Advertisement

Transport in Porous Media

, Volume 131, Issue 1, pp 157–174 | Cite as

Foam Flow and Mobility Control in Natural Fracture Networks

  • Bergit BrattekåsEmail author
  • Øyvind Eide
  • Sigbjørn Aasheim Johansen
  • Snorre Sande Vasshus
  • Andreas Grøteide Polden
  • Martin A. Fernø
Article
  • 109 Downloads

Abstract

We study the generation and flow of foam through rough-walled, fractured marble rocks that mimic natural fracture systems in carbonate reservoirs. Flow was isolated to the fracture network because of the very low rock permeability of the marble samples and foam generated in situ during co-injection of surfactant solution and gas. The foam apparent viscosities were calculated at steady pressure gradients for a range of gas fractions, and similar to foam flow in porous media, we identified two flow regimes for foam flow in fractures: a high-quality flow regime only dependent on liquid velocity and a low-quality flow regime determined by the gas and liquid velocities. Variations in local fluid saturation during co-injection were visualized and quantified using positron emission tomography combined with computed tomography.

Keywords

Foam generation in fractures Foam apparent viscosity Local fluid flow PET imaging 

1 Introduction

Fossil fuels are predicted to be a part of the energy mix for the next decades, and enhanced oil recovery (EOR) is necessary to supply the world energy demand. Approximately 60% of all known oil reserves are contained in carbonate reservoirs, which often exhibit significant reservoir heterogeneities, therein fractures (Roehl and Choquette 1985). Fractures combined with oil-wet or mixed-wet reservoir characteristics, often present in carbonate rocks, may cause primary and secondary recovery methods to recover less oil than expected. During water or gas floods, the injected phase will often prefer to flow through the fractures rather than entering into the matrix to displace oil, resulting in poor macroscopic and microscopic sweep, and early breakthrough of the injected phase in production wells. Early breakthrough of gas due to fractures is also highly relevant in developing carbon capture and storage (CCS) projects, where the object is to maximize CO2 storage in underground reservoirs. The effect of pure CO2 injections is greatly reduced when fractures occur, as the CO2 will tend to flow through the highly conductive fractures rather than the matrix (Fernø et al. 2015a, b). Reducing the mobility of gas in fractures, by usage of foams, has potential to greatly improve gas injections in fractured reservoirs, for both EOR and CCS applications. Foams are gas bubbles dispersed in a continuous aqueous phase separated by thin liquid films known as lamellae (Yan et al. 2006). Foam increases the apparent gas viscosity to improve sweep efficiency and oil recovery during gas injections, and has recently been suggested to provide mobility control in fractures and systems featuring large permeability contrasts (Kovscek et al. 1995; Seethepalli et al. 2004; Haugen et al. 2012), with a factor of up to 600 (Buchgraber et al. 2012). Foam generation in fracture systems was previously observed (Fernø et al. 2016; Brattekås and Fernø 2016). Snap-off appears to be the primary mechanisms for generating foam in fractures (Kovscek et al. 1995), and foam generated in fractures differs from foam generated in porous media by having a much larger bubble size. Kovscek et al. (1995) found that bubbles formed in fractures are roughly four times larger than bubbles in foam generated under the same conditions in Berea sandstone, because there are fewer snap-off sites in a fracture than in a porous medium. Foam in fractures is believed to behave as bulk foams (Sheng 2013); however, this might not apply to tight fractures.

Osterloh and Jante (1992) suggested the presence of two flow regimes for foam flooding in porous media: the high-quality, strong foam, regime and the low-quality, weak foam, regime. The regimes are divided by the gas fraction fg*, which position is dependent on the porous medium and the limiting capillary pressure (Pc*). In the strong foam regime (i.e., above fg*), liquid velocity alone controls the flow pressure gradient, while the gas velocity controls the pressure gradient in the weak foam regime. Plotting the pressure gradients as functions of liquid flow rate (x-axis) and gas flow rate (y-axis) forms the “L-plot,” named by the characteristic shape of the pressure contours (Alvarez et al. 2001). The gas fraction for the limiting capillary pressure, fg*, was found to range between 0.94 and 0.96 in porous media (Osterloh and Jante 1992). In fractures, the gas fraction for the limiting capillary pressure may be as high as 0.99 (Pancharoen et al. 2012).

We investigate foam generation and flow in rough-walled fracture networks in marble with varying aperture and at different length scales. Foam is often studied in micromodels or other artificial models of fractured or porous media and less investigated in real rock fractures. Marble is a metamorphic rock made by regional metamorphism of carbonate sediments and has the same surface and mineral composition as sedimentary carbonate rocks. Using marble material to study foam generation and flow through fractures is advantageous: The marble is impermeable with minor porosity, which limits fluid storage and flow in the rock matrix. Hence, fracture flow mechanisms may be studied without contributions from the matrix. The calcite fracture surface ensures liquid–solid interactions similar to sedimentary carbonate rocks, where foam can be applied for enhanced oil recovery (EOR), CCS or aquifer remediation. Surfactant solution and gas were co-injected in different fracture networks, constituting open, partially open and tight fractures. Foam flow in two-dimensional marble fracture networks was previously monitored by direct visual observations (Fernø et al. 2016). Foam was generated during co-injection in a fractured marble tile placed between two transparent plates, and foam propagation and sweep efficiency were monitored directly. In this work, we applied the same materials and fluid systems to generate stable foam with the aim of increasing the complexity of the fracture systems by using three-dimensional marble core plugs of diameters varying from 2.5 to 10 cm (1.5’’-4’’). Foam flow was monitored by differential pressure development, and we used PET-CT imaging to quantify foam behavior in situ. In situ imaging has been extensively used to analyze and quantify fluid transport in porous media. We have previously identified that flow diversion in fracture networks by foam or polymer gels may be visualized and quantified with imaging methods such as CT (Brattekås et al. 2013; Eide et al. 2015; Haugen et al. 2014), MRI (Brattekås and Fernø 2016; Brattekås 2018) and more recently PET-CT (Brattekås et al. 2017; Brattekås and Seright 2018). This paper investigates in situ foam generation and transport through a fracture network, where the marble rock material inhibits fluids diversion into the matrix. We present the approach used to study foam in fractures, and summarize observations and indications from several imaging experiments.

2 Methods and Materials

2.1 Natural Fracture Networks in Marble

Foam generation and flow rheology were studied in fractured marble cylindrical cores with different fracture geometries. Cylindrical core plugs were drilled from a large marble block, and the cores were fractured using the Brazilian test principle (Mellor and Hawkes 1971) to create fractures with rough surfaces that favor foam generation (Fig. 1). For the purpose of this study, we consider the rock matrix as impermeable and at experimental conditions applied flow occurred only in the fracture network. Hence, observed foam generation during co-injection test is from the fracture network only, without any flow or bubble generation in porous media. The fracture geometry varied depending on the fracturing process: A network of large, longitudinal open top and bottom fractures, with smaller interconnected fractures around the center, was generated using a short, tapered edge (shown in Fig. 1(1)). A longer tapered edge created a network of partially open fractures with larger aperture at the perimeter and smaller apertures toward the center. A network of closed, rough-walled fractures was generated with a uniform edge (uniform thickness, 2 mm thick).
Fig. 1

Left and middle: Fracturing device and three-step procedure: (1) a marble core plug was placed horizontally between the two sharp edges (top and bottom) along its entire length; (2) load was stepwise increased (10 bar increments) and developing fractures were confirmed visually to determine the fracture load pressure; and (3) after fracturing, the load pressure was released within seconds and the fractured core was carefully removed, assembled and wrapped in plastic foil. Right: a fractured marble core. The fractures were rough-walled (left image). The properties of the sharp edge controlled the aperture and depth of the open fracture close to the core circumference

2.2 Preparation of Fractured Marble Cores

Surface weathering and contamination from outdoor storage were removed by washing the marble slabs with a high-pressure water washer. Cylindrical core plugs were drilled out using a diamond coated drill bit, and cores were fractured as described above. The fractured marble systems were re-assembled to create two different fracture networks: tight or partially open. Fracture networks with increased tortuosity were generated by stacking up to six shorter fractured core plugs with alternating fracture directions. The fracture networks used in this work are characterized in detail below. The fracture networks were saturated by brine under vacuum and the fracture volume measured. Brine was thereafter injected at several different liquid rates while measuring the differential pressure. The hydraulic fracture aperture (h) was estimated from Buckingham using the following equation:
$$ \Delta P = \frac{12\mu vL}{{h^{2} }}, $$
(1)
where ∆P is the pressure drop across a fracture of length L when a fluid of viscosity µ is injected using a volumetric flow rate v per unit width of the fracture. Buckingham reported that this equation was valid for flow through slots of fine clearance and unit width, which is not the case in our spatially complex fracture networks. Nevertheless, Witherspoon et al. (1980) concluded that fluids behave similarly in deformable rock fractures, and we applied Eq. (1) to describe the hydraulic fracture aperture for the system and the analogue Darcy law for fractures:
$$ K = \frac{{h^{2} }}{12} $$
(2)
to calculate the fracture network permeability. Constant net overburden pressure and no-flow boundaries were applied for all injections and the geometries and hydraulic connectivity of the fracture networks did not change during tests. This was confirmed with CT imaging for three fracture networks (B, C and E, see Table 1).
Table 1

Overview and properties of the fracture networks

Fracture network

Description

Diameter (cm)

Length (cm)

hBuckingham (m)

Kfracture (D)

A (1.5i-3)

Tight

3.90

14.57

2.2 × 10−6

0.4

B (1.5i-4)

Tight

3.88

14.50

1.9 × 10−6

0.3

C (2i-1)

Partially open

5.08

14.90

14.3 × 10−6

17.4

D (2i-3)

Partially open

4.95

14.52

E (4i-3)

Partially open

9.39

26.60

5.1 × 10−6

2.2

2.3 Characterization of the Fracture Networks

Fracture networks B, C and E were characterized with positron emission tomography (PET) and X-ray computed tomography (CT) in terms of fracture aperture and degree of heterogeneity and are described in detail below. Two different PET-CT scanners were used. A Siemens Biograph TruePoint medical PET/CT scanner, previously described by Fernø et al. (2015a, b), with a high capacity in terms of weight and size of experimental equipment was used for fracture networks C and E. A preclinical 80 W Nanoscan PC imager, previously described by Brattekås et al. (2017), was used for tight fracture network B.

2.3.1 Tight Fracture Networks

The tight fracture networks A and B (see Table 1) constituted a stack of six short core plugs each with a single fracture (see Fig. 2). The inlet fracture (in core 1) was horizontal, and each subsequent fracture was rotated 90° resulting in alternating horizontal and vertical fractures. In addition, the space between each core constituted fractures, each one vertical and perpendicular to the direction of flow. The stacked systems were assembled using epoxy resin (Fig. 2(II)) and imaged with the 80 W Nanoscan PC imager PET-CT scanner, 30 μm3 voxel size. Local flow properties for fracture network B (Fig. 2(III)) were visualized with PET during a miscible brine/brine displacement: The fracture network was initially fully saturated with brine and displaced with an identical brine labeled with radioactive 18F-FDG. (18F was produced in an in-house cyclotron and used to synthesize 18F-fluorodeoxyglucose 18F-FDG, which is a water-soluble fluorine radioisotope with a half-life of t1/2 = 109 min.) Local spatial fracture geometries (Fig. 2(IV)) were identified with CT, and the pressure drop was measured to calculate fracture properties (see Table 1). Fracture network A was used for co-injections with gas fractions (fg) in increasing or decreasing order (Table 2), whereas network B was placed in the PET-CT scanner during co-injections to visualize local fluid saturations.
Fig. 2

Preparing and characterizing stacked, tight fracture network B: (I) each core was fractured individually; (II) six fractured core plugs stacked and assembled with epoxy resin to maintain the fracture network and fix no-flow boundaries during foam injection. The fracture through the first core plug was horizontal, whereas each subsequent fracture was rotated 90°; (III) PET image of the cores 1–4 when saturated with 100% brine labeled with 18F-FDG; (IV) CT image where the fracture in core 2 was vertical and the fracture in core 3 horizontal

Table 2

Experimental schedule for the different fracture networks

Fracture network

Description

Experimental schedule

A

Tight

Baseline and foam co-injections from fg = 1 → 0

Baseline and foam co-injections from fg = 0 → 1

B

Tight

Baseline and foam co-injections from fg = 1 → 0 in PET-CT

C

Partially open

Baseline and foam co-injections from fg = 1 → 0.6 in PET-CT

Baseline and foam co-injections from fg = 0.6 → 1 in PET-CT

D

Partially open

Baseline and foam injection at different const. vliquid with different vgas

Baseline and foam injection at different const. vgas with different vliquid

E

Partially open

Baseline and foam co-injections from fg = 0.4 → 1 in PET-CT

2.3.2 Partially Open Fracture Networks

The partially open fracture networks C and D were fractured using the procedure outlined in Fig. 1, resulting in large and highly conductive fractures that were aligned longitudinally. Local spatial fracture geometries in network C were visualized using CT (Siemens Biograph TruePoint PET-CT scanner, voxel size of 0.28 × 0.28 × 0.6 mm3). CT images (Fig. 3) showed longitudinal, open fractures along the bottom and top of the core, with one or several narrow, transverse fractures connecting the open fractures through the core center. A longitudinal cavity (referred to here as vug) was also present in the middle of the core, at normalized length L = 0.47–0.79. The average hydraulic fracture aperture was calculated from injection tests to be 14 µm, with a corresponding permeability of 17 D. The open, longitudinal top and bottom fractures conducted the majority of fluid flow during brine injection, visualized by PET imaging during a miscible water–radioactive water displacement.
Fig. 3

Fracture network C characterized with CT. 2D (XY-direction) slices for different normalized fracture widths. A 2D bird’s-view of the core (XZ-slice from the core middle) and a top-down 3D view of the core are shown in the right

Fracture network E consisted of three stacked and fractured marble core plugs (each core with nominal diameter 9.4 cm; length 8.8 cm). Local spatial fracture geometries (Fig. 4) were visualized by CT (Siemens Biograph TruePoint PET-CT scanner). A bird’s-eye view of the stacked cores (Fig. 4, left) and a selection of slices along the core length (Fig. 4, middle) show a larger vug between the first and second core in the stack. Due to its transversal orientation, the vug did not significantly influence flow. A three-dimensional mask, based on fracture size (threshold CT signal) and a high measured activity during initial waterflooding (threshold PET signal), was made (Fig. 4, right) and used for further analysis. The mask constitutes a network of large fractures; transverse fractures emerged during stacking of three individually fractured marble core plugs. Each longitudinal fracture was oriented perpendicular to the next, similar to tight fracture networks A and B previously described. These large fractures constituted the main fluid flow path. Fractures also existed outside of the fracture mask. Some fracture apertures were below CT resolution (0.48 × 0.48 mm), but short (< 2 mm long) wider fracture (up to 2 mm) was also identified. Most of these were disconnected from the main pathways and did therefore not influence flow to a significant degree.
Fig. 4

Fracture network E characterized with CT: 2D (XY-direction) slices at different lengths along the core. A 2D bird’s-view of the core (XZ-slice from the core middle) and a top-down 3D view of the core are shown in the left, with a 3D view of the network of large fractures on the right: This fracture mask was made by thresholding the CT and PET images, and excluded tight fractures elsewhere in the network

2.4 Foam Flow in Natural Fracture Networks

Co-injections with gas (Nitrogen) and brine or surfactant solutions were performed in the fracture networks. Injection tests with brine without surfactant present are referred to as baselines: Brine 1 (5 wt% NaCl, 5 wt% CaCl2*2H2O) was used in fracture networks A and B, whereas brine 2 (1 wt% NaCl) was used in fracture networks C, D and E. The brine composition was changed after initial injections using brine 2 in tight fracture systems, where the permeability was observed to increase for long-term injections due to calcite precipitation. In the partially open fracture networks C, D and E, significant changes in permeability were not observed. The surfactant (Huntsman SURFONIC® L24-22) was mixed in synthetic brine 2 (1 wt%) for injection in fracture networks C, D and E.

Foam was not pre-generated: The gas and surfactant solution were injected through separate inlets and formed foam in situ in the fracture network. Co-injections were performed at ambient temperature with back pressures of 1–6 bars to limit pressure fluctuations caused by gas compressibility effects (Rossen 1990; Buchgraber et al. 2012). Pre-defined gas fractions and injection rates were used to identify foam properties in fractures: The experimental schedule (injection rates, gas fractions and order of gas fractions) was slightly different for each fracture network and is outlined in Table 2. Differential pressure measurements and PET/CT imaging identified foam flow properties through the different fracture networks. The experimental setup used for co-injections is shown in Fig. 5.
Fig. 5

Schematic of the experimental setup used for co-injections. The main variable in the setup was the fracture network, which was either mounted in a Hassler core holder (as shown in the figure) or encapsulated in epoxy

3 Results and Discussion

3.1 Foam Behavior in Fractures

Flow regimes with in situ foam generation during co-injection were studied using fracture network D (partially open fractures, see Table 1) initially saturated by surfactant solution. Three constant liquid injection rates were used (9, 18 and 54 ml/h), and the gas injection rate was stepwise increased from 3 to 900 ml/h for each constant liquid rate when the differential pressure stabilized. The pressure gradient during constant liquid rate injections (Fig. 6) has a crossover between vertical and horizontal lines that identifies a change from high- to low-quality foam regime. These foam flow regimes were frequently reported in the literature for porous media, and a few reports in model fracture systems exist (Kovscek et al. 1995; Yan et al. 2006; Buchgraber et al. 2012; AlQuaimi and Rossen 2018), but are very limited in real fracture networks. The observed behavior indicates that foam in fractures behaves similarly to foam flowing in porous media. The critical gas fraction, fg*, i.e., the gas fraction identifying the transition from low- to high-quality foam regime, occurred between fg* = 0.82 ± 0.05, which is lower than reported fg* values for porous media (with typically fg* > 0.94).
Fig. 6

Differential pressure during co-injection for foam generation in fracture network D with liquid (x-axis) and gas (y-axis) injection rate during constant liquid rate co-injections. The experimental data points used to generate the contour plot are shown in the same color map as the contours

Foam fraction scans in tight fracture networks (A and B) were compared to partially open fracture networks (C and E) by calculating the apparent foam viscosity for a range of gas fractions (fg) and flow velocities (v) using an imbibition-like flow sequence with decreasing gas fractions from 1 to 0 (Fig. 7). Co-injections with surfactant solution and gas were compared to baselines without surfactant solution (brine only) using the same gas fractions. The apparent foam viscosities were higher than the baseline, indicating that foam generates in the fracture network. Several parameters influenced the pressure development including the co-injection rate, the gas fraction and the saturation history. In partially open fractures (C and E), the apparent foam viscosities increased with decreasing gas fraction between fg = 1–0.8 and decreased between fg = 0.7 to 0. Maximum apparent viscosity was identified at fg = 0.7–0.8 (fracture network C) and fg = 0.8–0.9 (fracture network E).
Fig. 7

Comparison of foam fraction scans for tight fracture networks (A and B) and partially open fracture networks (C and E). Foam generation during co-injection of gas and surfactant solution was observed in the real fracture systems without injecting pre-generated foam; hence, in situ foam generation was confirmed. The apparent foam viscosity in tight fracture networks (top, left) was constant for gas fractions 0.5 and below, and higher for lower velocities compared to the other fracture systems. Apparent foam viscosity in partially open fracture networks (C: top right; E: bottom left) increased for gas fractions below 0.6, whereas a shear-thinning behavior was observed for gas fractions 0.7 and higher. The velocity v (volumetric flow rate per unit width of the fracture) is included for each co-injection, and influenced stacked fracture networks (A, B and E) more than fracture network C, where foam apparent viscosity was similar for three velocities (12, 24 and 36) below fg = 0.6

Foam apparent viscosity in tight fracture networks (A and B) decreased at higher gas fractions (fg = 1.0–0.7) relative to lower gas fractions (fg = 0.5–0.0). Fracture networks A/B (tight) and E (partially open) did not have a continuous longitudinal fracture; rather, individually fractured core plugs were stacked to constitute a fracture system, where each longitudinal fracture was oriented perpendicular to the next (see Figs. 2, 4). In these fracture networks, the apparent foam viscosity was significantly influenced by velocity. In fracture network E, an increase in foam flow velocity by a factor of 1.5 led to an increase in foam apparent viscosity by a factor roughly between 1.6 and 2.2. In network A, a doubling of foam flow velocity increased the apparent foam viscosity by factors between 2 and 4. Due to the partially open fractures in network E, a significantly higher velocity was necessary to obtain high foam viscosities, compared to tight fracture networks. In network C, partially open fractures were oriented in the direction of flow and spanned the entire length of the core, and foam apparent viscosity was not significantly influenced by flow velocities at low to intermediate gas fractions (fg < 0.6). The observations summarized in this section and in Fig. 7 indicate that the type of fracture, specifically the fracture aperture and permeability, influence foam flow. In addition, the orientation of conductive fractures relative to the foam flow direction also seems to have an impact on foam strength, specifically its dependency on flow velocity.

3.2 Can Foam Flow be Visualized Using PET-CT?

In situ imaging was previously used to identify fluid diversion in fractured core plugs during foam injection to improve mobility control and sweep efficiency. Recent combination of advanced medical imaging and dynamic fluid flow identified PET-CT as a highly useful tool to identify fluid flow in fractures (Brattekås et al. 2017; Brattekås and Seright 2018). Here we use CT imaging to characterize complex fracture networks and PET images to quantify foam flow in the fracture networks during co-injections at different gas fractions, by labeling the injected surfactant solution with radioactive 18F-FDG. The PET signal scales linearly with the volume of surfactant solution present, and the PET images could thus be used to study the local spatial saturation of the aqueous phase.

The image resolution of the PET scanner used during most co-injections (Siemens Biograph TruePoint PET-CT scanner) restricted direct characterization of foam texture (e.g., bubble size) due to limited voxel size (0.28 × 0.28 × 0.6 mm3). The high spatial resolution in the Nanoscan PET-CT scanner (30 μm3 voxel size) may enable such investigations in vuggy marble tile systems, but the fracture aperture in network B was below resolution (see Table 1). PET imaging could, however, give a strong indication of foam forming in the fracture networks: Fig. 8 shows the first transversal fracture (in direction of flow, between stacked core plug 1 and 2) in network E during co-injection of gas and labeled surfactant solution. Gas and liquid were equally distributed in the fracture volume and not prone to gravity segregation, which indicates formation of foam ahead (upstream, close to the inlet) of this position.
Fig. 8

Foam distributed in a transversal fracture between stacked core plugs in fracture network E. Surfactant is distributed throughout the fracture, without gravity effects. The fracture was not smooth or of equal aperture. A higher surfactant saturation in some areas (warm colors) may be explained by fracture geometry

PET also enabled comparison of dynamic foam behavior for different gas fractions, i.e., observation of time-averaged foam behavior to quantify differences between the gas fractions. The PET signal is directly related to the volume of surfactant present, and all scans were normalized to the volume of surfactant injected to enable direct comparison. Thus, the acquisition times for images at higher fg were higher compared to lower fg, at the same volumetric flow rates. The decay of radioactive tracer (half-life 109 min) was accounted for in the image reconstruction software.

3.2.1 PET Imaging of Foam Flow Through Tight Fractures

The apparent foam viscosities during co-injections with surfactant solution in tight fracture network B (see Table 1 for fracture network properties) ranged between 20 and 30 cP, more than four times higher than the baseline (nominal 5 cP). Interestingly, PET imaging revealed that liquid (brine or labeled surfactant solution) was equally distributed in the fracture network in both cases (when measured in one dimension along system length; see Fig. 9). Due to the low fracture conductivity, gas and water were efficiently distributed in all fractures during baseline co-injections, and an improvement in sweep efficiency was not observed during foam generation. An increase in both the baseline and foam co-injection PET signal is observed with an increase in gas fraction. This is most likely an artifact associated with the normalization of continuous foam flow to injected surfactant volume; data for higher gas fractions are averaged over a longer period of time, during which more foam has passed through the fracture system.
Fig. 9

Normalized PET signal (linear with volume of surfactant solution present in region of interest) in fracture network B during co-injections of water/gas and surfactant solution/gas. Baseline (without surfactant solution present) was similar to co-injections with surfactant solution present that generated foam. The measurements were normalized to the PET signal attained in an initially fully water saturated fracture network

3.2.2 PET Imaging of Foam Flow in Partially Open Fractures

Fracture network E was constructed similar to fracture networks A and B, where several core plugs with longitudinal fractures were stacked resulting in fractures angled perpendicular to each other along the direction of flow. An important difference was the partially open fractures present in network E that influenced foam strength (Fig. 7). In addition, tight fractures spanned out from the main fracture network in some locations, where fluids could be diverted with improved foam strength.

Image reconstructions were first performed over a constant time interval for all gas fractions. As expected, the PET signal (Fig. 10) was higher for fg = 0.4 (highest liquid-to-gas ratio applied) and directly reflects foam quality, i.e., that the amount of surfactant injected is higher for lower gas fractions over the same time interval.
Fig. 10

Gamma counts for fracture network E when the PET signal is reconstructed for the same period of time for each gas fraction co-injected. A higher PET signal is recorded during injection at lower gas fractions, because more surfactant is injected. The peaks in PET signal are the vertical fractures between stacked core plugs, where a larger volume of fluids accumulates

To analyze differences in foam flow between gas fractions, the PET signal was reconstructed again (when the pressure gradient for each gas fraction was stable) to have the same surfactant solution volume injected for each gas fraction. The effects of foam on gas trapping and amount of surfactant solution present could then be visualized (Fig. 11) in the main conductive fracture network (represented by the 3D mask in Fig. 4). Fluid saturations inside the fracture mask could be compared to fluid saturations outside the mask to estimate the propagation of foam into tight fractures spanning from the main fracture network. This analysis was challenging for fracture network E because the PET signal-to-noise ratio was generally low outside of the main fracture network given by the mask and much lower in absolute number due to large difference in volume.
Fig. 11

Two-dimensional distribution of surfactant solution (labeled with radioactive 18F-FDG) in fracture network E at four different gas fractional flows: gas fraction 0.4 (fg04-1 to fg04-3), 0.5 (fg05-1 to fg05-3), 0.7 (fg07-1 to fg07-3) and 1.0 (fg1-1 to fg1-3). The majority of the surfactant solution was flowing through the larger fractures and the void between stacked core plugs. A visual difference between gas fractions was not evident

Qualitatively comparing the two-dimensional images shows that the surfactant saturation was relatively stable for each gas fraction, and profiles (normalized to gas fraction 0.4–1, Fig. 12) enabled a quantitative comparison. (Signal from perpendicular fractures between the stacked cores was removed because these initially dominated the signal due to large volumes.) The profiles confirm that there is relatively little change in the PET signal within each gas fraction, showing that foam flow was stable and that foam behaves similarly for a given gas fraction for different times with constant pressure gradient. A clear trend in the PET signal between gas fractions is not evident (Fig. 12, right), but the signal does change, indicating that some of the surfactant may flow into the tight fractures outside of the defined mask.
Fig. 12

Left: Normalized PET signal along the length of fracture network E for four gas fractions, normalized to gas fraction 0.4–1. Between three and five scans were obtained for each gas fraction. For each gas fraction, the surfactant saturation profile remained close to constant for each scan. Right: The relative change in PET signal (i.e., the amount of surfactant solution present) did not vary significantly for the gas fractions (0.4, 0.5, 0.7 and 1.0) used

Further analysis was performed to compare foam diversion for different gas fractions in regions where both vugs and tight fractures were present. Such regions were unfortunately mainly present close to the outlet in fracture network E (in the third fractured core in the field of view), where fluctuations in the back pressure may influence foam flow. Hence, a conclusive analysis could not be performed in this network. Fracture network C was therefore used for this investigation. Fracture network C was also characterized as partially open, but differed from the other fracture networks with the main fractures continuously oriented longitudinally, in the flow direction. In addition, tight fractures and vuggy features were present in most of the core; thus, foam diversion could be analyzed for the entire length of the fracture. Foam diversion (Fig. 13) was analyzed at different normalized lengths from the inlet: 0.32 (slice 27); 0.54 (slice 45); 0.81 (slice 67) and 0.93 (slice 77). The PET signal in the tight fractures was normalized to the entire cross section in each slice and for each gas fraction. Image acquisitions for network C were performed for 10-min intervals with 20-min pause between scans regardless of the pressure development. Reconstructions included the entire acquisition period for all gas fractions; the normalization process accommodates this, although complementary analyses (e.g., comparing foam flow between gas fractions for the entire fracture network) require a different post-processing routine. Fractional flow of gas was first increased from fg = 0.6–1 and then reduced back to fg = 0.6. Diversion of foam (i.e., increased surfactant signal) into tight fractures generally increased with increasing gas fraction along the core. At two length positions (0.54 and 0.93), very little hysteresis was seen, and foam diversion was similar for all given gas fractions regardless of the foam scan direction. The peak value at fg = 1 reflects that foam remains in the tight fractures, while the more conductive fractures are swept by pure gas (thus significantly reducing the PET signal).
Fig. 13

Foam diversion as a function of gas fraction for four different slices. Increased diversion of foam into tight fractures was observed with increasing gas fractions, which reflects the high apparent foam viscosities achieved at high gas fractions (Fig. 7). The high signal at gas fraction 1 (only gas injected, no surfactant with signal) is due to the flow of pure gas through the vug-like fracture features

4 Conclusions

The following observations were made during co-injections of gas and surfactant solution into marble fracture networks:
  • PET was used to verify foam generation, by observing foam accumulation in transversal fractures. Measured apparent viscosities showed that foam formed in situ in all presented fracture networks.

  • PET imaging was used to observe average foam properties over several different time intervals, and compare dynamic foam transport through fracture networks using different gas fractions. Average foam properties were stable for a given gas fraction over several different time steps for all fracture networks.

  • Natural fracture networks with large differences in fracture aperture were not easily analyzed by PET because very high signal in large fractures; thus, signal in closed fractures bears resemblance to noise.

  • Two foam quality regimes (one high-quality and one low-quality regime) were observed in partially open fractures, and the characteristic L-plot could be generated.

Notes

Acknowledgements

The authors would like to thank the Research Council of Norway for financial support under Grant Number 268216—Nanoparticles to Stabilize CO2-foam for Efficient CCUS in Challenging Reservoirs. The PET-CT imaging was performed at the Molecular Imaging Center (MIC) and was thus supported by the Department of Biomedicine and the Faculty of Medicine and Dentistry, at the University of Bergen, and its partners.

References

  1. AlQuaimi, B.I., Rossen, W.R.: Foam generation and rheology in a variety of model fractures. Energy Fuels Article ASAP (2018).  https://doi.org/10.1021/acs.energyfuels.8b02178 CrossRefGoogle Scholar
  2. Alvarez, J.M., Rivas, H.J., Rossen, W.R.: Unified model for steady-state foam behavior at high and low foam qualities. SPE J. 6(03), 325–333 (2001)CrossRefGoogle Scholar
  3. Brattekås, B., Fernø, M.A.: New insight from visualization of mobility control for enhanced oil recovery using polymer gels and foams. In: Romero-Zerón, D.L. (ed.) Chemical Enhanced Oil Recovery (cEOR)—A Practical Overview. InTech, London (2016)Google Scholar
  4. Brattekås, B., Seright, R.S.: Implications for improved polymer gel conformance control during low-salinity chase-floods in fractured carbonates. J. Pet. Sci. Eng. 163, 661–670 (2018)CrossRefGoogle Scholar
  5. Brattekås, B., Haugen, Å., Ersland, G., Eide, Ø., Graue, A., Fernø, M.A.: Fracture mobility control by polymer gel- integrated EOR in fractured, oil-wet carbonate rocks. In: Presented at EAGE Annual Conference & Exhibition incorporating SPE Europec. London, UK (2013)Google Scholar
  6. Brattekås, B., Steinsbø, M., Graue, A., Fernø, M.A., Espedal, H., Seright, R.S.: New insight into wormhole formation in polymer gel during water chase floods with positron emission tomography. SPE J. 22(01), 32–40 (2017)CrossRefGoogle Scholar
  7. Buchgraber, M., Castanier, L.M., Kovscek, A.R.: Microvisual investigation of foam flow in ideal fractures: role of fracture aperture and surface roughness. In: SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, Society of Petroleum Engineers (2012)Google Scholar
  8. Eide, Ø., Ersland, G., Brattekås, B., Haugen, Å., Graue, A., Fernø, M.A.: CO2 EOR by diffusive mixing in fractured reservoirs. PetroPhysics 56(1), 23–31 (2015)Google Scholar
  9. Fernø, M.A., Eide, Ø., Steinsbø, M., Langlo, S., Christophersen, A., Skibenes, A., Ydstebø, T., Graue, A.: Mobility control during CO2 EOR in fractured carbonates using foam: laboratory evaluation and numerical simulations. J. Pet. Sci. Eng. 135, 442–451 (2015a)CrossRefGoogle Scholar
  10. Fernø, M.A., Gauteplass, J., Hauge, L.P., Abell, G.E., Adamsen, T.C.H., Graue, A.: Combined positron emission tomography and computed tomography to visualize and quantify fluid flow in sediments. Water Resour. Res. 51(9), 7811–7819 (2015b)CrossRefGoogle Scholar
  11. Fernø, M.A., Gauteplass, J., Pancharoen, M., Haugen, A., Graue, A., Kovschek, A., Hirasaki, G.: Experimental study of foam generation, sweep efficiency, and flow in a fracture network. SPE J. 21(04), 1140 (2016)CrossRefGoogle Scholar
  12. Haugen, Å., Fernø, M.A., Graue, A., Bertin, H.J.: Experimental study of foam flow in fractured oil-wet limestone for enhanced oil recovery. SPE Reserv. Eval. Eng. 15(02), 218–228 (2012)CrossRefGoogle Scholar
  13. Haugen, Å., Mani, N., Svenningsen, S., Brattekås, B., Graue, A., Ersland, G., Fernø, M.A.: Miscible and immiscible foam injection for mobility control and EOR in fractured oil-wet carbonate rocks. Transp. Porous Media 104(1), 109–131 (2014)CrossRefGoogle Scholar
  14. Kovscek, A., Trethewaya, D.C., Persoff, P., Radke, C.J.: Foam flow through a transparent rough-walled rock fracture. J. Pet. Sci. Eng. 13(2), 75–86 (1995)CrossRefGoogle Scholar
  15. Mellor, M., Hawkes, I.: Measurement of tensile strength by diametral compression of discs and annuli. Eng. Geol. 5(3), 173–225 (1971)CrossRefGoogle Scholar
  16. Osterloh, W.T., Jante Jr., M.J.: Effects of gas and liquid velocity on steady-state foam flow at high temperature. In: SPE/DOE Enhanced Oil Recovery Symposium. Society of Petroleum Engineers, Tulsa, Oklahoma (1992)Google Scholar
  17. Pancharoen, M., Fernø, M.A., Kovscek, A.R.: Modeling foam displacement in fractures. J. Pet. Sci. Technol. (2012).  https://doi.org/10.1016/j.petrol.2012.11.018 CrossRefGoogle Scholar
  18. Roehl, P.O., Choquette, P.W.: Carbonate Petroleum Reservoirs. Springer, New York. Softcover ISBN: 978-1-4612-9536-5.  https://doi.org/10.1007/978-1-4612-5040-1 (1985)
  19. Rossen, W.R.: Theory of mobilization pressure gradient of flowing foams in porous media: I. Incompressible foam. J. Colloid Interface Sci. 136(1), 1–16 (1990)CrossRefGoogle Scholar
  20. Seethepalli, A., Adibhatla, B., Mohanty, K.K.: Physicochemical interactions during surfactant flooding of fractured carbonate reservoirs. SPE J. 9(04), 411–418 (2004)CrossRefGoogle Scholar
  21. Sheng, J.J.: Enhanced Oil Recovery: Field Case Studies. Gulf Professional Publishing, Waltham. ISBN: 9780123865458 (2013)Google Scholar
  22. Witherspoon, P.A., Wang, J.S.Y., Iwai, K., Gale, J.E.: Validity of cubic law for fluid flow in a deformable rock fracture. Water Resour. Res. 16(6), 1016–1024 (1980)CrossRefGoogle Scholar
  23. Yan, W., Miller, C.A., Hirasaki, G.J.: Foam sweep in fractures for enhanced oil recovery. Colloids Surf. A 282–283, 348–359 (2006)CrossRefGoogle Scholar

Copyright information

© Springer Nature B.V. 2019

Authors and Affiliations

  1. 1.Department of Physics and TechnologyUniversity of BergenBergenNorway

Personalised recommendations