This paper was prepared for presentation at the Unconventional Resources Technology Conference held in Houston, Texas, USA, 23-25 July 2018. The URTeC Technical Program Committee accepted this presentation on the basis of information contained in an abstract submitted by the author(s).
In this study, experiments were done on samples from the Marcellus, Woodford, and Eagle Ford shales. The experiments showed that samples from these formations were grossly water-wet, mixed-wet and oil-wet, respectively. The correlation of average wettability index with total organic carbon (TOC) showed that 5 wt% is the critical TOC content required to achieve connectivity and generate oil-wet pathways. Similarly, correlation of average wettability index with clay content showed that <10 wt% clay, samples are oil-wet and >65 wt%, they are predominantly water-wet, and between 10 and 65 wt% clay content, samples exhibited mixed wettability. The threshold values of 5 wt% TOC and 10 wt% clays represent the same volumetric fraction (~10%) of the rock. The figure of 10% can be thought of as percolation threshold for connectivity in shale rocks.
Scanning electron microscope (SEM) imaging done on representative samples (one per formation) was used to quantitatively assess the fraction of different pore types. The fractions of different pore types were in agreement with the observations from the macroscopic imbibition experiments. For instance, oil-wet Eagle Ford samples had a higher fraction of organic pores (22.5%) while water-wet Marcellus samples had a higher fraction of inorganic pores (40%). The samples from all the three shales had a high fraction of mixed-wet pores (Marcellus 57%, Eagle Ford 69%, and Woodford 68%). This knowledge of fractions of different pore types can be instrumental in modeling connectivity pathways.
Pore connectivity controls fluid flow in porous media. Well-established methods such as permeability measurements or mercury-injection porosimetry are used to evaluate pore connectivity. However, these methods have limits in unconventional shale formation mainly because of the presence of different-wettability pores with large capillary pressures associated with them. In the present study, we have used the sequential spontaneous imbibition of brine and dodecane to study pore connectivity in shales. The present study shows the presence of water-wet pores, hydrocarbon-wet pores, and mixed-wettability pores. The water-wet pores are the continuous phase, and are controlled by the amount and distribution of inorganic water-wet minerals. The hydrocarbon-wet pores are essentially contained within the organic matter and require a minimum amount of total organic carbon (TOC), ≈3 wt%, to form a connected network of hydrocarbon wet pores. The ≈3 wt% of TOC is a necessary, but not sufficient, condition for developing connectivity throughout organic bodies. The mixed-wettability pores appear to enhance the systems connectivity.
This study introduces a novel outlook on a shale-pore system and on the potential effect of pore compressibility on the production performance. We divide porosity of the system into accessible and inaccessible pores, and incorporate inaccessible pores with grains into the part of the rock that is not accessible. In general, accessible pores contribute to flow directly, whereas inaccessible pores do not.
We present a mathematical model that uses mercury-injection capillary pressure (MICP) data to determine the accessible-pore and inaccessible part of the rock (IRP) compressibility as a function of pressure. During MICP testing in a typical shale sample, the rock sample experiences conformance, compression, and intrusion as effective pressure increases. We characterize the compressibility value dependent on MICP data as a function of pressure. The calculated compressibility values for accessible pores generally appear to be much greater (two to three orders of magnitude) than those of IRP.
Next, we evaluate how calculated accessible-pore-compressibility values affect gas recovery in several shale-gas plays. Our results suggest that substitution of total pore compressibility with accessible-pore compressibility can significantly change the reservoir-behavior prediction. The fundamental rock property used in many reservoir-engineering calculations including reserves estimates, reservoir performance, and production forecasting is the total pore-volume (PV) compressibility, which has an approximate value typically within the range of 1×10-6 to 1×10-4 psi-1 (Mahomad 2014). By recognizing the part of the pore system that actually contributes to production and identifying its compressibility, we can substitute total pore compressibility with accessible-pore compressibility. The result changes the value by nearly two orders of magnitude.
The outcome of the paper changes the industry’s take on prediction of reservoir performance, especially the rock-compaction mechanism. This study finds that production caused by rock compaction is in fact much greater than what has often been regarded, which will change the performance evaluation on a great number of reservoirs in terms of economic feasibility.
Matskova, N. (Universite de Poitiers) | Pret, D. (Universite de Poitiers) | Gaboreau, S. (BRGM) | Cosenza, P. (Universite de Poitiers) | Brechon, R. (Universite de Poitiers) | Gener, I. (Universite de Poitiers) | Fialips, C. I. (Total E&P) | Dubes, G. (Total E&P) | Gelin, F. (Total E&P)
Non-conventional shale reservoirs are characterized by multi-scaled pore systems closely associated with a variable spatial distribution of mineral grains and organic matter. Therefore, only an integrated multi-technique approach can provide a quantitative balance in pore size distribution. Such a balance is not achievable when each of the different methods is applied on different samples that are randomly selected within the formation being studied. In this multi-technique study, the initial cores were first scanned by μ-tomography with an achieved resolution of 78.8 μm to visualize potential spatial heterogeneities at the core scale. This approach identified similar and homogenous areas of interest for localized bulk measurements. In particular, the sub-sample positions were selected, in the 3D views, in areas presenting (i) similar non-clay grain amounts, (ii) similar X-ray linear attenuation coefficients, and (iii) no cracks or macro-heterogeneities.
This approach was applied to seven core samples (dimensions of ~7x7 cm) from the Vaca Muerta formation (Argentina), taken from three different exploration wells in zones with various hydrocarbon types (condensate gas, dry gas, and oil). These samples were chosen on the basis of well log data and their petrophysical interpretations. Lithofacies with similar mineral distribution but contrasting physical properties (resistivity, porosity, and density) and organic matter content were selected.
Conventional bulk methods were used to characterize the pore network of the selected sub-samples: mercury intrusion porosimetry, gas adsorption, nuclear magnetic resonance spectroscopy and helium pycnometry. Some of these methods are generally applied to non-localized crushed powder samples, which is inappropriate for evaluating preserved microstructure from the micro- to the macro-porosity range. For this reason, most of the acquisitions were performed on preserved core blocks, including nitrogen gas adsorption experiments, which were performed with kinetic control of adsorption equilibrium. This innovative approach has provided, for the first time ever, a full set of truly comparable data, localized on 3D views, displaying quantitative balances of pore size distribution in shale reservoirs.
Development of source-rock resources relies on the rigorous knowledge of their petrophysical properties such as porosity, permeability, and hydrocarbon saturation. In parallel, a concise description of the wettability and pore structures is commended. This paper presents a detailed Nuclear Magnetic Resonance (NMR) T2 study of the wetting characteristics and pore structure in organic-rich source rocks from different locations including the Eagle Ford formation. Although these rocks are highly laminated and calcite dominated, our studies indicated that they have distinct different pore structure and connectivity, and differ in how TOC is dispersed within the rock fabric. We believe that the entailed findings could influence our thinking on how best to produce these shales, wellbore stability, drilling fluid selection and other asset development actions.
Source-rock samples with varied amount of total organic content (TOC) were drilled perpendicular or parallel to the laminations. The samples were cut into twin plugs which were sequentially saturated by spontaneous imbibition of 5% KCl brine and diesel (oil). The NMR T2 measurements were used to determine the fluid imbibition rate and amount, as well as the porosity associated with organic and inorganic components of the source rocks. The fracture apertures were obtained via an application of characteristic T2 cutoff times to the NMR T2 distributions. The mineral elements, phases and TOC of the rocks were measured using X-ray Fluorescence (XRF), X-ray Diffraction (XRD) and HAWK pyrolysis, respectively.
The prevalence of surface relaxation on the NMR dynamics was prominent as the transverse relaxation took place at time scales (T2 ≤ 100 ms) much shorter than their bulk values. The overall wettability of the samples showed a mixed character as the brine and the oil had been intimately imbibed. Nevertheless, the details of the wetting behavior of the Eagle ford samples and the other samples were different. For instance, Eagle Ford samples imbibed larger volumes of brine and faster than oil, on the contrary the other samples imbibed larger volumes of oil and faster than brine.
The apparent preference of oil on the other samples is attributed to their high TOC compared to the Eagle Ford samples. Upon imbibition in these samples, brine is observed to flow along the clay rich bedding planes. In fact, the interaction between brine and clay is identified to be the potential driver of the rock stability problems especially near the wellbore; however it is constrained by the type of residing clays. The discrepancies in the wetting traits are magnified by the presence of fractures which enhanced the network connectivity of both hydrophobic and hydrophilic pores or even across them. Furthermore, the fractures allowed the fluids to surpass the vertical bedding planes and thus accelerating the fluid distribution processes inside the pore space. The fracture apertures were found to range from 1 μm to 15 μm which are typical values for source rocks (
Decline curve analysis (DCA) is the most common method to forecast future production and to estimate ultimate recovery and reserves/well. The traditional form of DCA proposed by Arps is however restricted to boundary-dominated flow regimes. In unconventional shale plays, it is however likely that the transient flow regime may occur for the first few months or years of production. Consequently, the applicability of the traditional forms of DCA to early-time production data may not be appropriate.
This work is divided in to two main sections. In the first section, we apply some of the more recent decline curve models proposed for shale wells to production data acquired from the Woodford shale in Oklahoma. A comparison is then made between the different decline curve models in terms of their ability to replicate production history in a forecast mode. In the second section, we extend earlier work performed by other authors to compare well performance across different shale plays and over different time periods. The DCA presented in the earlier work utilizes a simple Arps decline; in this work, we utilize a composite decline curve that works for the linear transient flow regime and subsequent boundary dominated flows. Our work indicates that while the Arps decline curve analysis approach may be erroneous, in comparison to a more rigorous DCA, the errors are less than 20% in predicting EUR.
Interpretation of NMR relaxometry in organic-rich mudrocks still remains a challenge for petrophysicists. A reliable numerical simulation of NMR response in these rocks helps in better understanding of the NMR relaxometry data and the corresponding petrophysical properties (e.g., the wettability and fluid distribution in the pore space). The hydrocarbon-wetting or mixed-wetting characteristics of organic-rich mudrocks have significant impacts on hydrocarbon recovery from the rocks, but quantification of wettability using conventional well-log interpretation methods remains a conundrum. The objectives of this paper are (a) to introduce a NMR two-phase simulation method to model the NMR responses in organic-rich mudrocks and (b) to investigate the effects of wettability alteration and fluid distribution on NMR relaxometry using numerical simulations.
We simultaneously simulate the NMR responses from both water and hydrocarbon phases in organic-rich mudrocks using a random-walk algorithm. The main input to this two-phase NMR simulator is a digital rock matrix reconstructed from 3D pore-scale images of organic-rich mudrock samples, including water, inorganic grains, hydrocarbon, and kerogen. The pore-scale digital images are obtained from FIB-SEM (Focused Ion Beam Scanning Electron Microscope) images. The output of the NMR simulator is the NMR transverse magnetization decay in the rock matrix. The NMR
We successfully simulated NMR