IMPROVED RESERVOIR ACCESS THROUGH REFRACTURE TREATMENTS IN TIGHT GAS SANDS AND GAS SHALES Mukul M. Sharma The University of Texas at Austin Petroleum and Geosystems Engineering April 15, 2009
Outline Project Objective and Goals Timing, Project Participants, Major Milestones Value of the Research Project Impact Technical Overview Status of Current Technology Project Deliverables Progress to Date Technical Issues/Problems Encountered Summarize
Project Objectives and Goals Use stress reorientation models to quantify the role played by stress reorientation on refrac productivity improvement. Calibrate the findings with analysis of extensive field data Improve our ability to predict refrac production enhancement Candidate well selection Timing of refracs Improve refrac design based on findings.
Project Participants University of Texas at Austin Contact Mukul M. Sharma Professor of Petroleum & Geosystems Engineering Anadarko Petroleum Corp. Contact Jon David Caron Project Engineering Advisor Noble Energy Contact Michael Zoll Completions Manager Denver, CO BJ Services Contact Satya Gupta Senior Research Leader Tomball Technology Center Pinnacle Technologies Contact Steve Wolhart Region Manager
1 Research Management Plan 2 Technology Status Assessment Milestones 3 Data compilation for the Codell formation 4 Data compilation for the Barnett shale 5 Stress reorientation model implementation and runs for Codell re-fracs 6 Stress reorientation model implementation and runs for Barnett shale re-fracs 7 Evaluation of fractured well performance in the Codell, Barnett and horizontal wells 8 Candidate well selection based on poro-elastic model and field data analysis 9 Design of re-frac treatments in the Codell, and Barnett based on simulations, new fluids and proppants 10 Design of re-frac treatments in horizontal wells based on simulations, new fluids and proppants 11 Implementation of re-frac treatments in the Codell, and Barnett (new designs). 12 Post frac evaluation of re-frac treatments in the Codell, Barnett and horizontal wells 13 Workshop in Houston to discuss results 14 Final report with all the findings from the study
1 3 Project Timing Task Year 2 Year 3 4 5 6 7 8 9, 10
Value of the Research Project Impact Beating the decline curve in unconventional gas reservoirs requires continuous drilling and fracturing In a low gas price environment re-frac treatments offer a low cost alternative to drilling new wells. Performance of re-fracs is highly variable and must be made more reliable and predictable. This project aims to help accomplish that.
Status of Current Technology Refrac candidate well selection is based on: Statistical databases Heuristic rules of thumb Neural networks No systematic way of deciding on the timing of the refracs Current refrac treatment designs are done very much like the original fracs. Typically, no account is taken of stress reorientation or previously placed proppant.
Project Deliverables Monthly status reports. A final report on the results of the Defined Effort. Guidelines for selecting candidate wells for refracturing. Guidelines for selecting the appropriate timing of refracturing given a set of reservoir properties. New designs for better placement of proppants during refracturing operations. Guidelines for fracture placement and spacing in horizontal wells. Guidelines for avoiding fracture interference in wells with multiple fractures. New proppant placement strategies for horizontal well fractures. Detailed analysis and results for at least four refracture treatments in tight gas and gas shale wells.
Project Deliverables A report on statistical analysis of the refracture database in the Codell formation (2500 refracture treatments). Quantitative guidelines for when to use energized fluids when refracing depleted formations. Guidelines for when to use light weight proppants in refracture treatments. A web site with information about the project and updates as appropriate. A minimum of two presentations in local professional organization meetings; one each in Permian and San Juan Basin areas. At least one presentation at a RPSEA-directed event. An article discussing this project to at least one produceroriented trade journal. UT will provide technical results containing details and data to be utilized for determination of program impact as requested by RPSEA.
Proposed Tasks Task 4. Stress Reorientation around Fractured Wells: Implications for Re-fracturing Subtask 4.1 Data compilation in the Codell formation and the Barnett shale Subtask 4.2 Stress re-orientation around fractured wells in shales and tight gas sands Subtask 4.3 Models for stress reorientation in naturally fractured formations Task 5. Selecting Timing and Candidate Wells for Re-fracturing Task 6. Re-fracture Designs for Deviated and Horizontal Wells Task 7. Proppant Placement in Re-fracturing Treatments (Vertical and Horizontal Wells) Task 8. Use of Novel Proppant Placement Strategies in Re-fracturing Operations Task 9. Field Design of Re-Fracture Treatments in the Wattenberg Field Task 10: Design, Implementation and Evaluation of Field Fracture Designs
Task 4: Stress Reorientation Elastic, homogeneous and isotropic reservoir Biot s poroelasticity theory Constant pressure in vertical well and initial fracture Presence of bounding layers with different mechanical properties Pay Zone Bounding Layer Initial Fracture
Task 4: Stress Reorientation around a producing well
Analytical solution compares well with numerical solution Ref: Zhai, Sharma, 2004
Stress Reorientation Around Producers and Injectors Producer Direction of Maximum Stress Injector Stress Reversal occurs No Stress Reversal Angle of Stress Reorientation
Stress Reversal Region Producer Direction of Maximum Stress Isotropic point Angle of Stress Reorientation Fracture half-length Stress reversal region impacts direction of refracture measured in the field
Dimensionless Parameters (Berchenko et al., 1997; Siebrits et al., 1998; Rousell and Sharma, 2009) Dimensionless Time Dimensionless Stress Deviator 4ct κ t 4kt τ = = 4 = 2 2 Lxf S Lxf 2 1 α ( 1+ ν)( 1 2ν) μlxf + M ( 1 ν ) E Π= S 0 = S 0 σ = hmax σ hmin σ * ηp * α 1 2ν p Ri p wf ( ) 1 ν Dimensionless Fracture Height Ratio Dimensionless Shear Modulus Ratio γ = H L xf β = G b G r
Parameters Affecting the Stress Reversal Region The areal extent and timing of the stress reversal depend on: Fluid properties Reservoir characteristics Stress contrast Drawdown Thickness of the reservoir Mechanical properties of the bounding layers January 20, 2009 DOE Project Kick-off Meeting 18
Task 5. Selecting Timing and Candidate Wells for Re-fracturing The main results were recently published. Quantifying Transient Effects in Altered- Stress Refracturing of Vertical Wells, SPE 119522, Presented at the SPE Hydraulic Fracturing Meeting, Woodlands, 2009, Nicolas P. Roussel, Mukul M. Sharma. Work continues to include the effects of stresses induced by fracture creation Comparison with field data.
Task 5. Selecting Timing and Candidate Wells for Re-fracturing 0.25 λ max 0.2 Maximum areal extent of stress reversal L xf' / L xf 0.15 0.1 Time (months) Shale 0.05 Tight Gas τ max = 1.3 τ max = 1.15 τ max = 4.13 days months years Sandstone 0 0.001 0.01 0.1 1 10 100 1000 Optimum time for refracturing
Main Findings An approaching fracture will go: Away from a production well Toward an injection well Stress reorientation depends on: Drawdown Stress anisotropy Moduli Stress reversal does occur in fractured producers. We can now compute its, Spatial extent Timing
Field Data for Validation 3 wells where refracs worked and 3 wells where refracs did not work A complete dataset would include: Wellbore schematic Base map showing location of wells Details of frac and refrac jobs Logs (dipole sonic) Microseismic Gas flow rate before / after refrac January 20, 2009 DOE Project Kick-off Meeting 22
Task 6. Re-fracture Designs for Deviated and Horizontal Wells t = 0
Stress Reorientation for a Production, Injection Well Pair t = 0
Stress Reorientation for 1 Production, 2 Injection Wells t = 0
Stress Reorientation for 2 Production, 1 Injection Well t = 0
Task 7, 8. Use of Novel Proppant Placement Strategies in Re-fracturing Operations Status: Work is underway and we have some initial results.
Task 9, 10. Design of Re-Fracture Treatments in the Wattenberg Field Wattenberg field, D-J basin Codell formation Thin sandstone layer Low permeability, requires stimulation Refractured since 1998 Observations indicate that refracture performance is dictated by fracture-fluid viscosity profile (Ref: Miller, J. et al., 2004, SPE 90194) Fracture reorientation has been reported (Ref: Wolhart, S. et al., 2007, SPE 110034) Source: USGS
Objectives Use principal component analysis to determine the increase in production rate after a refracture treatment. Use stress reorientation models to study the role played by stress reorientation vs other factors such as GOR and depletion. Use these findings to recommend timing for refracs Create a statistical, predictive model for Production enhancement Candidate well selection
Data Set Refracture well data, approx. 4000 wells Anadarko, Noble Energy (1999-2008) Groups Well information Orig. frac treatment Pre-refrac data Refrac design Refrac treatment Rheology Water quality Job comments Refrac data Description Year Volume of gel and proppant during the first fracture Production information and number of perforations Gel loading, pad size, surfactant, etc Fluid injection, perforations Viscosity measurements, gel usage Water source, composition Problems during the job Production increment
Issues with Data Analysis Too many parameters (58) Incomplete data sets Wide range of refrac production increment -19.6 ~ 4756 BOE/mo Non-numeric data
Data Set Reduction Elimination Elimination criteria # missing entry > 500 Wells with any missing parameter values Reduced dataset Parameters: 58 43 Entries: 2154 1279 Param Value Missin g Entrie s MGA L 131.1 0 131.2 1 132.5 9 132.0 0 131.8 8 131.4 5 130.7 6 135.0 0 127.3 9 Refrac Treatment AVG. RATE New PERF Total PERF Leakof fcoef. Perf Fric tion Avg. PSI 13.60 19 28 0.0010 54 6550 13.50 30 43 0.0021 100 4400 13.90 0 60 0.0027 12 5300 13.00 28 34 270 4900 14.00 30 50 0.0013 0 7600 14.50 26 33 0.0013 40 4600 12.80 18 33 0.0013 210 6100 14.00 37 0.0013 10 4850 13.00 40 49 0.0013 667 6750 55 58 108 206 1560 Missing Values 154 8 62
Data Set Reduction Change of Statistical Properties Frequency 800 700 600 500 400 300 200 100 Frequency 400 350 300 250 200 150 100 50 0-2000 0 2000 4000 6000 8000 Production Increment (BOE/mo) Mean : 1282.6 Median : 1152.3 Std: 814.7 Data Set Reduction 0-1000 0 1000 2000 3000 4000 5000 Production Increment (BOE/mo) Mean : 1302.9 Median : 1161 Std: 754.5
Statistical Analysis Correlation (2) Observe the correlations between production increment and parameters 5000 5000 Production Increment (BOE/mo) 4000 3000 2000 1000 0 Production Increment (BOE/mo) 4000 3000 2000 1000 0-1000 0 1 2 3 4 Codell Phi*H Corr. value: 0.25-1000 0 0.2 0.4 0.6 0.8 1 Pre-Refrac Cum Rec. Factor Corr. value : 0.21
Statistical Analysis Correlation (1) Observe correlations between parameters Pre-Refrac Cum Rec. Factor 1 0.8 0.6 0.4 0.2 0 0 1 2 3 4 Codell Phi*H Corr. value : 0.13 Total Hardness ppm 700 600 500 400 300 200 100 0 0 50 100 150 200 Calcium ppm Corr value : 0.98
Statistical Analysis Linear Regression R 2 = 0.47 Median error = 42.5% Mean error = 95.9% Less number of dimensions can help improve Predicted production increment 4000 3500 3000 2500 2000 1500 1000 500 0 regression 0 500 1000 1500 2000 2500 3000 3500 4000 Actual production increment good bad
Parameter Reduction Correlation among parameters suggest a presence of highly correlated parameters Eliminate highly correlated parameters: 43 25 Orig. Frac treatment (5) Water quality (11) Pre-refrac (4) Orig. Vol (MGAL) Orig. Zone Total hardness Pre-refrac Cum Rec. Factor More entries recovered: 1279 1316
Statistical Analysis Linear Regression w/ Reduced Parameters R 2 = 0.31 Median error = 42.1% Mean error = 100.9% Requires a more sophisticated method for better fits Predicted production increment 4000 3500 3000 2500 2000 1500 1000 500 good bad 0 0 500 1000 1500 2000 2500 3000 3500 4000 Actual production increment
Statistical Analysis Principle Component Analysis Original dataset [m x n] mapped into new orthogonal vectors Variance of dataset can be captured with less dimensions Reduction of dimensions provides a better regression Covariance matrix [n x n] Eigenvectors [n x n] Choose eigenvectors [n x k] Dataset transformation [m x k]
Future Work Fill-in missing data instead of row elimination to increase the number of wells in data set. Work with Noble Energy on adding additional data (e.g. regional stress data) Data mining techniques K-means clustering Neural nets
Technical Issues/Problems Encountered Minor issues with getting contract in place. Data access issues have been resolved.
Summary of Progress to Date Stress reorientation due to poroelastic effects has been calculated for vertical, fractured and horizontal wells. Key parameters and conditions that control this stress reorientation have been identified. The optimum timing of refrac treatments has been computed for the first time. A data set of refrac treatments from the Wattenburg field has been reviewed and is being analyzed for statistical trends. Review of refrac treatment designs in progress.
I would like to Acknowledge: RPSEA for their support. Our partner companies (Anadarko, BJ Services, Noble Energy, Pinnacle) for collaboration and access to data. Members of the Fracturing and Sand Control JIP at the University of Texas at Austin (Anadarko, BJ Services, BP, ConocoPhillips, Halliburton, Schlumberger, Shell, Total) for providing the cost sharing for this project. Thank you Questions?