|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Accurate and frequent mud checking is essential for optimum drilling operations. Careful measurement and maintenance of drilling fluid properties (density and rheology) maintain the primary well control barrier and optimize fluid hydraulics and hole-cleaning ability. However, a full mud report while drilling is provided only once or twice a day. Moreover, the measurements are mostly performed using traditional equipment. Test interpretation is subjective and might be biased and difficult to verify independently.
In this paper, we present an automated mud skid unit (MSU), which performs continual drilling fluid sampling and measurements at variable temperatures. The unit provides the non-Newtonian rheological constants characterizing a yield-power law fluid as well as the real-time friction factor and critical Reynolds number using a pipe-viscometer measurement approach. Other important fluid properties such as pressurized density, oil/water ratio, and temperature are provided using high-quality in-line sensors. The unit is controlled by a programmable logic controller coupled with a Linux operating system for data analysis. The system sends real-time data to WITSML data servers and provides detailed mud reports to engineers working either on-site or remotely.
The MSU was deployed in the Permian Basin by an independent operator for automated mud monitoring during unconventional shale drilling operations. Rheology, density, and phase content measurements were compared with conventional mud reports provided by the on-site mud engineer. High accuracy (error<5%) was observed in mud rheology tests. The pressurized mud-density measurements provided by the MSU proved to be more accurate than nonpressurized mud balance measurements, which were affected by mud aeration. Moreover, the MSU provided mud check data 25 times more frequent than those generated by the mud engineer at temperatures of 50 and 65.5°C. Drilling-fluid-related issues, such as chemical overtreatment as well as sudden changes in mud density, rheology, and oil/water ratio, were reported immediately to the drilling crew. This paper provides details about the measurement technology as well as the results from the field deployment of the MSU.
Chu, Wei-Chun (Pioneer Natural Resources) | Scott, Kyle D. (Pioneer Natural Resources) | Flumerfelt, Ray (Pioneer Natural Resources) | Chen, Chih (Kappa Engineering) | Zuber, Michael D. (Pioneer Natural Resources)
Pressure communication is commonly observed in fractured horizontal shale wells, particularly at early times when wells are placed on production. In this paper we present a new technique, based on the diffusion exponent from the power-law model, to quantify connectivity in multistage-hydraulic-fractured wells with complex fracture networks. In addition to explaining the theory and analysis techniques, we present examples using measured bottomhole pressure (BHP) from the Permian Basin Wolfcamp Shale that illustrate the utility of this technique to better understand the relationship between completion size, well spacing, and well performance.
Using the concept of anomalous diffusion, Chen and Raghavan (2015) developed a 1D, fractional-order, transient diffusion equation to model fluid flow in complex geological media. They showed that anomalous diffusion, which can be caused by heterogeneities in the matrix or the fracture system, exhibits a power-law behavior. In addition to Chen and Raghavan (2015), Acuña (2016) demonstrated that variations in matrix block sizes, fracture conductivity, and drainage shape also exhibit power-law behavior. While the approach from these two studies is somewhat different, they each demonstrated that a generalized power-law model is often more appropriate than traditional linear- or radial-flow pressure-transient analysis techniques for unconventional shale reservoirs. Further, each work shows that the power-law response can be related to some form of heterogeneity in the drainage volume.
While traditional techniques for estimating well interference have been previously developed and applied in conventional reservoirs, in this paper we focus on quantifying the magnitude of pressure interference (MPI) in unconventional reservoirs, which commonly demonstrate a generalized, power-law pressure response that is different from radial or linear flow. The examples presented in this paper are for wells from the Permian Basin Wolfcamp Shale. Under the framework of power-law behavior, our technique involves plotting pressure-interference-test (PIT) data in terms of the Chow pressure group (CPG), which enables us to define an indicator of connectivity reflecting temporal and spatial effects. On each test, we derive a diffusion exponent reflective of the MPI. We will show among other things that multiple PITs over time often indicate degrading connectivity between wells.
From PIT analyses in Permian Basin Wolfcamp Shale, we were able to establish a relationship between MPI and well spacing. The first example demonstrates analyses of PITs between wells during the production phase and also shows how connectivity between wells diminishes over time. A second example applies the same analysis techniques to quantify interwell connectivity during the post-stimulation phase by analyzing a pressure falloff (PFO) after communication with other wells. A third example illustrates an application of desuperposition to remove the effect of a power-law pressure trend (PT) on interference tests.
Techniques to analyze PITs assuming radial or linear flow have been previously developed; however, Raghavan and Chen (2018) showed that apparent radial or linear flow could exist under anomalous diffusion for heterogeneous reservoirs. In this work, we present a technique for analyzing power-law PIT data, which is typical of most horizontally fractured shale wells. This model is a unique approach to understanding flow behavior, quantifying well interference, and analyzing and predicting well performance in unconventional reservoirs. Our examples, which are based on high-quality BHP gauge data, show how this technique could shorten the cycle time for operators to determine the well spacing for a given completion design.
Zhang, Shuang (Pioneer Natural Resources) | Tang, Hewei (Rice University) | Hurt, Robert (Pioneer Natural Resources) | Jayaram, Vikram (Pioneer Natural Resources) | Wagner, Jed (Pioneer Natural Resources)
The topic of fracture complexity is commonly evoked when discussing hydraulic fracturing of unconventional reservoirs. In this context, it is typically considered beneficial to successful stimulation, as it provides increased surface area, relative to single planar fractures. However, in the near-wellbore region (NWR), this same fracture complexity, commonly referred to as tortuosity, can be detrimental to successful placement of fluid and proppant. In the extreme, if not properly identified and mitigated, fracturing stages may need to be abandoned which leads to unstimulated sections of the wellbore and reduced completions efficiency. Yet, the ability to adequately quantify this phenomenon during stimulation remains limited.
In this paper, we show how modern diagnostic techniques can be leveraged to provide insight into this critical region. Specifically, we combine interpretations from both fiber optic distributed acoustic sensing (DAS) and external downhole pressure gauges (BHG) to improve the characterization of the NWR. This project was executed during the stimulation of a horizontal well located in the Wolfcamp formation within the Midland Basin. We first review observations from the external cemented in place fiber and the external pressure gauges.
The second section presents an investigation of fracturing net pressures trends identified with external pressure gauges. We apply traditional Nolte-Smith fracture diagnostics to analyze fracture propagation and near-wellbore proppant dynamics. The net-pressure investigation reveals that even in unconventional reservoirs, Nolte-Smith diagnostic plots are applicable, when external pressure gauges are available. We show that near-wellbore proppant screen-outs identified by the Nolte-Smith plot are independently identified by Distributed Acoustic Sensing (DAS) data.
In the third section we have attempted to develop a process to quantify near-wellbore tortuosity, where machine learning (ML) algorithm(s) were utilized to estimate the friction pressure induced by near-wellbore tortuosity. The training, testing and validation needed for machine learning algorithm(s) were based on utilizing DAS data, downhole gauge data, pumping schedule and post fracturing reports. The studies indicate that friction pressure due to tortuosity is initially high within the transient rate period and decreases to stable values later within the stage. The validation studies show promising performance of ML algorithm(s) for near-wellbore friction pressure estimation, even without downhole gauge data as inputs. It is expected that with further development of ML algorithms needing limited training data shall allow development of diagnostic tools for better prediction of bottom hole treating pressures in wells without the need of acquiring high frequency downhole data.
The paper also makes an attempt to validate the application of Nolte-Smith plot in unconventional reservoirs, especially in characterizing the NWR. Additionally, fluid communication between stages highlights the importance of the NWR on ensuring stage isolation. Finally, the applied ML algorithm for near-wellbore tortuosity pressure estimation is shown to have a reasonable generalization performance, which may serve as a diagnostic tool for completion optimization.
Gupta, Ishank (University of Oklahoma) | Tran, Ngoc (University of Oklahoma) | Devegowda, Deepak (University of Oklahoma) | Jayaram, Vikram (Pioneer Natural Resources) | Rai, Chandra (University of Oklahoma) | Sondergeld, Carl (University of Oklahoma) | Karami, Hamidreza (University of Oklahoma)
Summary Petroleum reservoirs are often associated with multiple target zones or a single zone adjacent to nonproductive intervals. Real-time geosteering therefore becomes important to remain in zone or to dynamically steer toward a target. This requires knowledge of the petrophysical/rock mechanical properties of the rock surrounding the bit. Although logging while drilling can provide this information, a cost-effective and almost-real-time solution is lacking. In general, there is a depth lag, and therefore, a time delay, between what the logging-while-drilling sub relays to the surface and the bit performance. This study focuses on relating drill-bit- and drillstringperformance data in a machine-learning (ML) workflow to predict the lithology at the bit while drilling. The method we are proposing offers several advantages in terms of cost and time savings for real-time geosteering applications, where going out of zone requires costly intervention. In this study, we have used a public data set from Volve Field on the Norwegian continental shelf. Within our proposed workflow, as a first step, logs sensitive to lithology [such as density, gamma ray (GR), and sonic] are grouped into three electrofacies. We also had access to core data, which helped us interpret the electrofacies in terms of mineralogy. The three electrofacies corresponded to quartzrich (sandstone/siltstone), clay-rich (shale), and carbonate-rich (limestone) lithologies. The next step is to predict the electrofacies using various measurement-while-drilling (MWD) variables, such as rate of penetration (ROP), weight on bit (WOB), and several others that are monitored in real time. Supervised classification algorithms were used to relate real-time surface measurements to lithology. The algorithms were able to predict lithology in test wells with more than 80% accuracy. These results, although encouraging, constitute a small step toward drilling-automation/advisory systems. The development of such systems can prevent costly out-of-zone drilling and minimize rig time and equipment use, thereby potentially reducing capital expenditures.
Gupta, Ishank (University of Oklahoma) | Devegowda, Deepak (University of Oklahoma) | Jayaram, Vikram (Pioneer Natural Resources) | Rai, Chandra (University of Oklahoma) | Sondergeld, Carl (University of Oklahoma)
Planning and optimizing completion design for hydraulic fracturing require a quantifiable understanding of the spatial distribution of the brittleness of the rock and other geomechanical properties. Eventually, the goal is to maximize the SRV (Stimulated Reservoir Volume) with minimal cost overhead. The compressional and shear velocities (Vp and Vs respectively) can be used to calculate Young’s modulus, Poisson’s ratio and other mechanical properties. In the field, sonic logs are not commonly acquired and operators often resort to regression to predict synthetic sonic logs. We have compared several machine learning regression techniques for their predictive ability to generate synthetic sonic (Vp and Vs) and a brittleness indicator, namely hardness, using the laboratory core data. We used techniques like multi-linear regression, lasso regression, support vector regression, random forest, gradient boosting and alternating conditional expectation. We found that the commonly used multi-linear regression is sub-optimal with less-than-satisfactory predictive accuracies. Other techniques particularly random forest and gradient boosting have greater predictive capabilities based on several error metrics such as R2 (Correlation Coefficient) and RMSE (Root Mean Square Error). We also used Gaussian process simulation for uncertainty quantification as it provides uncertainty estimates on the predicted values for a wide range of inputs. Random Forest and Extreme Gradient Boosting techniques also gave low uncertainties in prediction.
Rostagno, Ian (The University of Texas at Austin) | Yi, Michael (The University of Texas at Austin) | Ashok, Pradeepkumar (The University of Texas at Austin) | van Oort, Eric (The University of Texas at Austin) | Potash, Ben (Pioneer Natural Resources) | Mullin, Chris (Pioneer Natural Resources)
Pipe rocking is a process used during slide drilling to reduce friction between the drillstring and the wellbore. Pipe rocking is widely practiced in unconventional drilling operations, either conducted manually or through an automated system. Often times, the rocking regime adopted in the field is based only on experience and may not be at optimum, leading to higher friction with poor force transfer to the bit and reduced rate of penetration. In addition, non-optimum pipe rocking can lead to accidental connection back-offs and poor toolface control.
This paper introduces the first rocking simulator based on real time and contextual data to provide the driller with a robust recommendation of the optimum rocking regime, i.e. guidance on the optimum number of forward and reverse wraps in the drillstring and the time period in which to generate these wraps.
A model was developed to optimize the pipe rocking regime, determined by the specifics of rotating the drillstring at a certain RPM for a certain number of turns in forward and backwards directions. The objective was to keep the directional toolface constant while optimally reducing sliding friction between the drillstring and the wellbore. A torque and drag model was used to obtain the frictional forces between the drillstring and the wellbore. Drillstring dynamics was then simulated using a torsional damped wave equation applying finite difference approximations. Finally, the angular deformation as a function of time and measured depth for each drillstring element was calculated.
Static friction is an important performance limiter when slide drilling with a downhole motor. Pipe rocking can be used as a low-cost technique to break the static friction in a section of the well and thereby reduce its negative effect. Pipe rocking simulation was used to find the rocking regime that maximizes the section of the string under conditions of dynamic friction, without losing toolface control. The torsional damped wave equation was used as a drillstring dynamics model because it successfully accounts for the surface rotational energy that is dissipated as elastic energy stored in the drill pipe and friction against the wellbore. Simulations resulted in recommendations to the directional driller on the optimum pipe rocking regime to adopt. The methodology was applied on a historical data set consisting of more than 100 US land wells. It was observed that improper pipe rocking could lead to back-off events, poor toolface control and reduced force transfer to the bit. By minimizing friction, longer horizontal sections and reductions in tortuosity can be achieved. An advisory software program was developed to guide directional drillers on favorable pipe rocking regimes based on contextual and real time data.
Hydraulic fracturing stimulation designs are moving towards tighter spaced clusters, longer stage length, and more proppant volumes. However, effectively evaluating the hydraulic fracturing stimulation efficiency remains a challenge. Distributed fiber optic sensing, which includes Distributed Acoustic Sensing (DAS) and Distributed Temperature Sensing (DTS), can continuously monitor the hydraulic fracturing stimulation downhole and be compared with other monitoring technology such as microseismic. The DAS and DTS data, when integrated with the microseismic, highlight processes relevant to the completion design and allow for a better understanding and interpretation of each dataset.
This paper outlines a workflow to improve processing and interpretation of DAS and DTS data. In addition, an estimate of the slurry distribution can be made. These methods will be demonstrated for a horizontal Wolfcamp well in the Permian Basin. Here we compare key aspects of the microseismic, DAS, and DTS results in several fracture stages to understand the downhole geomechanical processes. In order to interpret the DTS data a thermal model is developed (using DTS data) to simulate the temperature behavior after pumping has ceased. A slurry distribution is obtained by matching the simulated temperature with the measured temperature from DTS. In addition, the DAS data signal is studied in the frequency domain and the dominant frequencies are identified that are mostly related to fluid flow and to reduce the background noise. This time frequency analysis enhances the ability to monitor and optimize well treatments.
After reducing the background noise, the acoustic intensity is correlated to the slurry distribution. The fluid distribution data from DAS and DTS are compared with the microseismic and near field strain to better understand the completion processes. We utilized fiber optic microseismic to better understand and compare it to conventional microseismic.
Finally, we highlight the dynamics of strain and microseismic signature as fluid moves from an offset well completion into the prior stimulated fiber well to better understand the reservoir and far field effects of the completion.
Expert-guided machine learning has been used to classify depositional facies from core photographs of the Wolfcamp, Bone Spring and Spraberry formations in the Permian Basin. Training sets of core facies were selected by a sedimentologist. A model was built using a convolutional neural network and then tested against core outside of the training set with a 98% accuracy. The system can yield a quit-look of core facies much faster than that of traditional methods.
Artificial Intelligence (AI) is a branch of computer science that creates intelligent machines that work and react like humans. Machine learning is a key part of AI and requires an ability to identify patterns in streams of inputs. Learning with adequate supervision involves classification, which determines the category an object belongs to. Today it is being extensively used in image and speech recognition. At present the application of machine learning is in its infancy in the area of geosciences for the oil and gas industry.
The objective of our research is to determine if machine learning can be used to fast-track identification of depositional facies from images of conventional core photographs. Normally, this work requires a sedimentologist to painstakingly describe a core that may take many weeks to incorporate with logs and other formation evaluation data. With over 200 cored wells having 50,000 feet of core in our Permian Basin projects, the task of core description is overwhelming.
Theory and/or Methods
In order to meet the objective the software needs to be trained to recognize the various depositional facies. This is done by employing a sedimentologist (expert) to guide the training with the AI specialist. The sedimentologist builds a training set from several cores through a particular formation (e.g. Wolfcamp). The training set is a set of images selected by the sedimentologist to cover the range of depositional facies and the variations seen in each facies (Figure 1). The training sets typically employ 20 to 40 images of each facies.
Meek, Robert (Pioneer Natural Resources) | Hull, Robert (Pioneer Natural Resources) | Woller, Kevin (Pioneer Natural Resources) | Wright, Brian (Pioneer Natural Resources) | Martin, Mike (Pioneer Natural Resources) | Bello, Hector (Pioneer Natural Resources) | Bailey, James (VSProwess)
Fluid and proppant are injected into a shale reservoir during a hydraulic stimulation, causing changes in rock properties. Over time fluid and pressure bleed off into the reservoir causing further changes. We measured these changes as well as the height of the hydraulic fracture at 1.5-hour intervals using single source point seismic recordings.
A distributed acoustic sensor (DAS) and pressure gauges were installed in a vertical well to monitor the hydraulic stimulation of several horizontal wells. In the vertical well we conducted microseismic recordings using geophones, tiltmeter measurements, strain measurements from DAS, distributed temperature sensor (DTS) readings, and several monitor walk-away time-lapse VSPs (vertical seismic profiles) along with repeated single offset source VSPs. The single source VSP was acquired every 1.5 hours over three days and was oriented so that the direct arrival passed through a single stage in one of the horizontal wells. We estimated the height of the p-wave velocity change due to the hydraulic fracture by measuring travel time changes in the direct arrival. The changes in height and velocity due to the deflation of the pressure over time was also measured. The fracture height was comparable with estimates from microseismic, DAS, and tiltmeters.
In this paper we describe a method to better highlight the geometry of altered rock from a hydraulic stimulation within the Spraberry Formation of the Midland Basin in West Texas. Pioneer Natural Resources is currently developing significant unconventional resources within the basin and methods like those noted here enable an understanding of fracture geometry and well interaction during hydraulic stimulation that are important in developing unconventional resources. By acquiring several different types of data, a more accurate picture of the fracturing process can be observed and field development and geomechanical models can be adjusted accordingly. The use of DAS/DTS fiber allows for a very cost-effective and rapid acquisition of vertical seismic profiles. Pioneer has used time-lapse, fiber-based VSPs in the past with good results (Meek, 2017). Meadows (1994) observed changes in travel time during a hydraulic fracture using geophones. Recently, Byerley et al (2018) described a time-lapse experiment to monitor a hydraulic fracture during each stage into a horizontal fiber. They observed that the time delay diminished over a few days. It is thought that this time-delay was caused by fractures opening during the completion and decreasing the velocity around the well bore. Fluid and pressure leaking off over time then results in an increase in velocity of the altered rock. Understanding this pressure build up and later diffusion is important to understanding the interaction of offset well fracture stages which may influence well spacing decisions. It is also useful in determining how long adjacent wells that were shut in during completion can be placed back on production. Beyond the use of microseismic, imaging the hydraulic completion from surface geophysical techniques has been challenging. As a result we have begun to utilize subsurface imaging techniques like VSPs to gain further insight into the dynamics of the stimulation. Here we demonstrate the usefulness of the VSP by recording data into the vertical fiber only. Unfortunately, with a horizontal fiber it is difficult to obtain the height and width of the fracture using reflection energy. Experiments are currently being conducted using downward continuation of reflection energy from horizontal fibers to image around the well bore (Fuller, 2019).
Three different types of analysis were performed on high-frequency bottomhole pressure data acquired in the Hydraulic Fracturing Test Site (HFTS; Ciezobka et al., 2018) program. The pressure data was made available by Laredo Petroleum Incorporated (LPI) and the Gas Technology Institute (GTI) through participation in the HFTS joint industry project. Rate transient analysis (RTA), pressure interference test (PIT) analysis, and reservoir pressure depletion analysis of production and pressure data were performed to better understand the performance of these hydraulically fractured Wolfcamp reservoirs of the southeastern Midland Basin. Unconventional RTA, PIT analysis, and reservoir depletion analysis of the HFTS pressure data provides three different perspectives to describe fracture systems in the formation. The study of these combined attributes of this unique dataset provides new insights about pressure communication and reservoir drainage of the Wolfcamp A and Wolfcamp B in the HFTS area.
Hydraulic fractures generated during multi-stage hydraulic fracturing operations often have complex geometries (Cipolla et.al, 2008). Estimating the dimensions of complex fracture networks is one of the biggest challenges of evaluating hydraulically fractured reservoirs. Utilizing high frequency bottomhole pressure (BHP) data, unconventional RTA provides a method to evaluate effective fracture dimensions with advantages of low marginal cost and simplicity. Chu et al. (2017) demonstrated the workflow to analyze multiphase rate transient data using examples of Permian Wolfcamp horizontal wells. In this study, a similar workflow is applied to BHP data collected from 11 Wolfcamp horizontal wells in the HFTS project.
High frequency BHP data collected during well interference tests can also be utilized to identify inter-well communication. Over the years, spacing between wells on a multi-well pad has been altered, along with fracture designs, to improve reservoir development efficiency. Larger fracturing treatments have been performed to increase well productivity. Interaction between nearby producing wells is more likely to happen with increasing fracture length and closer well spacings. Understanding the magnitude of fracture communication is therefore important for optimizing well spacing with fracturing treatment sizes. Communication between producing wells can be detected from pressure response at an observation well to significant rate changes at an active well, such as a shut-in (SI) or bring-online (BOL). The process is called a pressure interference test (PIT). PITs are widely used in conventional reservoirs to determine inter-well reservoir properties (Kamal, 1983). In unconventional shale reservoirs, analysis methods to understand the pressure interference test results have been developed in recent years. Sardinha et al. (2014) analyzed pressure interference between wells in Horn River Basin by calculating pressure hit percentage. Awada et al. (2015) identified the interference response time by looking at pressure derivatives. Roussel and Agrawal (2017) applied poroelastic geomechanical models to interpret pressure interference data and calculate fracture dimensions. In the HFTS project, two PITs were conducted among 11 horizontal wells at different times of production. Kumar et al. (2018) analyzed interference data from the first PIT by calculating field response times between source and observation wells. In this discussion, we follow the technique presented by Chu et al. (2018) for analyzing power-law PIT data to quantitatively diagnose well communication among HFTS wells. The magnitude of pressure interference (MPI) between communicating wells is calculated and compared for two PIT sequences conducted 18 months apart.