Introduction

The dependence on non-renewable energies brings several problems that now burden society, such as global warming and the depletion of fossil resources. It is, therefore, of vital importance the implementation of renewable energies, because they are clean, inexhaustible, and increasingly competitive. Geothermal resources are a great alternative, because they provide baseload power, without depending on weather conditions and with limited environmental impacts. In 2022, the total installed geothermal power generation capacity was 16,127 MW, mainly distributed among 10 countries: United States (3794 MW), Indonesia (2356 MW), Philippines (1935 MW), Turkey (1682 MW), and New Zealand (1037 MW); then, with less than 1 GW of installed power, Mexico, Kenya, Italy, Iceland, and Japan (Richter 2023). In South America, currently, the only conventional operating power plant is Cerro Pabellón in Chile in the Atacama Desert (Patel 2018). In Colombia, a co-production plant is generating electricity in an oilfield since 2021 (Franco et al. 2021), while the conventional geothermal resources associated with the numerous volcanic areas generated by the subduction of the Nazca plate under South America, have not being harnessed yet. In fact, there have been several challenges for the development of this energy during decades, mainly due to the lack of detailed information and studies, available professionals, high exploration risks, and initial investment required (Sacher and Schiemann 2010; Sanchez et al. 2011; UPME 2018; Moreno et al. 2020). However, in the last years, the Colombian government has been working to strengthen the regulatory framework associated with non-conventional energy sources, to promote their implementation through tax incentives and regulate the use of geothermal resources through technical requirements for the exploration and exploitation stages (Ley 2099 de 2021; Decreto 895 de 2022; Decreto 1318 de 2022; Resolución 40302 de 2022).

The Nevado del Ruiz Volcano (NRV) is one of the areas of major interest in the country (Mejía et al. 2014). The NRV is an active stratovolcano of the Parque Nacional Natural Los Nevados located in the Central Cordillera between the limits of the department of Caldas and Tolima. One of the most relevant advances in the NRV geothermal system characterization was the drilling of the first deep exploratory well-located on its western flank, where seven lithological units with hydrothermal alteration were identified, and a bottom hole temperature of approximately 200 °C was measured (Monsalve et al. 1998). More recent studies (Vélez et al. 2018; Moreno et al. 2018) mentioned the importance of secondary permeability in the metamorphic Cajamarca Complex, indicating that fractures and other discontinuities could act as preferential fluid transport pathways, as long as they are connected, and that they control various rock physical properties. It is, therefore, important to study the fracture patterns for the understanding of fluid circulation in hydrothermal systems, especially when geophysical data are not available (Singhal and Gupta 2010; Giordano et al. 2013; Maffucci et al 2015; Sanderson and Nixon 2015). Effective fracture permeability (or secondary permeability) is a key parameter in fluid flow in fractured rock formations, especially in hydrothermal geothermal reservoirs as they facilitate heat transfer (Brehme et al. 2016; Sævik and Nixon 2017; Wang et al. 2020; Silva et al. 2021). Secondary permeability depends directly on the geometry, connectivity and intensity of discontinuities (Cacace et al. 2013), which can be estimated from a topological characterization of fracture networks, as proposed by Sanderson and Nixon (2015). This methodology has been applied in several research studies in Canada, Chile, China, and Malta, among others, to determine the importance of fractures in groundwater flow (Miranda et al. 2018; Vélez et al. 2020; Nixon et al. 2020; Khan et al. 2021; Silva et al. 2021). Other methods have been used to estimate effective permeability: Jafari and Babadagli (2011) correlated fractal and statistical parameters to estimate the permeability of naturally patterned fracture networks; on the other hand, Giordano et al. (2013) implemented digital elevation models (DEM) for the determination of lineaments at macrostructural scale and proposed a recharge scheme and traced the groundwater fluid movements of the studied aquifer; furthermore, Smith et al. (2022) obtained 2D permeability tensors from the visualization, digitization, and analysis of fracture networks at small scale, to describe the flow through the rock.

This work aims to characterize the topology of the fracture networks, determine their connectivity at the microstructural and macrostructural level, and estimate the secondary permeability and hydraulic conductivity of the fractures from rock outcrop characterization. The results from the NRV zone, close to the city of Manizales (high influence of faulting), are compared with those obtained in the area located in the municipalities of Marulanda and Manzanares (low influence of faulting) to assess the effect of discontinuities. This research provides an assessment of the fracturing of the main lithological units in the study area (Cajamarca Complex, Quebradagrande Complex, Andesites unites, and PRE-Ruiz Lavas Group). The methodology described here can be replicated and applied to exploration studies in other geothermal areas around the world, to map structural complexity, which is a key element to understand fluid circulation in geothermal systems.

Geological and structural framework of the study area

The study area is located in the Central Andes Mountain Range, in the department of Caldas (Colombia). It includes two zones, the first is located west of the NRV (which is the area of geothermal interest, and it is influenced by faulting), while the second is located to the North–East, in the municipalities of Marulanda and Manzanares (MM), where there is less evidence of faulting (Fig. 1).

Fig. 1
figure 1

Map of the study area: a Nevado del Ruiz Volcano (NRV) and b Marulanda–Manzanares (MM). Coordinate system: WGS 1984 UTM 18 N

Full size image

The lithology in the NRV study area consists of several metamorphic and igneous units. The most relevant units are the Pelitic (Pes) and Basic (Pev), Quartzites (Pq), and Metasedimentites (Pmm) groups of the Cajamarca Complex (Paleozoic age). The Pes is composed of phyllites, sericitic, and micaceous schists, the Pev of green schists and amphibolites, Pq of quartzites and Pmm of metallimolites, metarenisks, slates, phyllites, and quartzites. Then, the Quebradagrande Complex is composed of a group of volcanic rocks (Kvc) and sedimentary rocks (Ksc). The Andesites unit (NgQa) correspond to andesitic–dacitic lavas, and basaltic in smaller proportion, and the PRE-Ruiz Lavas Group (CL-PRE), which is very homogeneous and is made up of strongly eroded massive rocks with variable concentration of mafic and felsic minerals. Other lithologic units present in the NRV include Pyroclastic Deposits (Qto) and Volcanic Mudflow Deposits (Qfl) (González 2001; Mosquera et al. 2013; Martínez et al. 2014).

The main lithologies of interest in the MM area are the Pelitic (Pes), Quartzite (Pq), and Marulanda Metasedimentites (Pmm) groups of the Cajamarca Complex, Pyroclastic Deposits (Qto), and Volcanic Mudflow Deposits (Qfl) (Mosquera et al. 2013). The lowest grade Cajamarca Complex metamorphism series (Pmm) is found between Marulanda and Pensilvania and is typical of dynamo-thermal regional metamorphism (González 2001). On the other hand, outcrops of the Pes’ member of the Cajamarca Complex are visualized in the NRV study area, where the rock is fine to very fine grained and the color is black to silver-gray depending on the graphite–mica content (González 2001; Mosquera et al. 2010).

The NRV study area has several fault systems, which are: N–S oriented fault system constituted by San Jerónimo, Laguna Baja, and Silvia-Pijao faults; NW–SE fault system constituted by Campoalegre, Río Claro, Nereidas, San Eugenio, Quebrada Negra, El Billar, Rio Molinos, and Villamaría—Termales faults. Finally, the NE–SW system constituted by Santa Rosa, Samaná Sur, Telaraña, and Palestina Faults (Mejía et al. 2012; Ceballos 2017; Ossa 2018; Macias and Granja 2020).

On the other hand, the MM study area is only influenced by Eden, Rio Hondo, and Palestina faults with NE–SW trend (Gómez et al. 2015). However, it is worth highlighting that there is limited structural information about this region, since it is not a focus of research by public or private entities.

Fracture network characterization

The characterization of the fracture network allows evaluating the connectivity from the topology, which is defined from the relationships between geometric objects and deals with the abstraction and generalization of spatial relationships that shape fracture connectivity (Sanderson and Nixon 2015). In two dimensions, the topology of a fracture network consists of lines, nodes, and branches (Fig. 2). Nodes are divided into isolated nodes (I) and two types of connecting nodes: crossing nodes (X) and abutting nodes (Y) (Manzocchi 2002; Sanderson and Nixon 2015; Nyberg et al. 2018). A “branch” is a line bounded by an isolated node (I) or a connecting node (Y or X). As the ends of branches are marked by nodes, branches are classified into three topological groups: isolated branches (I–I); partially connected branches (I–C, where C belongs to nodes X or Y); and doubly connected branches (C–C).

Fig. 2
figure 2

Nomenclature for fracture network characterization according to Sanderson and Nixon (2015). a Map of a fracture network at sampling point (P8NRV) and b topological characterization

Full size image

The characterization of a fracture network is made from the existing proportion of I, Y, and X nodes (Manzocchi 2002). The number and proportion of different types of nodes and branches can be used to simply calculate other statistics that quantify and assess connectivity (Nyberg et al. 2018; Sanderson and Nixon 2015). The average number of connections per branch provides a measure of connectivity, which requires the following mathematical model proposed by Sanderson and Nixon (2015).

To count the number of lines (\({N}_{L}\)) within a given area, Sanderson and Nixon (2015) propose the following relationship:

$${N}_{L}=\frac{1}{2}\left({N}_{I} +{N}_{Y}\right)$$
(1)

where \({N}_{I}\) is the number of nodes I and \({N}_{Y}\) the number of nodes Y.

Branches are characterized by having two nodes, hence the number of branches (\({N}_{B}\)):

$${N}_{B}=\frac{1}{2}({N}_{I}+3{N}_{Y}+4{N}_{X})$$
(2)

where \({N}_{X}\) is the number of X nodes.

The relation of the number of branches and lines is:

$$\frac{{N}_{B}}{{N}_{L}}=\frac{{N}_{I}+3{N}_{Y}+4{N}_{X}}{{N}_{I}+{N}_{Y}}$$
(3)

The average number of connections per node can be calculated as follows (Sævik and Nixon 2017):

$${C}_{N}=\frac{4{N}_{X}+3{N}_{Y}+{N}_{I}}{{N}_{X}+{N}_{Y}+{N}_{I}}$$
(4)

The average number of connections per line (\({C}_{L}\)) and the average number of connections per branch (\({C}_{B}\)), which have been used as a measure of connectivity in node counting. \({C}_{L}\) is given by (Sanderson and Nixon 2015)

$${C}_{L}=\frac{4({N}_{Y}+{N}_{X})}{{N}_{I}+{N}_{Y}}$$
(5)

and \({C}_{B}\) for:

$${C}_{B}=\frac{3{N}_{Y}+4{N}_{X}}{{N}_{B}}$$
(6)

\({C}_{B}\) provides a dimensionless number in the range 0–2.

Methodology

Microstructural and macrostructural characterization of the fracture networks was conducted using geographic information software (GIS), ArcGIS and QGIS, and the NetworkGT add-on developed by Nyberg et al. (2018). The information needed to conduct this analysis was collected from literature review and field work.

Microstructural analysis

Field work

In the NRV area, samples from the Cajamarca Complex—Pes (P2NRV, P3NRV, P8NRV, P9NRV, P11NRV and P12NRV), PRE-Ruiz Lavas Group (P4NRV, P5NRV and P6NRV) and Andesites unit (P7NRV) were collected in the Nereidas, Botero Londoño, and Cueva Pirineos sectors, while Quebradagrande Complex—Kvc samples (P15NRV and P16NRV) were taken along the Manizales-Chinchiná Road. In the MM area, Cajamarca Complex Pmm (P1MM, P2MM, P3MM, P4MM, P5MM, P6MM, P7MM, P10MM, P11MM and P12MM), Pq (P8MM, P9MM, P13MM, P13MM-C19 and P14MM-C19) and Pes (P14MM, P15MM and P16MM) samples were taken at the roadside between the municipalities of Marulanda—Pensilvania and Marulanda—Manzanares (Fig. 3).

Fig. 3
figure 3

Map of rock samples collected in the field and geological lithologies: a NRV study area and b MM study area. Modified from the sheets #187, #188, #206, #207, #225 and surface deposits map of the INGEOMINAS and SCG geological maps (Martínez et al. 2014; González 2001; González, 1980; Barrero and Vesga 1976). White spaces in the map correspond to lithologic units that have been not analyzed in this work

Full size image

Field work was based on the direct observation and measurement of fracture characteristics at outcrops influenced (NRV area) and not influenced (MM area) by faulting, respectively. The window sampling method (Zeeb et al. 2013) was applied with a sampling area of 1 \({{\text{m}}}^{2}\) to include the most representative features. Fracture length, mechanical aperture (\({a}_{m}\)), and orientation (strike and dip) were measured. The rock samples selected for thin-section petrography were P4NRV, P8NRV, P12NRV, P7MM, P8MM and P14MM.

Mineralogical composition

A 2 cm × 4.5 cm section was taken from the samples, the sections were cut longitudinally, and petrographic analysis was conducted with a NIKON ECLIPSE E − 20 0 petrographic microscope, counting 300 points per plate for classification. The metamorphic rocks were classified following the abbreviations for minerals according to the recommendations of the IUGS “Subcommission on the Systematics of Metamorphic Rocks”.

Characterization of the fracture network

The analysis of fracture network and its topological spatial distribution was performed by GIS using the NetworkGT plug-in. As input information, the toolkit requires a digitized and topologically consistent information, any inconsistency in the fracture networks was fixed with the “repair network” tool. The “nodes and branches” tool identify and extract the different types of nodes (I, Y, and X) and branches (I–I, I–C and C–C) from a network of fracture polylines within a sample polygon. It also identifies edge nodes (E), where a fracture intersects the sample area.

The “Topology Parameters” tool, measures and determine the characteristics of the network, using node and branch counts together with branch lengths, such as fracture abundance and dimensionless parameters describing the connectivity (e.g., average connections per branch and dimensionless intensity).

Fracture network connectivity was estimated comparing the following parameters with the values considered high permeability (Table 1). The average number of connections per branch varies between 0 and 2: a value of 2 indicates doubly connected branches, 1 partially connected branches, and 0 isolated branches (Sanderson and Nixon 2015; Turlin et al. 2019).

Table 1 Connectivity thresholds for fracture networks (Sanderson and Nixon 2015; Turlin et al. 2019)
Full size table

Determination of the secondary permeability and hydraulic conductivity

To calculate the permeability and hydraulic conductivity of fractures in a rock mass, the mechanical aperture \({a}_{m}\) (\(\mu m)\) was measured in the field. The hydraulic aperture (\({a}_{h}\)) can be determined from the mechanical aperture and the Joint Roughness Coefficient (JRC) (Lee et al. 1996); however, the hydraulic aperture can be considered equal to the mechanical aperture when it is not possible to determine the surface roughness of fractures and they tend to be open (Singhal and Gupta 2010; Zhao 2017).

The fracture permeability \({k}_{f} ({{\text{m}}}^{2})\) is known as

$${k}_{f}=\frac{{a}_{h}^{2}}{12}$$
(8)

The hydraulic conductivity \({K}_{f} ({\text{m}} {{\text{s}}}^{-1})\) of a rock mass was calculated from the hydraulic fracture aperture considering the parallel plate model:

$${K}_{f}=\frac{\gamma {({a}_{h})}^{2}}{12\mu }$$
(9)

where \(\gamma (N {{\text{m}}}^{-3})\) and \(\mu ({\text{Pa}} {\text{s}})\) are the specific weight and dynamic viscosity of water, respectively, at the temperature of interest. Considering that:

$$\gamma =\rho \cdot g$$
(10)
$$\mu =v \cdot \rho$$
(11)

where g is the gravitational acceleration (\({\text{m}} {{\text{s}}}^{-2}\)) and ν is the kinematic viscosity \(\left({{\text{m}}}^{2} {\text{s}}\right)\), which varies according to the water temperature of each study zone, it is possible to write:

$${K}_{f}=\frac{g {({a}_{h})}^{2}}{12 v}$$
(12)

It is important to note that the values obtained are not representative of the conditions at depth, since the fractures were measured at outcrops, where are exposed to weathering conditions that alter the physical properties of the rocks.

Macrostructural analysis

With ArcGIS and QGIS programs, satellite photographs, DEM, and bibliographic search, faults and lineaments in both study areas were identified at regional level (macrostructural), to compare the connectivity of NRV with that of MM area. The connectivity was determined using the NetworkGT tool from the branch connections (\({C}_{B}\)) represented in a map.

Results and discussion

Mineralogical composition

The petrographic results of the 6 rock samples selected (P4NRV, P8NRV, P12NRV, P7MM, P8MM, and P14MM) are presented in Table 2. These samples were chosen based on the two study areas that are contrasting in terms of fracture density. In the NRV study area, samples from the Cajamarca Complex—Pes (P8NRV, and P12NRV) were analyzed, because in previous studies, this lithology is considered being the formation hosting the geothermal reservoir (Velez et al. 2018; Moreno et al. 2018). On the other hand, the CL-PRE Ruiz sample (P4NRV) is located near the Nereidas well (Monsalve et al. 1998), which is an area of high geothermal potential. With respect to the MM study area, the lithological variations of the Cajamarca Complex (Pes, Pq, and Pmm) were analyzed to verify their differences with respect to the NRV study area.

Table 2 Description of sections of rock samples from the study area NRV (P4NRV, P8NRV, and P12NRV) and MM (P7MM, P8MM, and P14 MM)
Full size table

The lithological units of the rocks reported by González (2001) and Martínez et al. (2014) were confirmed. In the MM samples (P7MM, P8MM, P14MM) quartz content is higher than in the NRV samples; these rocks samples are harder and with lower degree of weathering.

Topological characterization of the fracture network

Table 3 shows the topological characterization of the fracture networks in the study areas. In the NRV (12 sampling points), the number of nodes varies between 11 and 59, while in MM (16 sampling points) it varies between 0 and 20. The average \({C}_{B}\) indicates that branches are mainly doubly connected in both study areas.

Table 3 Topological characterization of the fracture networks in the two study areas
Full size table

According to the connectivity thresholds in Table 1, the values of the parameters representing the connectivity of the fracture networks are illustrated in Figs. 4, 5, 6, and 7.

Fig. 4
figure 4

NRV–MM connectivity according to the 2D Fracture intensity and \({C}_{B}\) parameters. The red circle highlights the NRV points with the highest connectivity

Full size image
Fig. 5
figure 5

NRV–MM connectivity according to X/Y node ratio parameters and \({C}_{B}\)

Full size image
Fig. 6
figure 6

NRV–MM connectivity according to node proportion parameters X/I and \({C}_{B}\)

Full size image
Fig. 7
figure 7

NRV–MM connectivity according to node proportion parameters Y/I and \({C}_{B}\)

Full size image

According to Fig. 4, 83.3% of the NRV points exceed the connectivity threshold (\({C}_{B}\)>1), where 4 points present the highest connectivity with respect to \({C}_{B}\) and 2D intensity, which indicates that they have the highest branch connection (\({C}_{B}\)) and the highest sum of the fracture lengths in an area of 1 \({{\text{m}}}^{2}\) (2D intensity). These 4 points with elevated connectivity and intensity are Pes (P12NRV located near the San Jerónimo fault and P9NRV to the west of the Molinos fault), CL-PRE (P4NRV located in the Valle Nereidas sector) and Kvc (P16NRV located on the Villamaria-Chinchiná road). In the MM area, only 55.6% of the points exceed the thresholds. In addition, 2 points in the MM zone have a high 2D Intensity, but no branch connection.

In Fig. 5, 58% of the NRV points are above the connectivity thresholds, indicating that have a higher proportion of X nodes with respect to Y, and a high connection of branches. The 2 NRV points with the highest connectivity correspond to Kvc and Pes, with Pes having a higher branch connection than Kvc, while Kvc has a higher 2D intensity than Pes. 83.3% of Pes are connected. In the MM zone, 38.9% of the points are above the thresholds and 6 points belonging to Pmm do not present connectivity, indicating little fracturing in this lithological group.

As shown in Fig. 6, 58.3% of the points in the NRV zone are above the connectivity thresholds, showing a high proportion of X nodes with respect to I, and a high connection of branches. Meanwhile, in the MM zone, the value decreases to 50%.

41.7% of the points in the NRV zone (Fig. 7) are above the connectivity thresholds, indicating high proportion of Y-nodes with respect to I-nodes, and high branch connectivity. On the other hand, in the MM zone the percentage increases to 44.4%.

Considering all the information shown in Figs. 4, 5, 6, and 7, both zones present high connectivity. However, there is a different intensity of fractures, which is higher in the NRV study area. Therefore, it can be observed that the greater the presence of geological faults, the higher the level of fracturing. Similarly, the mineralogical composition may also play an important role in the fracture intensity: where a higher content of hard minerals (such as quartz) is observed, the greater is the resistance of the rocks to fracturing, as observed in the MM area, where there is a greater presence of lithologies with high quartz content, such as the Pmm and Pq group (González 2001).

Regarding the Pes group, the points sampled in the NRV presented a greater connectivity than those in the MM area. In addition, during field work it was observed that the rock samples were different in spite of being from the same lithological group: the NRV samples presented a higher degree of alteration, greater foliation, and a different mineralogical composition, which was verified by the petrographic analysis (Table 2). On the other hand, the PRE-Ruiz Lavas Group (NRV study area) presented a lower connectivity than Pes due to the low presence of connection nodes, but a considerable intensity of fractures.

Hydraulic conductivity and fracture permeability

In this study, the hydraulic aperture was considered equal to the mechanical aperture, because it was not possible to determine the fracture roughness, since they were observed at outcrop and there was no access to the other fracture planes. In addition, as mentioned by Singhal and Gupta (2010), those apertures can be considered equal when fractures are open (with mechanical openings between 0.5 and 2.5 mm), which is the case here. The hydraulic conductivity of the fractures was then estimated from Eqs. 7 and 12 using the mechanical fracture aperture measured at the outcrops, whose mean value was \(1.847\) \(mm\) for the NRV zone and \(2.267\) \(mm\) for the MM zone; the kinematic viscosity was set at \(1.308 \times {10}^{-6} \frac{{{\text{m}}}^{2}}{{\text{s}}}\) for the NRV zone (considering an average temperature of 10 °C) and \(1.141 \times {10}^{-6} \frac{{{\text{m}}}^{2}}{{\text{s}}}\) for the MM zone (considering an average temperature of 15 °C). Secondary permeability for the two zones was estimated from Eqs. 7 and 8 using the mechanical aperture. The values of hydraulic conductivity and secondary permeability are shown in Table 4.

Table 4 Average values of hydraulic conductivity and secondary permeability of the fractured medium
Full size table

In the NRV study area, the Cajamarca Complex is characterized by having a low primary permeability and fluid flow is dominated by secondary permeability (Vélez et al. 2018), which is provided by the high intensity of fractures as a consequence of the convergence of the fault systems present in the area. Consequently, as suggested by Vélez et al. (2020), it was corroborated that the secondary permeability of the Pes fractures in the NRV (\(3. 59 \times {10}^{-7} {{\text{m}}}^{2}\)) is much higher than the permeability of the matrix of the same lithology (\(1. 93 \times {10}^{-15} {{\text{m}}}^{2}\)), leading to heat transfer by advection and convection. However, other parameters, such as fracture wall roughness and flow channeling (Klepikova et al. 2021), heat source, thermal conductivity of the matrix, velocity of the fluid, and thermal dispersivity in the fracture (Jaoude et al. 2018) have an impact on heat transfer. Other authors have also shown that the structural control exerted on the groundwater flow by regional fault structures may be an important factor in influencing the convective component of heat transport (Chicco et al. 2019). It is, therefore, important to assess the relevant parameters to predict groundwater flow and heat propagation (Jaoude et al. 2018). According to Sævik and Nixon (2017), the connectivity of a fracture network is directly proportional to the effective permeability. Therefore, if the influence of fractures on fluid flow must be properly analyzed, not only the secondary permeability found by Eq. 8, but also the effective permeability must be considered (Figs. 4, 5, 6, and 7). The secondary permeability and hydraulic conductivity found in the Cajamarca Complex of MM is higher than that found in the NRV (Table 4), because Eqs. 7 and 12 are a function of the mechanical aperture and the values measured in the field were higher in this zone compared to those measured in the NRV. However, in the NRV study area the Pes visually has a higher foliation intensity (observed in the study area) which also influences fracture connectivity and favors fluid circulation (Kleine et al. 2016). Connectivity, as mentioned above, is high in both study areas, but fracture intensity is higher in the NRV area. Although the secondary permeability and hydraulic conductivity are higher in the MM area, if permeability, connectivity, and fracture intensity are analyzed together, the NRV area shows a greater potential for transporting fluids through fractures.

Consequently, proper characterization of fracture networks is a key point in the exploration and characterization of geothermal reservoirs when fluid circulation is controlled by secondary permeability since it is necessary to understand the flow pathways in hydrothermal systems. This is essential to achieve sufficient permeability to produce fluid flow rates that are suitable for the development of economically successful geothermal projects (Genter et al. 2010; Giordano et al. 2013; McClure and Horne 2014; Kneafsey 2018; Lepillier et al. 2020).

Macrostructural fracture networks

Figure 8 shows the connectivity based on branch connections at the macrostructural level in both study zones, together with the location of the hot springs present in the NRV study area. The green color illustrates the zones of low connectivity, the yellow color medium connectivity, and the orange and red colors high connectivity.

Fig. 8
figure 8

Connectivity map at macrostructural scale: a NRV study area and rose diagrams of the strike of fractures sampled in the field with NW–SE, NE–SW and N–S trends; b MM study area and rose diagrams of the strike of fracture sampled in the field with NW–S

Full size image

It is noteworthy that in the NRV study area there is a greater macro-structural connectivity, where connection and intensity of faults and discontinuities is higher. The points characterized by the highest connectivity are located at the intersection between Santa Rosa, Telaraña, Campoalegre, and San Jerónimo faults and in the Rio Claro sector, where Nereidas, Santa Rosa, and Samaná Sur faults intersect. However, the Pes samples collected in the sectors of high macrostructural connectivity do not necessarily match the values of higher connectivity at microstructural scale, such as one of the points sampled in the Rio Claro sector (P3NRV) is below the thresholds established in Table 1, although it shows high foliation intensity. On the contrary, the faults have a great influence on the direction of the fractures at the microstructural level, where the NW–SE trend is mainly reflected in the fractures, which is similar to that of faults Nereidas, Rio Claro, Quebradanegra, and El Billar faults. In the NRV area, most hot springs are located in zones of high macrostructural connectivity, which confirms that secondary permeability allows fluid flow (Fig. 8a). In a study conducted in northern Portugal (Ferreira et al. 2014), they also evidenced hot springs in fault zones with extensive fracture systems and suggest that the detection of fractures is a key aspect at the time of drilling the reservoir in order to be successful in the production of wells.

On the other hand, in the MM study area there are less orange and red zones (Fig. 8b), indicating lower connectivity than in the NRV study area. This low connectivity is due to a lower presence of faults and possibly to the mineralogical content of the rocks, since here the lithologies have higher quartz content which is recognized for its stability, resistance to mechanical and chemical weathering, and can increase rock strength (Keykha and Keykha 2013; Naqvi 2013). It should be noted that the MM area does not have the same amount of available structural information as the NRV, because it is located outside of the region of geothermal interest. Here, fracture networks at the microstructural coincides with the strike of the main faults in the area (Rio Hondo and Eden Fault).

Conclusions

This work presents the topological characterization of fracture networks at microstructural and macrostructural scale in two areas with different structural characteristics, located in the department of Caldas (Colombia). The first one is characterized by high influence of faulting to the northwest of the Nevado del Ruiz Volcano (NRV); the latter lies to the northeast, between the municipalities of Marulanda and Manzanares (MM). The main objective is to understand the influence of faulting over the connectivity of fracture networks and to quantify secondary permeability and hydraulic conductivity, which are key aspects to understand fluid flow for the development of geothermal projects in fractured reservoirs. Structural data and rock samples were collected at outcrops to perform the topological characterization of the fracture networks. It was found that both connectivity and fracture aperture directly influence the effective permeability of fracture networks, favoring the flow of subsurface fluids, which could be evidenced with hot springs that are mostly located in areas of high connectivity. Faulting has a great influence on the intensity, quantity, and strike of fractures at microstructural scale, but not necessarily on the interconnectivity of the fracture networks. The outcrops in both study areas show a high microstructural connectivity, but higher intensity of fractures was observed in the area with greater influence of faulting (NRV). On the other hand, the lithology of main interest in the NRV area (Pes, belonging to the Cajamarca Complex) was one of those that presented higher connectivity and fracture intensity, which, combined with the high foliation observed in the field work, increase its effective permeability.

Although it is improbable that the fractures at surface will show the same connectivity, intensity and aperture trend at depth, outcrops are generally easily accessible and together with connectivity measurements based on microstructural and macrostructural topology characterization (Sævik and Nixon 2017), constitute a key parameter to improve the construction of numerical and provide a valuable tool for preliminary geothermal explorations. With this surface fracture information, among other geophysical data, random statistical distribution models of fractures at depth can be developed (Rile, 2005; Darcel et al. 2013; Gutierrez and Youn 2015; Lei et al. 2017; Chen et al. 2023), and later could be validated using information from drilled boreholes, once available.

For further work, it is recommended to develop numerical models of groundwater flow and heat transfer that consider temperature-dependent fluid properties, to anticipate the behavior at reservoir conditions, as well as to apply statistical approached to predict the behavior of fracture connectivity and aperture at depth. Variation of fracture aperture can also be analyzed: roughness (Barton and de Quadros 1997) may be included in the models, as well as and a transient behavior, since mineral precipitation that can partially reduce fracture aperture during time (Jones and Detwiler 2016; Song et al. 2022).