PETROPHYSICS, VOL. 57, NO. 6 (DECEMBER 2016); PAGE 638–643; 8 FIGURES; 1 TABLE
TECHNICAL NOTE Normalizing Gamma-Ray Logs Acquired from a Mixture of Vertical and Horizontal Wells in the Haynesville Shale Chicheng Xu1, W. Sebastian Bayer1, Marcus Wunderle1, and Abhishek Bansal1 ABSTRACT This article introduces a practical workÀow of normalizing gamma-ray (GR) logs acquired from both vertical and horizontal wells penetrating different geological subzones in the Haynesville shale. Gamma-ray logs are ¿rst normalized in all vertical wells over the same stratigraphic zone. The normalized GR log in the nearest vertical well is then used to reconstruct synthetic GR logs along horizontal well paths using true stratigraphic projection method. The measured GR logs from the horizontal wells are then compared and normalized with the synthetic GR logs to enforce statistical consistency with the normalized GR logs in the associated vertical well. The workÀow delivers a set of normalized GR logs in a group of vertical and horizontal wells that provide a reliable basis for reservoir description and modeling. INTRODUCTION A gamma-ray (GR) log is available in almost every well drilled in the Haynesville shale. In many horizontal wells, LWD GR is the only acquired log that provides critical information for business decisions on drilling, fracturing, and completions. Due to the underlying tool physics, GR logs acquired under different drilling conditions and borehole environments need to be normalized to be quantitatively comparable across different zones and wells (Neinast and Knox, 1974; Aly et al., 1997). The GR log normalization procedure has been well established for multiple vertical wells penetrating the same stratigraphic zones; however, the workÀow for normalizing GR logs from multiple horizontal wells drilling through different geological subzones has not yet been reported. The standard log-normalization procedure cannot be directly applied to horizontal wells because different horizontal wells drill through different target subzones (Fig. 1). Consequently, the GR-log histogram
in horizontal wells is not supposed to be comparable with nearby vertical wells because it measures rock populations in different subzones.
Fig. 1—A typical drilling scenario in the Haynesville ¿eld. A vertical well is used to guide the horizontal well drilling using geosteering. The vertical well penetrates all subzones while the horizontal well only penetrates the top four subzones.
Figure 2 shows a ¿eld example in which the vertical well penetrates all subzones while the horizontal well only penetrates the top four subzones. Even in the same subzone, the horizontal well likely records more data from an individual subzone than it does the vertical well. Since the data density vs. subzone in horizontal wells is different as compared to the vertical wells, the GR log histograms are not comparable to each other either. Therefore, GR-log normalization work, for a mixture of vertical and horizontal wells, is an outstanding problem in many ¿elds. We propose a workÀow for normalizing GR logs, from multiple horizontal wells, to enable interpreters to use the GR log in horizontal wells for reservoir characterization. The workÀow was tested with a group of 16 horizontal wells and 8 vertical wells in the Haynesville shale.
Manuscript received by the Editor April 27, 2016; revised manuscript accepted September 6, 2016; manuscript accepted September 9, 2016. 1 BHP Billiton Petroleum, 1360 Post Oak Blvd., Houston, TX 77056, USA;
[email protected]
638
PETROPHYSICS
December 2016
Normalizing Gamma-Ray Logs Acquired from a Mixture of Vertical and Horizontal Wells in the Haynesville Shale
vertical offset from the top of HSVL, their GR readings are assumed to be same (GR_A = GR_B). So the GR reading from the vertical wells can be projected on to the horizontal well path using the geosteering and well survey information.
GR-LOG NORMALIZATION WORKFLOW FOR A MIXTURE OF VERTICAL AND HORIZONTAL WELLS
Fig. 2—Comparison of GR histograms from a pair of vertical and horizontal wells. The vertical well penetrates all subzones while the horizontal well only penetrates the top four subzones. The histogram of GR logs from the horizontal well shows much less variability because the horizontal well stays mostly in the same subzone instead of penetrating all subzones.
TRUE STRATIGRAPHIC PROJECTION METHOD Figure 3 illustrates how the GR log acquired in a vertical well is projected to its corresponding horizontal wells by using a true stratigraphic projection (TSP) method. Implementation of a TSP algorithm is straightforward so it will not be detailed in this paper. The basic assumption is a layer-cake reservoir model where reservoir properties are the same at a certain vertical true stratigraphic thickness (TST) offset depth to the horizon top. In a heterogeneous shale reservoir, this assumption is often not true; however, comparison between the acquired LWD GR log and the projected GR log is an indicator of how good the layercake model is. Before quantifying the GR log mismatch, normalization between the acquired LWD GR log and the projected GR log needs to be performed. That is to say, the GR log in a horizontal well has to be normalized with the vertical well ¿rst.
Fig. 3—Schematic illustrating the principle of the TSP method from a pilot well to the horizontal well to reconstruct GR log based on a layercake reservoir model. If two points (Point A and Point B) have the same
December 2016
Within the study area, all GR logs from vertical wells penetrating the Haynesville formation should be normalized following existing standard procedures (Neinast and Knox, 1974; Aly et al., 1997). A geologically and petrophysically consistent normalization procedure should follow the steps below: Step 1: Normalize GR logs in all vertical wells in the study area over the same stratigraphic interval. Step 2: Construct a synthetic GR log along each horizontal well path by interpolating the normalized GR log in the nearest vertical well using the TSP method (Fig. 3). Step 3: Normalize the raw GR log measurement to the synthetic GR log by enforcing a consistent statistical distribution (equalized mean and standard deviation). Step 4: Repeat Steps 2 and 3 for all horizontal wells. Step 5: Use normalized GR logs in all horizontal wells and vertical wells for further analysis and modeling. The normalization workÀow is summarized in the Àowchart shown in Fig. 4.
Fig. 4—Flowchart for the GR-log normalization workÀow in a mixture of vertical and horizontal wells.
PETROPHYSICS
639
Xu et al.
Figure 5 shows an example of normalizing a GR log from a horizontal well to the offset vertical well. Figure 6 shows the comparison of histograms of different GR logs. In this example, a signi¿cant difference exists between the normalized GR log and the raw GR log.
FIELD CASE—HAYNESVILLE SHALE The above workÀow was applied to a group of 16 horizontal wells and 8 vertical wells in the Haynesville shale (Table 1). We ¿rst applied a standard normalization procedure to all the vertical wells penetrating the Haynesville within the study area. Before normalization, GR-log histograms from all vertical wells show large variability in their mean and standard deviation (Fig. 7a). After normalization, GRlog histograms from all vertical wells exhibited a consistent mean and standard deviation (Fig. 7b). Table 1—List of 16 horizontal Wells and 8 Vertical Wells in the Haynesville Study Area
Fig. 5—An example of normalizing GR logs from a horizontal to a vertical well. Track 1, raw GR log measurement from the horizontal well; Track 2, reconstructed synthetic GR log using true stratigraphic projection from the nearby vertical well; Track 3, comparison of raw and synthetic GR logs along the horizontal well; Track 4, comparison of the normalized GR log and the reconstructed synthetic GR logs along the horizontal well.
The rightmost two columns show the shift of mean raw GR to achieve statistical consistency with the associated vertical well in both GAPI unit and percentage unit.
Fig. 6—Comparison of histograms of raw horizontal GR log (green), reconstructed GR log (red), and normalized horizontal GR log (blue).
After applying this procedure to all horizontal wells, GR logs in both vertical and horizontal wells in the same work area should be comparable on a quantitative basis; therefore, can be integrated into reservoir characterization.
640
We then applied the TSP method to construct synthetic GR logs in all horizontal wells and normalized them against the nearest vertical well. After normalization, we observed statistical consistency in the normalized GR-log histograms of all horizontal wells from each geological subzone (Figs. 8a to 8d). The rightmost two columns of Table 1 show the difference between the normalized GR log and the raw GR log. The maximum percentage difference can reach 22.7%. If this difference is not eliminated by normalization, GR logs in the horizontal wells will not be quantitatively comparable to each other.
PETROPHYSICS
December 2016
Normalizing Gamma-Ray Logs Acquired from a Mixture of Vertical and Horizontal Wells in the Haynesville Shale
Fig. 7—(a) GR-log histograms of all vertical wells in the study area before normalization. (b) GR-log histograms of all vertical wells in the study area after normalization.
December 2016
PETROPHYSICS
641
Xu et al.
Fig. 8—(a) Normalized GR-log histograms in the subzone HSVL 1 from all horizontal wells. (b) Normalized GR-log histograms in the subzone HSVL 2 from all horizontal wells. (c) Normalized GR-log histograms in the subzone HSVL 3 from all horizontal wells. (d) Normalized GR-log histograms in the subzone HSVL 4 from all horizontal wells.
SUMMARY We used the TSP method to generate a synthetic GR log along a horizontal well path using the nearest vertical well. Based on the TSP method, we successfully implemented and tested a new workÀow of normalizing multiwell GR logs acquired in both vertical and horizontal wells penetrating different geological subzones. The results of testing on a group of 16 horizontal and 8 vertical wells show that GR logs from each geologic subzone are reasonably normalized. The maximum percent difference between GR logs acquired in a horizontal well and its vertical pilot well can reach 22.7%. ACKNOWLEDGEMENTS The authors would like to thank the BHP Billiton
642
Petroleum Petrophysics team and the Haynesville asset team for their technical and data support. Great thanks also goes to the Geoscience function management team for reviewing and approving this article to be published. In particular, thanks to Dr. Juan-Mauricio Florez-Nino, Andy Brickell, Jim Seccombe, and Jorge Sanchez-Ramirez for providing constructive technical opinions to improve this paper. REFERENCES Neinast, G.S., and Knox, C.C., 1974, Normalization of Well Log Digitizing, The Log Analyst , 15(2), 18–25. Aly, A.M., Hunt, E.R., Pursell, D.A., and McCain, W.D., Jr., 1997, Application of Mutli-Well Normalization of Open Hole Logs in Integrated Reservoir Studies, Paper SPE-38263 presented at the SPE Western Regional Meeting, Long Beach, California, USA, 25–27 June. DOI: http://dx.doi.org/10.2118/38263-MS
PETROPHYSICS
December 2016
Normalizing Gamma-Ray Logs Acquired from a Mixture of Vertical and Horizontal Wells in the Haynesville Shale
ABOUT THE AUTHORS Chicheng Xu received his PhD degree in petroleum engineering at The University of Texas at Austin (2013). He received his BS in physics from the University of Science and Technology of China (2002) and his MPHIL in physics from the Chinese University of Hong Kong (2004). Before joining the Formation Evaluation Research Consortium group in the Department of Petroleum and Geosystems Engineering at the University of Texas at Austin, he worked at Schlumberger Beijing Geoscience Center as software engineer from 2004 to 2009. He worked as petrophysicist in BP America from 2013 to 2014 and currently works as petrophysicist in the Geoscience function team of BHP Billiton Petroleum. He had 20 technical papers published in conferences and journals and served in technical committees of SPE and SEG. Sebastian Bayer is a Senior Reservoir Geologist working as a reservoir integrator for Unconventional and Conventional projects. He has more than 10 years of industry experience focused on integrated-¿t-for-purpose reservoir modeling, static models designed with the Engineers to provide integrated solutions for dynamic simulation. He graduated from the University of Oklahoma with a MS degree focused in Structural Geology and Stratigraphy. He also holds a BS in Petroleum Geology from the Universidad Nacional de Colombia. His current interests focus on integrated reservoir characterization and implementation of discrete fracture networks (DFN) based on geologic scenarios to understand the matrix and fracture components of the system. Integration of well logs, petrofacies, seismic attribute volumes, and microseismic event clouds in relation to different geomechanical units and stress. His work is focused on single-well analyses and multiwell interference studies based on robust static parameters for projects to evaluate reservoir quality and production performance.
December 2016
Marcus Wunderle is a Geologist at BHP Billiton. He has 10 years of experience in the oil and gas industry spanning conventional development, unconventional exploration and development. Most recently he has been working on the Haynesville shale as a Geologist and Geomodeler. He is a graduate of Ohio University with an MS in geology, focusing on Geoscience Education, and a BS in Earth Science Education focusing on geology. His current work is primarily focused on interpreting and integrating all subsurface data into regional to well-scale geomodels for reservoir characterization. Abhishek Bansal graduated with a BSc in Mechanical Engineering from National Institute of Technology, Surathkal (2006) and an MSc in Petroleum Engineering from The University of Texas at Austin (2012). He worked for Halliburton as a Wireline ¿eld engineer (2006–2010) before joining UT AUSTIN as a research associate. He is currently working with BHP Billiton primarily focusing on multiwell integrated petrophysical analysis for conventional and unconventional reservoirs.
PETROPHYSICS
643