Gathering elemental data from rock material using X-ray florescence (XRF)-techniques is a relatively common practice in the Delaware Basin. As such, operators often amass very large datasets comprising thousands of sample points that superficially appear ideal for “big data” projects. However, varying sampling strategies between core and cuttings but also over the lifetime of an analytical program means that not all data can be directly or meaningfully compared.
In this paper we discuss the pros and cons of different sampling approaches when attempting to resolve the questions in the following areas: (a) regional or field wide correlation, (b) characterization of landing zones and (c) characterization of depositional environments. Additionally, we will also show how data can be successfully integrated from high resolution core analysis (2-12 inch vertical resolution) to low resolution (30ft-50ft) vertical cuttings.
Typically, XRF datasets from cores are gathered at a high resolution over a relatively thin interval to provide detailed facies and depositional environment information. However, as these datasets can show such small-scale variability, it often becomes difficult to integrate this data into more regional and broader scale findings. Conversely, cuttings can be sampled from a far higher number of wells and as such cover a much wider stratigraphic and geographic area, with sample density controlled by sample collection. Datasets gathered from cuttings are commonly collected for either more regional characterization or for landing zone identification. Due to the composite nature of cuttings, however, high resolution facies variability in sequences can be lost at a 20 or 30ft interval, making it tough to meaningfully define landing zones and integrate data from lateral wells using vertical cuttings data alone.
Using a series of Delaware Basin case studies, we firstly show how relative changes in elements and element ratios, from widely spaced cuttings material in vertical wells, are used to provide robust regional correlation frameworks that complement and refine stratigraphic tops picked from wireline log data. We will also highlight some of the difficulties in tying vertical cutting data with high resolution core data and strategies to resolve these issues. Finally, we will show how data acquired from core can provide detailed insight into landing zone characterization and insights into the rapidly changing depositional environments of the Delaware Basin, through Wolfcamp and Bone Spring time. This type of analysis can be easily integrated with data from lateral wells in order to understand and ultimately predict the type of lateral facies variation expected from different landing zones.
By systematically working through XRF datasets at different sampling resolutions we are able to utilize and integrate these datasets more efficiently. Ultimately, we can start to understand the optimal sampling resolution needed to solve different characterization or stratigraphic problems in different Delaware Basin formations and design more focused and “fit for purpose” analytical programs in the future.
Padin, Anton (Total Exploration and Production) | Pijaudier-Cabot, Gilles (Université de Pau et des Pays de l'Adour) | Lejay, Alain (Total Exploration and Production) | Pourpak, Hamid (Total Exploration and Production) | Mathieu, Jean-Philippe (Total Exploration and Production) | Onaisi, Atef (Total Exploration and Production) | Boitnott, Gregory (New England Research, Inc.) | Louis, Laurent (New England Research, Inc.)
Having a large number of layers in a reservoir model is computationally time-consuming, hence simulation of hydraulic fracturing in unconventional reservoirs usually rely on simplified, log-based models. In such models, vertical heterogeneities are upscaled to a few, averaged facies with homogeneous stiffness, stress, strength, toughness and natural fracture properties. In reality, however, unconventional reservoirs often contain singular heterogeneities and strong vertical and horizontal anisotropic properties that greatly affect fracture growth. The abundance of heterogeneities is believed to affect vertical hydraulic fracture growth (positively or negatively) due to stress differences, toughening effects at interfaces or the piling of thin lithologies with extreme, opposed stiffness or strength properties. From the simulation point of view, the challenge remains keeping a computationally-efficient but also representative (well-upscaled) model.
In this work, we focused on stiffness characterization, and particularly, on Young’s modulus calibration, and provide a concept-proof example for the Vaca Muerta formation, in Argentina. The current strategy in building stiffness models is to rely on sonic data to generate upscaled models with a few representative layers. A core acquisition program is normally put in place to calibrate the log-based model, and includes triaxial tests, where dynamic and static properties are measured at various stress conditions. Technically, characterization of each relevant lithology using these core measurements is possible, but given the degree of vertical heterogeneity, it would imply a very important logistical and economical effort. In addition, core plug selection is usually biased towards the stiffest rocks, leaving aside other facies, such as weak or ductile layers. As a result of these difficulties, core programs are usually limited to a few core plugs covering the target reservoir, leaving calibration of other units, and particularly of potential fracture barriers, unknown.
To address these difficulties, we concept-proved a core-to-log methodology that provides a fast calibration method for log-based elasticity. We measured rebound hardness in parallel to dynamic measurements of ultrasonic surface wave velocities (P and S) at the milimetric scale, then calibrated the results with discrete triaxial tests performed on plugs, representing all relevant lithological facies, and finally compared the results against log-based parameters. Our work shows that such integration helps at developing robust core-to-log elasticity relationships in the entire core length, eventually providing a proper foundation for better stiffness model prediction, at a fraction of the cost and time of traditional core acquisition programs.
Core analysis of shale via nuclear magnetic resonance (NMR) techniques presents a new set of challenges not existing in conventional cores. Specifically, the small pore sizes associated with shale leads to short NMR relaxation times making measurement of the pore size distributions difficult. In recent years, new NMR systems have been developed with the capabilities to measure these short relaxation times. It is also possible that these same samples have larger pores with longer relaxation times requiring signals to be acquired for a very long time. This means the Carr-Purcell-Meiboom-Gill (CPMG) NMR pulse sequence, will have thousands of NMR pulses at very high duty cycle (Time the pulse is on versus the time it is off). This large number of pulses can lead to heating of the samples of 20 to 30 C. This is especially problematic in shale samples where increases in experimental temperatures can lead to evaporation of liquids and/or phase changes in the organic matter present in the sample. In addition, the NMR signal itself is dependent on temperature leading to errors in the reported porosity of up to 10-15%.
In this work, we propose a modification to the CPMG pulse sequence where the NMR pulses are logarithmically spaced. This new spacing maintains the ability to measure the short relaxation times associated with the pore size distributions in shales while simultaneously reducing the heating of samples by applications of hundreds of thousands of NMR pulses.
A new CPMG pulse sequence with 512 pulses spaced logarithmically between 100 μs and 3600 μs was implemented. This sequence was repeated quickly in succession over several hours looking for changes in the observed signal due to heating. The sequence was tested with calibration samples, sandstone samples and shale samples. Each sample was then tested again employing a 500 ms long CPMG sequence with 2500 pulses linearly spaced at 100 μs. A comparison between the traditional CPMG sequence and the logarithmically spaced sequence was made.
The new CPMG sequence successfully eliminated any evidence of heating of the samples in the tests described above. For example, the calibration sample showed an approximate 5% reduction in observed signal over the length of the experiment due to heating when the traditional CPMG sequence was employed. When the same sample was tested using the logarithmically spaced CPMG sequence, there was no observed reduction in observed signal. Similar results were also observed for the shale and sandstone samples. A comparison of the raw data and the derived pore size distributions from each sequence for each sample was also made. This was to ensure that logarithmically spaced data was not degrading the ability to derive accurate pore size distributions.
Measuring pore size distributions accurately is one of the most important aspects of NMR core analysis. This work shows that measuring pore size distributions with CPMG sequences with short pulse spacings can result in a reduction in the observed signal due to heating and hence a reduction in the estimated pore volumes. Application of a pulse sequence which employs logarithmically spaced pulses eliminate the observed reduction in signal due to heating and results in the retrieval more accurate pore volumes. This methodology can be extended to all CPMG based NMR measurements such as T1-T2 maps and T2-Diffusion maps.
A nonupset pipe with an O.D. close enough to the I.D. of an inner pipe to cause hydraulic diversion. Used inside a screen during gravel packing to direct flow and gravel to the bottom of the screen and effect a tighter pack. Developing hydraulic diversion benefits with a wash pipe usually requires that the washpipe OD be at least 80% or the outer pipe or screen's ID.
Coring is essential to offshore exploration programs—but sometimes cores are taken from the wrong formation or return to surface in poor condition. One firm thinks it can solve these costly issues with a first-of-a-kind coring device that uses logging instruments that add accuracy and integrity. Pulled directly from the reservoir rock, core samples provide critical data used to determine how exploration should proceed. Until recently, core analysis remained old school, however, there is an ongoing transition to bring the process of core description into the digital age.
This paper compares the results of gas identification and lithology identification using pulsed-neutron spectroscopy in openhole and casedhole environments. Coring is essential to offshore exploration programs—but sometimes cores are taken from the wrong formation or return to surface in poor condition. One firm thinks it can solve these costly issues with a first-of-a-kind coring device that uses logging instruments that add accuracy and integrity. The rising cost of fracturing offers a significant incentive for finding ways to avoid unproductive rock. One entrepreneur says he can use standard well logs to target the slice of rock likely produce most, and avoid the rest.