Frac fluid delivery is selective in effect, so must fracture models. Here, a physics-based analytical model, called nine-grain model, is presented for production forecasting in multifrac horizontal wells in unconventional reservoirs, where the utilized formulation inherently enables defining three-dimensional non-uniform SRVs, selective frac-hits, and pressure- and time-dependent permeabilities. The model is validated by constructing case studies of liquid and gas reservoirs and comparing the results with numerical simulations. In cases with both production history and fracing-induced microseismic data available, the SRV's spatial structure is extracted using a hybrid four-level straight-line technique that links volumetric RTA estimations to morphometric microseismic analysis and entails plots of plasticity, diffusivity, flowing material balance and early linear flow. By applying our model to an oil well in Permian Basin, we demonstrate that the knowledge gained from the coupled microseismic-RTA contributes to resolving the non-uniqueness of RTA solutions. The proposed reservoir modeling procedure enables efficient incorporation of microseismic interpretations in modern RTA while honoring the SRV space-time variability, thus facilitates informed decision making in spacing design of wells and perforation clusters.
Frac-hits. A frac-hit can be defined as observing a perturbation in the well production rate and/or pressure that is induced by a child offset (or an infill) well, usually triggered by pressure sinks created around parent wells or high permeability lithofacies. A frac-hit that temporarily alters the parent well productivity is called a communication frac-hit, and those with long-term effects, generally caused by fracture interference, are referred as interference frac-hits. A frac-hit may also compromise the productivity of the child well itself since the existing pressure sinks distribute the fracing energy in a larger area and might lead to an asymmetric fracture growth around the child well. Besides the parent well operational condition, the microseismic monitoring of fracing can potentially indicate interference frac-hits as it reveals fracture overlaps and any preferential fracture dilation towards existing wells. Depending on the rock and fluid properties, well age, parent-child horizontal and vertical distances, and the spatial extent of Stimulated Reservoir Volume (SRV), the constructive (Esquivel and Blasingame 2017) or destructive (King et al. 2017, Ajani and Kelkar 2012) effects of frac-hits can be experienced by fractures, SRV or the entire drainage volume (stimulated and non-stimulated zones), usually by impacting rock multiphase fluid interfacial arrangements and/or changing dimensions of conductive fractures. Aside from prevention, thoroughly reviewed by Whitfield et al. (2018), it is essential to incorporate frac-hits into production forecasting models, which to date, is not yet as straightforward as their detection. Both types of frac-hits cause a change in the well productivity over time which is not necessarily correlated with pressure, and hence, complicate the reservoir modeling process.
We apply two different approaches to determine the effective connected fracture network, responsible to the Primary production zone, in two shale reservoirs in different basins, generated by hydraulic fracturing stimulation. The first method is a 3D topological approach that uses well know geological concepts of concepts of node and branch intersections of fracture planes. The primary zone is defined as the high enhanced permeability zone where large numbers of individual fractures intersect. To perform this analysis, microseismic events source mechanism and fracture planes are determined using moment tensor and stress inversions. The second method performs a Dynamic Parameter Analysis (DPA) of the collective behavior of the microseismic events to determine high deformation areas. We show that the DPA approach is validated by the topology approach, where high deformation areas as defined by high Plasticity Index (PI) is a proxy for the Primary zone and for areas of high fracture complexity and enhanced permeability. We observe different responses to treatment between the two case studies (Marcellus and Permian) which can be caused by different initial fracturing states, stress state or treatment parameters. Both methods provide similar and reliable results and either can be used to estimate the effective SRV and fracture zone dimensions which are generally smaller than the ones obtained with the full microseismic cloud generated during the reservoir stimulation. These more realistic SRV and DFN dimensions can be use as input in production forecasting geomechanical models, and to improve the design of the completion program.
Presentation Date: Tuesday, October 16, 2018
Start Time: 1:50:00 PM
Location: 208A (Anaheim Convention Center)
Presentation Type: Oral
Summary Released seismic energy related to active fault zones within petroleum reservoirs can affect infrastructure and assets on the surface, the environment and potentially personal safety. It also can affect the ability to optimally stimulate and produce from a reservoir. In the case of fractured reservoirs, enhanced oil recovery (EOR) injection and production activities lead to local stress changes and consequently to activation of preexisting faults and fractures. In this paper we propose a technique to monitor an active fault zone, analyze the dynamic stress behavior and use it as a potential methodology for optimization of EOR activities and to reduce the risk of large seismic event occurrences. Introduction Long-term microseismic monitoring of EOR field development, with appropriate acquisition configurations, provides critical information concerning the state of active geological structures in the area (e.g., faults and fractures).
Current reservoir modeling strategies attempt to characterize the discrete fracture network (DFN) around producing wellbores to better predict both short-and long-term production levels and estimated ultimate recovery. A variety of data sources are used in describing the DFN, including image logs, petrophysical logs, geologically mapped fractures in the region (when available), and regional stress information. For hydraulic fracture stimulations, there is also microseismic data recorded during the stimulation of some wells. The event distribution obtained through microseismic monitoring gives a sense of where fracturing is occurring and how the stimulation progresses from the treatment zone into the reservoir. By using a multi-array distribution of sensors, seismic moment tensor inversion (SMTI) analysis may be performed for microseismic data, providing direct evidence of the DFN stimulated during completion activities. By performing this advanced analysis, a microseismic dataset includes the location, size, and orientation of stimulated fractures, allowing for detailed characterization of the DFN. This paper describes a methodology for characterizing a DFN observed through microseismic monitoring, which is illustrated by application to an example dataset from a North American shale play. By examining relationships between fractures and extracting statistical trends from the distribution of fractures, we arrive at a useful multifaceted description of the DFN which provides improved input data for reservoir modeling and allows a better understanding of the changes in the reservoir due to stimulation.
ABSTRACT: By examining the collective behavior of microseismicity using Dynamic Parameter Analysis (DPA) the stress response associated with the hydraulic stimulation in the Midland Basin, Texas for stacked wells are examined to understand the critical role of stress and the potential for well performance as identified through increased stimulated surface area accessible for production. Based on our observations, a clear demarcation in stress behavior is observed based on treatment order, the ability of different formations to dissipate load, and the relative distance to the treatment zone itself. The role of pre-fractured and depleted zones are also examined and their effect on observed stress behavior is identified. By integrating DPA with operational influences, such as stimulation order, the potential exists for identifying the occurrence of enhanced slip of previously stimulated fractures leading to increased production. In low permeability targets with negligible leakoff rates, we suggest that fluid and proppant stresses can push the stimulated fabric into a critically stressed region. With continued stress loading, the local stress regime within depleted reservoirs appear to be perturbed and cause fractures to reactivate under stress. In our study this resulted in enhanced production associated with increased activation of surfaces in the upper treatment zone.
Geomechanical literature is replete with examples of critically stressed behavior in rock media and how fracture planes are susceptible to slip when resolved stresses exceed plane cohesive strength. Most studies have been conducted in a laboratory setting or within numerical models representing natural conditions. Most observations have been through passive seismological efforts to determine fault slip orientations. Very little, if any, direct observation of critically stressed behavior in hydraulic fracturing has been reported to the scientific community.
Throughout the hydraulic fracturing process, high pressure and proppant injection generates stress gradients within a reservoir. The rock mass responds to this change to dissipate and transfer stresses. Previous activities in the reservoir and variations in the completion program in addition to the structural and lithological properties of the target formations play critical roles in the adjustment of the rock fabric to the stress-induced disturbance and consequently the performance of the completion.
Ardakani, E. P. (ESG Solutions) | Baig, A. M. (ESG Solutions) | Urbancic, T. I. (ESG Solutions) | Kahn, D. (Devon Energy) | Rich, J. (Devon Energy) | Langton, D. (Devon Energy) | Silver, K. (Devon Energy)
The perforation strategy for a hydraulic fracture completion for an unconventional reservoir can have a very large influence on the overall success of the injection program at effectively stimulating that network. To evaluate differences in perf clustering methodologies, operators are frequently in need of observational evidence to suggest which strategy is most efficient. We present a paper where we look at a detailed analysis of microseismicity for different stages with different completion programs.
While event distributions tend to be the first and most frequently examined aspect of a microseismic monitoring effort, because the generation of a microseismic event is not immediately diagnostic of fluid-induced fracturing, the event clouds tend to overestimate the effective area of fracturing. In order to gain further insight into how microseismic events describe effective fracture growth, a deeper look at the waveforms through techniques like Seismic Moment Tensor Inversion (SMTI) and subsequent stress inversion can be effective. These steps are necessary to describe the discrete network of cracks, from the microseismic data. Using a fracture network topology approach, the network can then be characterized in terms of its ability to percolate fluids.
We compare how cracks behave for a regular geometric shot cluster (GSC) and a variable shot cluster (VSC) and assess variations in the stimulations. Both shot clusters were completed in consecutive stages of the same lateral. The mechanisms from the GSC stages show shear-dominant mechanisms with opening and closing components in roughly equal proportions, while the VSC stages have a higher concentration of shear-tensile opening failures. Furthermore, the GSC stages showed modest connectivity around the treatment well relative to the VSC stages, which showed significant growth of connected fractures away from the treatment well. Since the VSC stages also showed relatively more stable stress behaviour than the GSC stages, these observations suggest that stability in stresses allows for steady growth of the fracture network across the reservoir.
This type of higher-order analysis of microseismic data is critical to establishing value from this data stream in terms of completion evaluation. The recognition that each microseismic event is tied to the rupture of a crack in the reservoir allows for these types of comparisons to be made in a robust fashion and be tied to the underlying geomechanics that governs the type of response from one type of completion to the other.
In this work we describe a method for noise suppression that exploits the correlated nature of noise in time and space in attempt to transform the recorded noise into white, Gaussian noise. The application of this technique on microseismic data allows for the imaging of events at SNRs previously not possible. It is further shown to be highly robust when handling unexpected changes in the noise's behaviour. Introduction The separation of signal and noise has long been studied in seismology due to the detrimental effects of noise in seismic imaging and inversion (Wang et al.,2008; Maxwell, 2010; Forghani-Arani et al., 2012). This is of particular importance to surface microseismic monitoring where events are often below the noise level. Numerous studies have shown that noise is correlated in time and space (Normark, 2011, Schilke et al.,2014) as well as containing a mixture of stationary and non-stationary aspects (Birnie et al., 2015). An efficient noise suppression algorithm must take into account both the spatiotemporal and mixed-stationarity aspects of the noise.
Urbancic, Ted (ESG Solutions) | Baig, Adam (ESG Solutions) | Viegas, Gisela (ESG Solutions) | Thompson, John M. (Anderson Thompson Reservoir Strategies) | Anderson, David (Anderson Thompson Reservoir Strategies) | Rice, Craig (Apache Canada) | Martin, Lucas (Apache Canada)
Common approaches based on event locations have not been able to effectively identify the connected volume leading to production. Utilizing the collective statistical behavior of seismicity, including their source characteristics, the underlying dynamic response of the reservoir through their spatial-temporal interaction during stimulation can be identified. We have successfully been able to descriptively identify the role of interacting rock properties, fracture state, and stress state, and how they can be used to construct reservoir descriptions that specify where flow and production volumes are most likely to be located relative to the treatment wellbore. By utilizing production data for a multi-well pad program in the Duvernay formation, a calibration of the dynamic parameters with RTA has been established which reduces the uncertainty of possible reservoir descriptions for model-based production forecasting. The synchronization of these data, coupled with production logs, geologic information, and fracture state, lead to an understanding of the effective, non-uniform fracture properties across the lateral of a horizontal well. The coupling of these techniques demonstrates a practical and transparent approach to enhancing reservoir characterization and improving decisions for field development design and adaptations to in-field stimulations in near real-time. A workflow for production and microseismic data integration is presented. The case studied examined provides definitive locations of stimulated reservoir volume (SRV) from dynamic parameter analysis (DPA), which corroborates the observed well performance data behavior; namely well interference effects and relative performance differences between wells.
The challenges associated with assessing the effectiveness of a stimulation program in unconventional plays are well described in the literature (Cipolla et al. 2009 and Cipolla et al. 2011). Complexities due to well-developed preexisting fracture networks and their behavior during injection has in recent years been addressed by considering the microseismic response to the stimulation program. In many ways microseismicity has been considered to provide insight into frac growth and extent, however the simplified view that event distributions are representative of a producing volume has been rebuked by many studies (e.g. Urbancic and Baig, 2013, Huang et al., 2014). The distribution of microseismic events has been shown to be more of an estimate of the maximum potential producible volume for modeling purposes rather than actual productive volume.
Microseismic monitoring is increasingly used to describe the extent of hydraulic stimulations in unconventional reservoirs. The key to this reconstruction is the realization that a singular microseismic event is the result of a rupture of a crack, likely associated with pre-existing lineaments in the subsurface, where the final areal extent and failure of the rupture is controlled by the frictional characteristics of this surface. Building on this concept, we discuss how microseismicity does not occur in isolation, but through clustering properties of the microseismicity that allows us to characterize the deformation in the reservoir, and further define volumes within the reservoir that are more consistent with interpretations of fluid vs stress activation. We describe the collective behavior through a series of “dynamic parameters” that describe the ability for the reservoir to deform with the seismicity and transfer stress.
We connect these concepts of fluid-driven vs stress-triggered seismicity to volumes in the reservoir of different percolation potentials. Fluid-driven processes are of primary importance to tying the microseismicity to productive volume, but we suggest that the stress-induced processes may also play a significant role in identifying poorly- or well-connected crack networks and hence the stimulated volumes within the reservoir. As such, we can resolve the likely volumes of primary (initial) and secondary (longer-term) production through these clustering processes and ensuring the behaviors determined are consistent with more of a fluid-induced vs stress-triggered behavior. This ranking of volumes in terms of productivity is analogous to work done in predicting variations in enhanced permeability in different engineering workflows, with the added benefit of being able to show variability along the well. As such, we suggest that coupling the dynamic parameter response to estimating and ranking the geometries of volumes of different productivity provides a rigorous methodology to tie microseismicity to stimulated reservoir volume, allowing for credible predictions of accessible reserves to be made over a short timescale.
ABSTRACT: In order to understand how the rock mass evolves with extraction in the particularly highly-stressed environment of a sill pillar in a hard rock mine in North America, we investigate the utility of seismic tomography as an interpretive tool. We use a rich dataset of blasts to highlight the compressional wave variations in the rock. Although tough there are areas within the sill pillar that are prone to smearing, we are able to resolve subtle variations in velocity structure. By considering different time periods, we image temporal changes in seismic velocity that we relate to the stress state and damage in the rock. Specifically, we consider data recorded during three sequential time intervals associated with the excavation of a stope during the start of the second interval. We observe a high-velocity anomaly in the first interval that is located to the edge of the future stope. After mining, the high-velocity regions migrate to the other side of the stope and then disperse to the edges of the resolvable area. Equating high-velocity anomalies with high-stress gives us the ability to characterize the evolving stress state in the mine and potentially an approach that can be used to avoid hazardous situations.
Underground excavations in mines entail a detailed understanding of the stress state to mitigate the risk of rockbursts and other seismicity that may cause damage to infrastructure and potentially casualties to personnel. Depending on the overall mining methodology, the highly-stressed areas become more focused in sill and crown pillars. As a mine reaches the end of its life-cycle, the need to safely excavate these areas becomes critical, and techniques that constrain the stress state are essential to enable the safe extraction of the resource.
The rock mass responds to variations in the stress state of the mine; high-stress areas have the effect of closing microcracks in the rock mass and, thereby, increasing seismic velocities. By mapping the velocity variations as a function of space and time we can start to understand the stress variations associated with extraction. Tomographic imaging of mines has been used to relate to the stress distribution in mines in several studies (e.g. Young & Maxwell 1992; Maxwell & Young 1996; Silver et al. 2007; Ma et al. 2016). A number of studies have shown promise in using these techniques to monitor the stability of block caving operations (Westman et al. 2012; Mercier et al. 2015).