The productivity and economics of horizontal wells are governed by the ability of the transverse fractures to communicate efficiently with the wellbore, which is
strongly controlled by the conductivity of the proppant bed and the effectiveness of the fluid additives. These impact the relative permeability, the capillary pressure
and the effective conductivity in the proppant bed. When time at temperature, stress cycling, embedment, multiphase flow and non-Darcy effects are considered, the effective conductivity can be reduced 100-fold. Another investigated parameter is the impact of the wellbore location relative to the propagated fracture. If the
wellbore is high in the fracture, gravity segregation will cause liquid removal from the lower portion of the fracture to be very difficult. In low conductivity proppant
beds, capillary pressure will tend to retain high water saturations, thus lower conductivity even for the portions of the fracture above the wellbore.
Laboratory experiments have addressed these issues for proppants with a range of permeabilities from 10 to 100 Darcies; 100 mesh to 20/40 mesh and ceramics. The relative permeability to gas can be as low as 0.01, with as much as a 70-fold improvement when suitable proppants and additives are employed. Evaluation of the production performance of 240 wells, 98 with effective additives and 142 without, and covering a similar range of proppant types and sizes, shows a similar benefit to the wells' normalized 30 day recovery and gross value. These results clearly demonstrate that economic expediency can be detrimental to a well's ultimate value and hydrocarbon recovery.
Gupta, Jugal (Exxon Mobil Corporation) | Zielonka, Matias (Exxon Mobil Corporation) | Albert, Richard Alan (ExxonMobil Upstream Research Co.) | El-Rabaa, Abdelwadood M. (Exxon Mobil Corporation) | Burnham, Heather Anne (XTO Energy) | Choi, Nancy Hyangsil (ExxonMobil Upstream Research Co.)
Fracture nucleation and propagation are controlled by in-situ stresses, fracture treatment design, presence of existing fractures (natural or induced), and geological history. In addition, production driven depletion and offset completions may alter stresses and hence fracture growth. For unconventional oil and gas assets the complexity resulting from the interplay of fracture characteristics, pressure depletion, and stress distribution on well performance remains one of the foremost hurdles in their optimal development, impacting infill well and refracturing programs.
To this end, ExxonMobil has undertaken a multi-disciplinary approach that integrates fracture characteristics, reservoir production, and evolution of the stress field to design and optimize developments of unconventional assets. In this approach, fracture modeling and advanced rate transient techniques are employed to constrain fracture geometry and depletion characteristics of existing wells. This knowledge is used in finite element geomechanical modeling (coupling stresses and fluid flow) to predict fracture orientation in nearby wells.
In this paper, an integrated methodology is described using case studies for two shale gas pads. The study reveals a strong connection between reservoir depletion behavior and the spatial and temporal distribution of stresses. These models predict that principal stresses are influenced far beyond the drainage area of a horizontal well and hence play a critical role in fracture orientation and performance of neighboring wells. Strategies for manipulating stresses were evaluated to control fracture propagation by injecting, shutting-in, and producing offset wells. Collective interpretation of completion, reservoir depletion and changes in stresses explained varying performances of wells and enabled evaluation of infill potential on the pad. This workflow can be used to develop strategies for (1) optimal infill design, (2) controlling propagation of fractures in new neighboring wells, and (3) refracturing of existing wells.
Ultra-deepwater reservoirs are important unconventional reservoirs that have the potential to produce billions of barrels of hydrocarbons, and are usually high pressure and high temperature with relatively high permeability. One major challenge of an unconventional, ultra-deepwater reservoir is pumping an effective and robust fracture stimulation treatment. Hydraulic fracturing a high permeability reservoir (>100 md) can be different from hydraulic fracturing technology used in low permeability formations (<1 md) due to their difference in purpose. The main purpose of hydraulic fracturing a low permeability reservoir is to create a long, conductive path to enhance drainage area and ensure a commericially economic well. In a high permeability formation, hydraulic fracturing is predominantly used to bypass near wellbore formation damage, control sand production and reduce near wellbore pressure drop. Such a treatment is achieved by pumping a short fracture packed with high proppant concentrations and may also aim at achieving enough fracture length to increase productivity especially when reservoir fluid viscosity is high. To pump such a job and ensure long term productivity from the fracture, understanding the behavior of the proppant pack is critical.
A series of laboratory experiments have been conducted to study conductivity and fracture width with high proppant loading, high temperature and high pressure using a Cooke conductivity cell. In this study, proppant was manually placed between two core samples and fracture fluid was initially pumped through the proppant pack. Conductivity was subsequently measured by pumping oil through the manually placed proppant pack to displace the fracture fluid and simulate reservoir conditions; resulting fracture fluid clean-up and proppant pack performance were studied. High strength proppant, ideal for fracture stimulations with high closure stress, was used to study the effects of proppant fracture conductivity with different proppant loadings and closure stresses. Proppant crushing and fracture width were also measured and compared to proppant pack conductivity in certain cases.
Testing results while pumping oil through the proppant pack at reservoir conditions indicated almost immediate fracture fluid clean-up. Increasing proppant concentration in the fracture showed higher conductivity values in some cases, while increasing the effective closure stress during an individual test resulted in a significant loss in conductivity for all cases. Additionally, fracture width decreased with increased effective closure stress and time. Tests were also run to study the effect of cyclic loading and showed further degradation in conductivity and width.
In this study we have undertaken a systematic investigation of the interactive effects of the key parameters that affect the final conductivity of a propped fracture, including flow back rate, proppant loading, polymer loading in the fracture fluid, the presence or absence of breaker, closure stress, and reservoir temperature. Fracture conductivity for conditions representative of field conditions was measured using a dynamic fracture conductivity testing procedure in which a fracture fluid/proppant slurry was pumped through a fracture conductivity cell, and then shut in and closure stress applied. Water-saturated gas was flowed through the fracture for a period of time at each closure stress to mimic gas flow back during the early stages of production. In all experiments, the proppant used was 30/50 mesh ceramic proppant. We used a fractional factorial design methodology to determine the relative importance of the fracturing parameters varied. The fractional factorial design method examines the combined effects on conductivity of potentially interacting parameters, while minimizing the number of experimental runs required.
The effects of the investigated factors arranged in order of decreasing impact on conductivity are closure stress, temperature, flow back rate, polymer loading, proppant concentration and presence of breaker. Increases in closure stress, flow back rate, temperature and polymer loading were observed to have deleterious effects on fracture conductivity. In particular, at high closure stresses and high temperatures, fracture conductivity was severely reduced due to the formation of a dense proppant-polymer cake. Dehydration of the residual gel in the fracture appears to cause severe damage to the proppant conductivity at higher temperatures. Also, at low proppant concentrations, there is the increased likelihood of the formation of channels resulting in high fracture conductivities.
Nowadays, it is commonplace to say that acid fracture conductivity depends on the fracture face asperities. Does it really depend on it? Almost thirty years ago, someone wrote, "We believe the conductivity measured in these tests is mainly due to the smoothing of peaks and valleys on the rough fracture faces, and is independent of rock heterogeneities due to the small sample size.?? Moreover, almost one year ago, one wrote, "More asperities touch and deform as the closure stress increases. The channels become even shorter and fewer openings are left.?? Between these two extremes, the asperities came to be pointed out as an essential factor to generate acid conductivity. Many published results from small, wet sawed and leveled carbonate rock samples support such claim. We did the same. Our experimental investigation on small scale carbonate samples with sawn faces, both from outcrops and well cores, reconfirm the existence of three main acid patterns namely uniform, channels and roughness. The design of experimental apparatus prevented that those etching patterns were artifact patterns. The acid etching patterns determine different conductivity behavior under confining stress. However, hydraulic fractures are tensile fractures and they are naturally rough. In nature, there is no such thing as a leveled fracture face. Tensile fracture faces could be rougher than fracture face after acid reaction. In fact, the first experimental results show that after acid reaction, linear roughness of tensile fractures can be larger, equal or less than linear roughness before acid reaction. This paper presents experimental results and discusses the asperities paradigm.
Microseismic event analysis is a valuable source of information that can play a pivotal role in optimizing well completion and spacing. This analysis can be taken a step further with the generation of discrete fracture networks (DFNs) from microseismic events. While DFNs can be modeled with microseismic event locations only, source mechanisms inverted from near surface-acquired microseismic data provide greater constraints for the DFN model so that the orientation of failure planes responsible for events can be explicitly assigned. The differences between such DFN realizations based on event locations only and source-mechanism constrained DFN realizations are evident in areas with significant geological complexity.
Three iterations of a DFN model were produced from a microsesimic monitoring project in the Barnett shale. The fracture network of the first iteration is modeled stochastically using only basic geologic assumptions for the area and microseismic event locations and the orientations of trends formed by the events. The second iteration is refined by deterministically locating fractures in the model and defining the fracture orientations using a source mechanism determined from the microseismic point set. The third iteration uses the results from a mechanism scan on an event per event basis to determine the best source mechanism that fits the polarity reversal signature observed on the surface array.
Refining the model by determining the mechanism of individual events can identify multiple fracture orientations within the point set. In this data set two distinct mechanisms were identified, further analysis of which identified separate event energy distributions for the two mechanisms.
The changes in the model can be quantitatively evaluated with analysis of flow properties generated from the DFN and output to the stimulated reservoir volume (SRV). While changes in the SRV and total fracture volume for models presented in this study are most significant between the first two iterations, the total permeability change across the geocellular volume is significant between all three iterations.
Drilling, completing, and fracturing unconventional formation wells in North America are now commonplace and will play a major role in the future of natural-gas production on which the nation will depend. What is not as common, however, is to drill, complete, and fracture multiple lateral branches from a single main wellbore. Multilateral wells have been routinely drilled for a number of applications, and shale plays are a natural progression for its use. Augmenting a multilateral well with selective fracturing of each leg is as straightforward as fracturing a single horizontal well.
Using conventional equipment and techniques, a multilateral well (with any number of laterals) can accommodate any type of fracturing system and program with pressures up to 12,500 psi and complete isolation of the lateral junction(s). In this project, a plug-and-perf system was used to address 10+ intervals in each leg, with average stimulation pressures up to 9,000 psi. By employing multilateral completion systems in unconventional wells, operators can lower drilling and completion costs, lower risk, help avoid non-productive time, reduce well count and surface footprint by combining two or more targets into one well, and maximize the net-pay-per-foot drilled ratio.
This paper will discuss the implementation and execution of this project as the first dual lateral well by any operator in the Granite Wash at vertical depths over 12,000 ft and measured depths reaching beyond 17,000 ft. This well targeted two different sections of the Granite Wash (a complex series of sands, shales, and siltstones that run from the Northern Texas Panhandle into Oklahoma) from a single main wellbore with commingled production rates doubling typical single horizontal well performance. Additionally, significant cost savings were achieved in contrast to drilling two separate single horizontal wells.
"Stress shadowing," where the stress field around an induced hydraulic fracture reorients from its far field directions by up to 90 degrees, is a major factor in designing and executing multiple hydraulically fractured, horizontal well completions. This is especially true as the number of hydraulic fractures increase for a given lateral length. Often the number of fracture stages is determined by well analogues without considering how stress shadows alter fracture properties. In this paper, the main objective is to determine what properties are most important in determining the minimum distance needed between hydraulic fractures to avoid stress interference. A finite element model of a horizontal wellbore with a transverse hydraulic fracture is constructed in order to perform numerical simulations of the stress around the fracture. The model is used to perform sensitivities on various mechanical and reservoir properties to investigate how and why the stress field changed.
The simulation results show that the ratio of minimum to maximum horizontal stress is the most important parameter to know in order to determine the optimal fracture spacing. Changes in this ratio show an exponential change in fracture spacing, affecting spacing requirements by up to 81%. Poisson's ratio, Biot's parameter, and net fracture pressure were also important.
It can be concluded that fracture spacing cannot be determined by looking only at one or two properties. The fracture spacing must be determined by looking at all the important variables and identifying those that are most variable for the reservoir in question. The sensitivity of the "stress shadow" to various properties is an indication that obtaining good data is key to proper completion design.
This paper presents the results of an investigation into fracture growth pattern in three horizontal wells, each fractured multiple times. The completion system was selected such that it allowed recording of the bottom-hole pressure in two critical locations; at the frac port which was being fractured, and, within the previously fractured part of the same wellbore. Downhole cups isolated the two locations from each other.
The data show remarkable results. All fractures initiated axially and re-oriented to become perpendicular to the minimum principal stress (MPS). The re-orientation details varied widely between different fractures in the same well, and also between wells. In-spite of these variations, there was no communication within the formation between the multiple fractures. All fractures in the same well had the same value of MPS, and in fact nearly the same in all three wells. In one of the wells there was obstruction to proppant movement inside the fracture which caused increasing pressures during fracture extension. In one instance this resulted in screen-out very near the wellbore early in the treatment, and in another case inside the fracture and close to the end of the stage. Still, the high pressures encountered during these fractures did not cause communication with the previous fractures. The growth pattern in all fractures can best be described as off-balance, with no evidence of "complexity??.
Once a microseism is detected, its source location can relatively easily be identified if the velocity characteristic of the medium traversed by the recorded waveforms is known. Unfortunately, this is rarely the case. Velocity models are used to estimate, with some degree of confidence, microseismic event locations. This work shows how a simple modification to the velocity model, accounting for a 4.5-degree dip supported by geological data, significantly impacts the event final locations during a borehole-based hydraulic fracturing monitoring job. Overall geometry of the hydraulically-induced fracture system interpreted (e.g. height) is the most affected. For instance, when a preliminary event location is selected without introducing the observed structural component of the beds, these measurements could change by as much as fifty percent. For the reservoir engineer, sometimes unaware of the assumptions made at the microseismic processing level, these differences could imply major changes to the field development plans. These results underscore the importance of integrating all available data and implementing well known quality controls before using microseismic monitoring data for reservoir analysis.