1 / 34

Lessons Learned From Data Mining in Unconventional Reservoirs

Lessons Learned From Data Mining in Unconventional Reservoirs. Randy F. LaFollette Director, Applied Reservoir Technology Baker Hughes Pressure Pumping. Presentation Outline. Importance of Data Mining Data Sources Data Mining Methods Case Study Highlights. Problem / Solution.

leonaj
Télécharger la présentation

Lessons Learned From Data Mining in Unconventional Reservoirs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Lessons Learned From Data Mining in Unconventional Reservoirs Randy F. LaFollette Director, Applied Reservoir Technology Baker Hughes Pressure Pumping

  2. Presentation Outline • Importance of Data Mining • Data Sources • Data Mining Methods • Case Study Highlights

  3. Problem / Solution • Problem… • High well count, but… • Most with low-granularity data • Inconsistent production results • Multi-million dollar decisions to be made • Solution – Data Mining for Data-driven decisions

  4. Data Mining Tools – Past and Present R 4

  5. At the Beginning: The Variables • Reservoir Quality • Q f { k, h, Pres, m, re } • Proxied by well location • Well Architecture • Completion • Stimulation • Production Management • Production Metrics 5

  6. Production Result Time Dependence Barnett Shale Timeline H D V Max Gas Rate 1981 Completion Date 2009 6

  7. Well Architecture, Completion& Stimulation Time-Dependence

  8. Not the Barnett “Shale,” but close 8

  9. History: Spreadsheets and Cross PlotsDo Larger Treatments Yield Increased Production? 9

  10. Geographical Information Systems Mapping • 200,000+ wells in Fort Worth Basin • 12,000+ Barnett Horizontals • Provide for data-driven discussion of best practices 10

  11. Modern Analysis Techniques • Multivariate, non-linear, using boosted trees 11

  12. Available Data • Commercial data sets • Well history • Completion & stimulation practices • Monthly production • 3,300+ directional surveys • FracFocus data • In-house data sets • Collected, reviewed, put into a database • Quality Control Process • Statistical removal of outliers • Known limits & ratios examination 12

  13. Modern Analysis TechniquesBarnett Vertical Wells • Geographical Information Systems (GIS) 13

  14. Lowest 10% and Best 10% Barnett Hz Wells 14

  15. SPE 163852 Application of Multivariate Analysis and Geographic Information Systems Pattern-Recognition Analysis to Production Results in the Bakken Light Tight Oil Play Randy F. LaFollette, Ghazal Izadi, Ming Zhong, Baker Hughes 15

  16. Middle Bakken Light-Tight Oil Play, Montana and North Dakota 16

  17. Middle Bakken Eastern Montana vs. North Dakota 17

  18. Slide 18 Exploratory Data Analysis One-Column Format 18

  19. Slide 19 Related Variables, Data Clustering, Outlier Identification 19

  20. Slide 20 Transformed Scatterplot 20

  21. Middle Bakken, Max Monthly Oil, n=750 wells 21

  22. Slide 22 Middle Bakken, BO/ft Most Influential • CLAT • Location • Prop Qty • Fluid Vol • Prop Conc 22

  23. Lateral Length and Well Efficiency Log 10 Max Oil Rate 23

  24. SPE 168628 Application of Multivariate Statistical Modeling and Geographic Information Systems Pattern-Recognition Analysis to Production Results in the Eagle Ford Formation of South Texas Randy F. LaFollette, Dr. Ghazal Izadi, Dr. Ming Zhong SPE, Baker Hughes 24

  25. Slide 25 25

  26. Slide 26 Data Sources, QC, Focus • Public and proprietary • In house proprietary database • Commercial US Well Database • Well headers, location, architecture, completion, stimulation, production • Focus on oil wells (GOR <15,000 scf/bbl) 26

  27. Mineralogy/Rock Properties Eagle Ford 80 miles 30 miles 40 miles 27

  28. Eagle Ford Completion / Stimulation 28

  29. Slide 29 Max Monthly Oil Rate, Area 1 29

  30. Slide 30 Max Monthly Oil, Key Drivers, Area 1 30

  31. Slide 31 Max Monthly Oil, Partial Dependence Plots, Area 1 31

  32. Summary • Data sources, methods, tools, lessons-learned from unconventionals • Interpretation most complete using multivariate statistical methods • Reservoir quality, well architecture, completion, stimulation all significant production drivers • Data Mining for Data-driven decisions!

  33. Acknowledgements • The author thanks SPE and the Management of Baker Hughes for the opportunity to present this work to the global SPE community. • Thanks also go to my team members, past and present, for their hard work and insights • Dr. Ming Zhong • Dr. Ghazal Izadi • Bill Holcomb • Dr. Jorge Aragon

  34. Thank You!

More Related