Best Practices for Reserves Estimation in Unconventional Reservoirs —

1. Best Practices for Reserves Estimation in Unconventional Reservoirs — Present and Future Considerations Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

2. Slide 2 Advancement of Knowledge We advance our knowledge by application of mathematics and physical science, trial and error, and observation. Heckman (novice philosopher) Estimating Reserves in Unconventional Reservoirs: • It’s early • It’s different • It’s good to learn from others • It’s good to have guidance • It’s good to have a vision (www.eia.gov/analysis/studies/usshalegas/pdf/usshaleplays.pdf)

3. "It's Early" (Use Caution) Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

4. Slide 4 It's Early — Horizontal Wells as the Enabler (1/2)

5. Slide 5 It's Early — Horizontal Wells as the Enabler (2/2)

6. Slide 6 It's Early — Tight Oil Will Dominate in 10-15 Years (U.S.) From the Authorities: Oil • Tight oil (shale and chalk) production projected to rise sharply over next decade. • U.S. Energy Information Administration (EIA) AEO 2013 Early Release Overview. U.S. domestic crude oil production by source, 1990-2040 (millions of barrels/day) U.S. Energy Information Administration (AEO 2013 Early Release Overview)

7. Slide 7 It's Early — Lots of Room to Grow (U.S.) From the Authorities: Gas • Shale gas is projected to be the most significant U.S. growth play over the next ~30 years. • Considerable uncertainty in size and economics of shale gas resources. • Most shale gas wells have been drilled in the last few years, leaving considerable uncertainty regarding long-term productivity. • U.S. Energy Information Administration (EIA) AEO 2013 Early Release Overview. U.S. Dry natural gas production by source, 1990-2040 (trillions of cubic feet per year) U.S. Energy Information Administration (AEO 2013 Early Release Overview)

8. Slide 8 It's Early — There's Uncertainty… • For many unconventional reservoirs, (particularly shales) it's early in the… • Development life-cycle • Production life-cycle • Learning life-cycle We are in a period of significant uncertainty! • Production forecasts and ultimate recovery • Drilling and completion techniques • Optimum well spacing (interference) • Impact on company portfolio of opportunities • Project financing, investor confidence, regulatory harmony

9. "It's Different" (We May Not Know What We Don't Know…) Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

10. Slide 10 It's Different — Definitions Conventional Reservoirs Unconventional Reservoirs (Shale) ● Localized structural trap ● "Continuous-type" deposit ● External HC sourcing ● Self-sourced HC ● Hydrodynamic influence ● Minimal hydrodynamic influence ● Porosity important ● Porosity may not be important ● Permeability > 0.1 md ● Permeability << 0.1 md ● Permeability ≠ f(p) ● Permeability = f(p) ● Traditional phase behavior (PVT) ● Complex (HP/HT) PVT ● Minimal extraction effort ● Significant extraction effort ● Significant production history ● Limited production history ● Often late development life-cycle ● Early development life-cycle ● Few wells for commerciality ● Many wells for commerciality ● Base reserves on volumetrics ● Base reserves on analogs ● Assess entire prospect before drilling ● Prospect driven by drilling ● Boundary-dominated flow (months) ●No boundary-dominated flow ● Traditional reserves methods●Traditional reserves methods

11. Slide 11 It's Different — Challenges and Methodology • Challenges: • No industry standard tech-niques for assessing uncon-ventional exploration plays. • Stimulation is the major challenge (cost/technology). • Fractures (induced or natural) are critical producibility factors. • Success is judged based on production results. • Methodology: • Expect the unexpected. Well performance will vary, despite similar drilling/completion practices, well spacing, etc. • Unconventional plays are "statistical," many wells must be drilled to assess potential. • Drilling too few wells is likely to lead to a bad decision.

12. " It’s Good to Learn from Others " (Industry Survey: Development of "Most Likely" Production Forecasts from Unconventional Reservoirs) Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

13. Slide 13 Industry Survey Objective of the Survey: • Collect and communicate industry "Best Practices:" • "Most Likely " forecast (expected technical outcome). • Reasonable " Low Side " forecast. • Criteria: • Survey of individuals. • Anonymous. • Independent of regulatory classification systems.

14. Slide 14 Industry Survey Administration of the Survey: • Design: • Designed by project team. • Applicable to unconventional reservoirs. • Vetted and previewed survey with all participants. • Focused on current practices in developed areas. • Intended participants selected from all facets of E&P sector. • Structure: • Survey delivered by secure, web-based system. • 13 core questions (rate-time analysis/forecasting). • 17 optional questions (confidence/risks/undeveloped reserves/probabilistics). "My opinion? Are you sure I'm supposed to have one?" "User name and password…"

15. Slide 15 Industry Survey Participation: • Survey Particulars: • 39 surveys submitted. • Survey responses: • [33 responses] Shale Gas (i.e., Eagle Ford, Marcellus, etc.). • [17 responses] Shale Oil (i.e., Bakken, etc.). • [17 responses] Tight Gas (i.e., Cotton Valley, etc.). • [12 responses] Tight Oil (i.e., Spraberry, etc.).

16. Slide 16 Industry Survey Primary Techniques for Forecasting Existing Shale Gas Wells: • Arps' Method: • Single most often used method. • Often limited with a maximum value of the Arps' b-factor. • Combination: • Arps' method with alternative cross-check. • Implementation varies, but general practice is to constrain EUR. • RTA: • Relatively uncommon, need for higher quality data. • Requires time-rate-pressure data.

17. Slide 17 Industry Survey Forecasting Undeveloped Shale Gas Wells: • Analog Type Curve: • Most popular approach. • Need a relevant analog. • Need a big database. • Simulation: • Either basic or exhaustive datasets can be used. • Tailor model to needs: • Basic model for screening studies. • Full "mechanistic" models for evaluating well completion, well spacing, PVT, etc.

18. Slide 18 Industry Survey Time-Rate Analysis: • Comment: (from survey) • Arps' DCA methodology can be aggressive if unconstrained. "Work-around" is the "Modi-fied-Hyperbolic" (hyperbolic-exponential model), • Other models mentioned: • "Stretched-Exponential" • "Duong" (early-time) model • Suggestions: (from survey) • Double hyperbolic model (with terminal exponential). • Attempt to represent physical behavior of well-reservoir system (SRV + non-SRV). • Last Words: • What we have works (when constrained). • Some users want more models and better diagnostics.

19. Slide 19 Industry Survey Best Practices: • New Wells: • Require 3-24 months of production performance. • Exception — maintain "shape" of type curve and fit to actual IP. • EUR: • Cross-check estimates against various/alternate methodology(s). • Ensure "reasonableness" of results (confidence limits). • Reserves Methodologies: • Time-Rate (DCA/Arps) — existing plays. • Time-Rate-Pressure (RTA) — existing plays. • Analogs — new plays. • Reservoir simulation (mechanistic) — existing and new plays. • Probabilistic (stochastic simulations) — existing and new plays.

20. "It’s Good to Have Guidance" (Statistical Analysis of Public Production Data — Dry Horizontal Shale Gas Wells ONLY) [all data obtained from publicly available sources] Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

21. Slide 21 Statistical Analysis of Public Production Data [all data obtained from publicly available sources — Dry Horizontal Shale Gas Wells ONLY] Objectives of the Public Data Study: • Use Arps' "Modified Hyperbolic" relation: • Establish analysis limits: • 0.5 < b < 2.0. • Dlim = 10 percent (terminal exponential). • Abandonment Rate = 50 MSCF/D. • Regression Approach: • Automated data fitting using constraints given above. • All results quality-checked, outliers discarded. • Provide guidance and limits with graphical results: • P50 rate vs. time with observations to generate confident EUR. • Comparison of P10/P50/P90 EURfinal-values.

22. Slide 22 Statistical Analysis of Public Production Data [all data obtained from publicly available sources — Dry Horizontal Shale Gas Wells ONLY] P50 Well Gas Rate Trends: Comment: • Left plot yields time required to estimate EUR (~12-32 months). • The "hyperbolic" (or "constant b") flow regime is required to estimate EUR.

23. Slide 23 Statistical Analysis of Public Production Data [all data obtained from publicly available sources — Dry Horizontal Shale Gas Wells ONLY] P90/P50/P10 EUR Comparisons: (Modified Hyperbolic Model with 30 year max life) Comment: • Results are “auto-fitted” and should be considered reasonably accurate. • Results will vary when data are segregated by geological area, completion practices, spacing, etc. • Analyses represent an attempt to quantify the RANGE of values.

24. "It’s Good to Have a Vision" (Vision for Reserves Estimation — The Next 10 Years) Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

25. Slide 25 Vision — Models for Production Forecasting Models for Production Forecasting: • Most likely scenarios: • Statistical models will remain an alternative to reservoir models. • Analytical and numerical models will focus on SRV/beyond SRV. • Numerical models will use more microseismic and geomechanics. • Developed Reserves: Decline Curve Analysis (DCA) • (present) DCA models are useful to relate "SRV"-based reserves. • (future) Beyond the "SRV" will require a reservoir model. • Developed Reserves: Probabilistic Methods • Probabilistic (non-deterministic) reserves can provide insight. • Probabilistic models will account for changes in completions. • Probabilistic approach will continue to evolve... • Undeveloped Reserves: • Data mining methods will yield insight into EUR trends. • EUR = f(geomechanics, geology, and engineering data ). • Seismic can be/will be used to calibrate geostatistical models. • Geomechanical properties are/will be the weak link.

26. Slide 26 Vision — Integration of the SRV (1/2) Integration of the SRV: Fracture Models • Assessment of SRV Architecture: • [need] Modeling of SRV flow (block size + fracture conductivity). • [need] Improved modeling of hydraulic fracture growth models. • [challenge] Geomechanics and formation characterization. Comment: • We assume (Darcy) flow physics — and we force performance to match "old" models. • We assume we understand the nature of the (natural and induced) fracture systems. Microseismic Fracture Distribution Model Simulation Model

27. Slide 27 Vision — Integration of the SRV (2/2) Integration of the SRV: Matrix Contribution • Matrix Contribution within the SRV: • [need] Storage, transport, PVT for unconventional systems. • [need] Physical experiments — porosity, adsorption, permeability. • [need] Numerical experiments — molecular simulation/up-scaling. Comment: • Need to redefine the fundamentals. • Need more precise experiments. • Can we model instead of measure? Geology Nano-Scale Behavior Concept Model

28. Best Practices for Reserves Estimation in Unconventional Reservoirs — Present and Future Considerations Summary Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University

29. Slide 29 Summary and Conclusions Summary: • This work provides orientation to industry reserves practices via a comprehensive survey — the major conclusion of which is that industry will continue to utilize traditional reserves methods (with tighter constraints) while continuing to develop methods/practices which will provide a unique characterization of reserves for un-conventional reservoir systems. • This work also provides a statistical study derived from approxi-mately 25,000 shale gas wells. This work confirms the use of the Arps' "modified hyperbolic" model and provides statistical results (EUR's and other DCA results) on a play-by-play basis. Conclusions: • Industry Survey: • DCA will remain a primary reserves tool (and will evolve). • Definition of simulation models will likely rely more on analogs. • Reserves Analysis: • DCA is not fully representative (and must be constrained). • Time-Rate-Pressure analyses will become common (e.g., RTA).

30. Best Practices for Reserves Estimation in Unconventional Reservoirs — Present and Future Considerations End of Presentation Tracy HECKMAN, Anadarko Petroleum Corporation Grant OLSEN, Pressler Petroleum Consultants Kerry SCOTT, Pioneer Natural Resources Bernard SEILLER, Total Marcia SIMPSON, EXCO Resources Tom BLASINGAME, Texas A&M University