Brittleness is a major rock property for effective reservoir stimulation in unconventional reservoirs. Differentiating brittle from ductile rocks is key to performing efficient well location and completion. I calculated a brittleness index (BI) volume from surface seismic data, calibrated by well logs, in the Barnett Shale.
Completion effectiveness is a function of the interaction between multiple engineering variables (length of the horizontal wells, number of stages, number and size of the hydraulic fracture treatments in a multistage completion, volume of proppant placed, proppant concentration, total perforation length, and number of clusters) and the spatial variation between geological factors (permeability, porosity, maximum stress field, among others) in shale gas reservoirs.
I correlated a BI log from a well with core descriptions and mineralogy log information with lithological (gamma ray) and geomechanically (λρ and μρ) related well logs, building a non-linear relationship between these variables.
Using prestack simultaneous inversion, I derived geomechanical seismic attributes, such as: λρ and μρ seismic volumes; and I used them to predict lithology and geomechanical behavior in the reservoir. Additionally, I generated a pseudo gamma ray (GR) seismic cube using probabilistic neural network (PNN). I combined these seismic attributes, using the non-linear relationship developed from well logs, to generate a pseudo BI seismic cube.
I proposed a methodology to integrate well logs and seismic derived attributes, using non-linear relationships, to highlight and identify brittle zones in unconventional reservoirs. Finally, I correlated the resulting BI seismic volume with production and volume of proppant placed into the reservoir, validating the effectiveness of this technique.
I concluded that in order to generate a brittleness index seismic volume was necessary select a combination of geological and geomechanical seismic attributes. Also, non-linear relationships show better results than the linear methods to calibrate results with seismic data. Finally, refining BI results by facies definitions is necessary to correlate to geological results from core descriptions.
In 2004 the USGS completed an assessment of the undiscovered oil and gas resources within the Fort Worth Basin. As part of the study, they evaluated two assessment units within the Barnett-Paleozoic total petroleum system. The total undiscovered gas resource is estimated to be between 21.7 Tcf and 31.5 Tcf, with a mean estimate of 26.7 Tcf. Specifically, the Barnett Shale is the largest shale gas producer in the U.S. thanks to the pioneering efforts of Devon Energy (previously known as Mitchell Energy Corporation) which first drilled into the Barnett in 1981 (Smith, 2007).
The Fort Worth Basin is a shallow N-S elongated foreland basin, encompassing roughly 15,000 square miles in North Texas, formed during the late Paleozoic Ouachita Orogeny (Walper, 1982). As a result of the collision of the North and South American plates, the Fort Worth Basin is delineated in the East by the Ouachita Thrust Front, to the North by the Red River Arch, to the N-NE by the Muenster Arch, to the West by the Bend Arch, Eastern Shelf, and Concho Arch, and to the South by the Llano Uplift. Paleotectonic collision events in the Fort Worth Basin resulted in a NW-SE main stress field orientation at the time of the Barnett Shale deposition. However, the present day regional maximum stress direction in the basin is NE-SW, with local deviations in intensity and direction about the Mineral Wells and other minor faults (e.g. Simon, 2005).
The vertical section of study is 1,300 ft on average and consists of five units of limestone and shale formations, listed from top to bottom with their average thicknesses: the Marble Falls Limestone (160 ft), the Upper Barnett Shale (365 ft), the Forestburg Limestone (45 ft), the Lower Barnett Shale (510 ft), and the upper section of the Viola Limestone (225 ft).
Lithologically, Montgomery (2005) defines the Barnett Shale as a black siliceous shale with limestone and minor dolomite lithologies. Based on four cored wells, Singh (2008) identified and described nine lithofacies in the Barnett Shale: non calcareous mudstone, calcareous mudstone, calcareous laminae deposits, concretion zones, fossil-shell intervals, phosphatic deposits, flaser to hummocky cross-bedded deposits, dolomitic mudstone, and micrite facies flows. Figure I-1 shows the set of logs available for this study. In addition to conventional logs, elemental capture spectroscopy (ECS) logs were acquired in the area of the study, revealing important vertical and lateral mineralogy variations.
Figure I-1. Conventional log sets such as gamma ray, photoelectric factor, and resistivity were acquired in the area, and differentiate shale from limestone formations. Applying the same concept used by Singh (2008) and Perez (2009), I interpret the gamma ray log pattern (indicated by arrows in track 2) and their corresponding GRP (track 3). Track 4 shows the ECS log corresponding to Well A indicating that the mineral distribution along the wellbore agrees with Karastathis’s (2007) and Kale’s (2009) findings, where tracks 5, 6, and 7 show the individual mineralogy log results decomposed into clay, quartz, and calcite content, respectively.
I calculated the brittleness index using Jarvie et al.’s (2007) and Wang and Gale’s (2009) equations using the ECS log data points and show the results in Figure I-2 tracks 9 and 10, respectively. Comparing both BI indexes with the mineralogy logs, (track 4), I observe that the zones with high quartz and calcite content are more brittle than the regions with high clay content, which are less brittle (ductile).
where Qz is the fractional quartz content, Dol the dolomite content, Ca the calcit content, TOC the total organic carbon content, and Cly the clay content by weight in the rock. In future blog posts I will review in more details the definition of brittleness.
LINEAR VS. NON-LINEAR CORRELATION
Intuitively, as humans we imagine that the world behaves in a linear way. The question is, if we drill a well with a horizontal length of 1,000 ft and it produces 100 MCF, drilling a second well with double the horizontal length of the first one, do we expect double the production? The answer is: maybe. Why? Because the production from a well is related to several variables (geological, engineering, among others) that will affect the performance production of the well (Figure I-2).
Figure I-2. Diagrams showing the difference in concepts between linear and non-linear correlations, relating a variable and it corresponding response, in this case wellbore horizontal distance and production.
In this case, I selected three variables (logs) which are: gamma ray (geological) and λρ and μρ (geomechanical). Figure I-3 shows the crossplot of each individual parameter with the response, brittleness index (calculated using Wang and Gale’s (2009) equations). Notice that the correlation factor between each variable is relatively low, and very disperse.
Figure I-3. Crossplot between each variable and the predictor, and its corresponding correlation coefficient and rank.
Using a linear correlation approach, using principal components, I was able to predict an estimate of brittleness index (Figure I-4). Notice that this type of approach results in an overestimation in brittleness index.
Figure I-4. Brittleness index prediction using linear correlation results.
Using a non-linear correlation let’s us analyze the interaction between variables, and the contribution of each to the final solution. Figure I-5 shows the original variable (x axis) and its corresponding transform (y axis). In the case of the gamma ray notice a flat plateau at the end of the transformation. This behavior indicates that after a certain point (approximately 200 API) the GR variable does not contribute to the brittleness index. In order to understand the contribution of each of the geomechanical variables (λρ and μρ) to brittleness index it is necessary to refer to the λρ and μρ crossplot proposed by Perez (2013). He computed and plotted the λρ and μρ of the three most common minerals in the Barnett Shale: calcite, clay, and quartz using the moduli, densities, and velocities published by Mavko et al. (2009). Connecting the three vertices of each mineral generates a mineralogy ternary plot in λρ and μρ space.
From this it is possible to understand the contribution of each variable. In the case of lower λρ, the presence of quartz in the rock is higher, which can be translated to a higher contribution to the brittleness index. On the other hand, higher μρ indicates that the rock has higher amount of clay, reducing the contribution to the brittleness index. Using a non-linear approach we are able to increase the correlation coefficient between the measured and predicted BI values (Figure I-6).
Figure I-5. Crossplots showing the variable and its transform, indicating the contribution of each variable to the brittleness index.
Figure I-6. Brittleness index prediction using non-linear correlation results.
The same non-linear correlation used in the logs was used to correlate three volumes of gamma ray, λρ and μρ, resulting in a final brittleness index seismic volume (Figure I-7). The entire workflow is summarized in Figure I-8. This result can be used to perform a regional production analysis, in order to identify possible sweetspots.
Figure I-7. Brittleness index seismic volume, obtained of applying the non-linear regression analysis from logs, to three seismic volumes (gamma ray, λρ, and μρ).
Figure I-8. Workflow summary.
In order to generate a brittleness index seismic volume it was necessary to select a combination of geological and geomechanical seismic attributes. Additionally, I demonstrated that non-linear relationships show better results than the linear methods to calibrate results with seismic data.
Jarvie, D. M., R. J. Hill, T. E. Ruble, and R. M. Pollastro, 2007, Unconventional shale-gas systems: the Mississippian Barnett Shale of North-Central Texas as one model for thermogenic shale-gas assessment: AAPG Bulletin, 91, 475 – 499.
Kale, S., 2009, Petrophysical characterization of Barnett Shale play: M.S. Thesis, Mewbourne School of Petroleum and Geological Engineering: The University of Oklahoma.
Karastathis, A., 2007, Petrophysical measurements on tight gas shale: M.S. Thesis, Mewbourne School of Petroleum and Geological Engineering: The University of Oklahoma.
Mavko, G., T. Mukerji, and J. Dvorkin, 2009, The rock physics handbook: Cambridge University Press.
Montgomery, S. L., 2005, Mississippian Barnett Shale, Fort Worth Basin, North- Central Texas: Gas-Shale Play with Multi-Trillion Cubic Foot Potential: AAPG Bulletin, 89, 155 – 175.
Perez, R., 2009, Quantitative Analysis of Gamma Ray Parasequences in the Barnett Shale: M.S. Thesis, ConocoPhillips School of Geology and Geophysics: The University of Oklahoma.
Simon, Y. (2005), Stress and fracture characterization in a shale reservoir, North Texas, using correlation between new seismic attributes and well data: M.S. thesis, Department of Geosciences, University of Houston.
Singh, P., 2008, Lithofacies and sequence stratigraphic framework of the Barnett Shale: Ph.D. Dissertation, ConocoPhillips School of Geology and Geophysics: The University of Oklahoma.
Smith, T., 2007, Producing gas from shales: GEO ExPro, 2, 48 – 52.
Walper, J. L., 1982, Plate tectonic evolution of the Fort Worth Basin, Dallas Geological Society.
Wang, F.P., and J. F. W. Gale, 2009, Screening criteria for shale-gas systems: GCAGS Transactions, 59, 779 – 793.
What do you think? Leave a comment below.
Latest posts by Roderick Perez (see all)
- Seismic Brittleness Index Volume Estimation From Well Logs in Unconventional Reservoirs (Part III) – Brittleness From Well Logs - August 12, 2014
- Seismic Brittleness Index Volume Estimation From Well Logs in Unconventional Reservoirs (Part II) – Brittleness Definitions Review - July 10, 2014
- Seismic Brittleness Index Volume Estimation From Well Logs in Unconventional Reservoirs (Part I) - June 26, 2014