Physical property characterization of the Waipapa greywacke: an important geothermal reservoir basement rock in New Zealand

Greywacke basement rocks in New Zealand host conventional geothermal reservoirs and may supply important hotter and deeper geothermal energy resources in the future. This work combines petrological analyses and physical property measurements of Waipapa greywacke, a basement unit hosting New Zealand geothermal reservoirs, in order to understand better how structurally controlled flow networks develop and channel geothermal fluids within it. Results show intact Waipapa greywacke has high tensile and triaxial compressive strengths, and low intrinsic permeability (~ 10–21 m2). Permeability of intact Waipapa greywacke does not increase significantly during triaxial loading to failure and is accompanied by minimal changes ultrasonic wave velocities. These data taken together suggest that microcrack development during brittle deformation is very limited. Upon failure, the permeability increases by two orders of magnitude and shows similar permeability to tests performed on synthetic, single, mode I fractures in intact Waipapa greywacke. Permeability persists in Waipapa greywacke fractures under confining pressures of at least 150 MPa. It is concluded that Waipapa greywacke rocks will not allow fluid flow through the matrix of the rock and that substantial geothermal fluid flow will only occur through macrofracture networks.

. Furthermore, future New Zealand geothermal potential lies in the development of deeper and hotter reservoirs likely hosted within similar basement lithologies (Bignall 2011;Sheburn et al. 2003).
In basement geothermal reservoirs hosted in greywacke or igneous rocks, fracture and fault networks control fluid flow (Wood et al. 2001;Sutherland et al. 2017;Wallis et al. 2012;Browne 1980). As such, these resources are susceptible to changeable heat fluxes (Aliyu and Archer 2021), dynamic fluid flow regimes (Graf and Therrien 2009), changes arising from mineral precipitation (McNamara et al. 2015), and tectonic stress fields (Davatzes and Hickman 2010), all of which can exert influence on the physical and mechanical properties of these fractured reservoir host rocks. Information on the physical and mechanical properties of basement rocks in New Zealand is thus crucial to their future development as geothermal resources both at conventional development depths and at novel, deeper depths. Furthermore, such dynamic environments generate fluidrock interactions, which can alter the mineralogy and texture of reservoir host rocks, in turn modifying the physical properties, which govern their mechanical behaviour and subsequent structural network development (Siratovich et al. 2015;Heap et al. 2020a, b). In order to optimize and maintain structurally hosted geothermal resources, studies of their physical properties, their interactions, and their dynamic evolution due to fluid interactions need to be studied to understand the effects they produce in the subsurface reservoir (Gupta and Sukanta 2006;Di Pippo 2008;Grant and Bixley 2011).
To date, few data on the thermo-physical properties of New Zealand greywacke basement rocks exist (Richards and Read 2007;McNamara et al. 2014;Mielke et al. 2016). This study aims to enhance the understanding of the physical properties of these important geothermal reservoir rocks by presenting new data on the physical, mechanical, and elastic properties on the Waipapa greywacke terrane. This basement terrane is known to host geothermal reservoirs in the Ngawha Geothermal Field in Northland, and has the potential to host deeper geothermal resources in the Taupo Volcanic Zone (TVZ). Rock texture and microstructure analysis, combined with physical property determination (uniaxial compressive strength (UCS), tensile strength, triaxial compressive strength, static and dynamic elastic properties, porosity, density, seismic wave velocity, and permeability) of Waipapa Terrane greywacke samples presented here provide new information on the physical properties controlling brittle deformation in this lithology, the process necessary for it to act as a geothermal reservoir. Data presented here will contribute towards the improved construction of thermo-hydro-mechanical-chemical models of this basement terrane to assist in de-risking of future New Zealand geothermal systems and other similarly hosted geothermal reservoirs globally.

Geological setting
The Waipapa Terrane in New Zealand spans a significant portion of the northwest half of New Zealand's North Island, outcropping from the northwest border of the TVZ, and northwards into Northland (Fig. 1). Permian to Jurassic units comprise the terrane in a complex sequence of indurated and metamorphosed volcaniclastic sandstones and siltstones (Adams et al. 2009;Beetham and Watters 1985). Following Begg and Mazengarb (1996), in New Zealand the greywacke sequences are generally identified as medium to dark grey, coarse-to-medium grained, lightly metamorphosed sandstones. Grains are poorly sorted and consist of angular quartz and feldspar, and lithic fragments of metamorphic and igneous rocks. The intergranular material consists of clay minerals formed during induration or low-grade metamorphism. Greywacke sandstones may be interbedded with lightly metamorphosed mudstones (argillite), usually layers of clay, silt, or mud, generally dark to black in colour, yet occasionally red when there is a high content of iron bearing minerals. The proportions between mudstone and sandstone are spatially variable within the Waipapa Terrane. Geothermal expression in this unit is well documented at the Ngawha Geothermal Field in Northland, and potentially within the buried basement units underneath the modern (> 2 Ma) volcaniclastic deposits of the TVZ.
Locally, at the Ngawha Geothermal Field, Waipapa Terrane basement hosting the geothermal resource is composed of grey-green argillites and massive, quartzo-feldspathic greywacke sandstones which have experienced low-grade metamorphism [prehnitepumpellyite grade (Mayer 1968)], and are strongly hydrothermally2011 altered and veined with common fault breccia textures (Bayrante and Spörli 1989;Cox and Browne 1998;Sheppard 1986). Waipapa greywacke basement is not yet drilled within the TVZ, an extensional intra-arc basin formed as a consequence of oblique subduction of the Pacific Plate underneath the North Island of New Zealand (Sheppard 1986;Wilson et al. 1995). Torlesse Terrane greywacke basement rock has been penetrated by drilling in the Kawerau, Ngatamariki, Ohaaki, Rotokawa, and Tauhara geothermal fields (Wood et al. 2001;Adams et al. 2009;Cole and Spinks 2009;Cant et al. 2018;McNamara and Massiot 2016;McNamara et al. 2016a, b;Bignall et al. 2010), and at Kawerau Geothermal Field hosts geothermal resources. The suture between the Waipapa and Torlesse terranes is thought to occur somewhere within the TVZ underneath the infill units but a precise location and depth is undetermined (Adams et al. 2009;Milicich et al. 2013a). However, given that seismic evidence points to brittle deformation reaching potential depths of ~ 6 to 8 km (Bibby et al. 1995), Waipapa terrane greywacke holds significant potential as a reservoir for hotter and deeper geothermal resources within the TVZ (Bignall 2011).

Materials and sample preparation
Samples of Waipapa greywacke terrane were acquired from quarry outcrops ( Fig. 1) in order to test their physical and mechanical properties. Waipapa samples are sourced from Taotaoroa Quarry (TTGW), which is located between Cambridge and Matamata, ~ 8 km north of Karapiro Lake. While outcrops where samples are acquired show the Waipapa greywacke to be layered and fractured, the samples tested here contain no discernible anisotropy with respect to fabric. All the rock deformation data presented in this work were collected at the Rock Deformation Laboratory at the University of Liverpool (UK), and SEM (scanning electron microscope) images were collected using a JEOL JSM 7001F FEG-SEM in the SEM Shared Research Facility, Albert Crewe Centre for Electron Microscopy at the University of Liverpool. Twenty-five, 20 mm diameter × 50 mm height, cylindrical specimens were cored and squared from samples of Waipapa greywacke sandstones from Taotaoroa Quarry. Squareness of the cored samples was better than 0.01 mm or less (Paterson and Wong 2015). Ten of these cores were then further cut into 20 mm diameter × 10 mm height disks and utilized for tensile strength measurements, and two of these cores had a macroscopic Mode I fracture created in it to use for single fracture permeability measurements. Three thin sections of Waipapa greywacke from Taotaoroa Quarry were made for petrological observations using an optical microscope. Finally, the sample taken to failure during triaxial testing was impregnated with epoxy resin and cut parallel to the core axis and perpendicular to the shear fracture plane, polished and utilized for imaging of the deformed specimen in an XL30 Philips scanning electron microscope (SEM).

Connected porosity measurements and bulk density
Connected porosity estimates are determined from cores using a helium multipycnometer (MVP-D160-E, Quantachrome Instruments). The connected porosity was calculated for 17 samples of Waipapa greywacke. Bulk density was calculated for three Waipapa greywacke cores, using their mass and solid volume.

Tensile strength
Brazilian tests, conforming to ATSM standards (ATSM D3967-08), were performed on 20-mm-diameter disks of dry greywacke basement rock, with a thickness-to-diameter ratio (t/D) between 0.2 and 0.75. Loading rate was kept at 0.2 MPa s −1 . Tensile strength is calculated with the following equation: where σ t is the splitting tensile strength (MPa), P is the maximum force applied indicated by the machine (N), L is the thickness of the specimen (mm), and D is the diameter of the specimen (mm).

Uniaxial compressive strength
UCS experiments were carried out on ten dry greywacke samples from Taotaoroa Quarry. Axial and circumferential strain gauges were attached to the samples during testing. Samples were then brought to failure using a uniaxial press. The loading rate was not servo-controlled, but manually increased by ~ 0.2 MPa s −1 so that the tests lasted around 15 min. Static Young's modulus (E) and static Poisson's ratio ( ν ) are estimated from the gradient of a linear elastic region of the UCS test results (from 40 to 120 MPa).

Triaxial loading test
A single triaxial compressive test (σ 1 > σ 2 = σ 3 ) was carried out on a greywacke core in order to obtain the lithology's strength under confining pressure conditions of 20 MPa and a pore pressure of 5 MPa (effective pressure, σ eff = 15 MPa) to ensure repeatability of results delivered in McNamara et al. (2014). Additionally, the evolution of permeability and P-and S-wave velocity (V p and V s ) were monitored to establish the development of these properties during loading to failure. We used a triaxial deformation rig able to perform triaxial experiments up to 250 MPa confining pressure (servo-controlled), with 250 MPa pore pressure (servo-controlled), and ~ 1000 MPa differential stress with a load capacity of 300 kN (Faulkner and Armitage 2013) (Fig. 2). Figure 2 shows the layout of this apparatus, illustrating where the sample is situated within the pressure vessel. It also shows the internal force gouge that provides high-resolution measurements of the axial force applied to the sample via a screw-driven actuator below the vessel. V p and V s , and permeability measurements (by using water as fluid) of the sample are measured axially during this experiment. The position of the piezoelectric crystals (lead-zircon titanate) within the assembly is illustrated in Fig. 2. Axial loading proceeded at a displacement rate of 0.1 µm/s and paused at various intervals in order to obtain P-and S-wave velocity and permeability measurements (detailed in "Permeability measurements" section). The load point axial displacement reported is corrected for the Fig. 2 A schematic illustration of the sample assembly contained within the pressure vessel used for the experiments described. It also highlights the position of the piezoelectric crystals within the assembly that are used for the P-and S-wave velocity measurements (Faulkner and Armitage 2013) elastic distortion of the loading column, which is 75.5 kN mm −1 for the apparatus and so represents the axial displacement of the sample.

Elastic wave velocities and dynamic elastic moduli
Ultrasonic wave velocities (P-and S-wave) were measured along sample axes while loaded to failure in both uniaxial compression and triaxial testing (see Fig. 2 for rig design detail). These tests were carried out in order to understand how the wave velocities develop as microcrack networks develop under increasing stress, providing insight into the change in microcrack density in the samples during deformation (e.g. Schubnel et al. 2003). Values are determined under varying levels of axial stress during UCS testing, and under various levels of axial stress with constant confining pressures during triaxial testing. Physical property characterizations are summarized in Table 2.
Velocity data were used to calculate a dynamic Young's modulus ( E d ) and dynamic Poisson's ratio ( ν d ) for these samples, using the below equations and assuming isotropy (Kuttruff 1991): where ρ is the density (g cm -3 ). The change in the bulk sample density, calculated from the variation in volumetric change which occurs during uniaxial compression, was accounted for in the analysis.

Permeability measurements
To examine fluid flow properties of the Waipapa Terrane greywacke, permeability tests were carried out on both intact samples and samples containing an induced Mode I single fracture. Permeability measurements were performed using the pulse-transient method (Brace et al. 1968) with deionized water used as the pore fluid. When the measurement begins, the upstream and downstream reservoir pressures of the sample are equilibrated and equal. A small increase (< 1 MPa) of the upstream reservoir pressure or a decrease of the downstream reservoir pressure is then imposed to apply a pressure gradient across the sample. The small pressure difference imposed was to avoid any adiabatic pressure changes that could result from a sudden larger pressure change (Brace et al. 1968). The decay characteristics of this pressure pulse, monitored in the upstream or downstream reservoir or both, may then be used to obtain a value for permeability.
Permeability measurements were made during triaxial compression tests to monitor permeability evolution as intact greywacke approaches failure. Permeability measurements were made for each pressure step (every 20 kN, or 63.91 MPa increase in load) and the loading piston was held stationary during the measurement. The first permeability measurement was taken at 0 MPa (no load applied) over a period of approximately one hour, with subsequent measurements made at compressive stresses of There is the possibility of continued deformation of the sample during the hold periods needed for permeability measurements. Any significant change in the pore volume could result in pressure transients that might affect the permeability measurements. However, the reduction in stress during the periods where permeability was measured was typically less than 1 MPa, which implies a very small increment of axial strain (~ 4 µm) given the stiffness of the apparatus is 240 MPa mm -1 . Furthermore, the pore volume was measured throughout the hold periods to a resolution better than 0.1 mm 3 indicating that negligible pore volume changes occurred during the times of the permeability measurements.
In order to understand how the permeability of single fracture responds to varying stress conditions, two intact greywacke rock cores (20 mm diameter × 10 mm height) were placed in a Brazilian loading jig, then loaded to failure in order to produce a single fracture plane (Nara et al. 2011). These samples were then carefully placed in a polyurethane jacket and put into a triaxial testing apparatus to hydrostatically load the specimen and measure evolving fracture permeability. For both specimens, permeability was recorded initially at 40 MPa confining pressure, with subsequent measurements taken every 5 MPa up to 110 MPa. After that, permeability measurements were taken at 120, 130 and 150 MPa. These confining pressures are chosen to reflect depths in crust that greywacke serves as a geothermal reservoir in New Zealand, with 40 MPa being the shallowest, ~ 1.6 km depth, and going as deep as 6 km (150 MPa) to represent potential deeper greywacke hosted geothermal systems (Bignall 2011;Sheburn et al. 2003). The procedure used here follows that of Nara et al. (2011). Connected porosity has not been estimated during the triaxial test, pore volume change can delineate the connected porosity evolution during the triaxial test.

Petrology
From examination, Waipapa Terrane samples used in this study are composed of a medium-to-coarse grained (0.25-1.5 mm), greywacke sandstone consisting of abundant to common, subangular-subrounded, detrital quartz, plagioclase, and abundant lithic fragments (≤ 5 mm sizes) (Fig. 3). Lithic fragments are made of andesitic, basaltic, and rhyolitic lavas with granophyric textures, and siltstone. Some lava lithics show a trachytic texture. The greywacke sandstone is clast supported and matrix poor (Fig. 3A, B, F). Matrix is composed of indurated clay/silt. Chlorite, calcite, quartz, and occasional epidote are found as vein minerals. In general, the Waipapa samples are moderately altered to chlorite, leucoxene and clay with some weak epidote alteration. Plagioclase crystals in places are weakly altered to clays while detrital biotite crystals are relatively fresh. Within the rock andesite lithic fragments show some albitization. The groundmass is altered to chlorite and detrital plagioclase is partially altered to clay. Mineralization within observed microfractures in these samples indicates that these crack networks at one point operated as fluid flow pathways (Fig. 3C-E).
Greywacke samples also show a small amount of microcracks with an estimated crack density parameter of 0.51 using the methodology of O' Connell and Budiansky (1974). This method was applied using the relation for saturated materials containing circular cracks and using the Poisson's ratio of the uncracked material (taken to be that above crack closure at 40 MPa of axial stress) along with the Poisson's ratio before loading commenced. Table 2 provides an overview of the mechanical (tensile and unconfined compressive strength), elastic, and physical properties of intact Waipapa greywacke.

Tensile strength
A standard deviation in tensile strength for the Waipapa greywacke samples indicates some variability (Table 2), which is not outside the norm for these types of experiments (Gale and Holder 2008).

Uniaxial compressive strength (UCS) and static elastic moduli
For all UCS tests (Table 1), stress-strain curves show a slight concave-upwards trend at low loads, followed by a linear region up to failure (Fig. 4). No discernible yield point was observed before failure. Static Young's moduli range between 54 and 85 GPa and static Poisson's ratios range from 0.19 to 0.36 (Table 2).

Triaxial compressive strength
Permeability measurements (Table 1) performed prior to the start of the triaxial test provide an intact rock permeability of 1.65 · 10 -21 m 2 . Axial loading proceeded at a displacement rate of 0.1 microns/s, and the loading was stopped at regular intervals in order to obtain P-and S-wave velocity and permeability measurements (Fig. 9). The axial displacement reported is corrected for the elastic distortion of the loading column which is 180 kN/mm for the apparatus.
We consider the sample to be drained during the triaxial test. Although the permeability is low, the characteristic time (t = l 2 /k) for fluid diffusion is calculated to be on the order of ~ 40 min, given the sample length and approximating the sample storage capacity to be 4 · 10 -12 Pa −1 (porosity multiplied by fluid compressibility; see Faulkner et al. 2018). This value corresponds reasonably well to the time taken for the permeability measurements and suggests that the timescale over we ran the experiment, and the strain rate used, meant that the sample was drained during the test.
The loading curve in Fig. 5 shows a typical, though short, initial concave-upwards trend, then quasi-linear loading, the slope of which provides a static Young's modulus of ~ 55 GPa. No yield is observed as failure (314 MPa) is approached, and the sample fails while still under quasi-linear loading. Reloading of the sample after failure occurs along a shallower gradient indicating the fractured sample is more compliant, and a residual strength of ~ 120 MPa is recorded.
Throughout the triaxial test, as strain accumulates before failure is reached (Fig. 5), the permeability does not vary greatly. The small changes observed up to displacements of ~ 0.42 mm are likely a result of measurement variability. Only immediately before failure is a dramatic increase in permeability seen, where the permeability increases by ~ 4 orders of magnitude. During this part of the loading history, the pore volume shows a decrease of 2.73 mm 3 (Fig. 6). Most of this decrease occurred during the initial loading of the sample.
Following failure and unloading of the sample, the permeability is much lower than seen at peak stress. Following residual strength yield, and a small amount of shearing along the shear fracture created during loading, the permeability increases slightly again, although this change is within the uncertainty bounds of our permeability

Scanning electron microscopy observations
SEM observations (Table 1) of the shear fracture generated by the failure of the core during the triaxial test reveal it is filled with fragments of greywacke material (Fig. 7a). The upper extremity of the fracture, where the sample comes into contact with the loading platen, widens and has a triangular shape, filled with large greywacke fragments (≤ 0.675 mm) set in a finer matrix (≤ 200 µm sized cement grain). The typical aperture of the fracture is < 1 mm. A network of subsidiary fractures develops close to both sides of the main rupture. This subsidiary fracture network is variable and includes both relatively wide fractures and clusters of narrower cracks. The subsidiary fractures have variable morphologies that are broadly parallel and sub-parallel to the main fracture, and parallel to the loading direction. Very few subsidiary brittle deformation features are observed at greater distance from the main rupture in the tested sample. The synthetically produced fractures from Brazilian loading (Fig. 7b) show very little comminution and, in contrast to the shear fracture (Fig. 7a), few fragments contained within the fracture. Fracture walls are a mixture of tortuous and straight, and rough and smooth, and the fracture width is consistent (~ 25 μm). There is no development of a damage zone or subsidiary fractures around the main fracture (Fig. 7b). Similar postfailure conditions occurred for TTGW_11_3_5 sample (Fig. 10). Still, in Fig. 7b it is visible how the fracture formed is both intra-and inter-granular. Clearly there is a preference for development of the fracture through the matrix along grain boundaries more than through the grains, as the matrix is a weaker material, however the brittle failure undoubtedly affects both grains and matrix. V p , V s , and dynamic elastic moduli V p and V s values were acquired during both uniaxial and triaxial tests ( Table 1). The P-wave and S-wave data were measured along the sample axis. From uniaxial tests, V p varies from 6.072 to 6.410 km s -1 (Fig. 8), and V s varies from 3.423 to 3.605 km s -1 . From the triaxial test, peak V p is 6.334 km s -1 and peak V s is 3.647 km s -1 . Dynamic Poisson's ratio was in the range of 0.28-0.30 and dynamic Young's modulus was in the range 80 to 84 GPa. Figure 8 shows the trend of V p and V s for Waipapa greywacke during uniaxial testing. Data show an initial rapid increase in V p and V s as the differential stress is increased (Fig. 8). After that, V p and V s assume a sub-horizontal trend and show little changes up to the point of failure. In detail, all the V p curves show fluctuation with a general increasing trend, while, for the V s , in the beginning the trend is upward, then it levels out until failure. Data from the triaxial test reveal a similar trend between V p , V s , and axial stress (Fig. 9).

Single fracture permeability
As confining pressure increased a monotonic decrease in permeability is observed for the single, Mode I fracture in Waipapa greywacke (Fig. 10). In both experiments, a similar trend in permeability response is observed, with permeability measurements (Table 1) between 10 -16 m 2 and 10 -17 m 2 made at 40 MPa confining pressure, and  permeability measurements of ~ 10 -19 m 2 made at 150 MPa confining pressures. At no point in the experiments did the fracture cease to allow the transmission of fluid along it. We saw that even with 150 MPa confining pressure the permeability was still higher than measured for the intact samples ( Fig. 10 and Additional file 2: Figure S2).

Discussion
The physical properties of a lithology affect its ability to transmit fluid within it. The matrix permeability of a rock is the key factor influencing fluid flow in a rock type, which in turn is strongly influenced by porosity, pore connectivity and, for low-porosity lithologies, microfractures. Beyond matrix permeability a lithology's ability to transmit fluid is dictated by the presence of structures (Evans et al. 2005), and the formation of structural permeability within a lithology is related to the mechanical properties of the rocks that constitute it. We discuss here our findings in the context of how they help understand fluid flow within Waipapa greywacke basement units.

The role of microfractures in Waipapa greywacke mechanical strength
The Waipapa greywacke mechanical strength reported here is high. As a comparison, the mechanical strength of the Waipapa greywackes (Table 3) is higher than that found in other lithologies reported as mechanically strong such as Westerly granite [UCS strength ~ 200 MPa (Mitchell and Faulkner 2008;Heap and Faulkner 2008)], or the Rotokawa Andesite, another important geothermal reservoir rock in the TVZ (UCS strength of 60-211 MPa; Siratovich et al. 2012Siratovich et al. , 2014. In Waipapa greywacke porosity ranges from 0.841 to 1.304% (Table 3). These measurements of low porosity, coupled with the microstructural observations such as optical microscope (Fig. 3) illustrate the minimal primary porosity that exists within this lithology. The presence of microcracks in a low-porosity lithology often contribute little to the overall porosity, but can have an important impact on its mechanical strength, and thus the stresses at which it will experience brittle failure Eberhardt (Siratovich et al. 2012 Mean values are reported for density and connected porosity. Dynamic properties are identified in this table with the subscript "d" ( E d and ν d )  al. 1999;Faulkner et al. 2006;Lajtai 1998). Previous studies on low-porosity crystalline rocks have shown that the presence of microfractures is evident by deviation from linear elastic behaviour at stress levels well below the failure stress, as well as an increase in volumetric strain (Mitchell and Faulkner 2008;Zoback and Byerlee 1975). Both these observations lend support to the interpretation that, as loading progresses, existing microfractures start to propagate, leading to reduced compliance (related to deviation from linear elasticity) and volumetric strain increase. Microstructural observations of microcrack densities of 0.51, in addition to the lowporosity measurements, are likely related to the initial short "tail" on the triaxial experiment stress-strain graph, as they are closed up during initial loading. The loading curve (Fig. 4) indicates little evidence for dilatancy from the volumetric strain curve and there is an appreciable lack of any yield point. This indicates that the formation of any microcrack network prior to failure is very limited, and contrasts with other brittle deformation studies on low-porosity rock which show evidence of microcrack formation prior to failure (Mitchell and Faulkner 2008;Eberhardt et al. 1999;Zoback and Byerlee 1975). Moreover, the direct dilatancy measurements made via volumometry during the triaxial test also indicate little or no dilatancy prior to failure, though similar direct dilatancy measurements made during the brittle deformation of low-porosity anhydrite indicate dilatancy can occur and thus potentially affect sample porosity and permeability during deformation prior to failure (De Paola et al. 2009).

Property TTGW (this study) # of tests WAGW # of tests Rotokawa andesite # of tests
Elastic wave velocities can be strongly attenuated by the presence of microcracks in crystalline rocks Vinciguerra et al. 2005;Blake et al. 2012;Heap et al. 2014) such that microcracking can strongly reduce V p and V s . The mechanical test results from the Waipapa greywackes show, at the beginning of both the UCS tests and the triaxial loading test, there is a short-lived V p and V s increase. This finding is similar to findings on wave velocities measured during pressure increase on low-porosity basalts (Fortin et al. 2011). We suggest this short-lived V p and V s increase represents closure of the low microcrack density present in the intact greywacke sample. Further changes in V p and V s with increasing pressures are minimal and, importantly, show no appreciable decrease immediately before failure that might indicate the development of a microcrack network (Figs. 4, 8).
The lack of microcracks within intact Waipapa greywacke is interesting given this lithology is heavily fractured in situ. It may be that brittle damage is very localized around macrofractures, as we observed in our failed sample from our triaxial deformation experiment here (Fig. 7). Reloading of the failed sample shows a much shallower gradient than initial (pre-failure) loading, indicating that the Young's modulus has appreciably decreased and the fractured sample is much more compliant, consistent with the presence of a larger population of microcracks, particularly in the near vicinity of the microfracture.

Waipapa greywacke permeability
The results given in Table 2 are comparable to previous measurements on connected porosity of Waipapa greywacke (Mielke et al. 2016), and connected porosity and permeability measurements made under the same conditions for Torlesse greywacke [~ 1.6%, ~ 4.824 · 10 -22 m 2 (McNamara et al. 2014)], which also hosts geothermal reservoirs . These greywacke lithologies display the lowest porosity and permeability values measured to date across a range of New Zealand geothermal lithologies. For comparison, Rotokawa andesite has permeability values four to five orders of magnitude greater than the greywacke units (~ 10 -17 m 2 ) , and greater porosity values (Table 3), while lithologies from the Tahorakuri Formation, Matahina ignimbrite, Te Teko Formation, Tahuna Formation, and a range of TVZ andesite, dacite, and rhyolite lavas all have higher average porosity values (Mielke et al. 2016;Wyering et al. 2014).
The low porosity range reported here for Waipapa greywacke suggests a low density of inherent microstructures in this lithology, and those that are present are minimally connected. This is supported by the low permeability measurements (1.65 · 10 -21 m 2 ), the low reduction in volume observed under triaxial deformation experiments, and SEM images reported in this study. The evidence and arguments presented here for the low microfracture density of intact Waipapa greywacke have implications for the matrix permeability of this rock. The microcrack density does not have a great influence on the porosity, but the permeability is strongly affected by the presence of microcracks (Costa 2006;Chaki 2008), though this is strongly dependent on effective stress levels and whether the rock undergoes brittle failure or not (Faulkner and Armitage 2013;Zoback and Byerlee 1975).
In this study, Waipapa greywacke displayed no sign of significant dilatancy during deformation experiments (Figs. 4 and 6), and a lack of decrease in V p and V s before failure in both uniaxial and triaxial tests. This is complemented by the absence of enhanced permeability observed during loading to failure in the triaxial deformation experiment (Fig. 5).
It has been well demonstrated that faults and fractures in brittle rock are produced by the interaction and fusion of many microcracks (Mitchell and Faulkner 2008;Lockner et al. 1991;Lockner et al. 1992;Reches and Lockner 1994;Healy et al. 2006). When differential stress is applied to an intact rock sample, microfractures damage increases as the rock approaches failure, and the resultant increasing dilatancy will have direct impacts on both porosity and permeability. For example, permeability increases in Westerly Granite as it is triaxially deformed are recorded as the granite failure strength is approached due to increasing development of microcracks, which then increases further as stress is relaxed allowing more connectivity between variable oriented microcracks in the rock (Mitchell and Faulkner 2008;Zoback and Byerlee 1975). Even for the anhydrite it has been demonstrated that under triaxial conditions, after an initial compaction, dilatancy occurs producing an increase of porosity and permeability (De Paola et al. 2009). Instead for a diagenetically different rock like basalt, characterized by elevated porosity, increasing the effective pressure a constant decreasing in permeability and crack density takes place, drawing a quite logarithmic curve (Fortin et al. 2011). In this study, Waipapa greywacke displayed no sign of dilatancy during deformation experiments, both uniaxial and triaxial (Figs. 4 and 6), similar to results reported in McNamara et al. (2014). This lack of dilatancy signal in our experiments is supported by the lack of decrease in V p and V s before failure in both uniaxial and triaxial tests, and a lack of increased permeability observed during loading in the triaxial deformation experiment (Fig. 5). In terms of macrofracture generation within intact Waipapa greywacke, it appears that dilation of axially aligned microcracks and their subsequent coalescence must occur immediately before failure such that their impacts on other property measurements are not captured by our experimental work here. Concerning the contributions pre-failure deformation makes to Waipapa greywacke permeability, we conclude this is likely to be small, though suggest future experiments of this nature be performed at higher resolution along the stress-strain path. Our results suggest that Waipapa permeability is dominated by macrostructures, as no permeability is witnessed to be generated by dilatancy effects from deformation in this lithology, agreeing with similar conclusions made elsewhere (McNamara et al. 2014). Passelègue et al. (2018) demonstrate that permeability anisotropy is caused when the intact rock is loaded to macroscopic failure though the rock being deformed is essentially isotropic, indicating that permeability anisotropy begins to evolve only after a lithology is deformed, and that anisotropy will be aligned with respect to the stress state that generated the fractures. Considering the low permeability of intact Waipapa greywacke, permeability anisotropy is expected to be an important aspect of this lithology as a geothermal reservoir. Experimental data presented here confirm the introduction of permeability anisotropy within Waipapa greywacke. Rapid, large increases in permeability (around four orders of magnitude, from 10 -21 m 2 to 10 -17 m 2 ) at the point of brittle failure are noted from the triaxial compression experiment (Fig. 5) and the single fracture permeability tests (Fig. 10). Thus, permeability dramatically increases in Waipapa greywacke rocks as a result of brittle deformation, implying an anisotropic permeability will develop, oriented with respect to a given stress field.
Defining the orientation of such permeable anisotropy in Waipapa greywacke is not simple given the complex nature of the tectonics associated with this lithology in geothermal regions (McNamara et al. 2019). Within the TVZ the Waipapa greywacke is subject to NW-SE directed extension, with local variations (McNamara et al. 2019). Under contemporary tectonic conditions, fractures generated in Waipapa greywacke will predominantly align to this stress state (striking NE-SW, parallel to σ 2 ), thus defining the permeability anisotropy. However, the Waipapa greywacke has undergone deformation prior to TVZ rifting resulting in previous brittle deformation in a range of orientations (Rowland and Sibson 2004). We suggest that lateral permeability anisotropy in Waipapa greywacke will be aligned to the strike of the TVZ rift (NE-SW), in which both brittle structures formed during rifting, and brittle structures formed pre-rifting, but preferentially aligned for slip in the current stress field, control geothermal fluid flow. Indeed, a dominant NE-SW fracture strike orientation is shown within fluid flow zones in geothermal wells drilled into Torlesse greywacke basement at the Kawerau Geothermal Field (Wallis et al. 2012), establishing these fracture patterns do exist in greywacke basement type lithologies in the TVZ. Implications of our findings for geothermal fluid flow in the Waipapa greywacke reservoir Ngawha Geothermal Field are more difficult to determine due to a lack of direct structural or stress information from this location (Mongillo 1985). Reported structures have E-W, and NE-SW strike orientations and are associated with NW-SE directed extension in the region (Bayrante and Spörli 1989). Assuming this structure-stress arrangement, we can infer from out experimental results that a macrofracture related permeability anisotropy aligned NE-SW would form in the Waipapa basement of the Ngawha Geothermal Field.
In a geothermal reservoir where permeability is controlled by macrostructures, such as those in Waipapa greywacke, the ability for brittle structures to maintain fluid flow is important to the resource longevity and sustainability. This study examined the effect of single macrofracture behaviour on the permeability of Waipapa greywacke. Macrofracture closure pressures of synthetically generated fractures in Waipapa greywacke falls between 95 to 112 MPa. At these closure pressures, fluid flow across the fracture was observed to continue, developing a stable permeability around 10 -19 m 2 for confining pressures between 95 and 150 MPa (Fig. 10). Thus, even at closure, macrofractures in the Waipapa greywacke provide continued permeability (2 orders of magnitude higher than intact greywacke permeability). Permeability measurements on fractured greywacke under triaxial conditions also support persistent permeability in greywacke fractures post initial failure (Fig. 5). Similar large increases in permeability from inducing a single fracture are reported from dense volcanic rocks in geothermal regions (Lamur et al. 2017). Persistent permeability in Waipapa greywacke fractures under closure pressures may be facilitated by incomplete fracture closure due to the presence of fragments inside the fractures propping it open, asperity generation, permeability via secondary brittle damage generated around the main brittle fracture (Fig. 10). Our observations suggest all three possibilities may contribute in some way. A size range of wall rock fragments are observed within the triaxial fracture experiment, some of which are likely more than significant to prop open a newly generated fracture and facilitate ongoing fluid flow within it. The longevity of this propping is uncertain, as over time with ongoing fracture slip, these fragments may be constantly reduced in size and maintain less open space. With enough time and comminution of fracture fill material, combined with fluid-rock interaction, gouge may be generated and potentially seal the fracture completely to fluid flow. Asperity generation in fractures tested here will certainly play a role in ongoing fracture permeability after closure pressures are reached, as morphologically different fracture walls are observed in fractures generated in both triaxial and tensile tests. As with grain propping, there is a time limit to the ability for asperity to maintain fluid flow in a fracture as consistent slip will eventually grind these asperities down. The exact contribution secondary brittle damage makes to observed greywacke permeability is unknown, which makes it difficult to address in terms of their contribution to permeability under closure pressures. Given these secondary structures generated under the same experimental conditions as the primary brittle failure it is reasonable to assume they would behave similarly to the main fracture, though given their more varied orientation with respect to σ 1 , and the high level of intersection observed this is questionable and requires further study.
It should be noted that the results from the measurements on the single fracture are represented in terms of permeability to allow direct comparison with the permeability of the intact rock, as has been done in previous studies (Heap and Kennedy 2016;Wang et al. 2016;Pérez-Flores et al. 2017;Kushnir et al. 2018;Eggertsson et al. 2020). Strictly, reporting permeability for fracture flow is not correct, as the calculation assumes that fluid flow occurs over the cross-sectional area of the specimen, whereas for rocks with a low matrix permeability, fluid will predominantly flow through the fracture. Consequently, the "permeability" becomes scale dependent (Heap and Kennedy 2016). This can be illustrated by considering that if the sample diameter were doubled, the flow would also double, as the length of the fracture has increased by a factor of two. However, the cross-sectional area used in the permeability calculation would quadruple and the permeability would appear to half for the larger specimen. Hydraulic transmissivity, the product of the fracture permeability and its thickness, is a better way to represent the flow properties of fractures. This parameter cannot be directly derived from our pulse-transient measurements we made (Rutter and Mecklenburgh 2018) and could only be estimated by approximating the average aperture of our fractures. Given the uncertainty in the fracture aperture estimation, we did not make this calculation.
In summary, intrinsic microfracture density in Waipapa greywacke is low, and the development, coalescence, and growth of the microfracture network before failure is minimal. Consequently, microfractures play a key role in the formation of permeable structures within the Waipapa greywacke lithology only in proximity of the brittle failure, immediately prior of the pick stress. Macrofractures demonstrably increase permeability in Waipapa greywacke, contribute to long-lived permeability post-failure, and the permeability they develop within this lithology is anisotropic, and so permeability vectors in the greywacke units will be related to the local and regional stress conditions.

Conclusions
Our experiments indicate a high mechanical strength for intact Waipapa greywacke lithology that is partly due to a low density of inherent microcracks. This lack of microcracks contributes to the low intrinsic porosity and permeability of intact Waipapa greywacke. Our experiments confirm that the Waipapa greywacke lithology will only support fluid flow via the generation of macrostructures. Furthermore, these macrofractures maintain a level of permeability after stress conditions relax (fracture closure). Brittle fractures generated in Waipapa greywacke thus remain important permeable components in this lithology for an undetermined length of time after their formation. Such macrofractures controlled permeability is inherently anisotropic and as such the directionality of fluid transmission in the Waipapa greywacke will be strongly controlled by the arrangement of the local tectonic stress field as this controls orientation of microfractures development. Given that structures impart strong permeability anisotropy to Waipapa basement, any fluid flow modelling of geothermal systems within such geological units should account for this. Further studies on Waipapa greywacke mechanical strength that simulate deeper crustal levels, and flow experiments through single