1. Well test analysis of Shale gas wells JIP in New Developments in Well Test Analysis 12 September 2014 Irina Kostyleva, PhD student

2. Objectives test applicability of conventional well testing and reservoir simulation techniques to shale gas data find a physical explanation for the production rates behaviour in ultra-tight formations explain the rapid productivity decline of shale gas wells by means of physical principles achieve history matching of the pressure and rate data based on the discovered principles

3. Shale rock description 3 Self-sourcing, no conventional trap required (source rocks for conventional reservoirs) Large quantities of gas stored in pores and fractures, adsorbed gas Ultra-low matrix permeability Stress-dependent properties High brittleness, stiffness Mature, organic rich Massive stimulation is required to produce: horizontal wells with multiple hydraulic fractures increase well-reservoir contact area Haynesville shale: 10,500-13,000 ft depth 200 ft thickness

4. Agenda Production data analysis Semi-analytical simulations: applicability of conventional WTA Types of fractures, regions with different permeability Stress-dependent permeability for matrix and fractures Increasing skin at later times Pseudofunctions and deconvolution 4

5. Well trajectories, completion data (in production since 2008) Gas analysis (composition) Production history (surface gas and water rates) Bottom hole pressures recorded during the build-ups Entire pressure history (converted from tubing/casing wellhead readings) BU derivative analysis: Linear flow Late time stabilization Available data 5

6. BU derivative analysis BU derivative can be matched with a single fracture model – but not the entire history Negative skin is observed (-5 ~ -10) Steep productivity decline is observed 6

7. Flow regimes: multiply fractured horizontal well Semi-analytical models: Fractures radial flow Bilinear flow Inner linear flow Quasi-steady state flow Late-time outer linear flow SPE 149311 (Cheng, 2011) 7

8. Production data analysis 8

9. Multiply fractured horizontal well: BU 9 Nf=30 Nf=80

10. Hydraulic fracturing Ductile rocks / claystone Brittle rocks / siltstone Fractures createdTransverse vertical planar Complex network in multiple planes TreatmentGel/high viscosity fluidsSlickwater fracturing ProppantLarge proppant Low concentration of small proppant Geomechanics High ratio of max-stress to min-stress High Young’s modulus, low Poisson’s ratio ExamplesMarcellus, HaynesvilleBarnett Modelling LGR, symmetry element Complex planes/ unstructured grid 10

11. Trilinear model (Brown et al.) Enhanced permeability region (Stalgorova and Mattar) Multiple types of fractures: natural / hydraulic / stress- released / drilling-induced (Moridis et al.) Analytical solutions A single repetitive element is used due to system symmetry 11

12. Qcum vs. Normalised rate 12

13. Wells with decline 13

14. History matching: well B Increasing S at later times Fractures k(P) decline Matrix permeability decline Change of regimes is clearly seen on Qcum vs. rate plot

15. Well B Increasing S kf(P) decline km(P) d ecline Match quality 15

16. Well H Increasing Skf(P) decline Match quality 16

17. k(P) comparison 17 Rapid permeability decline to 50% of initial value Further decline (if present) has exponential form History is short to define whole trend in some of the wells Threshold values vary: 5,000- 9,000 psia Permeability decline starts earlier for shorter fractures as they are depleted faster

18. Skin comparison 18 Skin effect starts at 3,000-4,000 psia

19. Skin-effect starts acting at ~4,000 psia 19 Kf(P)Skin Kf(P) Skin

20. Skin effect 20

21. Parameters Group# fracsSpacing, ftdf, ftXf, ftK(P) S km, mDkf. mD B Decline + Skin 696513203 ++ 0.10.0001 PF706615106 ++ 0.0060.0001 D646917442 ++ 0.020.0002 H706313485 ++ 0.0120.0001 S5338217700 ++ 0.040.0002 Br4210513128 ++ 0.010.0002 SP4010020160 ++ 0.20.0004 OH Decline 3410816380 +- 0.10.0001 S8528316350 +- 0.20.0001 PL825716285 +- 0.010.0002 Mc49791783 +- 0.0070.0002 S9 Linear 756117300 -- 0.150.0002 A Drop 4477 E4584 G4886 Sh4884

22. Wells with rapid decline: constant Pwf 22 Sharp initial decline Pwf ~ 1100-1200 psia Short history for wells G, Sh Possible reservoir damage Well Sh Well E Well A A E Sh Well G G

23. Pseudopressure formulation 23 With these pseudo-functions the diffusivity equation becomes approximately linear Normalised pressure pseudo-function: Normalised time pseudo-function:

24. Deconvolution (Well B) 24 km kf

25. Conclusions Single equivalent fracture solution is inadequate due to fractures interference Interpretation of MFHW response with the single equivalent fracture solution gives very negative values of skin DD exhibit early linear flow behaviour and later PSS transition, BU derivative response shows strong production time effect The reservoir outside the stimulated volume is not flowing and true reservoir radial flow is not reached during any reasonable time Enhanced permeability regions and skin (salt deposit) explain the stages in productivity decline Pressure-dependent permeability relationships for the inner region were found from production data history matching Further damage (increasing skin) occurs at ~4,000 psia True deconvolved response is obtained if pseudo-functions employing k(P) relationship is used to linearize the system 25

26. Future work Documenting effect of different factors to illustrate that the combination of parameters is unique Explain the reasons why different patters occur in production data of the wells Access the uncertainty in parameters estimation Analyse deconvolution results 26

27. Thank you

28. Desorption A large part of gas desorbs only at low pressures (contributes to long-term gas production only) The effect of desorption cannot be detected in a single-well production test, but it is easily determined from geologic information Langmuir isotherm where V L is the Langmuir volume – maximum adsorbed volume of gas at infinite pressure P L is the pressure at which half of is adsorbed 28

29. Effect of permeability k=0.0001mD k=0.01mD k=1mD k=0.0001mD k=0.01mD k=1mD fracture linear flow dies out quickly in case of high permeability reservoir with high number of fractures. The transition to radial flow is shifted to the later times with decrease of permeability and not observed at all for low k High permeability: flow outside the fractures tips is contributing. Low permeability: flow stays strictly linear and any matrix flow from outside of the stimulated reservoir volume is negligible at early times 29

30. Well B Change of regimes is clearly seen on Qcum vs. rate diagnostic plot Increasing S at later times Fractures k(P) decline Matrix permeability decline Match quality 30