Offshore Pre-salt carbonates are the most relevant assets in Brazil. Since they have only been in production for the last two decades, their petrophysical characterization is of utmost importance to build reservoir models that can effectively estimate reserves and forecast production. The rock and fluid complexity of these carbonates increase the timeline and uncertainty of special core analysis programs. In this context, innovative technologies, such as digital rock analysis, can support and strengthen the petrophysical characterization of this challenging carbonates. In this work, we propose a methodology based on petrophysical trends to improve the definition of representative digital volumes of a reworked carbonate sample from the Pre-salt. In addition, digital static and dynamic properties were simulated at the pore scale and then upscaled for the whole plug. The porosity, permeability, formation factor and capillary pressure results showed a positive match with laboratory mercury injection capillary pressure and routine core analysis. The upscaling procedure was key to capture the dominant features at both the micro (pore) and macro (plug) scales commonly found in complex carbonate rocks. This workflow optimized the use of rock and fluid samples, reduced the timeline of the special core analysis program to 60% and supported minimizing the uncertainty of the traditional laboratory programs by improving quality control.

Identifying depositional facies and diagenetic modifications from BHI logs in sections without core is challenging due to heterogeneities in fabric of the carbonaceous rock, which are below the resolution of the BHI logs in these complex reservoirs. This is mainly due to different diagenetic processes resulting in different types of reservoir rocks (RRTs) with different petrophysical properties in the same geological facies. This work describes an innovative new workflow that uses a deep learning model to identify heterogeneities of fabric (textures) in complex carbonate reservoirs in sections without core; as input we use conventional (gamma ray, density, neutron, sonic), NMR, BHI (acoustic imaging) logs, core CT images and physical porosity and permeability measurements from plugs. This new workflow combines supervised deep learning and unsupervised machine learning methods, consisting of four steps: (1) Identification of textures in Core CT (2) Calibration of features of BHI logs in core section (3) Training a deep learning model with cropped images from BHI logs to extract texture curves (4) train a machine learning regression model with texture curves, WL logs and plug porosity and permeability values to propagate prediction of textures, porosity and permeability along the well. The texture curves generated by the ML model are in good match with the recognized core section textures (CT Images). The results show that the NMR Total Porosity has a good correlation with the porosity predicted by the regression model. In addition, the petrophysical data from plugs measured in the laboratory show a good match with the porosity and permeability predicted in the core zone. In the non-core zone, the predicted textures by the ML model relate to the heterogeneities of the carbonaceous rock fabric and the propagation of the petrophysical properties have good correlation with the predicted textures. In some cases, vuggy porosity combined with fractures do not allow correct porosity measurement with NMR or density tools due to the sensitivity of these tools to poor borehole conditions. In these cases, the NMR presents a fast relaxation of the T2 distribution due to the heterogeneity of these carbonaceous fabric and therefore the porosity and permeability measurements will be affected. The ML model uses CT Textures to calibrate BHI Crops in the core section zone and NMR and Basic logs as input, improving the recognition of the textural heterogeneities and refine the traditional Rock typing of these complex Pre-Salt carbonate reservoirs.

Advances in the fields of information technology, computation, and predictive analytics have permeated the energy industry and are reshaping methods for exploration, development, and production. These technologies can be applied to subsurface data to reliably predict a host of properties where only few are available. Among the numerous sources of subsurface data, rock and fluid analysis stand out as the means of directly measuring subsurface properties. The challenge in this work is to maximize information gain from legacy pdf reports and unstructured data tables that represented over 70 years of laboratory work and investment. The implication of modeling this data into an organized data store means better assessment of economic viability and producibility in frontier basins and the capability to identify bypassed pay in old wells that may not have rock material. This paper presents innovative and agile technologies that integrate data management, data quality assessment, and predictive machine learning to maximize the company asset value using underutilized legacy core data. The developed machine learning algorithms identify potential outliers, benchmark the valuable data against current industry standards, increase the confidence in data quality and avoid amplifying error in predicting reservoir properties. The workflow presented in the paper is expected to reduce uncertainties in subsurface studies caused by limited core data, improper analog selection, high cost, limited time for acquiring new cores, and long delivery times of core analysis data. The workflow reduces the requirement for subsurface formation evaluation rework as new data becomes available at later project stages resulting in optimized field development. The workflow enhanced by machine learning also improves the prediction and propagation of reservoir properties to uncored borehole sections. In conclusion, managing legacy core data and transforming it to generate new subsurface insights are critical step to establish a reliable database in support of business excellence and the digitalization journey. Innovative machine learning tools continue to unlock new values from legacy core data that significantly impact the entire reservoir life cycle including reserves booking, production forecasting, well placement, and completion design.

Direct numerical simulation at the pore scale on three-dimensional (3D) digital volumes obtained by high-resolution x-ray computed tomography (CT) images is a powerful tool for helping predict petrophysical properties. However, obtaining sufficiently high resolution and large field-of-view 3D CT images that capture multiscale heterogeneities in a single image is often not possible. Thus, methods for integrating and upscaling properties from multiple scales at various resolutions to plug scale are necessary. The methodology and results for upscaling capillary-dominated, two-phase flow in rocks based on multiscale CT images using trends are presented and discussed in this paper.

In this paper, we extend pore-morphology-based methods proposed by Hazlett (1995) and Hilpert and Miller (2001) to simulate drainage and imbibition in uniformly wetting porous media and add an (optional) entrapment of the (non -)wetting phase. By improving implementation, this method allows us to identify the statistical representative elementary volume and estimate uncertainty by computing fluid flow properties and saturation distributions of hundreds of subsamples within a reasonable time-frame. The method was utilized to study three different porous medium systems and results demonstrate that morphology-based pore-scale modeling is a viable approach to assess the representative elementary volume with respect to capillary dominated two-phase flow. The focus of this paper is the determination of the representative elementary volume for multiphase-flow properties for a digital representation of a rock. (C) 2016 Elsevier Ltd. All rights reserved.

Abstract Digital Rock Physics (DRP) has significantly evolved in the last few years and added invaluable contributions in improving core characterization and in providing high quality advanced SCAL measurements, emphasized through various studies/papers (SCA-2012–03 Kalam et al). This paper represents a unique DRP SCAL study that includes primary drainage capillary pressure (Pc) as part of Swi establishment and relative permeability (Kr) measurements done on four whole core (WC) samples from two different carbonate formations with a stylolite layer in between. The aim of the study was to evaluate how DRP results would compare with physical SCAL measurements done – on the same WC samples as a composite, as well as on plug samples from the same formations/layers – in a leading international core analysis lab in USA. The DRP results were up-scaled to the individual WC level and compared with the SCAL results from the corresponding layers. The DRP technology in this study also provided the capability of up-scaling the results to the WC composite which was used by the lab to assess the effect of the stylolite layer on the water flood. The comparison showed excellent matches between the physical and DRP-derived Pc and Kr data. The paper outlines the DRP methods used to determine the SCAL properties of the three formations. The laboratory measurements of SCAL properties took six years while the DRP work that followed blindly (without any knowledge of the laboratory results) was completed in six months. This demonstrates the effectiveness of the DRP technology in providing high quality SCAL data in a timely fashion regardless of sample size. Impact of possible wettability changes and sensitivities on one of the WC composite constituent component was also easily established unlike the high risk laboratory tests. This is the first water-oil displacement validation study results on reservoir whole cores of four inch diameter at full reservoir conditions using DRP.

Summary URTeC 1562626 As more organic rich mudstone resource plays are developed internationally, the need to understand flow potential and long term well performance increases dramatically. Many international locations have limited infrastructure for the economic development of these low permeability formations. Therefore, operators require comprehensive rock data and careful reservoir modeling to help reduce the risk of early-stage development. This paper describes the methods and results of a project designed to quantify the range of expected permeability and relative permeability in samples from a shale formation in Colombia. Porosity versus absolute permeability trends were determined for about 44 well samples using digital rock physics (DRP) methods. Results show rock quality that is equal or better than many prolific North American shales, including Marcellus and Eagle Ford. These samples average about 6% organic material content by volume. The total porosity range observed is from about 3 to 15%. For total porosity of 4% or above, the horizontal permeability is generally above 100 nanodarcy (nd). For porosity of 8%, horizontal permeability is typically 1000nd or more. From these 44 samples, several were selected for relative permeability analysis. Using a Lattice-Boltzmann numerical method, imbibition relative permeability computations (increasing fractional flow of water) were performed for oil-water systems for different scenarios including different contact angles ranging from oil to water wet, and different API values leading to different viscosity ratios.

Abstract A pilot study to evaluate the quality and validity of special core analysis (SCA) data from Digital Rock Physics (DRP) has provided results that are comparable to laboratory measurements. The DRP technique applied in this study employs the Lattice Boltzmann Method (LBM) for computing relative permeability (Kr(Sw)) and capillary pressure (Pc(Sw)) curves from high resolution digital pore structures obtained from micro-CT image data. The DRP processes, results, and comparisons with laboratory measurements on carbonate rock samples from different Saudi Arabian carbonate reservoirs are presented. DRP conventional core analysis (DRP-CCA) computations include porosity, permeability, formation factor, and dynamic elastic properties. DRP special core analysis (DRP-SCA) computations include Kr(Sw) and Pc(Sw). The translation of DRP-CCA and DRP-SCA determinations from imaged 4 mm subsamples to the 38 mm core plug-scale was achieved by upscaling the data for the various flow units and porosity structures in each plug. The number of flow units within each plug varied between one and four. The process of assembling plug-scale DRP-CCA and DRP-SCA properties is discussed. DRP-SCA results and laboratory measurements from similar rock types in the same wells are comparable and show inherent process and inter-lab uncertainties. The dynamic range of the computed relative permeability curves is superior to the laboratory measurements. The comparisons further showed the benefit of the DRP images and computations in capturing the detailed pore structure and fabric of the rock, especially in the capillary pressure responses. The DRP-SCA computations accentuate spontaneous imbibition and the transition to forced imbibition, a region that traditional laboratory methods may not adequately capture. Computations for different wetting conditions provide relative permeability data that cover all possible rock-fluid wettability states. Similar attempts in traditional laboratory experiments would be long, tedious and expensive. This work shows that DRP can provide satisfactory and complementary data for reservoir studies. The images are readily available and can be used for sensitivity studies. The workflow allows users to conduct their own validation tests, just as we have done, to determine the applicability of the method. Introduction In this work we consider the computation of porosity, conductivity, (relative) permeability and capillary pressure. These rock properties are of interest to petroleum engineers for characterizing a reservoir and measure the ability of the rock to transport fluids as well as the fluid pressure and saturation behavior exhibited through various hydraulic processes. These properties can be computed based on the pore space representation from CT or FIB-SEM based acquisition, physical models and their numerical implementation. Since rocks normally exhibit a strong multi-scale behavior, it is necessary to scan and compute properties on different scales. Scales can be differentiated by the resolution of the scans, but it is more useful to differentiate them conceptually. For example, there is a scale where the pore space is not directly visible at a certain resolution, but it is possible to identify different material regions that can be described by averaged properties. Physical processes can be described by equations like Darcy flow. We will call that scale the Darcy scale. On the other hand, there is the pore scale where the pore space is visible and the physical processes are described directly by pore scale physics. There can also be a mixture of both scales at a certain scan resolution. This includes large vugs embedded in one or several Darcy regions.

We propose a lattice Boltzmann model for immiscible two-phase Stokes flow with a local collision operator. The model is based on two different lattice Boltzmann automata, one for the flow field and one for an indicator function for the two different phases. The model is described in detail and verified by the following test-cases: a static bubble for the surface tension, a closed capillary tube for the contact angle and two phase flow in a concentric annulus for the viscosity ratio. In the appendix an asymptotic analysis for the derivation of the two-phase Stokes equation is given.

Digital rock physics (DRP) combines advanced 3D imaging techniques such as X-ray computed tomography (CT) scanning or focused ion beam scanning electron microscopy (FIB-SEM), segmentation algorithms to create a digital representation of the rock and advanced numerical methods to evaluate electrical, elastic, fluid flow and other properties of the rocks. A difficult problem in the numerical evaluation of relative permeability is to replicate the exact saturation sequences performed in SCAL experimental procedures including primary drainage and imbibition. In order to replicate these cycles, it is essential to define appropriate inlet and outlet boundary conditions to mimic the right flow field at the entrance and exit of a volumetric fraction of the plug, potentially located in any position inside the plug itself. Moreover, in order for a digital sample to be a representation of the whole plug, or only part of it in case of a plug with multiple flow units, it is important to make sure that the digital sample is a Darcian sample such that permeability can be defined and the sample is a volumetric representation of the plug. We present an approach to simulate fractional flow in a 3D digital rock by direct numerical simulation of the Stokes flow of two immiscible components through the rock. We use an improved method of the lattice Boltzmann method (LBM) to simulate the complex fluid movement through the rock that includes interfacial tension, wettability and viscous effects. Advanced boundary conditions are presented that allow the injection of varying fractional flow in a displacement process. A robust and simple way to verify Darcy’s law and to define a representative sample is presented. Primary drainage and imbibition cycles are performed on a carbonate sample, and the results are in good agreement with the experiments. The simulations are run on high performance computing (HPC) hardware to cope with the enormous computational load.

Permeability, unlike porosity, saturation, and lithology, is, arguably, impossible to infer from well data. Yet, it is the key parameter in reservoir simulation and also in well completion and stimulation decisions. The traditional method of obtaining permeability is in the physical laboratory where a pressure difference (ΔP) is applied to the opposite sides of a regularly shaped rock sample and the volumetric flux (Q) of the fluid is measured.

In this article a very efficient implementation of a 2D-Lattice Boltzmann kernel using the Compute Unified Device Architecture (CUDA™) interface developed by nVIDIA® is presented. By exploiting the explicit parallelism exposed in the graphics hardware we obtain more than one order in performance gain compared to standard CPUs. A non-trivial example, the flow through a generic porous medium, shows the performance of the implementation.