permeability exists in situ. Due to the well-developed industrial and private infrastructure in the Ruhr region, there is a great interest in utilizing geothermal energy sources, being it either mine energy storage (e.g., Hahn et al. 2017) or deep geothermal systems (e.g., Balcewicz et al. 2021), and simultaneously a strong necessity for understanding and monitoring possible movements on major discontinuities and related seismic risks.
The main aim of this study is to investigate the fracture behavior under present-day stress tensor of the Upper Carboniferous rock strata of the Ruhr region and assess its effect on fluid flow. To do this, we use the slip tendency analysis, where we resolve shear and dilational stresses along preexisting planar features, localized with borehole logging methods, and evaluate their slip and dilation potentials. Results of this analysis were compared with indicators of localized fluid flow based on observations of fluid loss during drilling operations, high-quality temperature log, and macro-scale observations of fault leakage. We re-evaluated available in situ stress data, coming from the coal mining  (DEKORP 1990) with major faults (marked in red) and the Münsterland 1 well (marked with a black triangle) intersecting the top of Devonian layers (after Drozdzewski 1988) industry, across the Ruhr region and derived linearized pore pressure and in situ stress gradients. We used results from destructive laboratory testing and field measurements to constrain the frictional properties of fractured rocks within the Ruhr region. Using the slip tendency analysis, the potential for shear slip along pre-existing discontinuities, from four exploration boreholes, with respect to the in situ stress tensor was investigated and reactivation potential of the penetrated planar features assessed. Based on results gained from this study, implications for utilization of deep geothermal energy in the region were discussed.

Geological Setting
The coal-bearing strata of the Ruhr region belong to the external fold and thrust belt of the latest stage of the Variscan orogen, located east of the Wales-Brabant Massif (Brix et al. 1988). The Variscan orogenic belt was formed during the Late Paleozoic convergence of the Euramerican and Gondwanaland continental masses (Ziegler 1990). The collision and convergence took place during the Devonian and Carboniferous geologic periods.
The Ruhr region is influenced by two main fault systems. The major faults strike in the NE-SW direction with dips ranging from steeply inclined to bed-parallel reaching lengths of up to 40 km. These reverse type faults have horizontal displacements in the order of tens to hundreds of meters with few reaching horizontal displacements of ∼ 2500 m . The reverse faults are dissected by multiple NW-SE oriented normal faults resulting in a horst-andgraben structure with few strike-slip faults of various orientations also being observed in the region (Brix et al. 1988). Folds, mainly NE-SW oriented, vary significantly in shape and dimension with wavelengths of up to 10 km, increasing towards the north, having amplitudes in the order of several hundred meters.
Extensive hard coal mining activities exposed thick molasse-type clastic sediments of the Upper Carboniferous age ranging from shales, silt-to coarse-grained sandstones with 2-2.5 m thick coal seams of varied strength, all heavily deformed by thrusting and folding (Bachmann et al. 1971). North of the Ruhr region, thick, in places up to ∼ 1800 m , Cretaceous strata overlays the Upper Carboniferous sequences (Drozdzewski 1993). Four deep exploratory boreholes, located north of the Ruhr region i.e., Münsterland 1 (Hesemann 1965), Versold 1, Isselburg 3 (Drozdzewski 1993), and Vingerhoets 93 (Eder et al. 1983) have all reached Devonian units. Carbonate layers of the Middle and Upper Devonian age, being part of the Devonian Reef Complex, are outcropping south of the Ruhr region close to the city of Iserlohn (Fig. 1). Based on the study by Balcewicz et al. (2021) from outcropping Devonian carbonate strata in Wuppertal, Hagen, and Honnetal, similar strike and dip of discontinuities to the ones from the Upper Carboniferous strata was observed, which could potentially indicate lack of significant in situ stress decoupling at depth.
As for today, there is, however, no direct proof (i.e., confirmation from exploratory drilling) of Devonian carbonates underlying the coal-bearing Carboniferous layers in the Ruhr region. Nonetheless, the interpretation of the DEKORP 2-N seismic line (DEKORP 1990) indicates strong reflections, corresponding to the high material contrasts, at approximately 5000 m depth (Fig. 1). These reflections, according The stress map of the Ruhr region with major fault systems (after NRW 2021), coal mines, boreholes investigated in this study, and borehole breakouts from Reinecker et al. (2008) (with increasing length of the line indicating higher quality of a stress indicator in accordance with Heidbach et al. (2018)) to Franke et al. (1990), Drozdzewski (1988), andDrozdzewski (1993), are thought to be Devonian platform carbonates encountered in deep boreholes north of the region.
Based on laboratory studies carried out on rock samples from the Ruhr region it can be concluded that low matrix permeability of < 10 −18 m 2 characterizes sandstones of the Upper Carboniferous age (Stöckhert 2015;Nehler et al. 2016;Brenne 2016;Alber and Meier 2020;Stoeckhert et al. 2020;Duda and Renner 2013) and permeability of < 10 −15 m 2 characterizes Devonian limestones and dolostones (Balcewicz et al. 2021). Matrix porosity of both rock formations does not exceed, on average, 5-7 % (Thielemann et al. 2001;Jorand et al. 2015;Stoeckhert et al. 2020) with local increases of permeability being connected to the weaker coal layers (Alber and Meier 2020) especially those induced by mining activities (Thielemann et al. 2001).

Past Geomechanical Studies
The in situ stress state of the Ruhr region was investigated in several geomechanical surveys, performed predominantly in it's northern parts with hydrofracturing and overcoring methods. From the 1970s to the mid-1990s, in situ tests were carried out primarily to optimize the design of crosscuts and mine layouts or evaluate stress state in coal bed methane wells (Braunher 1976;Rummel et al. 1993;Müller 1989;Muller et al. 1991;Stelling and Rummel 1992). Most of the measurements were limited to depths of interest for the coal mining industry (i.e., ∼ 1500 m ). Based on results from these surveys, magnitudes of minor, S hmin , and major, S Hmax , horizontal stress were acquired.
In the last 35 years, more than 1000 exploration boreholes were drilled in the northern part of the Ruhr region. In all of them, comprehensive logging programs were performed (Rudolph et al. 2010). This immense database allowed detailed analysis of borehole elongations and, as a result, estimation of S Hmax orientation, SHmax . A total length of 761 m of borehole breakouts from 3950 m of analyzed four-arm caliper logs in 40 boreholes spread across the Ruhr region were found by Müller (1989). The mean SHmax from borehole breakout from his analysis amounted to 149 ± 36 • . Reinecker et al. (2008), carrying out a similar study, observed borehole breakouts in 51 boreholes in the Ruhr region with a total length of breakout intervals of 3014 m. The length-weighted SHmax indicated by Reinecker et al. (2008) amounted to 168 ± 48 • , whereas the numberweighted SHmax to 160 ± 40 • .
Based on information from above-mentioned studies, Ruhr region can be separated into two main stress units in terms of SHmax with the boundary laying at the General Blumenthal mine (Fig. 2). The eastern part of the region is characterized by a transition from an NNW-SSE SHmax in its easternmost parts (i.e., mine Westfalen) to an NW-SE orientation in the westernmost parts (i.e., Haus Aden and Heinrich-Robert mines), which accordingly to Rummel et al. (1993) is due to the influence of regional faulting of the coal-bearing layers. The western part is characterized by SHmax ranging from the N-S in its central part (i.e., Lohberg and General Blumenthal mines), and reorientation from NNW-SSE to NW-SE in its most western parts (i.e., Friedrich Heinrich and Niederberg mines), which accordingly to Rummel et al. (1993) is due to the influence of the graben structure of the Lower Rhine Bay.

In Situ Stress State
The information about the magnitudes of the principal in situ stresses and pore pressure, P p , in the Ruhr region, collected throughout this study, was based on re-evaluated data from Wedewardt (1995), Braunher (1976), Rummel et al. (1993), Müller (1989), Muller et al. (1991), Stelling and Rummel (1992), unpublished reports by Solexperts GmbH (Rummel and Weber 1994) as well as on results from drilling operations and geophysical logging from 17 wells in the Ruhr, and adjacent, regions within the Carboniferous layers acquired in this study (well locations presented in Fig. 2). The mentioned dataset included (i) 115 measurements of S hmin from which depth was known for 80 measurements, (ii) 80 measurements of S Hmax from which depth was known for 49 measurements, (iii) 17 measurements of vertical stress, S v , based on geophysical logging, and (iv) pore pressure magnitudes acquired from 233 measurements of the density of water sampled from the local coal mines between 20 and 1470 m depth and fluid densities applied during drilling operations in the studied boreholes ( Fig. 2).

Slip and Dilation Tendency Analysis
The fluid flow along any discontinuity depends primarily on the in situ stress tensor and its positioning in the subsurface (Barton et al. 1995). Once the in situ stress tensor is well understood, acting shear, , and effective normal, n , stresses along any arbitrarily oriented planar feature can be resolved (Jaeger et al. 2007). Subsequently, in accordance with the Amonton's law, slip tendency, T s , defined as a ratio between and n , can be assessed (Morris et al. 1996) Discontinuities with T s approaching or exceeding frictional resistance for sliding, , are assumed to have increased likelihood for a shear movement The majority of discontinuity surfaces are not perfectly planar, but rather rough and undulating. In its undisturbed initial state, discontinuity surfaces mate to create the smallest possible apertures. However, once the in situ stress state is such that a discontinuity undergoes a degree of non-gouge forming microshear movement, asperites will slide over each other, resulting in discontinuity's self-propping effect and permeability increase (Esaki et al. 1999;Heffer and Lean 1993;Yeo et al. 1998). As a result, discontinuities with increased likelihood for a shear movement can be simultaneously assumed to be hydraulically active. On the other hand, discontinuities with low T s are considered to have low fluid flow potential (Barton et al. 1995;Morris et al. 1996) and are expected to be "locked" in the in situ stress state.
Based on Gudmundsson (2000) and Gudmundsson et al. (2002) it is known, that extensional fractures also act as fluid-conducting features. To investigate the tendency of a discontinuity to dilate in an acting stress state, dilation tendency, T d , being described as a potential for a given discontinuity to dilate under a given 3D in situ stress tensor was computed (Ferrill et al. 1999): where 1 and 3 are the maximum and minimum effective principal stresses, respectively.
To assess how close a given cohesionless discontinuity is to the critical state (i.e., failure), Coulomb's Failure Function, CFF, was computed (Zoback 2007): Negative CFF values describe stable planar feature, as acting are insufficient to overcome its resistance to sliding. In the case of CFF approaching or being equal to zero, frictional sliding on a pre-existing discontinuity is likely to occur.

Discontinuity Geometry
To evaluate the influence of the in situ stress tensor on the reservoir permeability, geometry of discontinuities (expressed by strike and dip angle) were acquired from dipmeter logs from three old deep vertical coal exploration boreholes. Although results from dipmeter logs are used primarily to indicate bedding planes, it is widely known that such logs provide information on faults, folds, disconformities, nonconformities, and pre-existing fractures (Rider 1986). As slip tendency analysis is a method applicable for any planar structure or feature (Finkbeiner et al. 1997), we use it for all discontinuities registered with dipmeter logs. We do not distinguish between bedding planes, faults, or fractures, but rather analyze the dataset as a whole in a statistical manner. In the fourth well, which is a recent shallow exploration well, we have used available results from an acoustic borehole imager tool, indicating pre-existing natural fractures. The four boreholes, penetrating the Upper Carboniferous strata of the Ruhr region, are located in four different sections of the studied region (Fig. 2), which allows for informed conclusions for the whole studied region. In addition to discontinuity geometry information, we use results from geophysical logging (i.e., borehole diameter, gamma ray, bulk density, and P-wave velocity logging) for preliminary lithology discrimination. The geophysical logging data was later smoothed for presentation. As no information from mud logging was available, we do not make an attempt to constrain detailed lithological profiles from the studied wells.

Permeability Indicators
For investigation of the relationship between localized fluid flow and registered planar features, verification data is necessary. Most often, such data include results from high-quality temperature logs, electrical images, Stoneley wave logs, flow-meter logs, or packer tests (Zoback 2007). In our case, only in one well high quality temperature log was accessible. This temperature log, available from a thermal response test, was used to determine the thermal properties of the formation rocks in situ. During the test, well was heated up for 84 hours and temperature was subsequently measured. To compute thermal anomalies, which are associated with fluid flow along a permeable discontinuities (Cornet 1989), we have calculated the temperature difference between the measured temperatures from the thermal response test and an average linearized temperature gradient computed from test results, following methodology presented in Ito and Zoback (2000). We make an assumption that any decrease of temperature during boreholes heating-up was related to an increase of permeability at the depth of a discontinuity.
In other three wells, due to the lack of any of the aforementioned verification data, we have used borehole observations based on fluid loss information registered during drilling operations. Fluid loss during drilling can be attributed to either (i) permeable matrix, (ii) large cavities, (iii) induced, or (iv) pre-existing discontinuities (Lavrov 2016). To initiate fluid losses, pre-existing discontinuities need to provide sufficient permeability for the drilling fluid to permeate. It means, that not every discontinuity will cause fluid loss, as they could be closed in the in situ stress state, filled with gouge, or mineralized. In the case of an open discontinuity, and especially for connected fracture networks, its capacity for accepting drilling fluid will increase and it will lead to loss of drilling fluid ranging from seepage to total losses. Drilling fluid loss data commonly contains a certain degree of error, as it is recorded primarily by the drilling crew in real-time during drilling operations. The loss of fluid is most often registered by indicating depth of the fluid loss and measuring volume of fluid lost into the adjacent formation. In our case, the latter was not available and only depth of the fluid loss was known. To eliminate the uncertainty connected to the actual fluid loss depth along the well and by taking into account possible depth mismatch, a moving window of 2 m for each fluid loss depth point registered during drilling was applied. Subsequently discontinuities detected within ±2 m of fluid loss depth were assumed to be responsible for the fluid flow at that depth.

In Situ Stress State
A compilation of in situ stress magnitude measurements across the Ruhr region is presented in Fig. 3a. Acquired data resulted in an average linearized P p gradient of 10.8 ± 1.0 MPa km −1 . An average linearized S v gradient amounted to 23.1 ± 1.3 MPa km −1 . Acquired in situ stress data resulted in an average linearized S hmin gradient of 16.1 ± 4.3 MPa km −1 , whereas the average linearized S Hmax gradient amounted to 33.6 ± 9.6 MPa km −1 . The S hmin ∕S v ratio for the Ruhr region amounted to 0.66 ± 0.18 and the S Hmax ∕S v ratio to 1.37 ± 0.39 (Fig. 3c). Considerable scatter of both S hmin and S Hmax values between ∼ 600 and ∼ 1200 m depth was observed (Fig. 3a). It is expected that such scatter could be related to the influence of the coal mining on the in situ stress magnitudes. Based on Fig. 3b, it can be observed that transpression (Dewey et al. 1998) with regime stress ratio, RSR, computed after Simpson (1997), amounting to ∼ 2 , was observed in the upper-most depths until approximately 600 m. Between depths of 600 and 1000 m, transition from transpression to a strike-slip regime was observed ( 1.5 < RSR < 2 ). Below that depth and until 2000 m, in situ stress measurements indicate predominantly strike-slip stress regime ( RSR ∼ 1.5 ). Both pure normal and reverse faulting regimes, can be considered as rather unlikely in the Ruhr region, as seen in Fig. 3b.   Fig. 3 a The present-day state of stress of the Ruhr region (shaded areas represent the uncertainty of pore pressure and principal in situ stress magnitudes; thick lines represent their average values; SD standard deviation); b regime stress ratio, RSR, from re-evaluated in situ stress measurements (NF normal faulting, SS strike-slip faulting, RF reverse faulting) computed according to Simpson (1997) ( n = 46 ); c re-evaluated in situ measurements presented on a stress polygon plot with mean total stress ratios and their standard deviation ( n = 46) ; d coefficient of sliding friction, μ, based on re-evaluated in situ measurements (line of best fit marked in red) computed according to Zoback and Townend (2001) ( n = 46 ); e absolute slip, T s , and f dilation, T d , tendencies for the Ruhr region based on extrapolated linear stress and pore pressure gradients from this study and average SHmax from Reinecker et al. (2008) with the direction of maximum T s (arrows indicate SHmax ) Figure 3d presents a compilation of in situ stress measurement from the Ruhr region, similar to the one presented in Zoback and Townend (2001), with theoretical relationships for various frictional coefficients, based on frictional-failure theory, represented with solid lines. As shown in this figure, ratio between maximum, 1 , and minimum, 3 , effective stress corresponds to a crust in a frictional-failure equilibrium state (Jaeger et al. 2007), where majority of coefficients of friction are falling between 0.5 and 0.85. This implies that the studied area is either critically stressed or close to the critical state in accordance with Coulomb frictional-failure theory Healy 1984, 1992) and coefficient of friction between 0.5 and 0.85 the most accurately describes rocks of the Ruhr region. Only a few data points indicated being slightly higher than 1 and smaller than 0.5. The line of best fit for the acquired dataset amounts to of ∼ 0.68 . The results presented in Fig. 3d agree with laboratory studies on the triaxially deformed Ruhr sandstone samples by Duda and Renner (2013), Duda et al. (2019), and Ahrens et al. (2018) shown in Fig. 4, where the friction coefficient of an intact and fractured Ruhr sandstone amounts to 0.73 and 0.67, respectively. Hunfeld et al. (2017), on the other hand, indicated of ∼ 0.5 for Carboniferous shales from direct shear tests. Stoeckhert et al. (2020), based also on direct shear test results, proved that of clean Ruhr sandstone sample with clay-filled gouge, can amount to value as low as 0.12. For the slip tendency analysis, we assume being in range between 0.5 and 0.7 in the Ruhr region.

Slip and Dilation Tendency Analysis
Figure 3e, f presents the lower hemisphere stereographic projection of absolute T s and T d values for the Ruhr region based on average linearized pore pressure and in situ stress gradients from this study and average SHmax for the Ruhr region from Reinecker et al. (2008). It can be observed that discontinuities most favorably oriented for failure in the in situ stress tensor, with a maximum T s of 0.79, are vertically dipping with strike angle of N6 • E and N134 • W (i.e., around 30 • from SHmax ). The NE-SW striking fractures have a low T s and are, thus, much less likely to experience frictional sliding. The dilation tendency is the highest for fractures striking along SHmax and lowest for the ones striking perpendicular to it. It can be, therefore, said that the NW-SE-and NNE-SSW-striking discontinuities will be the preferential fluid flow pathways in the Ruhr region. The NE-SW striking fractures, with much higher normal effective stresses and low T d , will be likely closed in the present-day stress state configuration and hydraulically dead. We confirm these findings below.

Permeability Verification
Discontinuity datasets from the four studied boreholes (locations in Fig. 2), as shown in Figs. 5, 6, and 7 were presented with (i) plots of fracture strike with well depth, (ii) threedimensional Mohr graphs and, and (iii) lower hemisphere stereographic projections of CFF normalized by a vertical stress. In each well slightly different SHmax , based on the nearby in situ stress measurement, was assumed.

Well A
The well A is located in the vicinity of the former Victoria mine, north of General Blumenthal and Lohberg mines. In this borehole, breakout analysis indicated SHmax of N127 • W . The orientations of discontinuities obtained from a dipmeter log revealed low dip angles ( < 20 • ) between depths of 660 and 980 m. Below that depth and until 1100 m moderate dip angles ( < 40 • ) were observed, whereas between 1100 and 1170 m, again low ( < 20 • ) dip angles were revealed. The strike direction until The given equations describe linear envelopes (thick black lines) in the Mohr diagrams, that is, Mohr-Coulomb failure and friction criteria for a intact and b fractured rock, respectively 870 m depth does not have any major trend and is highly scattered. Below that depth and until the end of the log, discontinuities strike predominantly in the NW-SE direction. It was revealed that ∼ 4% of the registered discontinuities have T s ≥ 0.5 , from which ∼ 43% . Critically stressed planes correlate well with bulk densities of ∼ 2.5 g cm −3 , P-wave velocities of ∼ 5 km s −1 , and gamma ray values below 100 gAPI, indicating confinement within sandstone layers. The rest of fluid loss zones at depths of 825 m, between 875 and 950 m, and at 1150 m correlate with coal-bearing formations, which can be observed by a sharp bulk density decrease, low P-wave velocity, and a borehole diameter increase, all being indicative of penetration throughweaker coal layers.

Well B
The well B is located between the central and eastern part of the Ruhr region between former General Blumental and Haus Aden mines. For this borehole, SHmax of N144 • W was derived from hydrofracturing tests carried out in the former Haus Aden mine. Based on results from a dipmeter log from the well B, it was revealed that discontinuities between depth of 560 and 580 m have significantly varied dips ranging from 10 to 60 • . Between depth of 580 and 640 m, high dip angles of approximately 50 • were revealed. Below that depth and until 730 m, dip angles < 20 • were observed.
The strikes between 560 and 580 m depth are significantly scattered without any major trend. Between 580 and 670 m discontinuities strike primarily in the NNW-SSE direction. Below that depth and until 730 m, registered planar features strike mainly in the NE-SW direction. In well B, ∼ 10% of discontinuities have T s ≥ 0.5 from which 63 % experienced fluid loss. Discontinuities with T s ≥ 0.5 that experi-  Results of the slip tendency analysis in the well B: a strike of registered planar features with a color map indicating T s (red outline and red dashed line indicate plane with T s ≥ 0.5 connected to a fluid loss; circle size represents increasing dip angle of a discontinuity. The actual dip angle value is presented in g); n = 155 ); b recorded fluid losses during drilling; c gamma ray log; d) P-wave velocity log; e bulk density log; f three-dimensional Mohr plot (blue circles indicate fluid loss; n = 155 ); g lower hemisphere stereographic projection of the CFF normalized by S v with registered planar features (blue circles indicate fluid loss; n = 155 ) assuming of 0.5 (arrows indicate SHmax ) layers (indicated by sharp bulk density and gamma ray decrease), which is especially visible at depths between 642 and 670 m, 680 m, and at 730 m. Based on a correlation of critically stressed discontinuities with bulk density log, all of the critically stressed planes are expected to be located within sandstone layers.

Well C
The well C is located in the most western part of the Ruhr region, close to the former Friedrich-Heinrich and Lohberg mines. The SHmax amounting to N135 • W was assumed for the well C based on results from hydrofracturing tests carried out in the Friedrich-Heinrich mine. Based on the results from a dipmeter from this well, it was revealed that discontinuities located between depth of 620 and 650 m, have highly varied dip angles, proving no major trend. Below that depth and until 825 m, low dip angles ( < 10 • ) were registered. Between 825 and 875 m depth, dip angle increased to 20 • . Registered discontinuities strike predominantly in the N-S direction. In the well C, only ∼ 3% of registered discontinuities have T s ≥ 0.5 . Three major fluid losses were localized in the analyzed interval i.e., between 625 and 650 m, 740 and 775 m, and between 830 and 875 m depth, with all Fig. 7 Results of the slip tendency analysis in the well C (presented on the left side; n = 230 ) and well D (presented on the right side; n = 525 ): a, e strike of registered planar features with a color map indicating T s (red outline and red dashed line indicate plane with T s ≥ 0.5 connected to a fluid loss; circle size represents increasing dip angle of a discontinuity. The actual dip angle values are presented in c and g); b recorded fluid losses during drilling; c, g lower hemisphere stereographic projection of CFF normalized by S v with registered planar features (blue circles indicate fluid loss) assuming of 0.5 (arrows indicate SHmax ); d, h three-dimensional Mohr plot (blue circles indicate fluid loss); f thermal anomaly registered while heating-up performed during thermal response test being correlated to the planes with high T s values. Six planar features have T s ≥ 0.5 and fluid loss was observed in three of them. These three planar features, located at depth of 662 m ( 169 • /64 • ), 834 m ( 178 • /55 • ), and 852 m ( 174 • /52 • ), correspond to the main fluid loss zones in the well. No geophysical logging was made available in this well to make a comparison with lithology.

Well D
The well D is located in the southern part of the Ruhr region within the city of Bochum. The SHmax amounting to N124 • W was assumed based on the drilling-induced fracture analysis performed in this well. Based on the acoustic borehole imager tool results, it was discovered that registered discontinuities have a highly varied dip angles ranging between 30 and 90 • until approximately 300 m depth. Below that depth they become more consistent and amount to approximately 60 • . The registered discontinuities strike predominantly in the NW-SE direction. A small group of discontinuities, striking perpendicularly to the NW-SE direction was registered between depth of 150 and 300 m. In the well D, ∼ 50% of registered discontinuities have T s ≥ 0.5 , from which ∼ 61% have thermal anomaly of ≥ 0.5 • C . Based on the analysis of the high-quality temperature log for indication of permeable discontinuities, it was discovered that fractures striking between N140 • W and N160 • W direction with high dip angles between 50 and 70 • are the most likely to be permeable i.e., are related to the highest thermal anomalies.

In Situ Stress State
There are significant limitations to the constrained in situ stress dataset and extrapolation of the constructed average linearized in situ stress gradients for the Ruhr region to depths greater than 2000 m. To remove uncertainties related to the state of stress below 2000 m depth, in situ measurements below that depth are advised. Moreover, due to the past and ongoing mine water pumping activates in the Ruhr region, water table levels in the subsurface have been significantly reduced. Currently average water table level in coal mines amounts to approximately 600 m (Brücker and Preuße 2020;Gombert et al. 2018), where in some cases water levels reduced to 1200 m were observed. Such significant differences in mine water table levels will cause pore pressure disturbances and, as a result, reactivation potentials change. As a consequence, the effect of P p changes on the in situ stress tensor in the Ruhr region cannot be regarded as marginal and shall be investigated prior to the development of a deep geothermal system. This is especially true for locations in the vicinity of the former coal mines and mine water pumping stations. It remains, also, not clear if or to what extend in situ stress magnitudes used to constrained the proposed stress model were affected by the coal mining activates.
The majority of SHmax data in the Ruhr region is of low quality (i.e., D-E quality category of stress indicator) and, thus, less reliable, with only few measurements being of higher (i.e., A-C quality category of stress indicator) quality (Heidbach et al. 2018). Few measurements, from which the majority being of low quality, indicate NE-SW SHmax , being more or less perpendicular to the regional SHmax (Baumann 1981). The reason for these "unexpected" SHmax re-orientations may be either due to (i) wrongly reported reading from the caliper tool, (ii) mistakenly picked or misinterpreted borehole breakouts (i.e., extended drilling-induced tensile fractures in weak layers, collapsed pre-existing open fractures or filter cake along caved zones picked instead of borehole breakouts) [Reiter and Heidbach (2014) references therein], or (iii) a near-isotropic horizontal stress state which would cause significant rotations of stress due to small local stress sources (Heidbach et al. 2007). The latter of which, based on the acquired stress data from the region, can be ruled out and with the first two explanations being considered as the most likely.

Fluid Flow and Discontinuity Analysis
The friction coefficient of the Ruhr region, as proven with in situ measurements and confirmed by laboratory studies, varies between ∼ 0.5 for clay-rich rocks and ∼ 0.7 for sandstones. Such high scatter indicates that pre-existing fractures and/or faults will be significantly more stressed in the clay-rich layers than in the sandstone layers. Based on the investigation performed in this study, the most hydraulically conductive are found to be the steeply dipping NNE-SSW and NW-SE striking discontinuities. The direction of these discontinuities, as confirmed with the permeability indicators on borehole scale, is also the preferential direction of the fluid flow in the Ruhr region. The majority of the critically stressed planes are confined within sandstone intervals.
Fluid flow and discontinuity analysis indicates that in all four analyzed wells there are (i) relatively few discontinuities that dominate the flow and (ii) relatively low number of planar features that are critically stressed. Nonetheless, the results of this analysis prove that the intersected critically stressed pre-existing planar features are hydraulically conductive and a match between critically stressed planes and permeability indicators can be considered as very good. This was indicated by fluid losses at depths corresponding to the critically stressed planes and is especially visible in the depth-related plots, where, intervals with planes having higher T s values represent intervals where fluid losses were registered. Although fluid loss data cannot be used to give accurate estimates on reservoir permeability, it is still useful in a qualitative sense for indicating fluid-conducting features. The increased resolution of the verification data, i.e., more sophisticated drilling fluid loss data, would additionally reduce the uncertainty during critically stressed plane evaluation. However, even without detailed information on individual pre-existing planar features and limitations of the verification data set, a correlation between the in situ stress state and preferential direction of fluid flow was found for the rock mass of the Ruhr region in wells A, B, and C. Especially good correlation is seen in the well D, where great majority of critically stressed planes are related to the high thermal anomalies, proving their increased permeability. In cases where a fluid loss was not related to a critically stressed planes, it may have been instead related to a permeable coal-bearing layers, induced hydraulic fracture, or extensive borehole deformations (i.e., wash-outs or wide breakouts) developed during drilling operations.
Results from this study agree with conclusions made by Rudolph et al. (2010) who, based on 3D numerical modeling efforts and results from geophysical logging from exploration boreholes across the Ruhr region, observed that in the vicinity of the NW-SE striking faults within Carboniferous sandstone formations increased permeability exists. The hypothesis of the NW-SE and NNE-SSW direction of permeability could be also confirmed by surface methane emissions study by Thielemann and Littke (1999), who concluded that the majority of the emitted gas at the surface, in the vicinity of the city of Bochum, is emitted alongside an NNW-SSE direction. Thielemann et al. (2001) observed that NW-SE striking Sachsen and Munster faults, located in the vicinity of the city of Hamm, were responsible for a few-meter-wide methane emission zones, along longitudinal fault axes, sourced at ∼ 800 m depth. Such an observation confirms the interconnectivity of permeable faults from the coal layers up to the surface. Reinewardt et al. (2009) showed that during the penetration of the NW-SE striking Krudenburg fault, which dips around 80 • to the west and has a thickness of ∼ 80 m , within the Prosper-Haniel mine in 1982, major water and mud inflows in volumes between 0.3 and 18 m 3 h −1 were observed at all fault sections causing major disruptions during mining operations. Attempts of sealing the highly permeable fault with grout were made with > 75, 000 m 3 water removed from the fault using relief boreholes and 5500 m 3 of grout injected into the fault via injection boreholes. High permeability was also observed during drilling through the NW-SE striking Blumenthal fault close to city of Recklinghausen (Pilger 1960). Fault penetration during coal mining activities created there a water ingress of 12 m 3 h −1 and strong methane outgassing. The aforementioned studies, although being on a much bigger scale, confirm the hypothesis of the preferential direction of permeability within the Ruhr region, as confirmed on a borehole scale and, thus, confirm the strong codependence of the present-day in situ stress tensor and fluid flow.
Other factors contributing to the discontinuity bulk permeability including discontinuity cohesion, filling material, aperture, roughness, or the degree of re-mineralization ( Barton et al. 1995) were not investigated in this study. These factors may create exceptions to the overall correlation between in situ stress tensor, discontinuities, and permeability. Based on pre-existing fracture filling analysis carried out in the well E (location in Fig. 2), also penetrating Carboniferous strata, it was discovered that sulfides are the most common fracture filling materials, followed by carbonate materials, and clay minerals. Al Ismail and Zoback (2016) proved that in clay-rich reservoirs, no direct relation between clay content and magnitude of permeability exists, with high permeabilities existing for both clay-and calcite-rich rocks. They indicated that as clay content increases, higher permeability reduction at increasing effective stress, in comparison with calcite-rich samples, exists. The permeability of clay-rich rocks is, thus, much more sensitive to effective stress, which emphasizes the importance of forecasting geothermal fluid production based on stress-dependent permeability in clayrich reservoirs. Only comprehensive data analysis including geophysical and temperature logging, wellbore imaging, in situ flow testing, and core studies would fully evaluate the importance of all aforementioned factors on the fluid flow across pre-existing discontinuities in the Ruhr region.

Implications for Geothermal Energy Utilization
Due to the scarce intrinsic matrix porosity and permeability of the formation rocks in the Ruhr region, it is expected that the future geothermal systems will rely primarily on either fracture network systems, faults, or carbonate rocks with a high degree of dolomitization or karstification. To take the most benefit of the hydraulically active discontinuities, during the development of a deep geothermal system, drilling a geothermal borehole in a potential reservoir layer in NE-SW or WNW-ESE direction will be the most advantageous in terms of reservoir secondary permeability. If drilling in this direction will be connected to the penetration of the major NW-SE-or NNE-SSW-striking normal fault, seismic activity can be considered as likely. This is especially expected in case of hydraulic stimulation efforts in any of these structures. Once penetration of the major NW-SE-striking fault will be necessary, fault segments with lower T s should be targeted. It is expected, that the NE-SW reverse faults are hydraulically dead and could act as a seal within a reservoir. To establish the best hydraulic connection between a geothermal doublet, wells could be drilled in either NW-SE or NNE-SSW direction to one another. Due to the potentially likely occurrence of critically stressed fractures and faults in the Ruhr region, it is advised to perform comprehensive 3D geomechanical modeling and implement real-time seismic monitoring using traffic light systems or seismogenic index models to minimize seismic risks related to the exploration of deep geothermal resources. In addition, more comprehensive laboratory studies on the frictional properties under reservoir-specific conditions are sought.

Conclusions
In this study, we investigated the relationship between the in situ stress state and reservoir permeability in a heavily faulted and folded Upper Carboniferous rock mass of the Ruhr region. Based on the collected in situ stress data, the stress state of the Ruhr region remains relatively homogeneous and can be described predominantly by a strike-slip regime with pure normal and reverse regimes being considered as unlikely. The P p gradient of the region amounts to 10.8 ± 1.0 MPa km −1 , S v gradient to 23.1 ± 1.3 MPa km −1 , S hmin gradient to 16.1 ± 4.3 MPa km −1 , and S Hmax gradient, with the highest uncertainty, to 33.6 ± 9.6 MPa km −1 . S hmin /S v amounts to 0.66 ± 0.18 and S Hmax /S v to 1.37 ± 0.39. in situ stress measurements below 2000 m depth are advised to better constrain the in situ stress tensor at greater depths in the Ruhr region. The slip tendency analysis of planar features located in four exploration wells suggests that only a small percentage of discontinuities registered by the dipmeter are critically stressed (i.e., favorably oriented for failure within the current stress field), whereas around half of all discontinuities registered by borehole imager tool are considered critically stressed. To establish relationships between orientations of planar features and fluid flow, we used data on both micro-and macro-scales. These data presented evidence that formation permeability is enhanced at depth intervals with brittle lithologies and critically stressed pre-existing discontinuities that are favorably oriented for failure within the current stress field. It was revealed that the steeply dipping NW-SE and NNE-SSW oriented discontinuities penetrated by the analyzed wells play an important role in providing fluid migration pathways in the low-permeability rocks of the Ruhr region. The critically stressed discontinuities are found to be better hydraulic conduits than discontinuities unfavorably oriented for failure (i.e., structures striking in the NE-SW direction) in the prevailing in situ stress tensor, which are considered to be closed and hydraulically dead. As it is expected that future geothermal systems in the Ruhr region will rely primarily on secondary permeability, drilling new geothermal boreholes in the NE-SW or WNW-ESE within a potential target layer will be the most beneficial in terms of reservoir bulk permeability.