Characterisation of the fracture- and karst-controlled geothermal reservoir below Munich from geophysical wireline and well information

Abstract

The Upper Jurassic carbonate aquifer in the German Molasse Basin (S Germany) below Munich is the focus of exploitation of geothermal energy. To implement geothermal wells, meaningful prediction of reservoir quality (e.g., volume, temperature, location of aquifers, porosity, permeability) is required. However, permeability of this aquifer is often highly heterogeneous and anisotropic, as in other karst- and fracture systems. Based on geophysical well logs from six wells, a 3D porosity model, and side-wall cores, we provide a comprehensive characterisation of the reservoir. We investigate the correlation between rock porosity and matrix permeability, and the impact of hyper-facies on fractures and karstification. We locate and analyse hydraulic active zones and compare them with hydraulic inactive zones within equivalent depth ranges, to characterise promising exploration targets. We show that fracture system parameters vary strongly between wells and within a single well. However, we observe local trends between the fracture systems and rock properties. For instance, fracture intensities and compressional wave velocity increase, while porosity decreases, in dolomitic reefal build-ups (massive facies). We observed substantial karstification dominantly within the massive facies. The main indicators for hydraulic active zones in the reservoir seem to be karstification, fractures, and fault zones. Although matrix porosity has neglectable impact on permeability, the identified hydraulic active zones appear more frequently in sections with higher porosity. We conclude, similar to previous studies, that the massive facies is a suitable exploitation target. Despite the favourable conditions within the massive facies, the strongest hydraulic active zones are nevertheless in the bedded facies, often considered as aquitard, directly below the top of the reservoir within the lithostratigraphic group of the Purbeck, at the transition between the Jurassic and the Cretaceous.

Introduction

The main geological requirements for a successful hydrogeothermal project are, e.g., sufficient permeability, temperature, porosity, surface area for heat exchange, and reservoir volume (Agemar et al. 2012; Bauer 2018; Bauer et al. 2019; Glassley 2015; Huenges 2010). In the Molasse Basin in Germany (GMB), these requirements are frequently met in the Upper Jurassic aquifer. This aquifer, a carbonate sponge-microbial reef belt, is successfully used to exploit geothermal energy (e.g., Agemar et al. 2012; Birner et al. 2012; Böhm et al. 2013; Fritzer et al. 2012; Homuth et al. 2015). The high permeability of the Upper Jurassic carbonates is often caused by fault zones, fractures (Cacace et al. 2013), and karst features; the latter playing a prime role (Birner 2013; Böhm 2012).

The geothermal gradient reaches up to about 40 K km⁻¹ in the northern parts of the basin and decreases to about 20 K km⁻¹ towards the southern border of the basin (Agemar et al. 2012; Fig. 1a). However, the depth of the Upper Jurassic aquifer, and thus its temperature, increases from north to south (Agemar et al. 2012; Przybycin et al. 2017; Stober 2014). Many of the active geothermal projects in Germany are located in this region, clustered around Munich (Agemar et al. 2014; Böhm et al. 2013; Moeck and Kuckelkorn 2015). SWM Services GmbH, Munich’s energy supplier, intends to use the geothermal potential to completely supply Munich with CO₂-neutral heating by 2040 (Moeck and Kuckelkorn 2015; Reinhold et al. 2018).

Nevertheless, even though the region has proven to be highly suitable for the exploitation of geothermal energy, and the temperature distribution is known to a sufficient degree (Agemar et al. 2012), the prediction of other reservoir properties remains challenging and is subject to difficulties (Bauer et al. 2019, 2017). In the GMB, these difficulties are mostly caused by highly variable depositional environments and diagenetic conditions, that, in turn, cause anisotropic- and heterogeneously distributed petrophysical rock properties that can, even over short distances, vary over a wide range (e.g., Birner 2013; Bohnsack et al. 2020; Homuth et al. 2015; Mraz 2019; Potten 2020; Wadas and von Hartmann 2022). In addition, previous tectonic events have formed fracture systems with parameters that can vary abruptly, e.g., across facies boundaries (Fadel et al. 2022; Homuth et al. 2015; Lüschen et al. 2014; Seithel et al. 2015; Wadas et al. 2023). One consequence of this is that karstification, which is dominantly controlled by pre-existing fractures, porosity, cavities, and petrological properties of the rocks, is also distributed heterogeneously (Goldscheider et al. 2010). Thus, the first-order parameter for geothermal exploration, permeability, must vary significantly, not only at basin, but also at reservoir scale. Small variations in this fundamental parameter, i.e. absolute values or directional properties, can decide between success and failure of a geothermal project (Bauer et al. 2019; Fadel et al. 2022). Accurate prediction of a fracture system and rock properties to determine permeability is thus crucial when planning geothermal projects.

In reservoir analysis, data coverage increases at the cost of data resolution. This is because the available tools either allow very detailed measurements on borehole scale (commonly mm to dm) or provide information for larger volumes at lower resolutions on a seismic scale (commonly tens of m to km, Bense et al. 2013; Howell et al. 2014). Thus, the prime parameters, permeability and total porosity, are only assessable within ranges that, with respect to the reservoir volume, do not allow to sufficiently characterise the geothermal reservoir in order to plan a meaningful production strategy, e.g., small changes in permeability (two orders of magnitude) can cause fluid channelling (Bauer et al. 2019). In spite of these difficulties, it is of the utmost importance to understand fluid flow. In a fractured carbonate reservoir, this also requires an understanding of dissolution along fractures.

In this study, based on the data from six boreholes at Schäftlarnstraße (Sls) in the middle of Munich, we aim to analyse and understand the fracture system, karst structures and facies, and their relationship to fluid flow in this reservoir. We use different well logs recorded in six deviated wells drilled within a small volume and a 3D porosity model that is based on amplitude inversion from 3D seismic by Wadas and von Hartmann (2022). Because we analyse six radially deviated wells (Fig. 1c, d, Table 1), we have the outstanding opportunity to analyse and compare relevant parameters in different directions. Thereby, we are able to minimise the directional bias of fracture orientations. We study the correlation and consistency of the fracture system and karst structures within the present-day stress field, and analyse how they correlate with lithology and facies. Finally, we search for parameters and their combinations that indicate hydraulic active zones (HAZ). Based on our reservoir characterisation, we identify promising exploration targets.

Site description and regional geology

Regional and reservoir geology

The studied geothermal reservoir is hosted in Jurassic sediments, north of the European Alps, within the Molasse Basin in Germany (GMB, Fig. 1). The GMB, with E–W and N–S extensions of about 700 km and 130 km, respectively, is a wedge-shaped foreland basin with the deepest part in the south. Basin formation started during the Oligocene and Miocene, as the Alpine Orogeny started. The GMB is filled with up to 5000 m of Cenozoic molasse sediments that overlay the Mesozoic strata, including the studied Upper Jurassic aquifer (Bachmann et al. 1987; Freudenberger and Schwerd 1996a, b; Lemcke 1988; Meyer and Schmidt-Kaler 1989).

Fig. 1

Tectonic map and study area. a Location of the study area in Germany and geothermal gradient of the Molasse Basin (drawn after Agemar et al. (2012)). b Sketch of the German Molasse Basin, showing the main tectonostratigraphic units and major fault lineaments. The location of the studied reservoir is marked by a black dot (drawn after Freudenberger and Schwerd 1996a; 1996b). c Map view of the trajectories of the deviated wells. Dashed red lines are traces of the Munich fault (ticks on downthrown side), which divides the reservoir into the northern footwall, the intermediate block, and the southern hanging-wall. d Three-dimensional view of the well paths (colours as in c). The depth at which the deviated wells enter the analysed reservoir section is marked in c and d by coloured dots. The black dots mark the drill site

During the Upper Jurassic, large parts of the European continent were covered by a shelf sea, the Tethys, in which sequences of carbonates and marl were deposited (Bachmann et al. 1987; Freudenberger and Schwerd 1996a, b). The lithostratigraphic classification of the Upper Jurassic Malm carbonates, based on distinct marl layers and ammonites, is subdivided into 10 units, from Malm alpha at the base up to Malm zeta 5 at the top (Meyer and Schmidt-Kaler 1990; Quenstedt 1858). At the end of this sequence, the regression of the Tethys led to the deposition of the brackish sediments of the Purbeck (Bachmann et al. 1987; Freudenberger and Schwerd 1996a, b; Meyer and Schmidt-Kaler 1990). After complete sea retreat, the sediments were exposed to weathering and karstification that took place over a period of 100 Ma, until the end of the Palaeocene. Karstification reaches depths of about 150 to 250 m below Top Malm (Bachmann et al. 1987; Goldscheider et al. 2010; Lemcke 1988). Ongoing N–S-directed compression due to the Alpine Orogeny caused basin subsidence and bulging of the Alpine foreland. This led to large-scale, roughly E–W striking normal faults, with displacements of up to 350 m (Bachmann et al. 1987; Bachmann and Koch 1983; Ziesch 2019). Most of these faults root in the basement and reach up into the Cenozoic (Bachmann et al. 1987; Lemcke 1988). One of the largest faults in the study area, the south-dipping Munich Fault, branches from west to east into a northern, NE–SW striking, and southern E–W striking part. These fault branches form a horsetail splay, characteristic for strike-slip fault regimes, and separate the reservoir into three tectonic blocks, the footwall, the intermediate block, and the hanging-wall (Fig. 1c). The normal throw on both fault branches is about 150 m (Ziesch 2019).

The Upper Jurassic sediments reach thicknesses of up to 600 m. They crop out in the Franconian and Swabian Jura (Fig. 1b), dip by about 2.5° to the south, and reach depths of over 5000 m at the Alpine front (Bachmann et al. 1987; Freudenberger and Schwerd 1996a, b; Lemcke 1988; Meyer and Schmidt-Kaler 1989).

Commonly, the Malm carbonates are distinguished into two hyper-facies; bedded- and massive-facies (e.g., Bachmann et al. 1987; Freudenberger and Schwerd 1996a; Koch et al. 1994; Pawellek and Aigner 2003; Fig. 2). The bedded facies consists of well-bedded, ammonite-bearing carbonates and marls that were deposited in inter-reef troughs and are classified as mud- or wackestones, after Dunham (1962). The massive facies, in contrast, has no-, irregular-, or indistinct bedding and consists of massive, low-porosity mounds or bioherms that are mostly composed of microbial crusts and siliceous sponges, and is classified, according to Dunham (1962), as rud-, float-, and grainstones. The top and slopes of the massive facies often consists of reworked debris. The onset of these debris limestones at the basin margins predates that in the central part of the basin and is known to have higher porosities compared to the other facies (e.g., Bachmann and Koch 1983; Birner et al. 2012; Homuth et al. 2015; Koch et al. 1994; Pawellek and Aigner 2003).

Fig. 2

Conceptual faciies characteristics and composition of the Upper Jurassic Malm strata below the Molasse Basin, separated into two hyper-facies: massive- and bedded-facies (drawn after Freudenberger and Schwerd 1996a)

The hydraulic characteristics of the Malm aquifer have been described to be controlled by fractures, matrix porosity, and karst features (Birner et al. 2012; Böhm et al. 2013; Fritzer et al. 2012; Homuth et al. 2015; Lüschen et al. 2014; Stier and Prestel 1991). Stier and Prestel (1991) showed that dolomites of the massive facies often have higher fracture intensities. The dolomites have a higher tendency to be karstified with respect to the other facies. Böhm et al. (2013) and Steiner and Böhm (2011) confirmed that good reservoir conditions can be expected from the dolomitised massive facies, i.e. they found a positive correlation between (a) the dolomite content and (b) the occurrence of massive facies and transmissibility. In addition to the dolomitised zones, the debris limestone, i.e. the areas along reef caps and slopes, are reported to have higher porosities (7–14% seismic, > 18% laboratory) compared to the reef cores (< 3% seismic; < 8% laboratory; Homuth et al. 2015; Wadas and von Hartmann 2022). Matrix permeability is reported to range between 0.001 mD and 100 mD, although only debris limestone reaches values above 10 mD (Homuth et al. 2015; Wadas and von Hartmann 2022).

The present-day maximum horizontal stress (SH_max) in the Molasse Basin is N–S oriented, rather homogeneous, and dominated by a strike-slip or thrust faulting stress regime (Heidbach et al. 2018; Reinecker et al. 2010). Accordingly, there is a low reactivation potential and it is assumed that roughly E–W striking faults and fractures have low hydraulic potential (Heffer and Lean 1993; Hestir and Long 1990; Seithel et al. 2015; Zoback 2007).

Site description

The geothermal plant “Schäftlarnstraße” (Sls) is located in the city centre of Munich (Fig. 1). The geothermal project is operated by the energy supplier “Stadtwerke München”, and construction started in April 2018. It is the largest inner-city geothermal project in Germany and one of the largest in Europe. Six wells, each around 3000 m deep (total vertical depth—TVD; Table 1), were drilled radially deviated from the drill site from 2018 to 2020. They comprise three production (Th4, Th1, Th2a) and three injection wells (Th3, Th5, Th6, Fig. 1c). In the upper part, about 800 to 1000 m TVD, the drill paths are nearly vertical. In the reservoir sections, the wells are deviated at lower inclinations (Table 1).

Table 1 Overview about the investigated depth sections, well orientations and deviations

The reservoir is divided by two faults into three different tectonic blocks (Wadas and von Hartmann 2022; Ziesch 2019; Fig. 1c), whereby each block contains two wells. Top Purbeck, i.e. the top of the reservoir, is 290 m deeper in the hanging-wall than in the footwall (Table 1). The wells were drilled to target either facies-related, i.e. reef bodies (Th3, Th5, Th6, Th2a), or for faults (Th4, Th1). Stadtwerke München estimate that this triple doublet can realize a productivity of more than 300 l s⁻¹ at a temperature of about 100 ℃. First well tests show that the productivity of the northern wells is higher and production temperature is lower than that of the other wells (Meinecke and SWM Services GmbH 2019).

Methods

We analyse the reservoir using mud logs, side-wall cores, various geophysical logs, and data taken from a 3D porosity model (Table 2). The data were analysed with the software WellCAD (ALT 2021). We focus on features that determine or are indicative of permeability, i.e. fractures, stress state, karst, porosity, lithological and facies characteristics, and hydraulic active zones (HAZ).

Table 2 List of tools used in this study

The routines we used to interpret the electrical images are described by Lai et al. (2018), Rider and Kennedy (2011), and Schlumberger (2004). We analysed the following structural parameters: bedding orientation, fracture system parameters, including orientation, minimum aperture, electrical resistivity of the fracture fillings (resistive—mineralised, conductive—open, partial—partially closed, Fig. 3b), and intensity (number of fractures along the well per unit length). Fracture apertures are likely overestimated due to, e.g., common enlargement of fractures close to the borehole wall, tool standoffs, infiltrated fluids, and the resolution of the CMI itself (Luthi and Souhaité 1990). Therefore, we use ratios and normalised fracture apertures to the largest observed fracture aperture for comparison. To correct fracture intensities for the line-sampling bias of inclined wells, we apply the Terzaghi correction with a minimum cut-off angle of 10° (Terzaghi 1965). To interpret the overall fracture intensity, it is important to take low-quality image sections into account (caused by, e.g., stick–slip, pad failure, pad mismatch, breakouts, washouts, mud cake). These sections, and sections with intense fracturing, which prevents measurement of individual fractures, are marked correspondingly in the figures. Additionally, we recorded the occurrence and orientation of borehole breakouts and drilling-induced fractures to determine the maximum horizontal stress direction along the different wells (SH_max; Zoback 2007; Fig. 3d). Furthermore, we mapped karstified well sections (Fig. 3c). Thereby, we distinguish between larger cavities (also visible in the calliper readings), vuggy or secondary porosity, and karstification along fractures and bedding contacts.

Fig. 3

Electrical image log (CMI) showing examples of the differentiated a facies (massive, bedded, areas that remain uncharacterised because of bad quality); b fractures (conductive, resistive, partial conductive); c karst features (massive, vuggy porosity, karst along fractures); and d stress indicators (borehole breakouts, drilling-induced fractures)

The analysed sedimentary features, i.e. the rock facies, were interpreted following the routines described in, e.g., Donselaar and Schmidt (2010), Lai et al. (2018), Steiner (2011), and Wilson et al. (2013). The hyper-facies of the rocks can be thus easily detected and we differentiate between massive- and bedded-limestone and dolomite (Fig. 3a). We also use the mud logs for lithological identification, based on the fine fraction (coarse fraction was used as additional aid) with a sampling interval of 5 m. Dolomite and calcite contents were determined by calcimetry.

To analyse porosity, we use three different methods (sonic-, seismic- and laboratory measurements). After a quality check of the sonic logs (Th1-2a, Th5-6), we used compressional wave velocities (${{\text{v}}}_{{\text{p}}}$) to estimate porosity and rock strength (Schlumberger 1991). Since we only have sonic logs for four out of the six wells, we additionally extract porosity values from a 3D-seismic porosity model. The porosity model is based on a full-stack seismic inversion of 3D seismic data and published in Wadas and von Hartmann (2022). The model is calibrated along the six wells using the sonic logs. We extract porosity along the six well paths within a radius of 1 m around the wells, and with a vertical sampling rate of 50 m. Even though the seismic porosity model has a lower data density compared to the sonic with 0.1 m, it allows to close data gaps within individual wells and to substitute the missing sonic measurements in Th2a and Th4. For comparison and validation of the porosity model, we measured He-porosity (helium pycnometer) on oven-dried side-wall cores (SWC) in the laboratory (AccuPyc II 1340). The SWC were also used to measure He-permeability (SYROPERM; DIN EN 1936, 2007).

We used flowmeter, temperature, and mud loss to determine hydraulic active zones (DVGW 2019; Schlumberger 1991; Steingrimsson 2013). Spinner revolutions of the flowmeter were measured during injection (Table 3) and are proportional to the fluid velocity. Sections with stable revolutions per second (RPS) values indicate hydraulic inactive sections, while sections with varying RPS indicate hydraulic active zones, i.e. zones that accepted fluid during injection.

Table 3 Summary of the flowmeter logging procedure

Results

The analysis of the reservoir characteristics is based to large parts on CMI logs, which allowed us to distinguish between the two hyper-facies types (bedded and massive facies; Fig. 3a) and to measure bedding dips. These logs also allow to identify and characterise fractures (conductive, resistive, partial conductive, Fig. 3b), karst features (macro, vuggy, along fractures, Fig. 3c), and maximum horizontal stress indicators (borehole breakouts, drilling-induced fractures, Fig. 3d).

Structural features

Bedding

All bedding dips southwards, with mean dip values in the range between 5° and 9°, with the exception of beds in Th1 (Fig. 4). In Th1, we identified two distinct dip directions, both differing from those observed in the other wells. In addition, the mean dips for these two sections are steeper. In the upper part of Th1, the bedding dips at about 13° to the NE. The first change in dip direction occurs between 2939 and 2976 m MD. Between these points, no bedding was observed. Below this depth section, the bedding dips 20° to the NW. Starting at 3073 m MD, the number of readings for bedding orientation decreases downhole and the orientation of the bedding changes multiple times from S to E to NW (Fig. 4). We note that these changes in bedding orientation coincide with the seismic interpretation of where the well either penetrates the fault that forms the boundary between the intermediate block and the footwall or where the well is in close proximity to the fault (Wadas and von Hartmann 2022; Figs. 1c, 4).

Fig. 4

Azimuthal walkout plots (graphical presentation of consecutive azimuthal orientation data) of bedding contacts from top to base. Arrows indicate the dip direction of bedding planes. Note that in Th1 dip direction is opposed to the other examples. The numbers in Th1 represent depths (in MD), in which distinct changes in bedding orientation occur. The dip direction of bedding is additionally shown as rose diagrams at the bottom. Rose diagrams have a 10° bin size and are linear scaled

Fractures

Fracture system parameters (orientation, intensity, and aperture) were analysed using CMI logs. We distinguish between electrically conductive (open), resistive (closed), and partially conductive (partly open) fractures (Fig. 3b).

Orientation

The vast majority of the 2599 mapped fractures are (sub)vertical, i.e. the dip angles are between 70° and 90°. The main strike direction of the whole fracture population is NNE–SSW (Fig. 5c). The strike of the fractures within the individual wells is as follows.

Fig. 5

Fracture system characteristics of the wells. a Rose diagrams show the strike of fractures for each individual well. Lines in rose diagrams show the wells’ azimuth trace. b Structural well correlation with the upper reference level of Top Purbeck. For each well, we show, from left to right, the depth in MD and TVD. The third column shows an image quality marker. Pink areas indicate intensely fractured zones, where not every single fracture could be measured. Areas highlighted in black indicate zones where bad image quality strongly affects or impedes the detection of fractures. The fourth column shows Terzaghi-corrected fracture intensity [n m⁻¹] from 0 to 17 counts per m MD. The fifth column shows the strike direction of fractures for different well sections using rose diagrams. Columns 4 and 5 together include the well deviation (0–90°). c Rose diagram with the fracture population of all wells. Rose diagrams have a 10° bin size and are linear scaled. In Th1, the dashed grey-shaded areas indicate zones with distinct changes in bedding orientation (see Fig. 4), where fault zones are assumed

When evaluating the different wells as a whole, we observe in Th4 and Th2a one well-defined fracture set that strikes NNE to NE. In the Th5, the main set strikes E–W and is accompanied by a small, but distinct N–S striking set. The fracture sets in Th1 and Th6 are characterised by a broad variation in strike, where the strike of the fractures in Th1 and Th6 is about NE–SW and NW–SE, respectively (Fig. 5a).

When examining the fracture orientations along the well path of each well individually, we can identify, with the exception of Th4, at least two different fracture systems in each well (Fig. 5b). In the upper part of the Th5, there is one fracture set with a strike direction of E–W. In the deepest part, below 2521 m TVD, an additional N–S-striking cluster appears. In Th4, over the complete well depth, the fractures strike NNE–SSW. The upper section of the Th1, which reaches down to 2697 m TVD, is dominated by a N–S striking fracture set. The following section is characterised by fractures with strike directions that cover a broad range between NW–SE and NE–SW, with a distinct maximum in NE–SW direction. Th6 can be, according to the fracture system, divided into three sections. The upper and lower parts are described by NW–SE striking main sets, while in the middle part, between 2665 and 2716 m TVD, fractures strike dominantly ENE–WNW. In the upper part of the Th2a, a distinct N–S to NNE–SSW cluster is apparent. However, the section below 2921 m TVD is characterised by a NW–SE and a subordinate NE–SW striking fracture set (Fig. 5b).

Mineralisation

Most fractures are classified as electrically conductive (open). Th5 has, with 58%, the highest amount of conductive fractures (Table 4).

Table 4 Summary of the electrical properties of the fracture fill (i.e. degree and proportion of mineralisation), Terzaghi-corrected fracture intensity, and image quality

Comparable to the whole fracture population, the main strike directions of open, partial open, and closed fractures differ among the various wells, and also within an individual well. However, the differences in fracture strike directions within individual wells are differently pronounced with respect to mineralisation. For example, in the Th2a almost no differences are visible, while variation in Th6 is strongly developed (Fig. 6a–c). Considering the whole reservoir, a roughly N–S-striking fracture strike direction dominates in the case of open and partial open fractures. The fracture main strike direction of closed fractures is shifted towards a more NE–SW direction (Fig. 6d).

Fig. 6

Fracture strike directions shown in linear-scaled rose diagrams with 10° bin size of all (grey), conductive (green), partial (blue), and resistive (red) fractures, for a wells in the footwall, b wells in the intermediate block, c in the hanging-wall, and d for the whole reservoir. In the upper rose diagrams, the lines show the wells’ azimuth trace

Fracture Intensity

Terzaghi-corrected fracture intensities (fracture counts per m MD [n m⁻¹]) vary strongly from well to well and within each individual well, and range from 0 to 17 counts per m MD. In Th5 in the footwall and in Th2a in the hanging-wall, the fracture intensities are highest in the upper reservoir section (Purbeck, Fig. 5b). In addition to the high fracture intensities in the upper reservoir section, we observed increased values in deeper well sections of the Th4 (~ 2820–2880 m TVD), Th1 (~ 2580–2620 m TVD), and Th6 (~ 2780–2800 m TVD). The overall highest maximum fracture intensity with 17 counts per m MD, occurs in the Th6 272 m below Top Purbeck (Fig. 5b, Table 4).

The image log quality does not allow for a complete fracture sampling throughout the wells, because of bad quality sections. Thus, the provided values for fracture intensity are underestimates. In addition, these estimates are afflicted by mapping problems in sections with very high fracture intensities, which are also a source of error. Ignoring bad quality sections, we observe that the mean fracture intensity increases from north to south, and varies between 0.66 and 1.49 counts per m MD in the Th5 and Th6. In the Th1, Th4, and Th5, there is a large number of highly fractured zones, in which detailed mapping of fractures is impossible. Thus, fracture intensities in these wells are probably higher than the determined values. Furthermore, the high amount of bad quality sections within the Th1 and Th2a falsify the mean fracture intensities (Fig. 5b, Table 4).

SH_max and fracture aperture

We measured borehole breakouts and drilling-induced fractures to evaluate the SH_max directions within the reservoir (Fig. 3d). The borehole breakouts occur predominantly in the E, W, ENE, and WSW-sections of the well and the drilling-induced fractures occur in the NNE and NNW sections. Both indicate that SH_max in the reservoir is roughly N–S oriented within ± 20° (Fig. 7a).

Fig. 7

Fracture aperture against orientation. a Rose diagrams showing the orientation of induced fractures (orange) and borehole breakouts (light blue) along the well, and strike direction of all fractures (white). Arrows show direction of SH_max derived from drilling-induced fractures—dark red, SH_max derived from borehole breakouts—dark blue. Line diagrams below show b count, c cumulative fracture aperture, d mean fracture aperture of open fractures versus fracture strike direction (0–180°). Fracture aperture is normalised to the widest fracture. Dashed horizontal lines show the mean value of the boreholes

Fractures classified as electrically conductive (i.e. open fractures) strike in a wide range between NW–SE and NE–SW or E–W and are thus often not (sub)parallel to SH_max (Figs. 6, 7). In detail, in Th4 and Th1 most open fractures are (sub)parallel to SH_max, but show, especially in Th1, a wide spread in orientation. In Th6 and Th2a (hanging-wall), the strike of the open fractures deviates by about 40° to 60° from SH_max. In Th5, many of the open fractures strike normal to SH_max (Fig. 7a).

With regard to the whole reservoir (left column in Fig. 7), the number of open fractures (Fig. 7b), their normalised cumulative aperture (Fig. 7c), and their normalised mean aperture (Fig. 7d), correlate with SH_max. Each of these parameters reaches its maximum value at the N–S strike direction, i.e. sub(parallel) to SH_max. When evaluating each well separately, this observation only holds for the wells in the intermediate block (blue lines).

Fracture aperture analysis was repeated by grouping fractures according to their normalised aperture (Fig. 8). This shows that conductive fractures with smaller apertures are more frequently E–W oriented, i.e. perpendicular to SH_max, than wider fractures (Fig. 8). In other words, fractures with large apertures are preferably oriented sub(parallel) to SH_max. This is reflected in the fact that the normalised mean conductive fracture aperture is lowest (0.29) in the well with the highest number of E–W striking fractures (footwall), and highest (0.37–0.4) in wells with the highest number of N–S striking fractures (intermediate block, Fig. 7b).

Fig. 8

Line diagrams showing counts of open fractures, grouped by normalised aperture, versus fracture strike direction (0–180°) for all wells

Karst

We find that in five wells, 514 m MD of well is affected by karstification (Fig. 3c). 475 m of this is in the massive facies, compared to 53 m in the bedded facies. In both cases, dolomite (330 m MD) is more affected by karstification than the limestone (183 m MD). In contrast to this, however, karstification in the northern well predominantly occurs in massive limestone (Appendix 1). We observed the highest percentage of macro-karst in the northernmost well, Th5 (3.7%), and the lowest in the southernmost well, Th6 (0.3%). Thus, macro-karstification decreases from N to S and from E to W (Table 5, Appendix 2a, d).

Table 5 Summary of karst characteristics (in % of measured depth) in the individual boreholes

A similar trend can be observed for the vuggy porosity, however, in contrast to macro-karstification, the southern well Th2a has the highest amount of vuggy porosity and thus does not fit the observed N–S trend. In contrast, karstification along fractures is highest in Th4 and Th1 in the intermediate block, i.e. in the wells that are in close proximity to the faults (Table 5, Appendix 2d).

We observe that in four of the five wells for which image logs are available, karst-affected fractures strike preferably N–S. The exception is Th1 with a mean NE–SW strike direction (Fig. 9).

Fig. 9

Strike direction of fractures associated with karst. Rose diagrams have a bin size of 10° and are linear scaled. Lines in rose diagrams show the wells’ azimuth trace

Rock properties

Lithology

The lithology of large parts of the reservoir has been determined using the cuttings description of the mud logs. The cutting report is, except for Th3 and Th5 in the footwall, almost complete. According to this, the encountered Upper Jurassic carbonates are composed of a succession of limestone, dolomite, dolomitic limestone (dolomite with 10–50% of limestone), calcareous dolomite (limestone with 10–50% of dolomite), secondary marl, and claystone (Table 6, Appendix 2a, b).

Table 6 Summary of the lithological (mud log) and facies (image log) composition of the individual wells

The amount of dolomite is highest in Th2a (hanging-wall) and lowest in the wells in the footwall. Whereas Th2a has the highest amount of dolomite (76%), the nearby Th6, in the same tectonic block, has, with 77%, the highest amount of limestone. The amount of dolomite increases from north to south and from west to east, while the amount of dolomitic limestone decreases, respectively. Pure limestone is more common in the western wells, independent of the tectonic block. We observe no trend for calcareous dolomite (Table 6, Appendix 2a, b).

Facies

The optical appearance of the image logs allows to determine the rock’s hyper-facies, i.e. the bedded and the massive facies (Lai et al. 2018; Steiner 2011; Fig. 3a). The rocks at the top of the reservoir belong mostly to the bedded facies, which we also identified at the base of the reservoir in the wells Th5 (footwall) and in the Th4 (intermediate block). The amount of bedded facies is higher in the wells of the footwall and decreases towards the wells of the hanging-wall in the south. Accordingly, the amount of massive facies increases from north to south (Table 6, Appendix 2a, c).

The rocks of the massive facies are more frequently dolomitised in the intermediate block and hanging-wall (25–45%), than in the footwall (18%). In contrast, the rocks of the bedded facies are more frequently dolomitised in the footwall (23%), than in the intermediate block and hanging-wall (3–6%). Comparable to the cutting description, the amount of dolomite seems to increase from N to S and from W to E (Table 6).

Compressional wave velocity and porosity

The interquartile range of the ${{\text{v}}}_{{\text{p}}}$ (compressional wave velocity) in the reservoir ranges between 5.15 and 6.15 km s⁻¹, with a mean of 5.63 km s⁻¹. The probability density estimate of the ${{\text{v}}}_{{\text{p}}}$ distribution is left-skewed, has a maximum at roughly 6.10 km s⁻¹, and a pronounced shoulder between about 4.70 and 5.50 km s⁻¹ (Fig. 10a, d).

Fig. 10

Probability density estimates and corresponding box-whisker plots of the compressional wave velocity (${{\text{v}}}_{{\text{p}}}$). The plots are grouped by a–d lithology, based on the cutting report; e–h hyper-facies, based on the image analyses; and i–l the different wells. Outliers are values that fall outside the range of Q1 ± 1.5*IQR

According to the rock’s lithology, the strong left skewness and the maximum density estimate at 6.10 km s⁻¹ of the whole dataset, is dominantly caused by the three lithologies: dolomite, calcareous dolomite (Fig. 10b), and limestone (Fig. 10c). When examining the ${{\text{v}}}_{{\text{p}}}$ of a specific lithology, we can show that the probability density estimate of the ${{\text{v}}}_{{\text{p}}}$ in the calcareous dolomite (Fig. 10b) has a second but distinct peak at 4.60 km s⁻¹. The probability density estimate for ${{\text{v}}}_{{\text{p}}}$ of the dolomitic limestone (Fig. 10c), shows also bimodality, with both peaks being in an intermediate range of the ${{\text{v}}}_{{\text{p}}}$ (5.05 and 5.90 km s⁻¹). The mean ${{\text{v}}}_{{\text{p}}}$ is highest for dolomite, followed by limestone, calcareous dolomite, and dolomitic limestone (Fig. 10d).

Plotting ${{\text{v}}}_{{\text{p}}}$ according to the hyper-facies (Fig. 10e, h) shows that the probability density estimate of the massive facies is described by a bimodal ${{\text{v}}}_{{\text{p}}}$ distribution. For the bedded facies in contrast, the probability density estimate is slightly left-skewed, with its maxima between the peaks of the massive facies (Fig. 10e). Based on the image log, we further differentiate between massive and bedded facies that are either dolomitic or calcareous (Fig. 10f–h). The dolomite within the massive facies has the lowest interquartile- and data range, a strong left-skewed ${{\text{v}}}_{{\text{p}}}$ probability density estimate and, with 6.06 km s⁻¹, the highest mean velocity (Fig. 10f, h). In contrast, the limestone within the massive facies is characterised by a bimodal probability density estimate with a stronger peak at lower ${{\text{v}}}_{{\text{p}}}$, the widest interquartile- and data range, and the lowest mean ${{\text{v}}}_{{\text{p}}}$, at about 5.27 km s⁻¹. The rocks of the bedded facies, regardless of being dolomitised or not, lie in between these “end members”. Comparable to massive rocks, dolomitisation of the bedded facies also leads to a higher ${{\text{v}}}_{{\text{p}}}$ (Fig. 10g, h).

The wells of the hanging-wall (Th6, Th2a) and the intermediate block (Th1) have a comparable, left-skewed estimate of the probability density distribution of ${{\text{v}}}_{{\text{p}}}$, with peaks at about 6.10 km s⁻¹. These three wells show less-pronounced plateaux at lower ${{\text{v}}}_{{\text{p}}}$, but with different heights and magnitudes. Notably the probability density estimates for Th6 and Th2a in the hanging-wall are almost identical (Fig. 10j, k). In contrast, the well of the footwall shows a distinctly different ${{\text{v}}}_{{\text{p}}}$ density estimate (Fig. 10i). The probability density estimate of the Th5 has its maximum at 5.10 km s⁻¹, is right-skewed, and has a second less-pronounced peak at 5.70 km s⁻¹.

The modelled porosities (extracted from Wadas and von Hartmann 2022) follow mostly the trends as expected from the sonic logs. This means that the probability density estimates are roughly mirrored, i.e. left-skewed probability density estimates of ${{\text{v}}}_{{\text{p}}}$ cause right-skewed probability density estimates of porosity (compare Figs. 10 and 11). We suggest that stronger deviations (e.g., for the porosity and ${{\text{v}}}_{{\text{p}}}$ distribution of massive dolomite; Figs. 10f, 11f) are due to a significantly lower data density, i.e. the spacing in the 3D porosity model is in the range of tens of metres and in the millimetre range for the sonic logs.

Fig. 11

Probability density estimates and corresponding box-whisker plots of porosity taken from the 3D model of Wadas and von Hartmann (2022). The plots are grouped by a–d lithology, based on the cutting report; e–h hyper-facies, based on the image analyses; and i–l the different wells. Outliers are values that fall outside the range of Q1 ± 1.5*IQR

The porosity along the wells, derived from the 3D model of Wadas and von Hartmann (2022), ranges from 0 to 15% and has a mean of 6% (Fig. 11a, d). When subdividing the rocks according to lithology, calcareous rocks have higher mean porosities (ca. 7%) compared to dolomitic rocks (ca. 4 to 6%; Fig. 11b–d). The bedded facies has a slightly higher porosity compared to the massive facies. It would seem that, independent of facies, dolomitisation leads to lower porosities (Fig. 11a–h). This also reflects in the probability density estimates of Th6 and Th2a, which are almost identical (Fig. 10k). However, mean porosity in the more dolomitised Th2a is lower (Fig. 11k, l, Table 6). Plotting porosity vs. the individual wells and tectonic blocks shows that porosity roughly decreases from north to south (Fig. 11i–l).

Laboratory measurements

He-porosities of the SWC (n = 16) were measured at 30 MPa confining pressure and lie between 0.2 and 17.5%. The porosity at the sample locations from the 3D seismic porosity model (Wadas and von Hartmann 2022) range from 1.7 to 11.9%. Even though the modelled- and the measured data do not precisely match, because they are representative of the different scales; they show comparable values. The measured porosities are higher compared to the modelled porosities and thus the 3D seismic porosity model can serve as a pessimistic estimate of reservoir porosity (Fig. 12a).

Fig. 12

Crossplots of petrophysical parameters. a He-porosity vs. modelled porosity (Wadas and von Hartmann 2022) despite different scales and methods, the values fall into the same range; b modelled porosity vs. permeability; c He-porosity vs. permeability. He-porosity and permeability were measured at 30 MPa confining pressure. We observe no correlation between permeability and porosity. As an example, the regression factor for a linear correlation is provided in (b) and (c). Note, in (b) and (c), except for one outlier, the data cover only a fifth of one magnitude. Thus, to avoid feinting a non-existing correlation between porosity and permeability, we decided to plot permeability on a linear-scaled axis

The mean He-porosity of the massive dolomite (mean: 4.3%) is lower than that of the massive limestone (mean: 7.5%). We observe porosities in the massive limestone and massive dolomite between 0.2 and 12.2% and 3.0 and 7.9%, respectively.

The permeability of the samples (< 1.4 mD) is extremely low. We observe no correlation (R² returned for linear, exponential, and logarithmic distributions < 0.2, outliers not considered) between porosity and permeability. Note that one of the lowest permeabilities was measured in the sample with the highest porosity (Fig. 12c).

Identification of hydraulic active zones

To identify hydraulic active zones (HAZ), we analyse spinner revolutions of the flowmeter log (RPS) recorded during injection, mud loss (ML) observed during the drilling process, and for the wells Th1 and Th4 (intermediate block), the base temperature (T).

In the four wells for which flowmeter logs are available (Th5, Th4, Th1, Th6), we observe a major HAZ at the top of the reservoir, about 24 to 35 m below Top Purbeck (HAZ-1). In HAZ-1, about 42 to 95% of the drilling fluid was lost. HAZ-1 is predominantly located in zones of the bedded facies with comparatively high fracture intensities (1.2–2.5 n m⁻¹). In the Th1 and somewhat less pronounced in Th5, with mean fracture intensities of 1.2 counts per m MD, large parts are intensely fractured. These parts are so densely fractured that complete sampling of the fractures was impossible and the true fracture intensity is thus higher. In Th1 and Th4, zones of karstification are observed as well. The mean ${{\text{v}}}_{{\text{p}}}$ lies between 5.12 and 5.44 km s⁻¹ and the mean porosity between 7.6 and 12.8% (see HAZ-1 in Fig. 13, Tables 7 and 8).

Fig. 13

Interpretation and well correlation of the hydraulic active zones (HAZ). The upper reference level is Top Purbeck (magenta line). From left to right, depth in MD and TVD. In the third column, the auxiliary curves (bit size, calliper log) and underlined lithology. In the fourth column, raw data for the analysis of the HAZ: mud losses (0–10 m³ seepage green, 10-15 m³ partial blue, > 15 m.³ severe-to-total mud loss red), spinner revolutions and temperature of the flowmeter measurements. The fourth column includes the resulting amount of injected volume and highlights the interpreted HAZs (interpreted based on blue: RPS, green: mud loss, orange: temperature anomaly)

Table 7 Summary of the interpreted HAZs of the reservoir with according facies and lithological characteristics and ${\text{v}}_{{\text{p}}}$

Table 8 Summary of the interpreted HAZs of the reservoir with according fracture system characteristics (only open and partial fractures are considered), amount of karstified reservoir sections, and porosity taken from the 3D porosity model of Wadas and von Hartmann (2022)

According to the drilling reports, the cumulative mud losses in the six wells range from 48 m³ (Th4) to 18,070 m³ (Th5) within the 8 ½’’ well section. Mud losses were most severe in the wells of the footwall and in Th2a of the hanging-wall. The onset of the severe mud loss in each of these three wells is between 156 and 186 m below Top Purbeck and interpreted as HAZ-2 (Fig. 13). In the wells of the intermediate block (Th4, Th1) and in the Th6 of the hanging-wall, we observed only minor mud loss. In flowmeter measurements, HAZ-2 can only be seen in the Th6 and lies 159 m below Top Purbeck. HAZ-2 is not visible in the wells of the intermediate block. HAZ-2, independent of the fact that it was interpreted based on mud loss or the flowmeter, belongs mostly to the massive facies. The mean fracture intensity is lower than in the HAZ-1 and lies between 0 and 0.3 counts per m MD. Except for Th6, HAZ-2 is associated with a high amount of karstification (17–48%). Comparatively low ${{\text{v}}}_{{\text{p}}}$ (5.00–5.41 km s⁻¹), together with the modelled porosity (6.4–9.2%) indicate a higher porosity compared to other well sections (see HAZ-2 in Fig. 13, Tables 7 and 8).

For the wells Th4, Th1, and Th2a, we interpreted additional HAZs (HAZ-3 to 10), which can only be observed in these wells (Fig. 13, Tables 7 and 8):

• In Th4, the flowmeter indicates a single injection zone in Malm Purbeck, however the base temperature show three additional temperature anomalies. These three anomalies may indicate HAZs at 73, 113, and 495 m below Top Purbeck (HAZ-3 to 5).

• In Th1, the strongest injection zone (HAZ-6) lies 212 m below Top Purbeck, which is close to the fault (c.f. Figure 5). Below 2662 m TVD, no flowmeter measurements are available. Based on mud losses, which occur close to faults, we suspect two additional HAZs 308 and 424 m below Top Purbeck (HAZ-7 to 8). The upper HAZ-1 coincides with a temperature anomaly.

• In Th2a, HAZ-9 to 10 were identified by mud loss and lie 279 and 367 m below Top Purbeck.

With the exception of the HAZ-3 (Th4), HAZs-4 to 10 lie in sections that exclusively belong to the massive facies. In HAZ-3, the bedded facies accounts for 44% of the well path. HAZ-3 to 7 are characterised by high fracture intensities and in HAZ 3, 9, and 10 karst was identified. The ${{\text{v}}}_{{\text{p}}}$ in HAZ-6 to 10 lies between 4.75 and 6.11 km s⁻¹. The exceptional high velocities belong to HAZ-7 and 10 and are in agreement with the low modelled porosities (1.3 to 2.9%). Also in HAZ-4, for which no sonic log is available, the modelled porosity is, with 0.7%, very low (Fig. 13, Tables 7 and 8).

Discussion

With the six wells, we have a high data density within the Upper Jurassic carbonate reservoir “Schäftlarnstraße” (Sls). The reservoir rocks consist of carbonates that were deposited in two different environments and are distinguished as two hyper-facies types. These are the massive facies that was deposited in the deeper parts of the carbonate platform in the form of reefal build-ups and the bedded facies that was formed in a lagoonal environment (Bachmann et al. 1987; Freudenberger and Schwerd 1996a; Koch et al. 1994; Pawellek and Aigner 2003; Fig. 2).

We now have the opportunity to evaluate, compare, and discuss parameters that significantly affect reservoir permeability with regard to their predictability, i.e. if they can be correlated between the different wells, and their impact on the interpreted HAZs. Finally, we draw conclusions about which parameter combinations indicate HAZs in the reservoir.

Exploration strategy

Recently the exploration in this area has mainly focused on reefal build-ups (massive facies) that underwent dolomitisation (Birner 2013; Böhm et al. 2013; Böhm 2012; Stier and Prestel 1991). The bedded facies, in contrast, is reported by the majority of authors to have a low permeability (e.g., Böhm et al. 2013; Bohnsack et al. 2020; Homuth et al. 2015; Stier and Prestel 1991). One exception is Wadas et al. (2023), who identified a potential exploration area within the bedded facies north to the Munich Fault, based on a seismic attribute analysis. This area is characterised by large clusters of small dolines. Independent of hyper-facies, increased permeabilities can be expected in the close vicinity of faults (e.g., Bense et al. 2013; Böhm et al. 2013; Caine et al. 1996; Fadel et al. 2022).

The six wells of the Sls project therefore targeted reefal build-ups and fault zones (Fig. 1c). The relative amount of dolomites belonging to the massive facies increases from N to S and from E to W, i.e. from 18% in the footwall to 25–45% in the intermediate block and hanging-wall (Table 6, Appendix 1, Appendix 2). Accordingly, the southern wells should be expected to have a higher productivity. The highest well productivity, however, is observed in the footwall (Meinecke and SWM Services GmbH 2019), where we observe the highest amount of bedded facies and the lowest dolomite contents. Birner (2013) made a similar observation for several wells in the greater Munich area. This shows that the distribution of hyper-facies and degree of dolomitisation are not sufficient on their own to indicate the quality of the reservoir and should be considered together with structural geological conditions to get a meaningful overall picture.

Predictability and correlations

Fracture system

Since fractures and fracture systems are the main providers of permeability, it is important to know their properties (e.g., orientation, aperture, intensity, and cementation or dissolution). Their parameters are known to be affected by the stress conditions over time, chemical processes, and sedimentological and lithological conditions (Bear et al. 1993; Gale et al. 2004; Hestir and Long 1990; Nelson 1985).

Fracture orientation

The tectonic activity that mainly affected the reservoir below the Molasse Basin started during the Late Cretaceous (Kley and Voigt 2008). Due to flexural bending of the European plate, despite the NNE–SSW-directed compressional stress state, normal faults were formed, parallel to the developing Alpine front (Bachmann et al. 1987; Bachmann and Koch 1983; Freudenberger and Schwerd 1996a, b). For the recent tectonic setting in the Molasse Basin, with the NNW-SSE-directed SH_max, three main fracture sets can be expected. (1) E–W striking extension fractures caused by flexure parallel to the Alpine front (von Hartmann et al. 2016), (2) N–S-striking extension fractures parallel to SH_max, and (3) N(N)E–S(S)W, and N(N)W–S(S)E striking, conjugate shear fractures. We observe all of these three main strike directions. However, they do not occur in a predictable manner, i.e. the sets do not occur consistently in all wells and/or at the same depth (Fig. 5). This observation is supported by, e.g., Fadel et al. (2022), Homuth (2014), Seithel et al. (2015), Wadas et al. (2023), and thus, even though data density is high and a degree of predictability exists, upscaling of fracture patterns to a larger volume is difficult.

Fracture intensity

Geological drivers that determine fracture intensity and spatial arrangement are, e.g., rock mechanics, geological history, sedimentology, facies, and relative position to fault zones (e.g., Bense et al. 2013; Caine et al. 1996; Laubach et al. 2009; Nelson 1985; Ortega et al. 2006). Thus, it is not surprising that we observed that fracture intensities vary greatly from 0 to 17 counts per m MD, over short distances (Fig. 5, Appendix 2). The observed increase of mean fracture intensity from N to S appears to be not arbitrary, but can be explained by (1) structural conditions, (2) the rock strength, and (3) the rock texture, with the latter two being facies dependent.

• 1) Regions exposed to higher stress concentrations, e.g., areas around faults, often exhibit higher fracture intensities (Bense et al. 2013; Caine et al. 1996). For instance, the hanging-walls of normal faults often undergo stronger deformation compared to the footwalls. In our study area, this was shown by Wadas et al. (2023), based on seismic attribute analysis and is further supported by retro-deformation techniques (Ziesch 2019). The observation of increasing fracture intensities from the footwall, over the intermediate block to the hanging-wall and increased fracture intensity close to the interpreted fault in Th4 confirms this correlation (Fig. 5, Table 4).

• 2) Tight and low porosity rocks are, in general, more prone to fracturing than weak and high porosity rocks (Nelson 1985; Ortega et al. 2006). Our observation is that fracture intensity correlates with increasing ${{\text{v}}}_{{\text{p}}}$ and decreasing modelled porosity (Fig. 5, Appendix 2a). In detail, dolomites that belong to the massive facies and are more common in the southern hanging-wall, are characterised by comparable high ${{\text{v}}}_{{\text{p}}}$ and lower porosity (Figs. 10, 11, Table 6, Appendix 2a). Thus, also fracture intensities increase from N to S.

• 3) From outcrop studies in Upper Jurassic rocks, it is well known that thinly bedded rocks are characterised by more-pronounced fracturing, i.e. fracture intensity increases with decreasing bed thickness (Nelson 1985; Ortega et al. 2006). This is also in agreement with our observations. In all wells, we observed the highest fracture intensities at the top of the reservoir, which belongs mostly to the bedded facies (Appendix 2a). Note that we observed higher fracture intensities in regions with frequent fluctuations in facies (metre scale). Such changes are more common in the southern wells of the reservoir (Appendix 1a) and are reported to promote fracturing processes due to differential compaction (Homuth et al. 2015). This observation, like the previous, can explain increasing fracture intensities from N to S.

In summary, we postulate that fracture intensity is controlled by the proportion of dolomite within the massive facies, the occurrence of bedded rocks at the top of the reservoir, and the position relative to the faults.

Fracture aperture, karst, and cementation

Fracture apertures are, to a large degree, controlled by their orientation to the current stress field. In particular, fractures that strike parallel to SH_max or with an acute angle between 20 and 30° to SH_max have been shown to have larger apertures compared to fractures that strike perpendicular to SH_max (Heffer and Lean 1993; Laubach et al. 2004; Singhal and Gupta 2010). We determine a roughly N–S-directed, maximum horizontal stress direction based on borehole breakouts and drilling-induced fractures (Fig. 7a). This orientation matches the regional stress field (Heidbach et al. 2018). We observe, however, that a correlation between fracture orientation and aperture exists in the Th1 (Fig. 7d), and to a lesser degree, in Th4. Fractures with low apertures follow the same pattern, in general, with a slightly increased amount with E–W strike (Fig. 8).

Regardless of an unfavourable stress field, fracture roughness and partial fracture mineralisation can also keep fractures open (Laubach et al. 2004). However, karstification along fractures is, according to our observations, almost entirely restricted to N–S striking fractures (Fig. 9). This supports the assumption above, that N–S striking fractures have and/or had higher permeabilities compared to fractures with other orientations. However, the Upper Jurassic reservoir was, after Bachmann et al. (1987), karstified during the Cretaceous. During this time, SH_max of the paleostress field, was NE–SW oriented (Reicherter et al. 2008), and thus not favourably oriented to support karstification along N–S striking fractures. Whereas the mismatch is not large enough to rule this possibility out entirely, we hypothesise that karstification along the fractures is more recent. The reasons are: (1) the present-day SH_max favours higher aperture and thus higher permeability in N–S-striking fractures and (2) the regional hydraulic gradient is caused by the southward dip of the rocks, and thus favours fluid flow along N–S-striking fractures that accordingly become wider and susceptible to karst processes. This begs the question whether the favourable orientation of higher fracture apertures was caused by the stress field or by hydraulic conditions.

Rock properties

${v}_{p}$, porosity, and mechanical rock properties

Porosity and mechanical rock properties are strongly related. Together they exert a major influence on the reservoir quality. While matrix porosity is an important value that controls storability, thermal properties, and if the pore volume is connected it improves permeability, but it has also a strong impact on mechanical properties regarding the initiation and growth of fractures and faults (Bense et al. 2013; Laubach et al. 2009; Nelson 1985; Price 1966). Thus, detailed knowledge of their values and spatial distribution within the reservoir is of great importance for the evaluation and exploitation of geothermal reservoirs. ${{\text{v}}}_{{\text{p}}}$ serves as a measure for both porosity and mechanical rock properties. In general, ${{\text{v}}}_{{\text{p}}}$ increases with decreasing porosity and increasing mechanical rock properties (Schlumberger 1991).

The wide spectrum of ${{\text{v}}}_{{\text{p}}}$ measured in the reservoir (Fig. 10, Appendix 2a) is thus indicative for the heterogeneity of the reservoir in terms of its petrophysical properties. This is supported by the porosity values modelled by Wadas and von Hartmann (2022; Fig. 11). With this heterogeneous distribution, the need arises for a meaningful subdivision of the reservoir that would be suitable for numerical models. It is essential for such a subdivision that it must find a balance between seismic and borehole scales, i.e. a balance between small and large-scale resolutions and a meaningful subdivision into units with the smallest possible variation of the parameters.

We decided to evaluate ${{\text{v}}}_{{\text{p}}}$ separately for four lithological and facies units (limestone and dolomite within the massive- or bedded-facies) that can be distinguished in seismic- and borehole analysis (Steiner 2011; Wadas et al. 2023). Of these groups, only the dolomite of the massive facies shows a low variability in ${{\text{v}}}_{{\text{p}}}$ distribution (Fig. 10). The other subgroups show a two to three times broader interquartile range (mean 50% of the data, Fig. 10). Thus, a strong heterogeneity in porosity and mechanical rock properties is apparent on both reservoir- and small scale, i.e. within individual sub-units. This is supported by the similar ${{\text{v}}}_{{\text{p}}}$ distribution measured within the closely spaced boreholes, Th2a and Th6, that share a similar facies structure, but are lithologically very different (Fig. 10, Appendix 2b, c, Table 6).

We determine that the following points about the porosity hold true for large parts of the reservoir:

• Porosity rarely exceeds 10%.

• Dolomitic rocks have, independent of hyper-facies, lower porosities, and higher dynamic properties with respect to limestones.

• In particular, dolomites of the massive facies are of low porosity and have high mechanical rock properties, such as Young’s modulus and compressive strength.

• Within the reservoir, porosity decreases while mechanical rock properties increase from N to S.

Dolomitisation and karst

Since dolomite is less soluble than calcite, karst should be predominantly observed within limestone. However, our study, like several others before (Birner 2013; Böhm et al. 2013; Stier and Prestel 1991; Wadas and von Hartmann 2022), shows that most karstification occurs in the dolomitic rocks of the massive facies (Appendix 1, Appendix 2). This seeming contradiction can be explained by the fact that grain-supported carbonates (massive facies—reefal build-ups) are less likely to compact, compared to mud-supported carbonates (bedded facies, Lucia 2004). Secondary dolomitisation requires fluid flow in the matrix, provided by a sufficient primary effective porosity, and thus the rocks of the massive facies are more prone to dolomitisation. The process of dolomitisation, which involves 12% volume loss and protects the rock against compaction, produces additional secondary porosity or retains former porosity (Lucia 2004; Weyl 1960). Thus, the process of dolomitisation is a self-energising process that supports permeability and thus karstification. In addition, incomplete dolomitisation supports karstification, i.e. relictic calcite grains or domains are preferably dissolved and can form “nuclei” for larger cavities. Thus, the spatial distribution of hyper-facies has a direct influence on the spatial distribution of karst.

However, when examining the three tectonic blocks of the study area, we observe that the footwall with the lowest dolomite content has the highest proportion of vuggy porosity and karst cavities (Tables 5 and 6, Appendix 2). While karst in the footwall is mostly within the limestone of the massive facies, it is concentrated in dolomites of the massive facies within the intermediate block and the hanging-wall (Appendix 1). This implies that conditions at the time of karst formation varied significantly, even in such a small volume. Notably, many processes can increase (dolomitisation, dissolution) or decrease (compaction, cementation, recrystallisation) porosity. Thus, the present porosity distribution with decreasing mean values from, limestone, over dolomitic limestone, and dolomite, to the calcareous dolomite (Fig. 11), does not necessarily reflect the initial porosity distribution.

Reservoir permeability

How matrix porosity evolves over time, as discussed above, has an important effect on recent reservoir conditions, i.e. karstification and dolomitisation. Our measurements on small reservoir samples, however, show that even He-porosities above 15% do not result in a permeability above 1.4 mD (Fig. 12c). Even distinct vuggy porosity in some of the plugs does not reflect increased permeability. The results are in agreement with Bohnsack et al. (2020), who show that matrix permeabilities of 100 mD or more are exceptions. Notably, 500 mD approximately marks the lower limit for successful exploitation of hydrogeothermal reservoirs in Bavaria (Fritzer et al. 2012). Thus, according to our study, the matrix porosity has a very little impact on total reservoir permeability, and must thus considered at best minor compared to the impact of karst and fractures.

Characteristics of HAZs

The main aim of our study was to identify HAZs; to characterise the conditions under which they occur and to find parameter combinations that allow high flow rates. To identify the HAZs, we analyse flowmeter and temperature measurements in combination with the fluid loss during the drill process.

We identified two main HAZs in different depth sections. Both are present in four out of the six wells. The greatest HAZ is HAZ-1, which has a 42 to 95% injection rate, starts at 24 to 35 m below Top Purbeck, and reaches down to 44 to 61 m below Top Purbeck. The second main HAZ (HAZ-2), has an injection rate of up to 28%, starts between 156 and 183 m, and ends at 182 to 225 m below Top Purbeck. In Th4, Th1, and Th2a we identified additional HAZs. These additional HAZs occur at different depths and are thus unlikely to correlate with each other (Fig. 13).

However, even though flowmeter logs are a standard tool to identify hydraulic active zones, interpretation of such measurements must be taken with care. Conditions that may cause the interpretation of a HAZ or alter their interpreted strength are, e.g., open holes, perforated liner, and the change of borehole diameter (Fig. 13, DVGW 2019). These conditions are also present at the Sls site and fall in some cases together with the identified HAZ-1. Nevertheless, Schölderle et al. (2021), who analysed HAZs in the well Th4 with flowmeter- and fibre optic-measurements, verify the occurrence of HAZ-1 at the top of the reservoir. We are therefore confident that HAZ-1, directly below the Top Purbeck, is a real feature of the reservoir. We recognised HAZ-2 mainly based on severe mud loss. Although the depth at which mud loss starts does not necessarily corresponds exactly with the location of the HAZ, because the HAZ might be blocked by drill cuttings or by the drill bit itself. The mud loss corresponding to HAZ-2 occurs in four out of six wells in a comparable depth range. The existence of a reservoir wide HAZ-2, is supported by our flowmeter-based interpretation of HAZ-2 in the Th6, that occur in the same depth range of severe mud losses of the other wells and observations by Fadel et al. (2022), who identified a HAZ at a comparable depth at a geothermal site south of Munich.

Our analyses, and the findings from Wadas et al. (2023), lead us to the conclusion that the dolomites of the massive facies do not provide, at least at this location, necessarily the best conditions for a HAZ. This is in contrast with Böhm (2012), Böhm et al. (2013), and Stier and Prestel (1991), who suggest this formation is the main exploitation target in this area. In fact, we located the strongest HAZs within the bedded facies at the top of the reservoir. This facies is commonly expected to be an aquitard (Böhm et al. 2013; Bohnsack et al. 2020; Homuth et al. 2015; Stier and Prestel 1991), and thus not considered an exploration target. Only the deeper HAZs are located in the massive facies and occur mostly in the parts that are not dolomitised (Fig. 13, Table 7, Appendix 2a).

That permeability can be significantly increased in the vicinity of fault zones (Bense et al. 2013; Caine et al. 1996) is well known and holds also true in the studied reservoir (Figs. 4, 13, Table 7 and 8). We observe that fracture-associated karstification occurs most frequently in the intermediate block, where the wells are either located close to the branching of the Munich Fault (Th4) or repeatedly penetrate the fault (Th1). Furthermore, HAZ-6, 7, and 8 are related to fault zones, with increased fracture intensities (Fig. 13, Appendix 2). However, our data do not let us decide whether the faults act as distributed conduits or as combined conduit-barriers with an impermeable fault cores (Caine et al. 1996). We consider that the individual tectonic blocks may be isolated hydraulic provinces.

Differences between hydraulic active and hydraulic inactive zones

The question—what makes a zone hydraulic active—still remains. To answer it, we compare depth sections in the wells that are hydraulically inactive (HIZ; Appendix 3), but coincide with the depth sections that host HAZ-1 and HAZ-2 (Fig. 13). We compare their properties with those of the corresponding HAZs.

The rocks that host the HAZ-1 belong mostly to the bedded facies, are mostly composed of dolomitic limestone or dolomite, and characterised by high fracture intensities (Fig. 14e-h, Tables 7 and 8). In the wells Th2a and Th3, we could not identify the HAZ-1 (Fig. 13). These two depth sections, are either strongly karstified (indicated by strong outbreaks in calliper readings; see Fig. 13) or have a high fracture intensity (Th2a; Fig. 14e, Appendix 3). Notably, along this HAZ-1 depth interval, we observe very high fracture intensities in Th2a. Th2a and Th3 differ from the other wells that contain the HAZ-1, in that they have comparatively low porosity (3.8–6.4%) and a high ${{\text{v}}}_{{\text{p}}}$ of 5.8 km s⁻¹ (Fig. 14, Appendix 3).

Fig. 14

Crossplots of parameters of the hydraulic active zones (HAZ—green) and parameters of the corresponding depth section where no HAZ could be identified (red). Plotted are parameters that are known to affect the reservoir quality, i.e. porosity against fracture intensity, macro-karst, massive facies, and dolomite content. a–d Show averages for the well parameters; e–h parameters of the HAZ-1; i–l parameters of the HAZ-2; and m–p parameters of the HAZ-3 to HAZ-10

The less-strong HAZ-2 is located in the massive facies, the prime target for the exploitation of geothermal energy in this area, which is highly variable in lithology. The rocks of the HAZ-2 comprise dolomites in the Th2a, limestone in the Th6, and calcareous dolomite in Th5. Thus, the HAZ-2 is not limited to the dolomites, which are characterised, according to our study, as rocks with increased fracture intensity and a high degree of karstification. We could not identify HAZ-2 in the wells Th4 and Th1. Here the most pronounced difference between HIZ-2 and HAZ-2 is that the hydraulic inactive sections do not contain macro-karstification. In accordance to HAZ-1, we can observe that the comparably low ${{\text{v}}}_{{\text{p}}}$ and high porosity occur predominantly in the hydraulic active zones (Fig. 13i–l, Tables 7, 8, Appendix 3).

However, in the HAZs-7 and 10, ${{\text{v}}}_{{\text{p}}}$ is comparatively high (5.96–6.11 km s⁻¹) and in the HAZs-4, 7, and 10 porosity is low (0.7–2.9%). Since other reservoir sections with outstanding low ${{\text{v}}}_{{\text{p}}}$ or high porosity do not necessarily form a HAZ, ${{\text{v}}}_{{\text{p}}}$ and porosity alone are not the overriding factor that identifies a HAZ (Fig. 13m–p, Appendix 2a). The same, however, applies to highly fractured- or karstified areas. Thus, we must assume that a dominant controlling factor does not exist. However, according to our results, supported by Bohnsack et al. (2020), key factors for HAZs seem to be both karstification and fractures, especially the combination of strongly karstified and fractured areas and higher porosities seems to favour the formation of HAZs. These attributes are most probably met in dolomites of the massive facies, which therefore is the most promising exploration target. However, our results demonstrate also that other seemingly less well-suitable lithologies and facies, also host HAZs (Figs. 13 and 14, Appendix 2).

Conclusions

We analysed permeability-determining properties within the reservoir of the Sls site using different geophysical well logs, laboratory measurements, and the 3D porosity model of Wadas and von Hartmann (2022). In our study, we localise and characterise hydraulic active zones and compare their properties with zones in equivalent depth ranges that are not hydraulic active. The aim is to characterise promising geothermal exploration targets. Our results coincide in several points with findings of previous studies, but also provide new insights, which we hope will help to exploit geothermal energy in the Molasse Basin in the future:

• The strongest HAZs are in the bedded facies at the top of the reservoir, a few tens of metres below Top Purbeck. We conclude that the Purbeck should be included as target formation and that a more detailed analysis of the Purbeck is necessary.

• Our study shows that the HAZs often have higher porosities. Our measurements show also that the reservoir matrix porosity contributes only to a neglectable degree to the bulk permeability. Permeability is, according to our findings, dominantly provided by karst and fractures.

• Fracture properties, e.g., orientation, aperture, even though highly variable in the reservoir, can be explained by the regional tectonic setting and/or hydrogeological conditions. The variability makes a detailed prediction of fracture system parameters, for such rock volumes as investigated here, difficult.

• Classification of the rocks according to lithology and hyper-facies allows for a more differentiated prognosis of rock properties in terms of rock strength, porosity, karst, and fracture intensity.

• We have applied numerous methods on a large amount of data in a comparatively small volume. Nevertheless, we could not find the particular indicators of HAZ.

Appendices

Appendix 1 Summary of the measured karst sections [m] for the different wells, depending on their occurrence in dolomite or limestone, within the massive or bedded facies

Karst [m MD]		Bedded facies			Massive facies
		Total	Dolomite	Limestone	Total	Dolomite	Limestone
Footwall	Th3	No data	No data	No data	No data	No data	No data
	Th5	17.84	16.76	1.08	92.04	38.63	53.41
Intermediate block	Th4	5.06	0.72	4.34	95.03	76.5	18.53
	Th1	9.59	8.74	0.85	152.08	111.96	40.12
Hanging-wall	Th6	4.77	2.08	2.69	42.14	25.52	16.62
	Th2a	7.77	0.57	7.20	87.50	48.97	38.53
Sum		45.03	28.87	16.16	468.79	301.58	167.21

Appendix 2 Combination of facies, lithological, petrophysical, and structural interpretation within the reservoir. a Well correlations, upper reference level—Top Purbeck. From left to right, the depth in MD and TVD, mud logs, facies interpretation, karst, ${{\varvec{v}}}_{{\varvec{p}}}$, modelled porosity, Terzaghi-corrected fracture intensity, interpreted HAZ. b-d Showing proportions of lithologies, facies, and karst for the different wells in top view

Appendix 3 Summary of the interpreted HIZs of the reservoir with characteristics of facies, lithology and structure (only open and partial fractures are considered) 

HIZ		1		2
Well		Th3	Th2a	Th4	Th1
m MD below Top Purbeck		24 to 61	24 to 61	156 to 225	156 to 225
Facies [%]	Bedded	No data	70	0	0
	Massive	No data	30	100	100
Lithology [%]	Dolomite	0	66	0	50
	Limestone	100	34	0	0
	Dolomitic limestone	0	0	100	45
	Calcareous dolomite	0	0	0	0
	No cuttings	0	0	0	5
Porosity	Min	5.52	1.51	5.67	0.89
	Max	6.49	6.2	9.23	9.77
	Mean	6.42	3.8	7.05	6.21
${{\text{v}}}_{{\text{p}}}$ [km s−1]	Min	No data	4.71	No data	4.35
	Max	No data	6.73	No data	6.49
	Mean	No data	5.80	No data	5.39
Fracture intensity [n m−1] and number	Mean	No data	3.42	0.36	0.22
	Max	No data	15.51	7.15	3.83
	n	No data	276.93	22.81	18.92
Mean aperture b [-]	Normalised	No data	0.31	0.42	0.47
IQ [%]	Intense fractured	No data	27	4.62	22
	Bad quality	No data	1	3.24	3
Strike direction		No data	NNE–SSW	N–S	NNE
Karst [%]	Macro	No data*	3	0	0
	Vuggy	No data*	4	0	0
	Along fractures	No data*	10	4	0

1. The fracture intensity is corrected after Terzaghi (1965)
2. *Due to strong calliper outbreaks, we expect larger karst cavities (see Fig. 12)