A COMPARISON OF HYDRAULIC VERSUS PROPELLANT FRACTURE GENERATION IN SHALE SYSTEMS

by

James C. Page A thesis submitted to the Faculty and Board of Trustees of the Colorado School of

Mines in partial fulfillment of the requirements for the degree of Master of Science

(Petroleum Engineering).

Golden, Colorado

Date______

Signed: ______James C. Page

Approved: ______Dr. Jennifer L. Miskimins Thesis Advisor

Golden, Colorado

Date______

______Dr. Craig W. Van Kirk Professor and Head Department of Petroleum Engineering

ii ABSTRACT

Approximately two-thirds of all oil and gas wells drilled in the United States are completed with some form of stimulation. Stimulations are used to bypass drilling induced damage and increase effective wellbore radii, in order to increase productivity.

Production from unconventional reservoirs including tight gas, coalbed methane and shales is becoming more common; however less research has been published on stimulation methods for these unconventional systems.

This thesis is a study of fracture propagation in a shale system. Several shale plays such as the Barnett shale in eastern Texas and the Bakken shale in North Dakota have become significant producers of natural gas in recent years. The Mancos shale is an organic rich, late Cretaceous shale that is approximately 2000 ft thick in the Douglas

Creek Arch area of Western Colorado, and is the area of interest for this study. It is the source rock for several conventional producing systems including the Weber, Mesaverde,

Cedar Mountain, Morrison and Dakota formations (Kirschbaum, 2003).

In particular, this thesis is a comparison of propellant and hydraulic stimulations in the Mancos shale. In December 2005, Wieland published a thesis entitled Laboratory

Testing and Finite Element Modeling of Propellant-Induced Fracture Propagation in

Shale Reservoirs. This thesis builds upon Wieland’s laboratory work with further laboratory experiments and analysis, as well as propellant and hydraulic field stimulation tests in the Douglas Creek Arch area of western Colorado.

iii The propellant fracture big block test conducted by Wieland is further analyzed by comparing the fracture geometry with unconfined compressive strength test data. The pressure responses recorded during inadvertent hydraulic fracturing of the Mancos block are also investigated. Mechanical properties of the core plugs tested by TerraTek using triaxial compression test equipment are also determined using acoustical methods under confining pressure. The acoustically derived mechanical properties of the core samples are compared with the mechanical properties of the Mancos formation as determined by acoustic logging.

As well as laboratory experiments, propellant field tests in three Mancos wells and a two-stage hydraulic fracture stimulation test are all analyzed. Treating pressures, radioactive tracer isotope data and well logs are all compared to evaluate the field tests.

The research presented in this thesis led to the following main conclusions:

• The propellant fracture propagation in the big block test was constrained by

relative strengths of the strata, not in-situ stress or layering effects.

• The inadvertent hydraulic fracture in the Mancos test block was caused by a flaw

that reduced the required breakdown pressure; the fracture then began propagating

once the minimum in-situ stresses had been overcome.

• The layering of the Mancos shale resulted in better height containment than

expected in the hydraulic field test stimulations. The mechanism of containment

(i.e. rock properties, in-situ stresses or layer effects) is still not fully understood.

iv TABLE OF CONTENTS

ABSTRACT ……………………………………………………………………….. iii

TABLE OF CONTENTS ………………………………………………………….. v

LIST OF FIGURES ………………………………………………………………... viii

LIST OF TABLES …………………………………………………………………. xiii

ACKNOWLEDGEMENTS ……………………………………………………….. xiv

Chapter 1 INTRODUCTION ……………………………………………………… 1

1.1 Available Mancos Shale Laboratory and Field Data ………………. 2 1.2 Research Objectives ……………………………………………….. 3 1.3 Research Contribution ……………………………………………… 4

Chapter 2 LITERATURE REVIEW ………………………………………………. 5

2.1 Mancos Shale ……………………...……………………………….. 5 2.2 Propellant Stimulations ……….……………………………………. 8 2.3 Fracture Modeling …………………...…………………………….. 9 2.4 Fracture Propagation in Laminated Shales ………………………… 10 2.5 Acoustic Logging ………………………...... ……... 12 2.6 Synthetic Logs …………………………...... ……… 14 2.7 Unconfined Compressive Strength .……...... ……… 14 2.8 Wieland’s (2005) Thesis .………………...... ……… 15

Chapter 3 SONIC CHARACTERIZATION OF MANCOS SHALE BLOCK AND WELL LOG CORRELATIONS ………………………….…………….. 21

3.1 Hand Held Velocity Probe Characterization ………………………. 21 3.2 Ultrasonic Core Measurements …..………………………………… 23 3.2.1 Experiment Methodology …..…..………………………………… 23 3.2.2 Results ……………………………………………………………. 25 3.3 Sonic Correlations …………………………………………………. 29

v Chapter 4 MANCOS BLOCK FRACTURE INVESTIGATION …………………. 38

4.1 Further Observations of the Propellant Fracture Face ……………… 38 4.2 S.E.M. Investigation of Quartz Band ………………………………. 44 4.3 Mancos Block Hydraulic Fracture Investigation …………………… 47 4.4 Hydraulic Fracture Pressure Response …………………………….. 48 4.4.1 TerraTek Report Summary ………………………………….…… 50 4.4.2 Hydraulic Fracture Pressure Analysis ……………………………. 50

Chapter 5 PROPELLANT FRACTURE STIMULATION FIELD TESTS ...…….. 53

5.1 Propellant Stimulation Design ……………………………………... 54 5.2 Stimulation Operations …………………………………………….. 59 5.3 Results ……………………………………………………………… 59 5.4 Conclusions ………………………………………………………… 60

Chapter 6 SLICKWATER STIMULATION FIELD TEST ….…………………… 62

6.1 Modeling …………………………………………………………… 64 6.1.1 FracCADE™ Simulation ………………………………………… 64 6.1.2 Synthetic Logs developed for GOHFER™ Simulation ………….. 68 6.1.3 GOHFER™ Simulation ………………………………………….. 72 6.2 Actual Stimulation …………………………………………………. 73 6.3 Tracer log …………………………………………………………... 77 6.4 Comparison with Modeling ………………………………………… 79 6.5 Traced Fracture Height as Compared with Logs …………………… 80

Chapter 7 DISCUSSION ……..………………………..…………………………… 87

7.1 Mancos Formation Correlation …………………………………….. 87 7.2 Propellant Block Test ...……………………………………………. 88 7.3 Inadvertent Hydraulic Fracture …………………………………….. 92 7.4 Propellant Field Tests ………….…………………………………… 93 7.5 Slickwater Field Tests ………….….…..…………………………… 93 7.6 Further Discussions ………………………………………………… 94

Chapter 8 CONCLUSIONS ……………………………………………………….. 96

8.1 Future Research …..………………………………………………… 97

NOMENCLATURE ……………………………………………………………….. 99

vi

REFERENCES …………………………………………………………………….. 100

APPENDIX A S.E.M. Images……………………………………………………. 104

APPENDIX B Sections of TerraTek Report Pertinent to Mancos Shale Block Test. 114

APPENDIX C Slickwater Treatment Reports …………………………………… 129

APPENDIX D Tracer log of Slickwater Fracture Treatment…………………….. 139

APPENDIX E Digital Copies ………………………………………….. Enclosed CD

vii LIST OF FIGURES

Figure 2.1: Douglas Creek Arch area of interest and its relationship to the Piceance basin and the states of Colorado and Utah. …………………………………. 6

Figure 2.2: Douglas Creek Arch stratigraphy (Modified from Kirschbaum, 2003). 7

Figure 2.3: Diagram of the original applied stresses relative to the pressure probes (left) and the redesigned applied stresses relative to the pressure probes (right). Modified from Wieland, 2005. ……………………………………………… 17

Figure 2.4: Data showing the unintentional hydraulic fracture of the Mancos sample. ………………………………………………………………………. 17

Figure 2.5: Initial pressure response propellant deflagration Mancos sample. …… 18

Figure 2.6: Unconfined Compressive Strength scratch test with photograph of core. 19

Figure 2.7: Graph showing the maximum pressure response and fracture growth plotted against model time. ……………………………………………..…... 20

Figure 3.1: Diagram and photo of velocity probe setup. Measurements of wave travel times were taken on the propellant fracture face as well as one side of the block. The oscilloscope photo shows a compressional wave. ………….. 22

Figure 3.2: Diagram and photo of core setup. …………………………..……….… 24

Figure 3.3: Calculated dynamic Young’s Moduli versus applied confining stress. 27

Figure 3.4: Calculated Poisson’s Ratio versus applied confining stress. ….………. 28

Figure 3.5: X-Y cross plot of Young’s modulus versus density at different confining pressures. ………………………………………...……………….. 30

Figure 3.6: Young’s modulus from dipole sonic log from Lower Horse Draw 2186 compared with Young’s modulus calculated using Equation 3.3. ………….. 32

Figure 3.7: X-Y Scatter plot of Young’s moduli from density correlation and Lower Horse Draw 2186 dipole sonic log. The correlation coefficient is 0.35. ……. 33

viii Figure 3.8: Young’s modulus from dipole sonic log compared with the shifted Young’s modulus values calculated using Equation 3.3 and a gamma ray curve. The best Mancos correlation can be seen between 4000 ft and 4400 ft. (Values are averaged over 5 ft intervals). ……………………………………………. 34

Figure 3.9: X-Y Scatter Plot of Young’s moduli derived from Equation 3.3 as compared with dipole sonic Young’s moduli for gamma ray counts greater then 125. The correlation coefficient is 0.37. ……………………………….. 35

Figure 3.10: X-Y Scatter Plot of Young’s moduli derived from Equation 3.3 as compared with dipole sonic Young’s moduli for gamma ray counts of less then 125. The correlation coefficient is 0.25. ………………………………. 35

Figure 3.11: X-Y Scatter plot of Young’s moduli with a gamma ray counts between 125 and 140. These cutoffs make the best correlation with a correlation coefficient of 0.43. ………………………………………………………...... 36

Figure 4.1: Photograph of the north half of the Mancos shale block showing the propellant fracture staining. Two layers labeled “1” and “2”, have significantly affected the fracture geometry. ……………………………….. 39

Figure 4.2: Outline of the propellant fracture staining correlated with U.C.S. scratch test. ………………………………………………………………………….. 39

Figure 4.3: Illustration of potential in-situ stresses and stress differentials between layers for Poisson’s ratio values between 0.19 - 0.22. ….…………………... 43

Figure 4.4: Low magnification of the quartz band sample showing lack of porosity and a well sorted grain size framework. The darker grains are quartz cemented with a mixture of materials including, dolomite, calcite and illites. Scope magnification: 250X. ……….…...……………………………………... 44

Figure 4.5: Typical detrital quartz grain in quartz band; height ≈ 0.05mm, length ≈ 0.025mm, meaning this would be defined as a coarse siltstone. Chart shows silicon and gold as the only elements present. Scope magnification: 1200x. ………………………………………………………………………… 45

Figure 4.6: Photograph of the north half of the Mancos shale block showing the hydraulic fracture. ………………………...………………………………… 47

ix Figure 4.7: Data recorded during the inadvertent hydraulic fracture. The recorded wellbore pressure response is marked in orange. Significant event occur at time 447 and 519 seconds as recorded by the pressure gauge...... ……. 49

Figure 4.8: Loading during hydraulic fracture stimulation. …………….…….....… 51

Figure 5.1: Hells Hole 9139x well logs. Gamma ray, resistivity, density and gas- while-drilling curves. The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007). .………………….. 55

Figure 5.2: Hells Hole 9131 well logs. Gamma ray, resistivity, density and gas-while- drilling curves. The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007)...... 56

Figure 5.3: Park Mountain 9025 well logs. Gamma ray, resistivity and density (gas- while-drilling curve not available in electronic format). The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007). ……………………………………………………………….. 57

Figure 5.4: Correlated cross section of the three propellant stimulated wells, stratigraphic picks by Fisher (2007). ………………………………..………. 58

Figure 5.5: Pressure build-up test results form the propellant stimulation of the Hells Hole 9139x. ……………….……………………………………………….... 60

Figure 6.1: Hells Hole 9139x well logs, gamma ray, resistivity, density and gas while drilling logs. The perforations for the two slickwater fracture stages are marked with purple boxes on the depth scale. ………………………….. 63

Figure 6.2: Stage I fracture profile and proppant concentration. ...………………… 67

Figure 6.3: Stage II fracture profile and proppant concentration. ……………...….. 67

Figure 6.4: Locations of the wells with dipole sonic logs and the Hells Hole 9139x. 69

Figure 6.5: Lower Horse Draw 2186 gamma ray, resistivity, neutron porosity, photoelectric absorption, density, compressional wave travel time and shear wave travel time logs. ………………………………………………..……… 70

x Figure 6.6: Hells Hole 9139x synthetic logs, compressional wave travel time, Poisson’s ratio, Young’s modulus, Biot’s constant, total closure stress and shale fraction. ……………………………………………………………….. 71

Figure 6.7: GOHFER™ Stage I fracture profile and proppant concentration...... 73

Figure 6.8: Stage I treatment plot for perforations 6,830 ft – 7,058ft...... 74

Figure 6.9: Stage II treatment plot for perforations 5,994 ft – 6,243 ft...... 74

Figure 6.10: Stage II Nolte-Smith plot. …...... 76

Figure 6.11: Sections of the tracer log showing the two slickwater stages pumped. The blue represents the antimony tracer used to tag the pad in each stage, yellow represents the scandium used to tag the proppant in Stage I and the red represents the iridium used to tag the proppant in Stage II...... 78

Figure 6.12: Hells Hole 9139x gamma ray curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals)...... 81

Figure 6.13: Hells Hole 9139x synthetic Poisson’s ratio curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals)...... 82

Figure 6.14: Hells Hole 9139x synthetic Young’s modulus curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals)...... 83

Figure 6.15: Hells Hole 9139x synthetic closure stress curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals)...... 84

Figure 6.16: Hells Hole 9139x gamma ray curve with the top of Stage 1 fracture height growth labeled (0.5 ft values)...... 85

Figure 6.17: Hells Hole 9139x gamma ray curve with the top of Stage 2 fracture height growth labeled (0.5 ft logged API measurements)...... 86

Figure 7.1: Lower Horse Draw 2186, standard logs plus Poisson’s ratio curve. Several silt sections within the Mancos are labeled. The silts all have significantly lower Poisson’s ratio values then the shales...... 90

xi Figure 7.2: Photograph of the north half of the Mancos shale block showing the propellant fracture staining. The two layers labeled “1” and “2” have significantly higher unconfined compressive strength. It is the mechanical properties of these two layers in particular that controlled the propellant fracture growth...... 91

xii LIST OF TABLES

Table 2.1 Geomechanical Test Results ...... 18

Table 3.1 Transit Time Measurements ...... 24

Table 5.1 Perforation Intervals ...... 59

Table 6.1 Key Input FracCADE™ Simulation Parameters for Stage I ...... 65

Table 6.2 Key Input FracCADE™ Simulation Parameters for Stage II ...... 65

Table 6.3 Simulated Pump Schedule for Stage I ...... 66

Table 6.4 Simulated Pump Schedule for Stage II ...... 66

xiii ACKNOWLEDGEMENTS

First and foremost I would like to thank Dr. Jennifer Miskimins, my advisor, for

her support, time, ideas and guidance. Without Dr. Miskimins, my master’s experience

would not have been as rewarding as it was.

I would like also to acknowledge and thank Dr. Fleckenstein and Dr. Graves for

serving as committee members. In particular, I would like to thank them for their input

and advice which significantly impacted the direction of research and completeness of the project.

Additionally, I would like to thank Dr. Batzle of the Geophysics department and

Mr. Skok of the Geology department for their time, use of equipment and guidance.

I also wish to thank EnCana Oil and Gas (USA) - Peter Loeffler, Bob Weaver,

and Richard Champion for their support. Thank you for bringing the Douglas Creek Arch

Mancos shale project to the Colorado School of Mines.

xiv CHAPTER 1

INTRODUCTION

Approximately two-thirds of all oil and gas wells drilled in the United States are completed with some form of stimulation. Stimulations are used to bypass drilling induced damage and increase effective wellbore radii in order to increase productivity. Without stimulation, many wells would be uneconomical. Stimulation types include hydraulic fracturing, acid fracturing, matrix acidizing and propellant fracturing. The majority of stimulations performed worldwide are hydraulic fracture stimulations. However, hydraulic fracture stimulations have their shortcomings.

Hydraulic fracturing requires the introduction of foreign fluids and chemicals into target formations which can often adversely affect flow into the wellbore. In formations where fracturing fluid reactions are a serious issue, one solution is propellant fracturing. Propellant fracturing is a process which uses a combined solid fuel and oxidizer to create the rapid expansion of gasses, generate pressure and overcome formation stresses, thus driving a fracture out into the rock. Propellant fracturing requires no foreign liquids be introduced to the formation, therefore minimizing potential issues with clay swelling and skin. The major disadvantage to propellant fracturing is that it does not carry proppant into the fracture. Instead, propellant fracturing relies upon shear slippage or spalling to prevent the fracture from fully closing back on itself, leaving a conductive path back to the wellbore.

In December 2005, Wieland published a thesis at the Colorado School of

Mines entitled “Propellant-Induced Fracture Propagation in Shale Reservoirs.”

Wieland conducted two laboratory-scaled propellant fracturing experiments, initially

deflagrating propellant in a Colton sandstone block and then deflagrating propellant

in a Mancos shale block. Wieland used the pressure response data and the mechanical

properties of the blocks to model the fracture propagation using a finite element

modeling package.

In conjunction with Wieland’s laboratory experiments, EnCana Oil and Gas

(USA) followed up with a series of Mancos shale stimulation field tests in an area of

the Douglas Creek Arch of Western Colorado. Both propellant and hydraulic

stimulation technologies were tested.

Much research has already been published concerning how hydraulic fractures

propagate in conventional reservoirs. In contrast, this thesis is a study and comparison

of hydraulic and propellant fracturing in the Mancos shale, a potential unconventional

hydrocarbon reservoir. Laboratory and field data from both propellant and hydraulic

stimulation treatments are all compared.

1.1 Available Mancos Shale Laboratory and Field Data

Wieland’s (2005) Mancos block test was designed to study propellant

deflagration; however, while pressurizing the block’s wellbore in preparation for the

propellant test, the block was inadvertently hydraulically fractured. Logically, the hydraulic fracture propagated in the direction of maximum horizontal stress. In order to continue his experiment, Wieland applied additional stress perpendicular to this fracture plane, effectively swapping the directions of minimum and maximum horizontal stress. The propellant was deflagrated and a propellant fracture was

2 propagated perpendicular to the hydraulic fracture. This series of events created a

unique data set with propellant and hydraulic stimulation data available in the same

Mancos shale block. Unfortunately, since the hydraulic fracture of the block was

unintentional, only a partial data set is available for analysis from this component.

Following the laboratory tests, several field tests were conducted in the

Mancos shale of the Piceance Basin, Colorado. Propellant field tests were conducted

in the Mancos sections of three wells, Hells Hole 9131, Hells Hole 9139x and Park

Mountain 9025. A pump-in test was conducted in the Mancos section on the Hells

Hole 9131, as well as a two-stage slick-water fracture treatment in the Hells Hole

9139x. A standard suite of logs including gamma ray, resistivity and neutron-density

was available for each well. Additionally, dipole sonic and formation imaging logs

are available for several offset wells.

1.2 Research Objectives

This research addresses the differences in propellant and hydraulic fracturing

of the Mancos shale. Conclusions are drawn as to how the different fracturing mechanisms propagate differently. This leads to a better understanding of when the

techniques should be used in shales and what results should be expected. The

information necessary to draw these conclusions comes from addressing these five

research objectives:

1) Measure and further characterize the propellant fracture growth in the Mancos

shale block test by integrating mechanical properties from core plugs, unconfined

3 compressive strength data, measurements of the fracture growth and pressures

recorded during the experiment.

2) Split the block in the plane of the hydraulic fracture and characterize the hydraulic

fracture.

3) Analyze the propellant field tests using information from Wieland’s finite element

modeling, the stimulation records and well logs.

4) Analyze the hydraulic fracture field tests using stimulation records, tracer log data

and well logs.

5) Compare the results and conclusions of the hydraulic and propellant tests in both

laboratory and field settings.

1.3 Research Contribution

Organic rich shales, such as the Mancos, hold a substantial volume of hydrocarbons. However, research into shale stimulation technologies has not been as thoroughly investigated as stimulation technologies in conventional reservoirs. With further understanding of possible stimulation techniques, additional shale prospects may become economical resources.

This research provides one of the first published analyses of propellant fracturing in shale systems and allows for a comparison of propellant and hydraulic fracturing stimulations in shale systems. This information can lead to a better understanding of where the different methods should be applied. The research also further analyzes the only published big block propellant fracturing test, thus increasing understanding of propellant fracture propagation in shales.

4 CHAPTER 2

LITERATURE REVIEW

This thesis is a study of several different Mancos shale stimulations within an

area of the Douglas Creek Arch, previously conducted laboratory Mancos shale

stimulation experiments and additional tests to allow correlations to be drawn

between the two. Thus this literature review covers several topics including the

Mancos shale, propellant fracturing, stimulation modeling, fracture propagation in

laminated shales, acoustic logging, synthetic logs, unconfined compressive strength

and the laboratory experiments conducted by Wieland in 2005.

2.1 Mancos Shale

The Douglas Creek Arch is an anticline located between the Piceance and

Uinta basins. The formation of interest in this research is the Cretaceous Mancos shale, specifically acreage lying within a development contract area outlined on

Figure 2.1. The Mancos shale lies between the Dakota sandstone and the Mesaverde group (also sandstone) (Figure 2.2). The shale is several thousand feet thick with some interbedded sands. The darkest and most organic rich shale lies at the base of the Mancos; often referred to as the Mancos B. In the Douglas Creek Arch area, the

Mancos B is approximately 400 ft thick. Unlike the more continuous Mancos, the

Mancos B is very thinly laminated with fine grained argillaceous quartz sandstone containing 10-20% carbonates (McLennan et al., 1983). Figures 2.1 and 2.2 illustrate the location of the area of interest and the stratigraphy of the Douglas Creek Arch.

5 2N 100W 2N 104W 2N 103W 2N 102W 2N 101W 25E

1N 100W 1N 104W 1N 103W 1N 102W 1N 101W 25E Douglas Creek Arch Development Unit 1S 100W 1S 104W 1S 103W 1S 102W 1S 101W 25E

2S 101W 2S 100W 2S 104W 2S 103W 2S 102W 25E

3S 101W 3S 100W 3S 104W 3S 103W 3S 102W 25E

4S 104W 4S 103W 4S 102W 4S 101W 4S 100W 25E

5S 104W 5S 103W 5S 102W 5S 101W 5S 100W 0108,999 25E FEET

6S 105W6S 104W 6S 103W 6S 102W 6S 101W 6S 10 25E

15.5S 26E 7S 105W7S 104W 7S 103W 7S 102W 7S 101W 7S 1

16S 26E

8S 105W8S 104W 8S 103W 8S 102W 8S 101W 8S

17S 26E 9S 102W 0 9S52,284 101W 9S 104W 9S 103W 9S FEET 2N 3W 2N 2W 18S 26E

Figure 2.1: Douglas Creek Arch area of interest and its relationship to the Piceance basin and the states of Colorado and Utah.

6 UtUt a a h h Col Col o o r r a a do do

WASATCH Fm. SS WASATCH Fm. SS

ROLLINS SS COZZETTE SS ROLLINS SS CORCORAN SSCOZZETTE SS SEGO SS CORCORAN SS SEGO SS faces LLOYD SS e Shore ces LLOYD SS esaverd Shorefa M saverde Me MANCOS A, B, & SILT MANCOS A, B, & SILT MANCOS SHALE MANCOS(Source) SHALE (Source)

DAKOTA Fm. SS DAKOTA Fm. SS

MORRISON Fm. SS MORRISON Fm. SS

Figure 2.2: Douglas Creek Arch stratigraphy (Modified from Kirschbaum, 2003).

7 When oil prices were relatively high in the early 1980’s, the Mancos was targeted as a potential play (specifically within the Dragon Trail Field of the Piceance basin). Wells were hydraulically stimulated in the Mancos shale intervals in order to produce natural gas. Certain wells had larger gas production than others and it was found that local geological factors such as the presence of faulting and in-situ stresses could account for production differences. Drilling damage and stimulation design were also found to often affect production. At the time, conventional fracture stimulations using cross linked gel were being pumped and marginal improvements were seen from pumping larger volumes of proppant (McLennan et., 1983).

The Mancos shale had not previously been targeted within the particular area

of interest, however the shallower Mesaverde formation (sourced by the Mancos

shale) and the deeper Dakota sandstones both produce natural gas from more

conventional petroleum systems (Figure 2.2). Presently, natural gas prices are again

relatively high and the natural gas production from wells within the area of interest is

declining due to depletion. With the huge success of shale plays such as the Barnett

shale in eastern Texas and the Bakken shale in North Dakota, operators are paying

more attention to potential shale plays such as the Mancos.

2.2 Propellant Stimulations

Although relatively new propellant stimulations were a logical progression in

the development of stimulation technologies. Originally wells were perforated using

actual bullets and the first stimulations involved the use of explosives such as dynamite. Later research by Schmidt et al. (1982) found that wellbore explosions

8 could cause plastic deformation in the rock surrounding the wellbore, creating a

“stress cage”. Stress cages retain relatively little permeability as the fractures

generated in the process fully anneal. Schmidt et al. found by using propellant, the

rate of energy transfer to the wellbore could be reduced, avoiding the formation of

stress cages. Rather than explode, propellants deflagrate (rapidly burn). Propellant

stimulations are only used as a stimulation method in a few select applications

because unlike a convention hydraulic stimulation, proppant can not be used to hold

open fractures. Any retained permeability comes from the fractures not fully

annealing. More frequently, propellant stimulations are used in pre-hydraulic fracture

treatments or near wellbore cleanup.

2.3 Fracture Modeling

Understanding how hydraulic fractures are generated and modeling this

process is an important part of understanding how fluids will flow in a reservoir.

Griffith first published a solution for the width of an elliptical crack in 1920; and, in

1945, Sneddon developed what is known as the “penny-frac” model. In 1955,

Khristianovich and Zheltov introduced a 2D model where width was assumed

proportional to height; and in 1961, Perkins and Kern introduced a 2D model where

width was assumed proportional to length. Adapted versions of these two constant

height 2D models are significant parts of today’s simulators. More recently, pseudo

3D and 3D models have been developed; pseudo 3D models typically use the 2D

plane strain solutions but allow for variable height. True 3D models are the very latest

in developing technology. To be truly 3D, a model needs to be able to handle the

9 heterogeneous nature of the reservoir, i.e. bedding planes, planes of weakness or

incipient failure, and pre-existing natural fractures and fissures (Barree, 2006).

Good fracture modeling requires the integration of information from all different scales to better characterize the reservoir. Wellbore logs are used for the majority of the characterization, however additional information such as core measurements and previous stimulation records are used to calibrate models. For example, density and sonic logs can be used to calculate relative mechanical properties of rocks, i.e. dynamic properties. Actual mechanical tests on cores can be used to relate the dynamic and static properties. Data from previous stimulations or tests is also integrated into fracture models. For example, valuable information such as fracture closure pressure can give a good estimation of the minimum field stress

(Chardac et al., 2005).

2.4 Fracture Propagation in Laminated Shales

It is widely accepted that reservoir stress, rock elastic properties, reservoir

pressure variations, leakoff, natural fractures and fluid proppant schedule all play a

significant role in governing fracture geometry. In 1982, Warpinski et al. published several papers covering mineback and laboratory experiments that he conducted.

Warpinski proved that small stress contrasts, i.e. several hundred psi, across adjacent rock strata were sufficient to restrict height growth. Warpinski’s laboratory experiments concluded that in-situ stress and pore pressure differentials were more important in controlling fracture geometry than the mechanical properties of the rocks being tested. However other additional factors are believed to also significantly

10 impact fracture geometry in layered formations (Warpinski et al., 1982 and

Miskimins and Barree, 2003).

Daneshy (1978) was the first to really consider fracture propagation in

laminated formations. As with Warpinski, Daneshy found that the mechanical

properties of formations did not seem to be the most dominant factor affecting height

growth. Daneshy investigated the strength and nature of the interfaces between rock

layers. He observed that “weak interfaces are likely to stop fracture propagation,

regardless of the relative properties of the formations.” Strongly bonded interfaces

eventually will allow fracture extension through them (Daneshy, 1978).

Recently, Daneshy’s observation that the bonding between formations can

significantly affect fracture growth has been supported by numerous field studies.

Cases of excellent height containment in formations with no stress contrast have been

documented. Using tools such as pump-in tests, microseismic mapping, tiltmeter

mapping and radioactive tracing, cases of height containment unexplainable by in-situ stresses and rock properties have been documented. Instead, the height containment is believed to be caused by layer interface affects. Papers by Wright, et al., (1999) and

Miskimins and Barree (2003) all agree that most simulations do not adequately account for layer or interface effects.

Plane-strain and surface integral solutions for fracture width are the basis for most fracture simulators. These solutions make inherent assumptions that all rocks are fully elastically-coupled, allowing all stresses to interact. Laboratory experiments and microseismic mapping indicate that shear failures and shear slippage does occur during the fracturing process. Thus, the assumption of coupled formations is often

11 inadequate when considering thinly laminated beds. Barree and Winterfield (1998)

published work where they modeled the affect of shear dampening and slippage at

bed boundaries. They concluded that by modifying a surface integral it is possible to

improve fracture simulations in these laminated formations (Barree and Winterfield,

1998).

Minimal work has been published concerning the propagation of propellant

fractures in layered media such as shales, other than the experiments conducted in

Wieland’s thesis work (Section 2.8). This thesis builds upon his experimentation and

findings.

2.5 Acoustic Logging

The measurements of travel times of compressional and shear waves through

formations as recorded by an acoustic logging tool can be used to determine elastic

rock moduli such as compressibility (c) (reciprocal bulk modulus (K)), shear modulus

(G), Young’s modulus (E) and Poisson’s ratio (ν). Although equations for

determining these moduli referred to as dynamic moduli have been developed, it is

important to understand these moduli are determined from the velocity of sound

waves. Sound wave velocity is a function of porosity, effective stress, mineralogy, fluid saturation, vugs, fractures, permeability, temperature, anisotropy and wettability

which can all effect the results when compared to actual rock static measurements.

Thus the relationship between static and dynamic properties for reservoir

characterization is generally defined by a best fit correlation (Jorden and Campbell,

1986).

12 Young’s modulus is a measure of the relationship between stress and strain

(Equation 2.1). It can be considered the opposition to an extensional stress or the material’s “stiffness”. Dynamic Young’s modulus is defined by a relationship that includes density, compressional wave travel time and shear wave travel time as show in Equation 2.2.

/ AF =Ε (2.1) ∇ / LL

where, E = Young’s modulus F = force acting on area, A A = area ΔL = change in length L = original length

2 2 ρ s − TT P )43( =Ε 2 * 22 Ts − TT Ps )(

(2.2)

where, E = Young’s modulus ρ = density

Ts = shear transit time Tp = compressional transit time

Poisson’s ratio is defined as the ratio of lateral-to-axial strain under conditions of axial loading (Equation 2.3). As with dynamic Young’s modulus, dynamic

Poisson’s ratio is a function of acoustical wave travel times as shown in Equation 2.4.

∇ / dd ν = (2.3) ∇ / LL

where, ν = Poisson’s ratio Δd = change in diameter d = diameter ΔL = change in length L = original length

13

2 2 − TT Ps )2( ν = *5.0 22 −TT Ps )(

(2.4)

where, ν = Poisson’s ratio Ts = shear transit time Tp = compressional transit time

2.6 Synthetic Logs

Synthetic logs are a tool becoming more available with the advancement of technology and modeling. These artificial logs allow a wellbore to be analyzed, even though a set of logs is missing or incomplete. To create synthetic logs, data available in certain offset wells is extrapolated to wells where that data is missing by using information that is consistent in both wells. For example, basic logs that are run in every well such as a gamma ray log can be used to correlate data derived from relatively expensive acoustical logs which are only often run in a few test wells. This technique allows data that only becomes available with the drilling of a new well to be extrapolated throughout a more mature field. Advances are being made in the way these correlations are developed. They include the assignment of electrofacies, log responses specific to geological facies and the use of neural networks (Miskimins, et al., 2002 and Rolon, et, al., 2005).

2.7 Unconfined Compressive Strength

Sample Mancos rock strength data was available from the propellant block test conducted in Wieland’s work (2005). Rock strength is defined as the amount of

14 axial load a sample can withstand before failing. Unconfined compressive strength

(U.C.S.) can be measured in a lab. This is the strength of the material when it is not subjected to any confining pressure. Relating a samples U.C.S. with well logs is currently only interpretable by analogy since logging tools measure rock properties at field conditions, i.e. under confining pressure.

Chardac et al. (2005) describe U.C.S. as the most fundamental and useful description of rock strength, since it is its peak load-bearing capacity. However,

U.C.S. is an index rock property rather than an intrinsic rock property as it is a function of sample geometry. This means that the U.C.S. data collected by Wieland shows the relative strengths between layers of the Mancos shale, but that these values are not absolute values of strength (Chardac, et al., 2005).

2.8 Wieland’s Thesis

Wieland (2005) conducted two large scale laboratory propellant tests in order to investigate how propellant fractures propagate. Wieland deflagrated propellant charges in a Colton sandstone block and a Mancos shale block and recorded various pressure data so that the stimulations could be modeled.

The experiments were conducted at TerraTek in Salt Lake City, Utah. The poly-axial test frame used can hold a block measuring 30 in x 30 in x 36 in and apply horizontal pressures up to 10,000 psi and overburden pressures of 12,000 psi. Thus, the blocks to be tested were cut to 30 in x 30 in x 36 in.

Pressure probes were embedded within the samples as shown in Figure 2.3.

These probes collected data at 1,000 samples per second. The flapjacks applying the

15 confining stress collected pressure data at 1,000 samples per second, and two different wellbore pressure probes collected pressure data at 1,000 and 41,000 samples per second. Using representative field data, Wieland determined that the overburden pressure should be 4,500 psi and the horizontal confining stresses should be 1,600 psi and 1,300 psi.

During the pressurization of the Mancos shale block wellbore, using a 3%

KCL solution with red dye, the block was inadvertently hydraulically fractured prior to the planned test. Logically the fracture grew perpendicular to the minimum horizontal stress. In order to continue the propellant fracture experiment, the minimum and maximum horizontal stresses were swapped and the pressures were increased to 1,600 psi and 2,850 psi, respectively. The inadvertent hydraulic fracture meant that only two pressure probes, 3 and 6, were in the plane of the propellant fracture. Figure 2.3 shows the loading of the Mancos block and locations of the wellbore probes. Figure 2.4 is the pressure data recorded during the inadvertent hydraulic fracture. In Figure 2.4 one can see the relatively constant stresses being applied to the block (T-B stress, N-S stress, E-W stress), the pressurization and release of pressure in the wellbore (30KSI-P) and the pressure increase in each of the intersected probes after the release of pressure form the wellbore. This will be discussed further in Chapter 4. Figure 2.5 is the pressure data recorded during the propellant test, and one can see the cyclical rapid development and release of pressure at the wellbore and each intersected probe over a time period of 0.3 seconds.

16

Figure 2.3. Diagram of the original applied stresses relative to the pressure probes (left) and the redesigned applied stresses relative to the pressure probes (right). Modified from Wieland, 2005.

5000

T-B Stress N-S Stress 4000 E- W Str es s 30KSI-P Probe 1 3000 Probe 2 Probe 4 E- W N-S Wellbore Probe 5 2000 Fljk Fljk Pressure (psi) Pressure

1000

0 450 550 650 750 850 950 Time (sec)

Figure 2.4: Data showing the unintentional hydraulic fracture of the Mancos sample. Modified from Wieland, 2005.

17

Figure 2.5: Initial pressure response propellant deflagration Mancos sample. Modified from Wieland, 2005.

Several other characterization experiments were also conducted at TerraTek.

The mechanical properties of several core plugs from the test block were measured using compression test equipment, the results of which can be seen in Table 2.1.

Additionally an unconfined compressive strength scratch test was performed on the top 2.5 ft of the wellbore core. Figure 2.6 shows these results along with a photograph of the core along the x-axis.

Table 2.1: Geomechanical Test Results

Effective Bulk Confining Pore Young’s Sample Compressive Poisson’s Formation Density Pressure Pressure Modulus ID Strength Ratio (gm/cm3) (psi) (psi) (psi) (psi) Man-Top 2.552 0 0 8685 1,425,000 0.22 Mancos Shale Man-Mid 2.561 0 0 10,020 1,508,000 0.19 Man-Btm 2.565 0 0 9245 1,447,000 0.20

18 45000

40000

35000

30000

25000

UCS (psi) UCS 20000

15000

10000

5000

0 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 Length (ft)

Figure 2.6: Unconfined Compressive Strength scratch test with photograph of core.

Wieland (2005) generated 2D Abaqus finite element models of the Mancos shale propellant deflagration in order to investigate fracture length. The model had eight planes for potential fracturing that could become de-bonded when a certain pressure was overcome. The model used the following average mechanical properties taken from the block: effective compressive strength of 9137 psi, Young’s modulus of 1.46*10-6 psi and Poisson’s ratio of 0.2. Wieland modeled the volumetric influx of gas as the propellant deflagrated. An example of the results of his modeling can be seen in Figure 2.7. The modeled wellbore pressure can be seen relative to the length of the two independently modeled fracture wings (Wieland, 2005).

19 5000 Pressure (psi) 180° Plane (0.01 ft) 4500 0° Plane (0.01 ft)

4000

3500

3000

2500

2000

1500 Mass flow to fracture

Maximum Pressure (psi), FractureLength (0.01 Ft) 1000 while fracture pressurizes

500 Fracture Generation 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Time (sec)

Figure 2.7: Graph showing the maximum pressure response and fracture growth plotted against model time. Modified from Wieland, 2005.

20 CHAPTER 3

SONIC CHARACTERIZATION OF MANCOS SHALE BLOCK AND WELL LOG CORRELATIONS

As described in the literature review, Chapter 2 Section 2.8, experiments to determine many of the mechanical properties of the Mancos shale block were previously conducted at TerraTek in Salt Lake City, Utah. In order for this laboratory data to be used in field characterization, correlations needed to be developed. The static mechanical properties attained at TerraTek using compressional test equipment had to be compared with properties derived from acoustic, electrical and radioactive well logs of the Mancos formation.

Several different processes were tested to develop a correlation between the

Mancos samples and well logs. Initially, attempts were unsuccessfully made to characterize the entire block using a velocity probe. Only by using equipment available at the Colorado School of Mines to measure the acoustical properties of smaller core plugs were satisfactory results attained. A correlation based on density was developed relating the Young’s moduli of the tested core samples to the Young’s moduli as calculated from dipole sonic log data of the Lower Horse Draw 2186 well.

3.1 Hand Held Velocity Probe Characterization

By measuring the time it takes for compressional and shear waves to pass through an object, mechanical properties such as Young’s modulus and Poisson’s ratio can be determined. To measure the velocity of these waves on outcrops, Batzle and Smith (1992) developed a tool called the hand held velocity probe. The probe

21 works by measuring the time taken for an emitted wave to travel between two transducers pressed against the outcrop. This process allows for the collection of a lot of sample data in a relatively short period of time.

Measurements were taken using this probe every inch along one side of the

Mancos shale block. Figure 3.1 is a diagram and photo of the experiment setup.

Compressional wave travel times were consistently recorded, however little success was achieved passing a shear wave through the block. Both the propellant fracture face and a side of the block were tested to no avail. Micro-fractures prevented the shear waves from reaching the receiving transducers. There are several plausible reasons for why these fractures exist, they may be naturally occurring, caused by damage during cutting the block or caused by the evaporation of fluids from the block over time.

(Batzle and Smith, 1992)

Figure 3.1: Diagram and photo of velocity probe setup. Measurements of wave travel times were taken on the propellant fracture face as well as one side of the block. The oscilloscope photo shows a compressional wave.

22 3.2 Ultrasonic Core Measurements

Since acoustic wave travel times could not be measured on the surface of the block, acoustic measurements were taken on core plugs that had be obtained by

Wieland during the initial block testing. The approach for taking acoustic measurements of core plugs is similar to the velocity probe. In this case, the core is placed between two transducers, an ultrasonic pulse generator is used to send compressional and shear waves through the core and the travel times are measured with an oscilloscope. Using equipment in the Center for Rock Abuse directed by Dr

M. Batzle of the Geophysics Department and the Colorado School of Mines these acoustical measurements were taken under confining pressures. Re-applying confining stress allows for the shear wave to travel though the sample. Although the acoustic properties are derived the same way as with the hand held velocity probe, the experiment setup becomes far more complicated when applying confining pressures.

The experiment requires calibrating all of the equipment, preparing the cores, applying a hydraulic confining pressure and making the measurements. The process for this experiment is detailed in the next section and the result in Section 3.2.2.

3.2.1 Experiment Methodology

The first step was to calibrate a set of transducers by measuring their acoustic impedance; this allows the acoustic impedance of the core sample to be found by subtracting the acoustic impedance of the transducers from the total acoustic impedance of the system. By measuring the travel times of P-waves and S-waves

23 through several aluminum blocks of varying lengths, the travel times for the P-waves and S-waves to travel only through only the transducers was determined.

To prepare the cores for measurement, both ends of the cores needed to be perfectly smooth and level. This was achieved using a very fine grinding wheel. The exact lengths and masses of the cores were recorded. Since the TerraTek core samples were 1 inch in diameter and the transducers were 1½ inches in diameter, the core samples were encased in rubber tubing allowing the transducers and core to be sealed as one piece, as can be seen in Figure 3.2.

Figure 3.2: Diagram and photo of core setup.

As the same set of transducers was used for each sample, one core had to be prepared and tested before preparing the next core. A core was placed between the transducers and a small amount of viscous coupling (molasses) was used to ensure a good acoustical connection between the transducers and the core sample. One transducer was mounted parallel to the bedding planes of the core sample, while the other was mounted perpendicular. This allowed for shear waves to travel perpendicular and parallel to the bedding planes.

24 Each of the transducers had an o-ring around it; these o-rings were coated in silicone vacuum grease before the entire sample was encased in heat shrink tubing.

This prevented the core from coming in contact with the hydraulic oil being used to apply the pressure. An additional wire clamp was also placed on top of the heat shrink tubing as an further sealing method.

The sample was then placed in the pressure vessel and connected to the pulse generator and oscilloscope. Hydraulic oil was pumped into the pressure vessel to apply confining pressure to the sample. For the purpose of this experiment, the core plugs were left unsaturated as was the case in the TerraTek experiments to allow for direct comparisons. The acoustical measurements were taken at increasing pressures and the process was repeated for the additional cores.

3.2.2 Results

Table 3.1 shows the travel times of the different waves through the core samples. Measurements were taken at 1,000 psi increments and the travel times were measured in microseconds.

Table 3.1: Transit Time Measurements

Pressure Sample 1 Sample 2 Sample 3 (psig) P (μs) S1 (μs) S2 (μs) P (μs) S1 (μs) S2 (μs) P (μs) S1 (μs) S2 (μs) 0 5.082 8.815 9.415 5.582 9.715 9.815 4.582 7.915 9.215 1000 4.782 8.015 8.615 4.682 8.315 8.615 4.482 7.715 9.115 2000 4.682 7.815 8.315 4.582 8.115 8.415 4.282 7.515 9.015 3000 4.582 7.715 8.015 4.582 8.015 8.215 4.182 7.415 8.815 4000 4.582 7.615 7.915 4.482 7.715 8.015 4.182 7.315 8.715 5000 4.482 7.515 7.815 4.382 7.715 7.915 4.082 7.315 8.615

25 In order to determine the Young’s moduli and Poisson’s ratios, travel times were converted to acoustic impedances by dividing the travel time by sample length.

Equations 3.1 and 3.2 are the same equations found in Chapter 2, but include units and conversion factors. These equations were used to find the Young’s moduli and

Poisson’s ratios.

2 2 10 ρ s − TT P )43( =Ε 2 **10*34.1 22 Ts −TT Ps )(

(3.1)

where, E = Young’s modulus (psi) ρ = density (g/cc)

Ts = shear transit time (μs/ft) Tp = compressional transit time (μs/ft)

2 2 s − TT P )2( ν = *5.0 22 − TT Ps )(

(3.2)

where, ν = Poisson’s ratio Ts = shear transit time (μs/ft) Tp = compressional transit time (μs/ft)

Since shear wave velocities were found parallel and perpendicular to the bedding planes, two different sets of mechanical properties were calculated. The fast shear wave, S1, travels parallel to the bedding planes, and the slow shear wave, S2, travels perpendicular to the bedding planes. When comparing derived acoustic values with logging tools, S2 values need to be used because acoustic logging tools predominantly measure velocities perpendicular to the bedding planes (due to

26 transducers being stacked vertically in the wellbore). Figure 3.3 shows the calculated

Young’s moduli for both the fast and slow shear waves relative to applied confining stress.

7.00E+06

6.50E+06

6.00E+06

5.50E+06

5.00E+06

Young's Modulus (psi) Modulus Young's 4.50E+06

4.00E+06 Sample 1 - S1 Sample 1 - S2 Sample 2 - S1 Sample 2 - S2 3.50E+06 Sample 3 - S1 Sample 3 - S2

3.00E+06 0 1000 2000 3000 4000 5000 6000 Confining Stress (psig)

Figure 3.3: Calculated dynamic Young’s Moduli versus applied confining stress.

Figure 3.3 demonstrates expected exponential trends for two of the three core samples (Jorden and Campbell, 1986). However the Young’s modulus values for

Sample 3 (S1 and S2) have higher than expected Young’s modulus values at 0 psig, caused by faster than expected travel times. This is most likely due to experimental error. As confining pressure were applied to the sample, the transducers were pushed against the core sample, ensuring good acoustic coupling. For this reason, the measurements at 0 psig were taken after all of the confining pressure had been bled

27 off. Faster than expected wave velocities were most likely caused by the confining pressure not being fully bled off before the measurements were taken.

Figure 3.4 is a plot of calculated Poisson’s ratio versus confining pressure.

Poisson’s ratio is the ratio of lateral to axial strain under conditions of axial loading.

As can be seen in the figure, this material property is not a function of pressure.

0.380

0.360

0.340

0.320

Sample 1 - S1 0.300 Sample 1 - S2 Sample 2 - S1 Sample 2 - S2 0.280 Sample 3 - S1

Poisson's Ratio Poisson's Sample 3 - S2

0.260

0.240

0.220

0.200 0 1000 2000 3000 4000 5000 6000 Confinning Stress (psig)

Figure 3.4: Calculated Poisson’s Ratio versus applied confining stress.

In Wieland’s thesis, he modeled the fracture generation process of the experiments conducted at TerraTek. He showed that the fractures propagate for a relatively short period of time and the fracture void is then pressurized by gases from the propellant deflagration (2005). The velocity at which fractures can propagate is limited by the compressional wave velocity as this is the maximum velocity at which stresses can be transmitted (Suarez-Rivera, 2007). The acoustic tests found that this velocity at a pressure of 4,500 psi was approximately 15,160 ft/s as compared with

28 the much slower compressional wave velocity caused by the production of gases during the propellant burn of 2,200 ft/s (Passamaneck, 2007).

3.3 Sonic Correlations

Often acoustic travel time data is not available in many wells. One solution is to develop synthetic logs which correlate information from offset wells, which is discussed further in Chapter 6. In the event synthetic logs can not be created, the simplest way to estimate the mechanical properties of a formation is to use mechanical property correlations based on density and lithology (Barree, 2006).

Using the bulk density of the core plugs and mechanical properties calculated from acoustical impedances, it is possible to create density–mechanical property correlations. Equation 3.3 is a correlation that was calculated relating the density of the Mancos shale core samples with the acoustically derived Young’s moduli (R2 =

0.77).

E 6 ρ ×−××= 1010.851097.36 6 (3.3)

where, E = Young’s modulus (psi) ρ = density (g/cc)

Figure 3.5 shows density versus acoustically determined Young’ modulus for the cores collected at TerraTek under different confining pressures. The data was used to develop Equation 3.3. The fact that this correlation is only based on data from three core plugs provides significant doubts as to the accuracy of the three core plugs provides significant doubts as to the accuracy of the correlation especially when considering the heterogeneous nature of the Mancos formation. While the correlation

29 7.0E+06

R2 = 0.7741

6.5E+06 4000 psi 3000 psi 5000 psi Linear (4000 psi ) 6.0E+06

5.5E+06 Young's Modulus (psi) 5.0E+06

4.5E+06

4.0E+06 2.42 2.43 2.44 2.45 2.46 2.47 2.48 2.49 2.5 Bulk density (g/cc)

Figure 3.5: X-Y cross plot of Young’s modulus versus density at different confining pressures.

does serve as a general trend, additional data points would improve its reliability. At the time the acoustical experiments where being conducted, attempts were made to cut additional cores from the Mancos block. Unfortunately the block had sufficiently dried causing it to crack. The combination of the cracks in the block and the interaction between the clays and the cutting fluids made taking additional cores at this time impossible. Both oil and water were used as lubricating fluids in attempts to cut additional cores.

Equation 3.3 can be used to estimate Young’s moduli from density data. In order to compare these laboratory findings with field data, suitable logs had to be acquired. The Lower Horse Draw 2186 well is a producing Dakota wellbore that has a full suite of logs available including dipole sonic (Figure 3.6). The well is located in

30 the northwest section of the area of interest (Figure 2.1), Township 2S, Range 103W,

Section 26.

Using the density log and Equation 3.3, estimated Young’s modulus values were created for the Mancos section (3200 ft - 5300 ft) of the Lower Horse Draw

2186 well. Figure 3.6 shows these estimated Young’s modulus values calculated from the density log as compared with the Lower Horse Draw 2186 Young’s modulus values calculated from the dipole sonic log.

As can be seen in Figure 3.6, the Young’s moduli values calculated using

Equation 3.3 and are significantly higher than the dipole sonic measurements, although some similar trends can be seen. These higher estimated Young’s modulus values are primarily caused by the fact that the acoustical experiments were not conducted with saturated cores, thus the densities of the core samples were significantly lower than the field measured values. The average density of a dry core sample was only 2.45 g/cc as compared with the average logged density of this

Mancos section of 2.58 g/cc.

Figure 3.7 is a cross plot of the two different Young’s modulus values: the values calculated using the Equation 3.3 correlation and the measured acoustic values. A linear trend can be seen between the two with a correlation coefficient of approximately 0.35. Considering the extremely heterogeneous nature of the Mancos shale and the small sample size used to generate the correlation, this low coefficient is not unexpected.

31 Young's Modulus (mpsi) 0 2000 4000 6000 8000 10000 12000 14000 16000 3200.00

3700.00

4200.00 Depth (ft)

4700.00

5200.00

Dipole Sonic Measurements Density Correlation

Figure 3.6: Young’s modulus from dipole sonic log from Lower Horse Draw 2186 compared with Young’s modulus calculated using Equation 3.3.

32 1.60E+07

y = 1.1661x + 6E+06 R2 = 0.3512 1.40E+07

1.20E+07

1.00E+07

8.00E+06

6.00E+06 Young'sMoudlus Density Correlation 4.00E+06

2.00E+06

0.00E+00 0.00E+00 1.00E+06 2.00E+06 3.00E+06 4.00E+06 5.00E+06 6.00E+06 7.00E+06 LHD 2186 Young's Log Measurment

Figure 3.7: X-Y scatter plot of Young’s moduli from density correlation and Lower Horse Draw 2186 dipole sonic log. The correlation coefficient is 0.35.

With known variations in lithology within the Mancos shale, a gamma ray log was used as a basic means to distinguish between consistent shales and sandy shales.

Figure 3.8 shows the values of both the dipole sonic calculations and the density calculations of Young’s modulus. The density correlation curve has been shifted over the dipole sonic curve by subtracting the difference between the averages of the two different curves (it was necessary to compensate for the differences between laboratory measurements and field measurements of acoustical impendence such as saturation and temperature). The graph shows that the correlation is clearly better in some sections of the Mancos than others. As can be seen from Figure 3.8, a gamma ray count of 125 API units was selected to split the Mancos shale. Figures 3.9 and

3.10 are scatter plots showing the correlation coefficients for shales with a count above and below 125 API units.

33 Young's Modulus (mpsi) 0 1000 2000 3000 4000 5000 6000 7000 8000 3200

3700

4200 Depth (ft)

4700

5200

0 50 100 150 200 250 300 350 400 450 500 Gamma Ray (API)

Shifted Density Correlation Dipole Sonic Measurements Gamma Ray Log

Figure 3.8: Young’s modulus from dipole sonic log compared with the shifted Young’s modulus values calculated using Equation 3.3 and a gamma ray curve. The best Mancos correlation can be seen between 4000 ft and 4400 ft. Values are averaged over 5 ft intervals.

34

1.60E+07

1.40E+07 y = 1.7735x + 3E+06 R2 = 0.3772 1.20E+07

1.00E+07

8.00E+06

6.00E+06 Young's Density Correlation

4.00E+06

2.00E+06

0.00E+00 0.00E+00 5.00E+05 1.00E+06 1.50E+06 2.00E+06 2.50E+06 3.00E+06 3.50E+06 4.00E+06 4.50E+06 5.00E+06 LHD 2186 Young's Modulus from DiPole Sonic Measurments

Figure 3.9: X-Y scatter plot of Young’s moduli derived from Equation 3.3 as compared with dipole sonic Young’s moduli for gamma ray counts greater than 125 API. The correlation coefficient is 0.37.

1.60E+07

y = 0.8274x + 7E+06 1.40E+07 R2 = 0.2468

1.20E+07

1.00E+07

8.00E+06

6.00E+06 Young's Density Correlation

4.00E+06

2.00E+06

0.00E+00 0.00E+00 1.00E+06 2.00E+06 3.00E+06 4.00E+06 5.00E+06 6.00E+06 7.00E+06 LHD 2186 Young's Modulus from DiPole Sonic Measurments

Figure 3.10: X-Y scatter plot of Young’s moduli derived from Equation 3.3 as compared with dipole sonic Young’s moduli for gamma ray counts of less than 125 API. The correlation coefficient is 0.25.

35 Low gamma ray counts suggest sandier shales. When the gamma ray count was less then 125 API, the compared Young’s moduli had a lesser correlation coefficient of 0.25. This suggests that the correlation is poorer for sandier sections of the Mancos. This is not surprising since the core plugs tested had very little quartz banding. Figure 3.9 shows the same correlation but for sections of the Mancos with gamma ray counts greater than 125 API. This produced a closer correlation with a coefficient 0.38.

By manipulating the gamma ray cutoffs, the sections of Mancos where the correlation works best was found. Using a gamma ray range of 125 to 140 API units resulted in a correlation coefficient of 0.43, as can be seen in Figure 3.11. This correlates predominantly to the Mancos section at depths between 4000 ft and 4400 ft.

1.60E+07

1.40E+07 y = 2.0867x + 2E+06 R2 = 0.4348

1.20E+07

1.00E+07

8.00E+06

6.00E+06 Young's Density Correlation

4.00E+06

2.00E+06

0.00E+00 0.00E+00 5.00E+05 1.00E+06 1.50E+06 2.00E+06 2.50E+06 3.00E+06 3.50E+06 4.00E+06 4.50E+06 5.00E+06 LHD 2186 Young's Modulus from DiPole Sonic Measurments

Figure 3.11: X-Y Scatter plot of Young’s moduli with a gamma ray counts between 125 and 140 API. These cutoffs make the best correlation with a correlation coefficient of 0.43.

36 It has been found that with the limited data available, certain lithological- mechanical property trends can be developed; however, much more data would be needed to adequately characterize the 2000 ft of Mancos shale in the Douglas Creek

Arch area. Fortunately, acoustical logs are available for several wells in the area of interest. Thus in order to interpret the mechanical properties of specific wells, synthetic mechanical property logs and not lithological trends will be used as explained in Chapter 6.

37 CHAPTER 4

MANCOS BLOCK FRACTURE INVESTIGATION

Wieland’s (2005) thesis focused on the implementation of the propellant fracturing experiments conducted at TerraTek and the modeling of the propellant fracturing process. Once the block had been split along the propellant fracture face,

Wieland made some general observations as to the geometry of the propellant fracture before commencing the Abaqus finite element simulation. Additional interpretation of the propellant fracture face was not completed, and the inadvertent hydraulic fracture was not analyzed. The Mancos block was not split along the plane of the hydraulic fracture, and the pressure data was not interpreted. This chapter addresses these issues.

4.1 Further Observations of the Propellant Fracture Face

Figure 4.1 is a photo of the propellant fracture face taken shortly after the block was split along the propellant fracture plane. The purple fluid staining was left during the propellant fracturing process and shows the propellant fracture geometry.

To further Wieland’s characterization of the propellant fracture face, a plot of the fracture growth was compared with the unconfined compressive strength (U.C.S.) test carried out on the core from the block. Figure 4.2 visually illustrates the correlation between unconfined compressive strength and fracture geometry. Two peak values of

U.C.S. can clearly be seen in Figure 4.2. These peak values correspond to two lighter color layers, labeled “1” and “2”, on Figures 4.1 and 4.2. Layer 1 is approximately 1 inch thick, and Layer 2 is only 0.5 inches at its thickest.

38 2

1

Figure 4.1: Photograph of the north half of the Mancos shale block showing the propellant fracture staining. Two layers labeled “1” and “2”, have significantly affected the fracture geometry.

North Half of Mancos Shale Block U.C.S. Scratch Test UCS (psi) 100 150 200 250 300 350 400 450 5 000 00 00 00 00 00 00 00 00 0 0 . 0 0

0

Casing . 2 5 0. 50 0. 75 1. 0 2 0 L e n 1 g .

3ft 2 t h 5

( f t) 1. 50 1 1. 75 2. 0 0 2. 2 5 2 . 50

Figure 4.2: Outline of the propellant fracture staining as seen in Figure 4.1, correlated with U.C.S. scratch test of the Mancos shale block’s core.

39 Only the top 2.5 ft of the Mancos block’s wellbore core was subjected to a

U.C.S. test. U.C.S values of 0 were actually caused by the breaks in the core which was tested in several sections. Wieland records observations of this one inch “white quartz band” (Layer 1) in his thesis, observing that it influenced the fracture geometry

(Wieland, 2005). Layer 1’s unconfined compressive strength is approximately 35,000 psi.

Layer 2 is a significantly thinner band and is located 6 inches above the first quartz band. This layer is thickest on the east side of the wellbore and thins out west of the wellbore. At its thickest, this layer is only 0.5 inches thick. Even though this layer is much thinner than Layer 1, Layer 2 has a measured unconfined compressive strength of just over 40,000 psi, 5,000 psi greater than that of Layer 1. On the east side of the block where the layer is thickest, the band clearly dominates fracture growth. However on the west side where the layer has thinned out to less then a 1/8 of an inch, the layer has a much lesser effect on fracture geometry.

Clearly the propellant fracture did not grow in a penny-like fashion (semi- circular shaped fracture wings) out from the wellbore. The different lithological layers constrained the fracture growth. As explained in the literature review, it is understood that stress contrasts, rock properties and shear slippage can all govern fracture geometry. Visual inspection of the block shows no evidence of shear slippage leaving only rock mechanics, in-situ stresses or a combination of both as the mechanism for controlling fracture geometry.

40 Figure 4.2 demonstrates that the fracture growth is restricted in Layers 1 and

2, which have the highest unconfined compressive strength values and are thus

“tougher”.

Closure stress is a function of rock properties, reservoir properties and tectonic stresses. Equation 4.1 can be used to calculate in-situ stress (Barree, 2005):

ν P = []D ( γαγ ) ++− ( ) EPDPD +++ σεγα (4.1) c −ν )1( offptvvobtv txoffptvh

where,

Pc = closure pressure (psi) ν = Poisson’s ratio

Dtv = true vertical depth (ft)

γ ob = overburden stress gradient (psi/ft)

γ p = pore pressure gradient (psi/ft)

αv = vertical Biot’s poroelastic constant

α h = horizontal Biot’s poroelastic constant

Poff = pore pressure offset (psi)

ε x = regional horizontal strain E = Young’s modulus (psi)

σ t = regional horizontal tectonic stress (psi)

Many of the terms in Equation 4.1 drop out when considering the Mancos shale test since the block was dry (i.e. no significant pore pressure) and no regional stresses or strains were present. Equation 4.1 can be simplified to equation 4.2:

ν P = []D γ (4.2) c −ν )1( obtv

where,

Pc = closure pressure (psi) ν = Poisson’s ratio

Dtv = true vertical depth (ft)

γ ob = overburden stress gradient (psi/ft)

41 The overburden stress is calculated by adding the 4,500 psi applied overburden pressure to the stress caused by the mass (and gravity) of the rock above.

Overburden stresses in the Mancos block range from 4,500.0 psi at the top of the block to 4,503.3 psi at the bottom (assuming the density of the Mancos block is 2.56 g/cc, the average density of the sample cores measured by TerraTek). When calculating closure pressure, the value of (static) Poisson’s ratio will have a significant effect. Values of Poisson’s ratio measured by TerraTek ranged from 0.19 -

0.22. Unfortunately, the exact Poisson’s ratio values of Layers 1 and 2 are undeterminable as representative samples could not be collected as the layers are too thin. The values measured at TerraTek were taken using core plugs collected perpendicular to the bedding planes. As a result, these values are actually averages of at least several different layers, and it would be reasonable to assume the range of

Poisson’s ratio values within layers of the block is actually larger than 0.19 - 0.22.

Assuming that the Poisson’s ratio of layers in the Mancos block only ranged between

0.19 - 0.22, stress differentials between layers could be as large as 214 psi. When a slightly greater range of 0.17 - 0.23 is assumed, however, the stress differential could be as high as 422 psi. Figure 4.3 illustrates the potential variability of in-situ stresses between layers by assuming random Poisson’s ratio values ranging between 0.19 and

0.22 for 0.5 inch thick layers (the range measured at TerraTek of core plugs that included multiple layers).

42 0

4

8

12

16 In-Situ Stress

20 Stress Differential Depth (inches) Depth

24

28

32

36 0 200 400 600 800 1000 1200 1400 Stress (psi)

Figure 4.3: Illustration of potential closure pressures and differentials between layers for Poisson’s ratio values between 0.19 - 0.22.

Figure 4.3 shows that even the tight range of Poisson’s ratio values between

0.19-0.22 causes significant stress differentials between different layers. These different stresses are created by the different mechanical properties of the layers.

Without knowing the Poisson’s ratio values of Layers 1 and 2 and the adjacent layers, it is not possible to calculate the stress differential between them, however a differential on the order of magnitude of several hundred psi is plausible. More than likely, the quartz banding has a significantly lower Poisson’s ratio than the adjacent shales, and thus has a lower in-situ stress. This is discussed further in Chapter 7. In order to more fully understand the quartz banding, a sample (from Layer 1) was analyzed using a scanning electron microscope.

43 4.2 S.E.M. Investigation of Quartz Band.

Scanning electron microscopes (S.E.M.) are capable of showing very detailed images at much greater magnifications than conventional light microscopes by using electrons rather than light waves. In order to better understand the quartz banding that controlled the propellant fracture geometry, a sample from the larger of the quartz bands was analyzed using an S.E.M. Figure 4.4 is a picture at relatively low magnification of the quartz band sample. The dark grains are the detrital quartz grains that make up the framework of the band. They are held together by a number of different cementing materials. No porosity can be seen in the sample.

Figure 4.4: Low magnification of the quartz band sample showing lack of porosity and a well sorted grain size framework. The darker grains are quartz cemented with a mixture of materials including, dolomite, calcite and illites. Scope magnification: 250X.

44 Figure 4.5 is a more magnified view of a single quartz grain surrounded by cement. Additionally, the chart shows the elements detected by the S.E.M.. In this case the elements present are silicon and gold.

Figure 4.5: Typical detrital quartz grain in quartz band; height ≈ 0.05mm, length ≈ 0.025mm, meaning this would be defined as a coarse siltstone. Chart shows silicon and gold as the only elements present. Scope magnification: 1200x

45 The S.E.M. detects quartz with a silicon and gold signature because an extremely fine gold layer was applied to the surface of the sample during the preparation process to act as a conductor. The S.E.M. cannot detect the first eight elements on the periodic table (including oxygen, the other element in quartz) because of the relative weakness of the X-ray emitted by the replacement of an electron within the elements.

The Wentworth scale is used to define sediments by the diameter of grain sizes. Sandstones are defined as having grains ranging range between 1/16 mm – 2 mm and siltstones have grains ranging between 1/16 mm- 1/256 mm. This would define the quartz band as a relatively coarse siltstone.

Figures 1 through 9 in the Appendix A are additional S.E.M. pictures of different grains including authegenic and detrital dolomite, calcite and illites. Figure 9 in Appendix A shows authegenic dolomite and quartz cementing material. This cementing material adds to the strength of the band as seen by the U.C.S. test.

Additionally, many of the quartz grains showed evidence of colloidal fracturing.

During sample preparation, these quartz grains broke rather than the stronger cementing material, verifying the toughness of the band as recorded by the U.C.S. test.

Clearly there are significant differences between the layers of quartz banding and the bounding shales. These differences, in particular rock strength, and not in-situ rock stress controlled the fracture geometry.

46 4.3 Mancos Block Hydraulic Fracture Investigation

The Mancos block had not previously been split along the hydraulic fracture plane. Before the block was split, a fracture in the wellbore 90º offset from the propellant fracture face was visible. The fracture covered the full length of the open wellbore region and extended at least 1 inch into the cased section. After all the necessary information was recorded from the propellant fracture face, the block was split along the plane of the hydraulic fracture. Figure 4.6 is a picture of the propellant fracture face and wellbore with the presumed hydraulic fracture labeled.

Stained Wellbore

Hydraulic Fracture

Figure 4.6: Photograph of the north half of the Mancos shale block showing the hydraulic fracture.

47 As with the initial splitting of the block along the propellant fracture plane, a hammer and chisel was used to split the block along the hydraulic fracture. Initially driving the chisel into the block perpendicular to the bedding planes was relatively difficult. There appeared to be no plane of weakness created by the hydraulic fracture.

Driving the chisel in caused the block to shear along the bedding planes, causing the shale to crumble. After chiseling approximately four inches into the block, the block split almost to its base. This was obviously caused by the chisel intersecting the hydraulic fracture. This splitting process confirms a fracture had previously broken the rock in this plane.

In contrast to the staining left after the propellant deflagration, no evidence of

KCL water staining was found once the block was separated to reveal the hydraulic fracture face. The deflagration of the propellant generated a substantial amount of heat and the staining material was carried into the fracture by the gases created. This process essentially scorched the stain onto the fracture face. In contrast, the dyed wellbore fluid was pumped into the hydraulic fracture at room temperature and the fluid did not seep far into the shale block due to its impermeability. Additionally, it was a full 18 months after the fracturing event that the block was split along the hydraulic fracture, and by that time any evidence of staining had likely faded.

4.4 Hydraulic Fracture Pressure Response

Figure 4.7 shows the pressure probe responses during the hydraulic fracture.

The graph shows the fairly constant horizontal stresses, the wellbore pressure response and the pressure probe responses when they were intersected by the fluid.

48 The graph does not show the constant overburden pressure of 4,500 psi. Due to the fact that this fracture propagated during the setup for the propellant deflagration, the high-speed wellbore probe did not record any data. All of the data that was saved was recorded at a sample rate of one sample per second.

2500

N-S Stress E-W Stress 30KSI-P Probe 1 2000 Probe 2 N-S Flapjack E-W Flapjack Isolation Valve Closed Probe 4 Probe 5 519 Sec

1500

Wellbore Probe Pressure (psi) 1000 Probe 5

447 Sec

500

Probe 4 Probe 2

Probe 1

0 350 400 450 500 550 600 650 700 750 800 850 Time (sec)

Figure 4.7: Data recorded during the inadvertent hydraulic fracture. The recorded wellbore pressure response is marked in orange. Significant event occur at time 447 and 519 seconds as recorded by the pressure gauge.

49 4.4.1 TerraTek Report Summary

The following is a list of events that occurred during the inadvertent hydraulic fracture. It is paraphrased from the final TerraTek report, a copy of which can be found in Appendix B.

1. Confining pressures were gradually applied to the block (4,482 psi

overburden, N-S 1,599 psi and E-W 1,301 psi).

2. The borehole pressure was increased to a pressure of 1,300 psi. During this

time a servo-controlled hydraulic pump injected unknown volumes of fluid to

reach the 1,300 psi. The report suggests a fracture initiation occurred at 447

seconds (450 psi) as the borehole pressure leveled off momentarily at this

point (while fluid was still being injected).

3. The servo continued to inject fluid until the 1,300 psi wellbore pressure was

reached.

4. 35 seconds after the wellbore pressure reached 1,300 psi, pressure probe 5 was

intersected by the hydraulic fracture

5. Probe 4, Probe 1 and Probe 2 started to increase in pressure, 9 sec, 27, sec and

148 seconds after Probe 5 was intersected.

4.4.2 Hydraulic Fracture Pressure Analysis

As can be seen in figure 4.7, the first probe to record a pressure response is labeled Probe 5. This is highly improbable when one considers the loading diagram

Figure 4.8. Pressure Probe 4 should have been intersected before Pressure Probe 5.

50 The most reasonable and likely explanation for this is that the transducer wires were switched during setup, although this cannot be verified.

Figure 4.8: Loading during hydraulic fracture stimulation.

Clearly borehole integrity was lost during the pressurization. At time 447 seconds, an event occurred that temporarily reduced the rate at which the wellbore pressure was increasing. This may have been a near wellbore breakdown caused by a flaw in the block. However, this was not the initiation of a hydraulic fracture as neither the breakdown pressure nor minimum horizontal stress had been surpassed. If a fracture had been initiated it would not have taken the KCl fluid 175 seconds to travel 3.5 inches and intersect the first pressure probe.

Breakdown pressure is the pressure that must be exceeded to initiate a fracture. Breakdown pressure is greater than the fracture closure pressure because of near wellbore stress concentrations. Breakdown pressure can be written in terms of effective stress as shown in Equation 4.3.

51 Pb = σ H −min −σ ''3 H −max − +Tp (4.3)

where,

Pb = breakdown pressure (psi)

σ 'H −min = minimum horizontal stress (psi)

σ 'H −max = maximum horizontal stress (psi) p = pore pressure (psi) T = tensile strength of formation (psi)

Using 1,300 psi and 1,600 psi for the minimum and maximum horizontal stresses, and assuming insignificant pore pressure and tensile strength, Wieland calculated a conservative breakdown pressure of 2,300 psi, which correlated with available field data (2005).

More than likely, the “hydraulic fracture” started propagating at time 518 seconds (pressure 1,267 psi), because at this point the rate of increase in wellbore pressure dropped again. Clearly Wieland’s calculated breakdown pressure of 2,300 psi was not surpassed. There are several possible explanations for why the breakdown pressure did not have to be surpassed. The wellbore could have had a naturally occurring flaw, the block could have been damaged during the coring process or the block could have been damaged during the loading process. Alternatively, assumptions as to the values of minimum and maximum horizontal stresses at the wellbore being the same uniform stresses applied to block may be inadequate. Figure

4.3 showed that the fracture closure pressure that needed to be exceeded to propagate a fracture after breakdown most likely varied from approximately 1,100 psi to 1,250 psi.

52 CHAPTER 5

PROPELLANT FRACTURE STIMULATION FIELD TESTS

The Douglas Creek Arch area, as outlined in Chapter 2 (Figure 2.1), has been a prolific natural gas producing area for many decades. As described in Chapter 2, the

Mancos shale is a thick organic rich shale sandwiched in between two conventional sandstone producing systems, the Dakota and Mesaverde group. The Mancos shale has very low matrix permeability, less than 0.0001 md, however it is known to be naturally fractured. In order to connect a wellbore to the natural fracture network, some sort of completion and stimulation is necessary.

Propellant stimulations were initially decided upon for testing the Mancos because of considerable advantages over conventional stimulations. For relatively little cost as compared to conventional hydraulic fracturing, it is possible to treat

2,000 ft of pay in one or two stages. The propellant simply has to be loaded in the wellbore after perforations have been shot and deflagrated. The rapid expansion of gas and the pressurization of the system break down the formation and fractures are driven out. The main drawback to propellant stimulation is that proppant cannot be used to prop open the fracture, thus any retained permeability is caused by the fracture not fully closing back on itself due to fracture face erosion, roughness, spalling or shear slippage.

Within the study area for this project, over 2,600 pre-existing wells were identified. As a result, there were plenty of options from which to select Mancos re- completion candidates. Initially, the list of wellbores was filtered by criteria dictated

53 by the operator. These included wells that did not already penetrate the Mancos shale, wells that did not have adequate cement over the Mancos interval and wells with significant current production. The other major factor considered in initially determining which wells should be selected was wellbore integrity. Many of the older wellbores were not suitable for re-completions due to natural deterioration over time, thus only wells completed within the previous five years were considered.

With the number of candidate wells significantly reduced, drilling records were analyzed for high gas shows while drilling. Three wells were selected including the Hells Hole 9139x, Hells Hole 9131 and Park Mountain 9025. Figures 5.1-5.3 are the gamma ray, neutron density, resistivity and where available digital total gas logs for each well. Figure 5.4 is a correlated cross section of the three wells.

5.1 Propellant Stimulation Design

The perforation intervals for each well were picked based on increased levels of gas recorded on the total gas drilling log. Figures 5.1 – 5.3 are the well logs for

Hells Hole 9139x, Hells Hole 9131 and Park Mountain 9025, respectively. Table 5.1 lists the perforations for each well. The Mancos was only targeted in the lower sections of the Hells Hole 9131 and Park Mountain 9025 due to cement constraints.

The Hells Hole 9139x had been cemented above the entire Mancos interval and therefore all of the significant gas shows during drilling were targeted. Because over

600 ft separated some of the perforations, the propellant stimulation was designed for two stages. Both the Park Mountain 9025 and Hells Hole 9131 have 5 ½” casing, while the Hells Hole 9139x has 4 ½” casing.

54

Figure 5.1 Hells Hole 9139x well logs. Gamma ray, resistivity, density and gas-while- drilling curves. The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007).

55

Figure 5.2 Hells Hole 9131 well logs. Gamma ray, resistivity, density and gas-while- drilling curves. The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007).

56

Figure 5.3: Park Mountain 9025 well logs. Gamma ray, resistivity and density (gas- while-drilling curve not available in electronic format). The purple boxes represent propellant stimulation perforations and the lines are picks by Fisher (2007).

57

llant stimulated wells, stratagraphic picks by Fisher wells, stratagraphic picks llant stimulated Figure 5.4: Correlated cross section of the three prope Figure 5.4: Correlated cross section (2007).

58 Table 5.1 Perforation Intervals

Wells Perforation Depths (ft) – 4 S.P.F. Hells Hole 9139x Upper Stage 5224-5267, 5434-5471, 5991-6019, 6035-6085, 6164-6243 Hells Hole 9139x Lower Stage 6830-6911, 6937-6960, 7038-7058 Hells Hole 9131 6718-6738, 6821-6849 Park Mountain 9025 6695-6710, 6715-6770, 6810-6878

5.2 Stimulation Operations

In order to re-complete in the Mancos, the three wells each needed to be isolated from the lower Dakota sandstone which was producing in each well. The wells were killed with 6% KCl fluid, the production tubing was pulled and cast iron bridge plugs were set above the Dakota and cemented in place. The Mancos was then perforated with deep-penetrating charges at 4 shots per foot. Originally the propellant stimulation design had called for propellant charges 4 inches in diameter, however under advisement from the service company the operator ran only 1.375 inch charges.

This significantly reduced the amount of propellant shot, thus reducing the amount of energy available for fracture generation. Once the propellant had been deflagrated, the stimulation crew rigged down and a workover rig was used to run in the hole with tubing and swab the well to initiate production.

5.3 Results

During the perforation process of the upper stage of the Hells Hole 9139x, pressure increases were recorded of 125 and 225 psi after perforating interval 6,184 ft

-6,204 ft and 6,184 ft – 6,204 ft, respectively. No other increases in wellbore pressures were recorded for the three other stages. Swabbing of the Park Mountain

59 9025 and Hells Hole 9131 did not enable these two wells to flow. However, the Hells

Hole 9139x did flow some gas at low rates, before it was shut in for a buildup test. A bottom hole pressure gauge was run, the results of which can be seen in Figure 5.5.

The bottomhole pressure slowly built to a 1,375 psi over the course of two weeks before the gauge was pulled.

Figure 5.5: Pressure build-up test results form the propellant stimulation of the Hells Hole 9139x.

5.4 Conclusions

From a production and economic point of view, the propellant stimulation of these three wellbores was a failure. The stimulations either failed to connect with a significant gas-bearing natural fracture network, or if a connection was made, the fractures fully annealed without leaving conductivity to the wellbore.

The original simulation work presented in Wieland’s thesis (2005) predicts a

37 ft half-length fracture using 4 inch charges in field scale simulations for these test wellbores. Adapting the models for the 1.375 inch diameter charges that were

60 eventually run only simulates a half length of 15 ft. This estimation of half-length is most likely optimistic since perforations and near wellbore damage is not accounted for in Wieland’s model. Ultimately, reducing the diameter of the propellant used had a significant effect on the amount of energy that was actually delivered to the wellbore. Pump-in tests discussed in chapter 6.2 suggest that the 1.375 inch charges failed to break down the perforations. Considering the gas production from the Hells

Hole 9139x and the undersized propellant stimulation, propellant stimulations cannot be ruled out as a potential stimulation technique. Any future test should again be conducted in a well where the entire Mancos section can be tested, (as with the Hells

Hole 9139x) and the stimulation should be designed to include larger diameter charges to maximize fracture half length.

.

61 CHAPTER 6

SLICKWATER STIMULATION FIELD TEST

After poor production results from the round of propellant stimulations tests discussed in Chapter 5, a new completion strategy was pursued. Initially, the propellant stimulations had been chosen since thousands of feet of pay could theoretically be stimulated. Analysis of the mechanical properties of the shale section as determined using dipole sonic logs, led to the identification of two zones within the

Mancos shale that would be suitable for hydraulic stimulation. These zones had lower

Poisson’s ratios, lower densities, and cleaner gamma ray signatures, all suggesting relatively higher quartz content while still being shale. In addition to being more conducive to stimulation, these zones also had relatively high gas shows whilst drilling. The operator determined that one of the three propellant stimulated wells should be re-stimulated in these new zones with the hydraulic stimulation. This prevented the loss of an additional producing Dakota wellbore. The Hells Hole 9139x was chosen since it was the only well to record a pressure increase during the propellant stimulation tests. Figure 6.1 shows the logs for the Hells Hole 9139x. The two zones to be stimulated were at 6,830 ft -7,058 ft and 5,994 ft – 6,243 ft.

With the target being shale, but wanting to pump a hydraulic stimulation, a slickwater stimulation was decided upon. Slickwater stimulations are low viscosity treatments that are often used in low permeability formations to re-activate natural fractures (Woodworth, 2006). A KCl based slickwater fluid was selected to minimize potential clay swelling. Previous experience had proven that low producing pressures

62 Stage 2 Perforations

Stage 1 Perforations

Figure 6.1: Hells Hole 9139x well logs, gamma ray, resistivity, density and gas while drilling logs. The perforations for the two slickwater fracture stages are marked with purple boxes on the depth scale.

63 were to be expected, thus an energized stimulation was selected in order to maximize fracturing fluid flow-back and clean-up. CO2 and N2 are typically used in energized fracturing. CO2 was chosen for this stimulation due to it lower cost and the fact it is pumped as a liquid reducing treatment wellhead pressures.

6.1 Modeling

Once it had been decided to pump an energized slickwater fracture stimulation, a series of stimulation models were created. Two different simulation software packages were used. Schlumberger’s FracCADE™ was used to simulate the two stages for the two different zones using basic inputs from the Hells Hole 9139x well logs. Additionally Dr. Barree of Barree and Associates ran more complex modeling using synthetic logs and the simulation program GOHFER™. The

GOHFER™ simulation was only conducted for stimulation Stage I (6,830 ft -7,058 ft).

6.1.1 FracCADE™ Simulation

FracCADE™ model inputs were developed using the gamma ray, neutron- density and resistivity logs that had been run in the Hells Hole 9139x and general shale correlations. Tables 6.1 and 6.2 show the average values assumed for the modeled parameters including fracture gradients, in-situ stresses, Young’s moduli,

Poisson’s ratios for the perforated and non-perforated sections.

64 Table 6.1: Key Input FracCADE™ Simulation Parameters for Stage I

Zone Top Shot Number of Frac In-situ Young’s Poisson's (name) MD Density Perforations Grad. Stress Modulus Ratio (ft) (shot/ft) (psi/ft) (psi) (psi) Barrier 6700 0 0 0.95 6427 7.60E+06 0.35 Shale 6830 4 324 0.83 5703 4.00E+06 0.27 Barrier 6911 0 0 0.90 6232 6.00E+06 0.30 Shale 6937 4 92 0.80 5559 4.00E+06 0.27 Barrier 6960 0 0 0.90 6299 6.00E+06 0.30 Shale 7038 4 80 0.78 5497 4.00E+06 0.27 Barrier 7058 0 0 0.95 6753 7.60E+06 0.35

Table 6.2: Key Input FracCADE™ Simulation Parameters for Stage II

Zone Top Shot Number of Frac In-situ Young’s Poisson's (name) MD Density Perforations Grad. Stress Modulus Ratio (ft) (shot/ft) (psi/ft) (psi) (psi) Barrier 5800 0 0 0.95 5602 7.60E+06 0.35 Shale 5994 4 100 0.78 4685 4.00E+06 0.27 Barrier 6019 0 0 0.90 5424 6.00E+06 0.3 Shale 6035 4 200 0.80 4848 4.00E+06 0.27 Barrier 6085 0 0 0.90 5512 6.00E+06 0.30 Shale 6164 4 316 0.83 5149 4.00E+06 0.27 Barrier 6243 0 0 0.95 5978 7.60E+06 0.35

The pump schedule was designed assuming a pump rate of 25 bbl/min, with proppant concentration increasing from 0 PPA to 2.5 PPA. Approximately 100,000 lbs and 115,000 lbs of proppant were simulated for Stages I and II, respectively.

Tables 6.3 and 6.4 are the pump schedules for the two stages.

65 Table 6.3: Simulated Pump Schedule for Stage I

Stage Pump Stage Gel Prop. Prop. Name Rate Fluid Conc. Type and Mesh Conc. (bbl/min) Volume (lb/mgal) (PPA) (gal) Pad 25.0 10000 10.0 0.0 0.5 PPA 25.0 7000 10.0 20/40 Sand 0.5 1.0 PPA 25.0 12000 10.0 20/40 Sand 1.0 1.5 PPA 25.0 20000 10.0 20/40 Sand 1.5 2.0 PPA 25.0 20000 10.0 20/40 Sand 2.0 2.5 PPA 25.0 7000 10.0 20/40 Sand 2.5 Flush 25.0 1637 20.0 0.0

Table 6.4: Simulated Pump Schedule for Stage II

Stage Pump Stage Gel Prop. Prop. Name Rate Fluid Conc. Type and Mesh Conc. (bbl/min) Volume (lb/mgal) (PPA) (gal) Pad 25.0 11000 10.0 0.0 0.5 PPA 25.0 7000 10.0 20/40 Sand 0.5 1.0 PPA 25.0 14000 10.0 20/40 Sand 1.0 1.5 PPA 25.0 22000 10.0 20/40 Sand 1.5 2.0 PPA 25.0 23000 10.0 20/40 Sand 2.0 2.5 PPA 25.0 7000 10.0 20/40 Sand 2.5 Flush 25.0 1437 20.0 0.0

Figures 6.2 and 6.3 are graphical depictions of the simulated stimulations for each stage. The simulations calculated three separate fracture lobes for each stage corresponding to the three sets of perforations present in each stage.

66 EnCana Oil and Gas FracCADE* Hells Hole #9139X ACL Fracture Profile and Proppant Concentration Stage 1 WF120 50Q CO2 05-31-2006 6700

6800 < 0.0 lb/ft2 0.0 - 0.2 lb/ft2

ft 0.2 - 0.4 lb/ft2 - 0.4 - 0.6 lb/ft2 6900 0.6 - 0.7 lb/ft2 Depth 0.7 - 0.9 lb/ft2

Well 0.9 - 1.1 lb/ft2 1.1 - 1.3 lb/ft2 1.3 - 1.5 lb/ft2 7000 > 1.5 lb/ft2

7100 5000 6000 7000-0.30 -0.15 0 0.15 0.30 0 500 1000 1500 Stress - psi ACL Width at Wellbore - in Fracture Half-Length - ft

*Mark of Schlumberger

Figure 6.2: Stage I fracture profile and proppant concentration.

EnCana Oil and Gas FracCADE* Hells Hole #9139X ACL Fracture Profile and Proppant Concentration Stage 2 WF120 50Q CO2 05-31-2006 5900

6000 < 0.0 lb/ft2 0.0 - 0.2 lb/ft2

ft 0.2 - 0.4 lb/ft2 - 0.4 - 0.6 lb/ft2 6100 0.6 - 0.8 lb/ft2 Depth 0.8 - 1.0 lb/ft2

Well 1.0 - 1.2 lb/ft2 1.2 - 1.4 lb/ft2 1.4 - 1.6 lb/ft2 6200 > 1.6 lb/ft2

6300 4500 5500 -0.3 -0.2 -0.1 -0 0.1 0.2 0.3 0 600 1200 1800 Stress - psi ACL Width at Wellbore - in Fracture Half-Length - ft

*Mark of Schlumberger

Figure 6.3: Stage II fracture profile and proppant concentration.

67 6.1.2 Synthetic Logs developed for GOHFER™ Simulation

In addition to the FracCADE™ simulation, Dr. Robert Barree of Barree and

Associates was brought in to assist with the design of the stimulation. Using

GOHFER™ software, Dr. Barree was able to take dipole sonic log information from three nearby wells, Hells Hole 9124, Hells Hole 9122 and Lower Horse Draw 2186, and use the density, gamma ray and resistivity logs available in all of the wells to create a synthetic dipole sonic log for the Hells Hole 9139x. As well as creating a synthetic dipole sonic log, Dr. Barree created theoretical logs of lithology, mechanical property curves (Biot’s constant, Poisson’s ratio and Young’s modulus) and formation stresses.

Figure 6.4 is a map showing the locations of the four wells, Hells Hole 9139x,

Hells Hole 9124, Hells Hole 9122 and Lower Horse Draw 2186. Figure 6.5 is an example of the available dipole sonic and triple combo log information available in the Lower Horse Draw 2186. Figure 6.6 shows several of Dr. Barree’s generated synthetic curves for the Hells Hole 9139x well. (Figure 6.1 includes the original Hells

Hole 9139x well logs.)

68 14 13 21 22 23 24 E 22 23 24 19 20 TAIGA MOUNTAIN

23 24 28 27 26 25 27 26 25 30 29

26 25 33 34 35 36 34 35Hells Hole36 31 32

35 36 4 3 2 1 3 2 1 6 5 Hells Hole

1 9 10 11 12 10 11 12 7 Hells8 Hole 9139x RABBIT MOUNTAIN

12 17 16 15 14 13 2S15 104W14 13 18 2S 103WLower Horse Draw 2186

13 21 22 23 24 22 23 24 19 20 LOWER HORSE DRAW

24 28 27 26 25 27 26 25 30 29

36 25 31 32 33 34 35 34 35 36 DRAG

36 4 3 2 1 3 2 1 6 5

1 9 10 11 12 10 11 12 7 8

12 17 16 15 14 13 3S15 104W14 13 18 3S 103W

13 21 22 23 24 E 22 23 24 19 20 MIS Zone: LOWER_MANCOS_ALLMISSOURI - 2ND_ROUND_CANDIDATE CREEK [RF] IS09 PRESENT ,629 24 28 27 26 25 27 26 25 30 29 MISS Log: TG IS PRESENT MISSOURI CREEK FEET

PETRA 03/01/2006 9:18:03 AM Figure 6.4: Locations of the wells with dipole sonic logs and the Hells Hole 9139x.

69 70

Figure 6.5: Lower Horse Draw 2186 gamma ray, resistivity, neutron porosity, photoelectric absorption, density, compressional wave travel time and shear wave travel time logs.

e, Poisson’s ratio, Young’s modulus, e stress and shale fraction.

logs, compressional wave travel tim Biot’s constant, total closur Figure 6.6: Hells Hole 9139x synthetic

71 6.1.3 GOHFER™ Simulation

Barree (2006) used the synthetic logs created for Hells Hole 9139x to generate a much more detailed and accurate set of fracture model inputs. The FracCADE™ models used a single value for each required input per simulated layer based on general shale correlations and the triple combo suite of logs. Instead, Dr. Barree used inputs that were calculated every half foot and averaged over ten foot simulation intervals based on the correlated dipole sonic log information.

As with the FracCADE™ simulation for Stage I, the stimulation was designed for approximately 100,000 lbs of total proppant. Dr. Barrree simulated bottomhole pressures during the fracturing process and reduced the pump rate to 20 bbl/min to keep within pipe pressure limitations (FracCADE™ 25 bbl/min). The simulated pump schedule also differed slightly calling for proppant concentrations ramping up to 3.0 lb/gal.

Figure 6.7 is the graphical depiction of the GOHFER™ simulation of Stage I.

The significantly more detailed simulation inputs and slightly different pump schedule led to a noticeably different model output. The fracture length was only predicted to reach 430ft and have a propped half-length of 150 ft-200 ft as opposed to over 1,000 ft and 900 ft respectively with the FracCADE™ simulation. Instead of the three fracture lobes predicted by the FracCADE™ model, only one lobe was predicted in the GOHFER™ modeling. Both simulations agreed the height of the fracture growth would be restricted relatively near to the top and bottom sets of perforations.

72

Figure 6.7: GOHFER™ Stage I fracture profile and proppant concentration.

6.2 Actual Stimulation

In June of 2006, the two-stage energized slickwater stimulation was pumped in the Hells Hole 9139x well (full stimulation report Appendix C). The designed volumes of fluids and proppants were pumped, however the slurry rates were predominantly lower than the scheduled 25 bbl/min. Higher treating pressures meant that the pump rates had to be reduced to keep the treating pressure below the burst pressure of the wellbore casing. The burst pressure rating of the 4½”, 11.6# casing was 7,780 psi, thus rates were reduced to keep the treating pressure at approximately

7,000 psi. Figure 6.8 and Figure 6.9 are treatment plots for the two stages, and each plot shows treating pressure and slurry rate.

73 10000 25 Treating Pressure Slurry Rate

9000

8000 20 6395 psi

7000

6000 15

5000

4000 10 Slurry Rate Slurry (bbl/min) Treating Pressure (psi) Pressure Treating

3000

2000 5

1000

0 0 11.522.533.5 Time (hours)

Figure 6.8: Stage I treatment plot for perforations 6,830 ft – 7,058ft.

10000 35 Treating Pressure Slurry Rate

9000 30

8000 6102 psi

25 7000

6000 20

5000

15 4000 Slurry Rate (bbl/min) Rate Slurry Treating Pressure (psi)

3000 10

2000

5 1000

0 0 00.511.522.5 Time (hours)

Figure 6.9: Stage II treatment plot for perforations 5,994 ft – 6,243 ft.

74 Figure 6.8 shows the lines being pressure-tested to over 8,000 psi, the stimulation beginning and the breakdown pressure at 6,935 psi. After breakdown, the treating pressure becomes highly erratic, and according to the service company there were difficulties bringing the CO2 online. Throughout the stimulation, the slurry rate had to be continually adjusted in order to maintain a maximum treatment pressure of around 7,000 psi (10 – 12 bbl/min). The treating pressures for Stage II (Figure 6.9) were much more stable, and again, directly after breakdown there was an issue with bringing the CO2 online, but once this problem was resolved, the job went smoothly with pump rates of 13 – 14 bbl/min.

Nolte-Smith plots can be used to analyze the treatment pressures during the fracture stimulation. Log of net treating pressure, i.e. bottom hole fluid pressure minus total fracture closure stress is plotted against the log of injection time. Nolte

1 and Smith found that slopes of /8 to ¼ indicate lateral fracture growth with restricted height growth, while steeper slopes typically indicate screenouts. Furthermore, zero slopes indicate stable height growth or fissure opening, and negative slopes indicate unstable height growth (Barree, 2006).

For this stimulation, bottomhole pressure gauges were not available.

Therefore, bottomhole fracture pressure was calculated from treating pressure and had to account for pipe friction, perforation friction and near wellbore tortuosity. Due to the irregular treatment pressure of Stage I and unknown CO2 rates, analysis of net fracture pressure was not feasible. However by maintaining a slurry rate of approximately 13.5 bbl/min for Stage II, treating pressure was kept much more consistent. Figure 6.10 is the Nolte-Smith plot for Stage II.

75 10000

m < 1/4

m = 0

1000 Net PressureNet (psi)

100 110100 11:23 am Time (mins)

Figure 6.10: Stage II Nolte-Smith plot.

In Figure 6.10 the net treatment pressure shows two distinct periods of fracture growth during Stage II. A slope of zero typically infers stable height growth, however a period of constant pressure could also infer the opening of pre-existing natural fractures. Both are plausible scenarios in this case (Gidley, et al., 1986).

Midway through the stimulation, at approximately 40 minutes, the net treating pressure increased, and the resulting slope of less than ¼ from 40 minutes to 80 minutes would infer lateral fracture growth.

The breakdown pressure of each stage is of particular interest as these zones had previously been stimulated with the undersized propellant treatment. The predicted fracture half lengths of 15 ft (Wieland, 2005) may have been somewhat optimistic. However, when breaking down the formation, one would expect to see a

76 lower than normal breakdown pressure as the propellant should have broken down the perforations and through any near wellbore damage.

A previously conducted pump-in test of the Mancos shale in the Hells Hole

9131 provides the best estimate for breakdown pressure. In this test, the formation broke down at 6,100 psi at a depth of 6,810 ft – 6,849 ft as compared with a Stage I breakdown pressure of 6,395 psi at a depth 6,830 ft -7,058 ft. This would suggest that the propellant test shot one year prior to the hydraulic stimulation had no significant lasting effect. (Stage II broke down at a pressure of 6,102 psi at a depth of 5,994-

6,243 ft)

6.3 Tracer Log

Radioactive tracers are tools that can be used to determine the placement of radioactively tagged proppant. A spectral gamma ray tool is used to measure the gamma rays emitted from radioactive elements. Different radioactive elements emit gamma rays with different energies, and the different energies (signatures) can be differentiated between by the logging tool (Hecker, Houston and Dumas, 1995). For this stimulation, three radioactive isotopes were used. Each stage’s pad was tagged with an antimony tracer, and the proppants of Stage I and Stage II were staged with scandium and iridium isotopes, respectively. A copy of the entire tracer log can be found in Appendix D. Figure 6.11 shows two sections of the tracer log corresponding with the two stages. Due to fill in the bottom of the well, only the top set of perfs from Stage I could be logged.

77

Stage II

Stage I

Figure 6.11: Sections of the tracer log showing the two slickwater stages pumped. The blue represents the antimony tracer used to tag the pad in each stage, yellow represents the scandium used to tag the proppant in Stage I and the red represents the iridium used to tag the proppant in Stage II.

The tracer log shows that significant height containment occurred during the fracturing process. The tracer log indicates that the hydraulic stimulation of Stage I did not grow above the top set of perforations at 6,830 ft and that the majority of the proppant was placed through the middle perforations at 6,850 ft to 6,810 ft. For Stage

II, the tracer log shows that the hydraulic fracture grew 14 ft above the top set of perforations. Both the pad and the proppant reached 5,980 ft where the fracture was abruptly contained. The pad of Stage II does appear to have grown down 47 ft below the lowest set of perforations although the proppant was contained to the lowest set of perforations.

78 The spectral gamma ray tool counts the gamma rays emitted by the different isotopes used to tag the stages. Factors that affect the number of gamma rays counted include the amount of radioactive isotope present, the proximity to the radioactive material and the age of the radioactive material. Since the spectral gamma ray tool predominantly records near wellbore radiation, one can not determine if the fracture grew across all of the sets of perforation in Stage I or the lower sets of perforations in

Stage II. The tracer log does show that even in the near wellbore region, the fracture grew across the upper two sets of perforation of Stage II to form one fracture.

6.4 Comparison with Modeling

Both the FracCADE™ and GOHFER™ modeling suggested significant height containment for each stage. However the tracer log revealed even better height containment than predicted. The FracCADE™ simulation predicted individual fracture lobes around each sets of perforations, an outcome clearly driven by a drastic contrast in properties between very few input layers. The tracer log shows that in at least the upper set of perforations, the fracture grew across both sets of perforations.

The GOHFER™ simulation, which had less drastic contrasts between inputs because smaller 10 ft simulation intervals were used (as opposed to intervals of over 100 ft), modeled the fracture growth across the entire perforated interval as one lobe. Whether the individual fractures of each stage amalgamated further from the wellbore is unknown, however it is clear that significant amounts of proppant were not placed between perforations, at least in the near wellbore region. With the actual stimulation having better containment than predicted by GOHFER™ simulation, either greater

79 width or length could be possible. However with the fractures growing between perforation intervals as seen with the upper sets of perforations in Stage II, significantly less length or width could be expected than modeled in the FracCADE™ simulations.

6.5 Traced Fracture Height as Compared with Logs

In order to determine what mechanism (in-situ stress, rock properties or shear slippage) controls the fracture height growth in the hydraulic fracturing of the Mancos shale, the tracer log results were compared with the synthetic logs created for the well. Figures 6.12 - 6.15 show the gamma ray log, synthetic Poisson’s ratio, synthetic

Young’s modulus and synthetic closure stress curves with the traced fracture heights marked on them. Figure 6.16 and Figure 6.17 are a close look at the fracture tops as compared with the gamma ray log.

Figure 6.12, 6.13, 6.14 and 6.15 show the fractures all cease at significant increases in gamma ray, Poisson’s ratio and closure stresses and cease at significant decreases in Young’s moduli. This shows the fractures all contained within significant changes in lithology. This could be caused by in-situ stress, rock properties or shear slippage.

Taking a more detailed look at the gamma ray log (values every 0.5ft) Figures

6.16 and 6.17 show thin layers of lower gamma ray counts right at the interpreted top of the fracture. It is possible that these thin, tough layers of quartz siltstone are containing the fracture and not the in-situ stresses of the shales above.

80

Gamma Ray 50 70 90 110 130 150 170 190 5800

5900

6000 Stage 2 - Traced Fracture Stage 2 - Perforations

6100

6200 Stage 2 - Traced Fracture

6300

6400

Depth (ft) 6500

6600

6700

6800

Stage 1 - Traced Fracture Stage 1 - Perforations 6900

Upper Limit of one Stage 1 - Traced Fracture 7000

7100

Figure 6.12: Hells Hole 9139x gamma ray curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals).

81 Poisson's Ratio 0.15 0.2 0.25 0.3 0.35 0.4 0.45 5800

5900

6000 Stage 2 - Traced Fracture Stage 2 - Perforations

6100

6200 Stage 2 - Traced Fracture

6300

6400

Depth (ft) 6500

6600

6700

6800

Stage 1 - Traced Fracture Stage 1 - Perforations 6900

Upper Limit of one Stage 1 - Traced Fracture 7000

7100

Figure 6.13: Hells Hole 9139x synthetic Poisson’s ratio curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals).

82 Young's Modulus (mmpsi) 012345678910 5800

5900

6000 Stage 2 - Traced Fracture Stage 2 - Perforations

6100

6200 Stage 2 - Traced Fracture

6300

6400

Depth (ft) 6500

6600

6700

6800

Stage 1 - Traced Fracture Stage 1 - Perforations 6900

Upper Limit of one Stage 1 - Traced 7000 Fracture

7100

Figure 6.14: Hells Hole 9139x synthetic Young’s modulus curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals).

83 Closure Stress (psi) 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 5800

5900

6000 Stage 2 - Traced Fracture Stage 2 - Perforations

6100

6200 Stage 2 - Traced Fracture

6300

6400

Depth (ft) 6500

6600

6700

6800

Stage 1 - Traced Fracture Stage 1 - Perforations 6900

Upper Limit of one Stage 1 - Traced Fracture 7000

7100

Figure 6.15: Hells Hole 9139x synthetic closure stress curve with tracer log fracture height growth labeled (values averaged over 5 ft intervals).

84 Gamma Ray 50 70 90 110 130 150 170 190 6700

6800 Depth (ft)

Stage 1 - Top of Traced Fracture

6900

Figure 6.16: Hells Hole 9139x gamma ray curve with the top of Stage 1 fracture height growth labeled (0.5 ft logged API measurements).

85 Gamma Ray 50 70 90 110 130 150 170 190 5800

5900 Depth (ft)

Stage 2 - Top of Traced Fracture

6000

Figure 6.17: Hole 9139x gamma ray curve with the top of Stage 2 fracture height growth labeled (0.5 ft logged API measurements).

86 CHAPTER 7

DISCUSSION

The research presented in this thesis has been conducted on many scales. A correlation was developed to relate samples of Wieland’s (2005) Mancos shale test block to well logs of the Mancos formation. The mechanical properties of the Mancos block Wieland used to conduct his big block test have been further analyzed in order to better understand propellant and hydraulic fracture propagation in shales.

Learnings from Wieland’s analysis of the big block tests and additional experimentation has been used to analyze propellant and hydraulic field stimulation tests of the Mancos shale within an area of the Douglas Creek Arch in Western

Colorado.

7.1 Mancos Formation Correlation

Analysis of the core plugs collected from the Mancos shale block enabled a correlation between density and Young’s modulus to be developed. The correlation was then compared to field data. The correlation was developed using several core plugs from a single 3 ft section of Mancos shale, and they were tested under confining pressure but were not saturated and the tests were conducted at room temperature.

This limited sampling cannot be considered a representative sample of the entire 2000 ft of Mancos, however by comparing calculated Young’s modulus values from the density correlation with those as determined using dipole sonic logs, it was found that the correlation was more effective in certain sections of the Mancos.

87 A gamma ray log was used as a tool to distinguish between different lithologies within the Mancos. It was found that the correlation was less effective in sandier sections of the Mancos below a gamma ray cutoff of 125, where the correlation coefficient was 0.25. The correlation was found to work most effectively between a gamma ray range of 125 – 140, where the correlation coefficient was 0.44.

Significantly improving this correlation would be difficult. Additional Mancos samples over the entire interval would be necessary, and the mechanical properties of individual layers as opposed to the average properties of several layers within a core plug would need to be determined. Even after that, the correlations would still need to be made using well logs that measure rock properties at vertical resolutions of six inches, which as seen in the Mancos block could incorporate tens of significantly different layers.

7.2 Propellant Block Test

The first large quartz band (Layer 1) as noted by Wieland (2005) dominated the fracture geometry of the propellant test. It was far more resistant to fracturing than the bounding shales, as was seen in Figure 4.1. Analysis of the U.C.S. data led to the identification of a second quartz band (Layer 2) that also had a significant effect on the fracture geometry even though it was much thinner.

In order to determine the mechanical properties of these quartz bands, attempts were made to core additional plugs from the Mancos shale block. Due to the fissility of the shale and interactions between the clays and lubricating fluids, the samples would disintegrate and additional cores could not be collected. Without knowing the exact mechanical properties of the quartz band and the adjacent layers, it

88 is not possible to accurately determine the in-situ stresses and stress differential between siltstone and shale layers.

However generalizations can be made as to whether a significant stress differential between the quartz layers and the bounding shales existed. Figure 7.1 shows the Poisson’s ratio curve for the Lower Horse Draw 2186 well. In Figure 7.1, several silts are noted. All the silts have significantly lower Poisson’s ratio values than the shales. For example, between 4700 ft and 4800 ft the silts have an average

Poisson’s ratio value of 0.245 as opposed to 0.295 for the adjacent shales. This suggests that the silt band within the test block should also have a Poisson’s ratio value significantly less than its bounding layers.

Additionally, correlations have been published showing that lower porosities and increased quartz content correlate with lower Poisson’s ratios (Kumar, 1976).

The S.E.M. analysis of the quartz band found almost no porosity, suggesting a low

Poisson’s ratio.

A number of factors would suggest that the siltstone quartz banding has a lower Poisson’s ratio than the bounding layers. Equation 7.1 has shown that layers with lower Poisson’s ratios have lower closure stresses. Lower closure stress often means the formations are more prone to fracturing, however that is not the case in this situation.

ν P = []D γ (7.1) c −ν )1( obtv where,

Pc = closure pressure (psi) ν = Poisson’s Ratio

Dtv = true vertical depth (ft)

γ ob = overburden stress gradient (psi/ft)

89 LOWER HORSE DRAW UN 2186 T2S R103W S26 PEFZ PR 14 0.1 0.3 GR [gAPI] AHT90 [OHMM] RHOZ [g/cm] TG [percent] 0 200 150230 2000

2800

2900

3000

3100 MANCOS_B_BASE 3200 LM_UPPER_LASTCYCLE 3300

3400

3500

3600

3700 LM_TOP_SANTONIAN_GRZONE

3800 LM_RED_MARKER 3900

4000

4100

4200

4300

4400

4500 LM_BASE_SANTONIAN_GRZONE 4600

4700 LM_ORANGE_MARKER 4800 LM_BASE_ORANGE_SILT

4900

5000 5100 LM_BLUE_MARKER 5200 LM_BASE__BLUE_SILT LM_GREEN_MARKER 5300 5400 LM_TOP_GREEN_SILT LM_TUNUNK 5500

5600 DAKOTALM_TURONIAN_UNC 5700

5800

5900

6000

6100

Figure 7.1: Lower Horse Draw 2186, standard logs plus Poisson’s ratio curve. Several silt sections within the Mancos are labeled. The silts all have significantly lower Poisson’s ratio values then the shales.

90 The fracture geometry created by the deflagration of propellant in the Mancos block was clearly not governed by in-situ stress. The block showed no evidence of shear slippage even though the strength of the bonding between the quartz banding and the shale layers was relatively week. Thus the mechanical properties of the quartz bands must have controlled the fracture geometry. Both the unconfined compressive strength test and the cementing of the quartz grains as seen in the S.E.M. photos confirm the toughness of the quartz layers which controlled the fracture geometry as seen in Figure 7.2.

2

1

Figure 7.2: Photograph of the north half of the Mancos shale block showing the propellant fracture staining. The two layers labeled “1” and “2” have significantly higher unconfined compressive strength. It is the mechanical properties of these two layers in particular that controlled the propellant fracture growth.

91 The strength of the quartz siltstone bands is derived from the excellent cementing as was measured by the unconfined compressive strength, future testing could determine the relative tensile strengths of these layers as hydraulic fractures propagate under tensile failures.

7.3 Inadvertent Hydraulic Fracture

The event recorded by Wieland during the pressurization of the wellbore for propellant testing was not fully representative of a hydraulic fracture in the field since a breakdown pressure greater than the fracture propagation pressure due to near wellbore stresses was not exceeded. Flaws within the block meant that once the fracture propagation pressure was exceeded, fluid began leaking off. The calculation of in-situ stress explained in Chapter 4 shows that minimum in-situ stress could be as low as 1000 psi, thus the fracture propagation pressure was somewhere between 1000 psi – 1300 psi. The lower rate of increase of wellbore pressurization at 519 seconds

(1290 psi) was most likely caused by the fracture propagation pressure being surpassed and fluid leaving the wellbore. The decrease in rate of wellbore pressurization would have been more pronounced if the target wellbore pressure had been greater then 1300 psi.

Since no evidence of staining could be found along the plane of the hydraulic fracture, the geometry of the fracture could not be established. The propagation of this fracture was governed by the same mechanism that allowed the fracture to propagate without overcoming a significant breakdown pressure. For example if there was a pre-existing flaw or a crack was created during the loading of the block, the fluid will have traveled the paths of this flaw rather than propagate a new fracture.

92 7.4 Propellant Field Tests

During the design of the propellant field tests, the amount of propellant to be used was significantly reduced. The propellant used was 1.375 inches in diameter as opposed to the 4 inch propellant originally suggested. This reduced the volume of propellant used by almost 90 percent. Modeling of the propellant stimulations after the fact using Wieland’s model shows that the expected fracture half length is at most

15 ft. Wieland’s model probably predicts larger than actual fracture lengths because it assumes an opened wellbore and not a cased hole. As a result, it does not account for friction losses or near wellbore perforation damage.

Proof that the propellant stimulations were ineffective was obtained when a slickwater fracture stimulation was pumped into one of the zones that had been treated with the 1.375 inch propellant. The propellant fracturing should have at least broken down the near wellbore damage, however the actual breakdown pressure of the hydraulic fracture was several hundred psi greater than the breakdown pressure of the same zone in an offset well that had not been treated previously with propellant.

7.5 Slickwater Field Tests

As opposed to the propellant field test where little could be determined about the development of any fracturing, the slickwater fracturing of the Hells Hole 9139x provided a wealth of information. During the treatment, pressures could be monitored and radioactive tracers allowed the placement of the treatment materials to be determined afterwards.

Analysis of the treatment pressures showed higher than expected in-situ stresses. The Nolte-Smith plot for Stage II showed two distinct periods of growth, the

93 first a period of stable height growth or fissure opening, and the second a period of later fracture growth. The tracer log proved that height growth was limited to within feet of perforated intervals for each stage. However with the layers of the Mancos shale being so fine, and interpretation of fracture containment accurate to only within feet, it was not possible to determine if thin high strength siltstone strata or thicker shales with higher in-situ stress confined the fractures.

The two different models used to design the fracture stimulations used significantly different inputs and generated significantly different projected lengths.

The actual fracture stimulation lengths are probably somewhere in-between the two models since the tracer log suggests that neither the single lobe predicted using

GOHFER™ nor the extreme containment between perforations predicted in the

FracCADE™ simulations were correct.

7.6 Further Discussions

Further analysis of Wieland’s propellant fracturing experiment has shown that fracture growth was controlled by the rock properties and not the in-situ stresses. The two siltstone layers identified were extremely well cemented giving the layers great strength, which ultimately restricted fracture growth. Unfortunately, analysis of the hydraulic fracture growth was not possible because the block was not permanently stained. Thus certain questions remain as to whether the propellant and hydraulic fracture growth are controlled by the same mechanisms.

The propellant field tests conducted in the Douglas Creek Arch Mancos shale, were ineffective in stimulating gas production, however it has been shown that stimulation was undersized and additional testing is necessary. In contrast, the two-

94 stage slickwater stimulation effectively created a fracture into the formation creating a larger effective wellbore radius. Analysis of the tracer log showed better than expected height containment. The height containment was caused by the heterogeneous Mancos shale, presumably rock properties but possibly also stress contrasts within the formation, or layer effects; additional hydraulic big block testing could answer this question.

95 CHAPTER 8

CONCLUSIONS

Significant data, from microscopic images to large scale field tests, have been incorporated into this thesis. By analyzing information at every scale, it has been possible to draw conclusions as to how propellant and hydraulic fractures propagate in the Mancos shale. These conclusions are:

1. Analysis of laboratory testing of the Mancos shale has proven that in

this system the fracture propagation of a propellant fracture was

governed principally by rock mechanical properties, in particular the

relative strengths of the strata.

2. The fine layering of the Mancos shale resulted in better height

containment then predicted for hydraulic slickwater stimulation.

3. An inadvertent hydraulic fracture occurred in the Mancos test block at

some pressure between 1100 psi - 1300 psi. This range was

significantly lower than the calculated breakdown pressure meaning

that the fracturing fluid pressure did not have to overcome increased

near wellbore stresses due to a flaw in the block.

4. The effectiveness of using propellant as a stimulation technique in the

Mancos shale of the Douglas Creek Arch remains unknown since the

field tests conducted in the three wells Hells Hole 9131, Hells Hole

9139x and Park Mountain 9025 used inadequate volumes of

propellant.

96 5. Rock strength controlled the fracture geometry of the big block

propellant fracturing, whereas lack of rock strength or the existence of

a failure point resulted in the inadvertent hydraulic fracture.

6. The data is inconclusive as to the mechanism that contained the

slickwater hydraulic fracture in the field test. It is possible that high

strength siltstone layers, in-situ stresses, or layer effects such as shear

slippage restricted height growth.

8.1 Future Research

This thesis is a study of the available stimulation data for the Mancos shale within a developmental area of the Douglas Creek Arch, Colorado. Laboratory data collected by Wieland (2005) has been compared with field information and testing.

As more data becomes available, further conclusions will likely be realized.

Suggestions for future work potential in this area are as follow:

1. Further analysis of the stratigraphy of the Mancos shale would allow

for a better understanding of how fractures propagate within it. Ideally,

the section should be cored so that the locations, frequencies and

properties of high strength quartz bands such as the one that controlled

the geometry of the propellant fracture in the test block can be

understood.

2. Further “intentional” big block hydraulic fracture testing of the

Mancos shale would make it possible to determine the specific

containment mechanism for hydraulic fracturing within the Mancos.

97 3. Analysis of the mechanical properties of more Mancos samples would

lead to a better understanding of the varying in-situ stress contrasts

between laminations.

4. Additional field propellant tests are warranted to assess the viability of

using propellant to stimulate the Mancos shale, since the tests

discussed most likely failed to breakdown the perforations due to

undersizing of the propellant system.

Clearly the Mancos shale remains a formation of interest, with high T.O.C.’s

(Fisher, 2007), gas shows while drilling and hydrocarbon liquids produced from the slickwater stimulation not discussed in this thesis. Due to this interest and the potential that it has for producing hydrocarbons, the Mancos shale formation should continue to be evaluated and examined in future research plans.

98 NOMENCLATURE

A = area

Dtv = true vertical depth, ft E = Young’s modulus, psi F = force acting on area, A L = Length, ft P = pressure, psi

Pb = breakdown pressure, psi

Pc = closure pressure, psi

Ppore = pore pressure, psi

ΔPp = pore pressure gradient, psi/ft PPA = pounds per gallon added PPG = pounds per gallon ν = Poisson’s ratio T = tensile strength of the formation, psi

Ts = shear transit time, μs

Tp = compressional transit time, μs vp = velocity of compressional wave, ft/s vs = velocity of shear wave, ft/s ρ = density, g/cc

αv = vertical Biot’s poroelastic constant

αh = horizontal Biot’s poroelastic constant

ε x = regional horizontal strain

σHmax = maximum horizontal stress, psi

σHmin = minimum horizontal stress, psi

σovb = overburden stress, psi

σ t = regional horizontal tectonic stress, psi

γ ob = overburden stress gradient, psi

γ p = pore pressure gradient, psi

99 REFERENCES

Barree, R. and Winterfield, P., Effects of Shear Planes and Interfacial Slippage on Fracture Growth and Treating Pressures. SPE Paper 48926, Presented at the Annual Technical Conference and Exhibition in New Orleans, LA, 27-30 Sept. 1998.

Barree, R. (2006). Fracturing. Industry Short Course Notes.

Barree, R., Barree and Associates. (2006). Personal Communications.

Batzle, M., and Smith, B., (1992). Hand-held Velocity Probe for Outcrop and Core Characterization. Proceedings of the 33rd US Symposium on rock Mechanics, Eds. Tillerson and Wawersik, Santa Fe, NM, p. 949-958.

Chardac, O., Murray, D. Carnegie, A. Marsden, J. A Proposed Data Acquisition Program for Successful Geomechanics Projects. (2005) SPE Paper No.93182, Presented at the 14th SPE Middle East Oil and Gas show and Conference held in Bahrain, 12 -15 Mar. 2005.

Daneshy, A.A. (1978). Hydraulic Fracture Propagation in Layered Formations. SPE Paper 6088, Society of Petroleum Engineers Journal Feb. 1978: p. 33-41.

Fisher, R., Masters student Colorado School of Mines. (2007). Personal Communications.

Fisher, R. Stratigraphic, geochemical and geomechanical analysis of the lower Mancos Shale, Douglas Creek Arch, northwest Colorado, U.S.A. (2007) MS thesis, Colorado School of Mines, Golden, Colorado.

100 Gidley, J., Holditch, S., Nierode, D., and Veatch, R., Recent Advances in Hydraulic Fracturing. (1989) SPE Monograph Volume 12.

Griffith, A.A. (1921). Phenomena of Rupture and Flow in Solids . Philosophical Transactions of the Royal Society of London, A221, 163-198.

Hecker, M., Houston, M., and Dumas, J., Improving Completion Designs In The Hugoton Field Utilizing Multiple Gamma Emitting Tracers. (1995) SPE Paper 30651, Presented at the Annual Technical Conference and Exhibition in Dallas, TX, 22 - 25 Oct. 2002.

Jorden, J., and Campbell, F., Well Logging II Electric and Acoustic Logging. (1986) SPE Monograph Volume 10.

Khristianovitch, S., and Zheltov, Y., Formation of Vertical Fractures by Means of Highly Viscous Fluids. (1955) Proc., World Pet. Cong., Rome 2, 579-86.

Kirschbaum, M., Geologic Assessment of Undiscovered Oil and Gas Resources of the Mancos/Mowry Total Petroleum System, Uinta-Piceance Basin Province, Utah and Colorado, Chapter 6 of Petroleum Systems and Geologic Assessment of Oil and Gas in the Uinta-Piceance Province, Utah and Colorado. (2003) U.S. Geological Survey Digital Data Series DDS-69-B.

Kumar, J., The Effect of Poisson’s Ratio on Rock Properties. (1978) SPE Paper 6094, Presented at the 51st Annual Fall Technical Conference and Exhibition in New Orleans, LA. 3 - 6 Oct. 1976.

McLennan J.D., Roegiers J.C., Marx W.P. (1983) The Mancos Formation: An Evaluation of the Interaction of Geological Conditions, Treatment Characteristics and Production. SPE/DOE Paper No.11606.

101 Miskimins, J. and Barree, R,. Modeling of Hydraulic Fracture Height Containment in Laminated Sand and Shale Sequences. (2003) SPE Paper 80935 presented at the SPE Production and Operations Symposium held in Oklahoma City, OK, 22-25 Mar. 2003.

Miskimins, J., Hurley, N., and Graves, R., A Method for developing rock Mechanical Property Logs using Electrofacies and Core Data. (2002) SPE Paper 77783, Presented at the Annual Technical Conference and Exhibition in San Antonio, TX, 29 Sept. – 2 Oct. 2002.

Passamaneck, R., Colorado School of Mines. (2007). Personal Communications.

Perkins, T., and Kern, L., Widths of Hydraulic Fractures, (Sept, 1961) JPT 937-49; Trans., AIME 222.

Rolon, L., Mohaghegh, S., Ameri, S., and Gaskari, R. Developing Synthetic Well Logs for the Upper Devonian Units in Southern Pennsylvania. (2005) SPE Paper 98013, Presented at the SPE Eastern Regional Meeting in Morgantown, WV 14 -16 Sep. 2005.

Schmidt, R., Warpinski, N. and Cooper, P., In Situ Evaluation of Several Tailored- Pulse Well-Shoot Concepts. (1980). SPE Paper 8934, presented at the 1980 SPE/DOE Symposium on Unconventional gas Recovery held in Pittsburg, PA, May 18-21, 1980.

Sneddon, I. The Distribution of Stress in the Neighborhood of a Crack in an Elastic Solid. (1945) Proceedings of the Royal Society of London, A187, 229-267.

Suarez-Rivera, R., TerraTek. (2007). Personal Communications.

102 Warpinski, N. R., Clark, J.A., Schmidt, R.A. and Huddle, C.W. Laboratory Investigation on the Effect of In-Situ Stresses on Hydraulic Fracture Containment. (1982) SPE Paper No.9834, Society of Petroleum Engineers Journal, Jun. 1982: p. 333-340.

Warpinski, N., Schmidt, R. and Northrop, D. In-Situ Stresses: The Predominant Influence on Hydraulic Fracture Containment. (1982) SPE Paper 8932, Journal of Petroleum Technology Mar. 1982: p. 653-664.

Wieland, C. Laboratory Testing and Finite Element Modeling of Propellant-Induced of Fracture Propagation in Shale Reservoirs. (2005) MS thesis, Colorado School of Mines, Golden, Colorado.

Woodworth, T. Proppant Placement in Low-Viscosity Slickwater Type Fracturing Fluids. (2006) MS thesis, Colorado School of Mines, Golden, Colorado.

Wright, C., Weijers, L., Davis, J. and Mayerhofer, M. Understanding Hydraulic Fracture Growth: Tricky but not Hopeless. (1999) SPE Paper 56724, Presented at the Annual Technical Conference and Exhibition in Houston, TX, 3-6 Oct. 1999.

103

Appendix A: Very low magnification of quartz band sample. Scope magnification: 25X.

105

Appendix A: Low magnification of sample showing lack of porosity, similar grain size framework. The darker grains are quartz, cemented with a mixture of materials including, dolomite, calcite and illites. Scope magnification: 250X.

106

Appendix A: Typical detrital quartz grain, height ≈ .05mm, length ≈ .025mm, meaning this would be defined as a coarse siltstone. Scope magnification: 1200x.

107

Appendix A: Dolomite grain. Scope magnification: 800X.

108

Appendix A: Limestone (calcite) grain. Scope magnification: 2700X.

109

Appendix A: Illite grain. Scope magnification: 3000X.

110

Appendix A: Potassium feldspar in dissolution. Scope magnification: 2000X

111

Appendix A: Detrital dolomite surrounded by authegenic dolomite. Scope magnification: 1000X.

112

Appendix A: Authegenic dolomite with silica. Scope magnification 1000X.

113 Final Report

Results of High-Energy Propellant Fracturing Tests Conducted on Large Scale Blocks of Rock in the Laboratory for EnCana Oil and Gas (USA), Inc with Cooperation of the Colorado School of Mines

Prepared for:

EnCana-Bob Weaver EnCana-Peter Loeffler CSM- Jennifer Miskimins CSM-Craig Wieland

Prepared by:

TerraTek, Inc. 1935 South Fremont Drive Salt Lake City, Utah 84104

Report TR5 July 2005

- 1 - Test Objective: The test objective was to conduct two high-energy propellant fracturing tests in two different large-block rock types, including Colton sandstone primarily for check out of the testing procedures and propellant and Mancos shale for simulation of field conditions and for borehole pressure and fracture initiation and propagation determination using a rapid burning propellant supplied through the Colorado School of Mines. The tests determined the pressure response inside the borehole and the resulting fracture initiation and propagation pressures, the number of fractures, the fracture orientations, etc. by monitoring pressure in the pressure probes inside the rock and extensive post-test examinations of the blocks.

Testing Equipment: The tests were conducted in the TerraTek Drilling and Completions Laboratory polyaxial stress frame. A sketch of the test set up is attached along with a photograph of the polyaxial stress frame. The center nominal 2” inch diameter hole in each block was cored using a servo-controlled drill rig. The borehole in the center of the blocks were sealed with glued stainless steel sleeves top and bottom to eliminate end effects and isolate fracture initiation in the open borehole in the center portion of the sample. Pressure build up inside the casing was transmitted to the rock in the open section of the borehole. Small notches were placed in the borehole wall to simulate actual field perforations or fracture initiation sites inside the borehole. The propellant and connecting initiation wires were lowered into the center of the borehole, the borehole filled with dyed 3% KCl, and then the leads from the initiation wires were attached to a feed-thru at the top of the casing. The feed-thru cap also sealed the top of the casing. Pressure probes were installed in the interior of the blocks to monitor pressure build up as fractures intersected the probe holes (see attached sketch for borehole notch and pressure probe locations). The prepared blocks of rock were placed inside the stress frame and stressed in the two horizontal directions and in the vertical direction using high pressure flatjacks. A computer data acquisition system recorded the three flatjack pressures, dynamic borehole pressure using two different types of pressure transducer and the six pressure probe pressures versus time.

- 2 - Test Specifications: Rocks: Colton sandstone 7600 psi UCS, 10.9% porosity, 0.04 md Mancos shale 9800 psi UCS, 7.9% porosity, <0.001 md Rock Size: 30” x 30” x 36” high Minimum horizontal stress: 1300 psi (1600 psi flatjack pressure) Maximum horizontal stress: 1600 psi (1975 psi flatjack pressure) Vertical stress: 4500 psi (5422 psi flatjack pressure) Borehole fluid: 3% KCl dyed red Case OD: 2 inch Casing ID: 1.5 inch Casing/Borehole volume: 1500 cc Rupture Disk in Borehole: 12,000 psi burst Instrumentation: Monitor borehole pressure (with both a 10,000 psi piezoelectric dynamic pressure transducer with a 2-10,000 Hz frequency response and a standard 30,000 psi strain gage type pressure transducer with 1,000 Hz frequency response), three flatjack pressures, pressure probes Data recording: Before, during and after test record data at 1 data point per seconds and during the fracturing test both at 500 points/sec for 60 seconds and 41,000 data points per second for 20 seconds

Test Set Up: The 30” x 30” x 36” block of Colton sandstone and Mancos shale had a 2” diameter hole cored in the center and five small notches were placed in the center of the borehole oriented at 60 degrees in a spiral pattern to each other at +4, +2, 0, -2 and -4 as shown on the attached sketch. Steel sleeves 12” long were glued in each end leaving a 12” open borehole section. Six pressure probes were glued in the block as shown in the attached sketch. These pressure probes were evacuated and filled with water and the water was pressurized to about 40 psi. After installing the propellant cartridge in the center of the open section of the borehole, 3% brine dyed red was placed in the borehole. The flatjacks were then pressurized to 5422 psi T-B, 1975 psi N-S and 1600 psi E-W to achieve block stresses of 4500 psi T-B, 1600 psi N-S and 1300 psi E-W. The borehole was pressurized to 1300 psi using

- 3 - a servo-controlled intensifier in pressure control. As will be discussed in the results section, the initial block stress/borehole pressure condition resulted in a hydraulic fracture being formed in the N-S direction and also in the discovery that the propellant would not initiate at borehole pressure greater than 100 psi. The flatjack pressures were then reversed to establish flatjack pressures of 5380 psi T-B, 1852 psi N-S and 3459 psi E-W which closed the existing hydraulic fracture and applied block stresses of 4465 psi T-B, 1500 psi N-S and 2802 psi E-W. At a borehole pressure of 50 psi, the propellant was successfully initiated and a propellant driven fracture was propagated in the E-W direction.

Following the test, the Colton sandstone block was cut horizontally in half to expose the fracture pattern and the Mancos shale block was split vertically along the N-S hydraulic fracture and the E-W propellant driven fracture.

- 4 -

- 5 -

- 6 -

- 7 -

- 8 - General Observations and Problems Associated with the Tests The following is an attempt to briefly describe general observations and problems associated with the tests and suggestions for improving future testing of this type.

General Observations: The Colton sandstone block test was run as a check out test and it turned out to be quite valuable in “lessons learned” for making improvements for the main Mancos shale block test. For example, after the propellant fracturing testing in Colton sandstone, the size of the propellant was increased and the top and bottom ends of the propellant was restrained by perforated tubes to keep the burn concentrated to the center of the block (open borehole). Also, because of initial problems of leakage past the top sleeve, care was taken to fill any opening in the sleeve OD and borehole ID with glue. . In the Colton sandstone block, a large fracture was made in the minimum stress direction (N-S) and two or three smaller fractures were started in other directions, but were confined to radiate out only an inch or two from the borehole.

As will be discussed, the Mancos shale block test had its own set of problems including electrical shorting to ground of the propellant initiation mechanism inside the propellant when subjected to pressurized brine above 100 psi which resulted in no initiation until the borehole pressure was reduced to 50 psi at which time successful initiation of the propellant was accomplished. Also, when pressurizing the borehole in the Mancos shale block, a hydraulic fracture initiated and propagated in the N-S direction. As a result, it was necessary to reverse and increase the block stresses so the propellant fracture could be initiated in the E-W direction. The Mancos shale block also experience observed crushing resulting in gas by-pass through the top borehole sleeve at the time of initiation of the propellant The Mancos shale block was inadvertently hydraulically fractured in the minimum stress direction (N-S) as borehole pressure was increased to 1300 psi and then to 1600 psi, which completed the hydraulic fracture. After reversing and increasing the block stresses, a large propellant driven fracture was created at 50 psi borehole pressure.

Problems Associated with Colton Sandstone Block Test: The following problems encountered during the Colton sandstone block test are summarized below: 1. Irregular Borehole and Initial Leak through Top Borehole Sealing Sleeve: The core barrel used to drill the borehole through the center

- 9 - of the Colton block was new and not broken in (relatively dull). Because of the relatively high clay content of Colton sandstone, the bit balled up and deviated so that the borehole diameter was 2” at the top and as great as 2.5” in the middle of the block. This resulted in a bad glue job which only sealed the top 2 or 3 inches and when the borehole pressure was first increased, the sleeve leaked at about 250 psi. At first, it was thought that the leak might have been caused by a mistake in tolerances in the upper borehole seal, but after correcting this problem, the sleeve leak persisted. When the sleeve leak was verified, a larger diameter core barrel was used to over-core around the sleeve and a thicker sleeve was machined and glued into place. Also, any gaps seen in the top part of the bottom sleeve was filled with glue. Recommendations: In the future, our standard 1.5” diameter core barrel will be used to core a much more uniform borehole. The sleeve ID would then be changed from 1.5” diameter to 1.37” diameter. 2. Propellant Size and Lack of Positive Electrical Seal: When the propellant was initiated, the pressure rise time was apparently less than expected and it was felt by the propellant supplier that the size of the propellant needed to be increased for the Mancos test and also end constraints needed to be placed top and bottom of the propellant to help confine the burn in the center open hole of the borehole. As a result, the size of the propellant (increased length) and end constraints were done using steel, perforated tubing top and bottom to constrain the burn to the center portion of the block. Also, the top of the propellant cartridge was not sealed, so TerraTek added a feed-through in a fitting at the top of the propellant. In addition, some continuity to ground was observed when brine was placed inside the borehole. Although this was a concern at the time, this partial continuity to ground did not prevent the propellant from being initiated at a borehole pressure of about 1200 psi. As noted earlier, for the Mancos test the propellant initiation contacts did short to ground at a borehole pressure greater than 100 psi. Recommendations: Apparently, work needs to be done on the propellant initiation system to provide a positive seal to operate under brine and pressurized borehole conditions. Also, end constraints may need to be designed into the propellant cartridge. 3. Length of Open Hole in Pressure Probes: For a propellant fracturing test, where the direction of the fracture might deviate from the center of the block, the existing 2” open section of the pressure probes

- 10 - proved to be too small. As a result, during the propellant fracturing test, only one pressure probe was intersected because the fracture deviated from center. Also, in impermeable rocks like the Colton sandstone and Mancos shale, it is difficult to verify that a pressure probe is not plugged, since you are unable to force air through the probe and into the rock. Recommendations: The open hole at the end of the probe should be increased from 2” to 5 or 6” long. Also, when installing the pressure probes, a wire should be inserted out the end of the probe and into the open hole and then withdrawn before the glue dries completely to prevent any possible plugging.

Problems Associated with Mancos Shale Block Test: The following problems encountered during the Mancos shale block test are summarized below: 1. Lack of Positive Electrical Seal of Propellant: This problem resulted in a mis-firing due to an electrical short and consequently the need to lower the borehole pressure to less than 100 psi to get the electrical short to go away enough to initiate the propellant. Recommendations: Apparently, work needs to be done on the propellant initiation system to provide a positive seal to operate under brine and pressurized borehole conditions. 2. Premature Hydraulic Fracturing of the Mancos Shale Block: As noted earlier, the Mancos shale block was fractured in the minimum stress direction at borehole pressure of 1300 psi or less. Unfortunately, the intensifier stroke or displacement was not being recorded, because there was no idea that a hydraulic fracture would have been created. As a result, only the borehole pressure and pressure probe data can be used to evaluate when the hydraulic fracture in the N-S direction occurred. In hindsight, because of the inhomogeneous and laminated nature of the Mancos shale and posttest observations that the Mancos shale block may have been crushed and started to fail at the block stresses applied, it is felt that these factors may have contributed to the initiation and propagation of a hydraulic fracture when the borehole pressure of 1300 psi (followed by 1600 psi) was equal to or exceeded the minimum horizontal block stress of 1300 psi. Recommendations: In future testing of this time in Mancos shale, it may be advisable to lower the block stresses as well as the borehole pressure. Also, it would be best to use block stress rather than flatjack pressure when applying the horizontal and vertical stresses (taking into account flatjack efficiency) so there is no confusion of the relative levels of

- 11 - borehole pressure to block stress. If this had been done, we would likely have avoided pressurizing the borehole above 1300 psi to 1600 psi.

Colton Sandstone Block Test Conditions and Results:

Colton Test Results Overall Events: After bringing the Colton sandstone block stresses up to the initial conditions of T-B 4544 psi, N-S 1603psi and E-W 1300 psi, borehole pressure was increased to 1300 psi and the valve isolating the intensifier from the borehole was closed. At this time, the borehole pressure began to bleed-off due to low permeability which resulted in a slight amount of fluid flow into the near wellbore region. The decision was made to increase the borehole pressure to 1600 psi at which time the valve was again closes. The borehole pressure bed off from 1600 psi to 1185 psi by the time the computer high rate data acquisition system was started and the firing box switched on to initiate the propellant burn. A plot of the 1 data point per second data (Encana1.xls), shows the events of block stress stabilization, borehole pressure bleed down, propellant initiation, probe P5 being reached by fracture and the natural bleed down after the propellant was initiated. The block size and location of the pressure probes and borehole notch locations are also shown for reference. The second plot shows the natural pressures bled off over a 30 minute period after the propellant initiation and the associated pressure decreases of the various parameters.

- 12 - EnCana Test 1-Propellant Fracturing Test in Colton Sandstone Block Overall Events (1 Data Point/Second)

5000

4500

4000

3500 1) Initial Block Stresses and Borehole Pressures: 3000 T-B 4544 psi N-S 1603 psi T-B Stress E-W 1300 psi N-S Stress 2500 Bore 1185 psi E-W Stress 2) Propellant Initiated and N-S Bore Pressure 2000 Fracture Reached P5 P5 Pressure

1500

4) Natural Pressure Bleed-off 1000

500

Block Stresses, Borehole Pressure and Probe 5 Pressure (psi) 3) P5 Reached by Fracture 0 0 20 40 60 80 100 120 140 160 180 200 Time (sec)

EnCana Test 1-Propellant Fracturing Test in Colton Sandstone Block Pressure Bleed-down (1 Data Point/Second) 5000

4500

4000

3500 Pressure Decline 1063 Seconds After Propellant Initiation: T-B Stress 4544 to 4302 psi or 242 psi decrease N-S Stress 1603 to 1437 psi or 166 psi decrease 3000 E-W Stress 1300 to 1240 psi or 60 psi decrease T-B Stress Bore Press 1751 to 142 psi or 1609 psi decrease N-S Stress P5 Pressure 1722 to 112 psi or 1610 psi decrease 2500 E-W Stress P5 Pressure Bore Pressure 2000

1500

1000

500 Block Stresses, Borehole Pressure and Probe 5 Pressure (psi)

0 0 200 400 600 800 1000 1200 Time (sec)

- 13 - Colton Block Test Results at Propellant Initiation:

The following two plots show the results recorded by the moderate rate computer data acquisition system (500 data points/second for 60 seconds) and the next two plots show the high rate computer data acquisition system (41,000 data points/second for 20 seconds. The following pressure events occurred with time:

Time (sec) 2.200 2.240 2.256 2.278 2.284 2.400 Time Change (msec) 0 40 56 78 84 200 Dynamic PT psi 1309 4630 1446 Bore PT psi 1167 3974 2034 P5 PT psi 29 269 4046 2009 T-B Stress psi 4544 4523 N-S Stress psi 1604 1693 E-W Stress psi 1299 1676

EnCana Test 1-Propellant Fracturing Test in Colton Sandstone Block 500 Data Points/Second for 60 Seconds

5000

4500

4000

3500

3000 Bore Propellant Dynamic Initiation Probe 5 2500 T-B Stress N-S Stress 2000 E-W Stress Pressure and Stress (psi) 1500

Pressure Bleed-Off 1000

500

0 0 5 10 15 20 25 30 35 40 45 50 55 60 Time (sec)

- 14 -

EnCana Test 1-Propellant Fracturing Test in Colton Sandstone Block 500 Data Points/Second for 60 Seconds

5000 Dynamic PT Peaked at 2.24 Bore PT Peaked at 2.256 sec 4500 P5 PT Peaked at 2.284 sec 4000

3500

3000 Bore Press Oscillations in P5 Dynamic Possibly Due to Probe 5 2500 Propellant Gas T-B Stress N-S Stress 2000 E-W Stress Pressure and Stress (psi) 1500

1000 Propellant 500 Initiation at 2.2 sec P5 PT Initiated Pressure at 2.278 sec 0 2.00 2.05 2.10 2.15 2.20 2.25 2.30 2.35 2.40 2.45 2.50 2.55 2.60 2.65 2.70 Time (sec)

- 15 -

- 16 -

Mancos Shale Block Test Conditions and Results:

Mancos Test Results Overall Events: The following plot shows the sequence of events for the Mancos shale test. After bringing the Mancos shale block stresses up to the initial conditions of T-B 4482 psi, N-S 1599 psi and E-W 1301 psi, using the intensifier in pressure feed-back mode, the borehole pressure was increased to 1300 psi. As pointed out earlier, the volume displaced by the intensifier was not being monitored during this test, therefore, the servo-controller in pressure feed-back would have injected whatever volume was needed to increase the pressure inside the borehole. At 451 psi, the borehole pressure leveled off momentarily which might indicate the intensifier had to inject more volume and that there could have been the initiation of a hydraulic fracture or leak into the shale at that pressure. The pressure continued to be increased to 1298 psi. About 35 seconds after the 1298 psi borehole pressure was reached, pressure probe P5 started to increase in pressure indicated that the hydraulic fracture had intersected P5. Shortly thereafter, P4, P1 and P2 started to increase in pressure 9 sec, 27 sec and 148 seconds, respectively after P5 was intersected

- 17 - by the fracture. About 64 seconds after 1298 psi had been reached, the valve isolating the borehole from the intensifier was closed and the borehole pressure fell rapidly to 444 psi. The isolation valve was closed again and the intensifier was used to bring the pressure back to 1310 psi. Not realizing that P5, P4 and P1 had been intersected by the fracture and duplicating what had been done in the Colton sandstone test, the borehole pressure was increased to about 1600 psi and shortly thereafter the intensifier ran out of fluid and injection into the hydraulic fracture stopped and the pressure bled down rapidly.

At this point in the test, a hydraulic fracture running in the N-S direction had been created inadvertently and all four pressure probes P5, P4, P1 and P2 had been intersected, the intensifier had run out of fluid and the borehole pressure had bled down to 294 psi. It was decided to initiate the propellant but when the firing box was activated, the propellant was not initiated. It was discovered that there was a short of ground in the initiation wires and so the borehole pressure was bled down to zero. At zero borehole pressure, continuity was restored. Several additional times, borehole pressure was raised and lowered to determine that the short to ground occurred at about 100 psi borehole pressure. The decision was made to both reverse the horizontal stresses (instead E-W being the minimum stress, N-S became the minimum stress) and the E-W horizontal stress was increased substantially to help close the existing hydraulic fracture and attempt to initiate the propellant at a borehole pressure less than 100 psi. During the reversal and increase in horizontal stresses, it was noted that there was a great deal of interaction between the horizontal stresses and that it was difficult to finally reach the desired block stresses. This reaction of the block to these stress changes may indicate that permanent deformation and some damage to the block was taking place. It should also be pointed out that as the E-W block stress was being increased, the pressure in all of the pressure probes increased. Probes 2 and 4 showed the largest increase, P5 a moderate amount of increase and P1 the least increase. This was an indication that the existing hydraulic fracture was effectively closed and fluid inside the fracture was trapped and pressurized by the closing fracture.

Once the block stresses of T-B 4465 psi, E-W 2802 psi and N-S 1500 psi had been established and borehole pressure was at 50 psi, the propellant was successfully initiated. The pressure spike in the borehole also resulted in pressure spikes in the block stresses, the intersection of P3 and P6 and small increases in P2, P4, P5 and P1. Also, at the time of initiation, gas from the

- 18 - burning propellant exited the block through a crushed zone in the rock around the outside of the top glued steel sleeve and borehole pressure bled off rapidly to 0 psi. From this point, the block stresses and probe pressures began to bleed off naturally. After 14 minutes of bleed off time, the data acquisition computer was stopped, the data rate changed to 1 data point per 60 seconds and restarted so the pressure bleed off phase of the test could be monitored overnight. The second of the following two plots show this over night pressure bleed off.

- 19 - EnCana Test 2-Propellant Fracturing Test in Mancos Shale Block Overall Events (1 Data Point/Second)

5000

4500

4000

3500 13) Had Difficulty Stabilizing Block Stresses T-B Stress 11) Reversed Stresses and Increased Difference N-S Stress 3000 Between E-W and N-S E-W Stress Stresses to Close Existing Fracture 30KSI-P Probe 1 2500 2) Borehole Probe 2 Pressure 3) N-S Fracture Probe 3 Established in 5) Borehole 8) Attempted to Initiate 10) Pressurized Initiated and Pressure Propellant but It Did Not Borehole and 15) Propellant Probe 4 2000 Pressure Reached P5 Initiated and E-W Control Established Burn Due to Electrical Observed Again Short to Ground Electrical Short to Fracture Reached Probe 5 Ground at About P3 & P6 Probe 6

1500 16) Block Stresses and Probe Pressures Bled-off 7) Intensifier Out of Naturally 1) Block 9) Reduced Borehole 1000 Fluid & Pressure Stresses Pressure to Zero at 12) P1, P5, Bleed-down Established Which Time Electrical P4 & P2 Block Stresses, Borehole and Pressure Probe Pressures (psi) Short to Ground Went Increased Away 500 4) Borehole Blocked In & Bled-off 6) P4 & P1 Reached P2 Reached 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Time (sec) 14) 50 psi Bore at Initiation

EnCana Test 2 Propellant Fracturing Test in Mancos Shale Over Night Bleed-Down of Block Stresses and Probe Pressures and Bore at 0 psi (1 Data Point per 60 Seconds)

4500

4000

3500

T-B Stress 3000 N-S Stress E-W Stress 2500 Probe 1 Probe 2 2000 Probe 3 Probe 4 Probe 5 1500 Probe 6

1000 BlockStresses and ProbePressures (psi)

500

0 0 10000 20000 30000 40000 50000 60000 Time (sec)

- 20 - Mancos Block Test Results at Propellant Initiation:

The following two plots show the results recorded by the moderate rate computer data acquisition system (500 data points/second for 60 seconds) and the next two plots show the high rate computer data acquisition system (41,000 data points/second for 20 seconds. The following pressure events occurred with time:

Time (sec) .0463 .0520 .0547 .0672 .0718 Time Change (msec) 0 5.65 8.38 20.94 25.51 Max. Dynamic PT psi 5559 Max. Bore PT psi 7641 Start Pressure at P3 PT psi XXX Pressure at P6 PT psi XXX T-B Stress psi 4465 N-S Stress psi 1500 E-W Stress psi 2802

- 21 -

EnCana Test 2-Propellant Fracturing Test in Mancos Shale Block 500 Data Points/Second Sampling Rate

7000

6000

5000

4000 Bore Dynamic 3000 Probe 3 Probe 6

2000

1000

0 Borehole Pressures and Probe 3 & 6 Probe Pressures (psi)

-1000 0 6 12 18 24 30 36 42 48 54 60 Time (sec)

EnCana Test 2-Propellant Fracturing Test in Mancos Shale Block 40,000 Data Points/Second Sampling Rate

8000

7000

6000

5000

Probe 3 Probe 6 4000 Dynamic Bore Pressure (psi) 3000

2000

1000

0 0.04 0.05 0.06 0.07 0.08 0.09 0.1 Time (s)

- 22 - - 23 -

- 24 - Geomechanical Test Results

Several tests were done on the cores of the Mancos shale and Colton sandstone blocks to characterize the properties of the rocks.

Colton Sandstone Block

The Colton sandstone block was tested to determine permeability and porosity. Three samples taken from the center core were tested, and the results are recorded in the table below.

Sample Sample Ambient Dry Bulk Grain Gas Sample Length Diameter Porosity Density Density Permeability Number (in) (in) (%) (g/cc) (g/cc) (md) COL T 1.973 0.995 8.83 2.43 2.670 0.099 COL M 1.973 0.994 9.11 2.42 2.667 0.099 COL B 1.973 0.994 9.68 2.41 2.673 0.127

Samples of the core were also tested using TerraTek’s Triaxial Compression test equipment. The following table contains these results.

Effective Bulk Confining Pore Young’s Sample Depth Compressive Poisson’s Formation Density Pressure Pressure Modulus ID (ft) Strength Ratio (gm/cm3) (psi) (psi) (psi) (psi) Col_T N/A 2.424 0 0 10,295 1,424,000 0.25 Colton N/A Sandstone Col_M 2.412 0 0 10,125 1,390,000 0.24 Col_B N/A 2.402 0 0 10,075 1,444,000 0.24

Mancos Shale Block

Samples of the core were testing using TerraTek’s Triaxial Compression test equipment. These results are shown in the table below.

Effective Bulk Confining Pore Young’s Sample Depth Compressive Poisson’s Formation Density Pressure Pressure Modulus ID (ft) Strength Ratio (gm/cm3) (psi) (psi) (psi) (psi) Man-Top N/A 2.552 0 0 8685 1,425,000 0.22 Mancos Shale Man-Mid N/A 2.561 0 0 10,020 1,508,000 0.19 Man-Btm N/A 2.565 0 0 9245 1,447,000 0.20

- 25 - The Mancos Shale core was characterized using TerraTek’s Tight Rock Analysis (TRA™). The results are found in the following table.

Tight Rock Analysis

Gas Bound A-R A_R Dry Bound Pulse-Decay Porosity Water Gas Mobile Oil Filled Clay Sample Bulk Grain Grain Hydrocarbon Permeability Depth (% of Saturation Saturation Saturation Porosity Water ID Density Density Density Saturation @500NEW BV) (% of PV) (% of PV) (% of PV) (% of (% of (gms/cc) (gms/cc) (gms/cc) (% of BV) (md) BV) BV) 1 top 2.546 2.638 2.711 7.53 50.50 46.39 3.11 3.49 0.19 5.34 0.000913 2 middle 2.546 2.608 2.679 6.44 59.29 37.03 3.68 2.38 0.30 4.77 *0.002488 3 bottom 2.550 2.668 2.729 7.75 39.80 57.16 3.04 4.43 0.19 3.59 *0.004333

*Note: Samples are likely microfractured due to bedding laminations

- 26 - Unconfined Compressive Strength Determination via Scratch Testing

Continuous profiles of unconfined compressive strength (scratch test) were obtained for representative core samples from the Colton sandstone and Mancos shale blocks. These profiles provide a high resolution (100 points/mm) representation of the variability in strength along the length of the core. The fundamental sources of strength variability are changes in texture (grain size distribution, presence of laminations, changes between depositional sequences resulting in bedding), and changes in mineral composition (depositional or diagenetic). The strength profile provides a continuous evaluation of these textural and compositional changes as reflected by their effect on a material property (UCS). The value of the measurements is to provide information regarding the frequency and magnitude of the variability in strength, and by inference, the variability on all other properties (mechanical and petrophysical).

The Colton Sandstone core was broken into several pieces during the coring. Two of the pieces, labeled Samples 2 and 3, were tested. Graphs of the results along with photographs of the core are included on the following two pages. Results from continuous measurements on Colton sandstone show unconfined strengths in the range from 7000 psi to 15000 psi. The red traces represent the high resolution measurements. The blue traces represent the low-resolution (filtered) measurements. The latter help visualizing the strength variability at inch-scale. The variability in strength along the core (Core Samples 2 and 3) is apparent in the images. In Sample 2 the upper section of the core is markedly variable and contains the weakest and strongest sections of the core. In Sample 3, the strength is reasonably uniform except towards the bottom of the core (0.22 to 0.3 ft).

Results from continuous measurements on Mancos shale show unconfined strengths in the range from 10000 psi to 35000 psi. The red traces represent the high resolution measurements. The blue traces represent the low- resolution (filtered) measurements. The latter help visualizing the strength variability at inch-scale. The large variability in strength in Mancos shale is associated to the visually apparent bedding. The bed sections that are rich in silica (white interlayer) are considerable stronger than the clay-rich darker inner-beds. In addition, because of this predominantly laminar heterogeneity, Mancos shale exhibits strong anisotropy and thus a strong relationship between material properties (strength and moduili) and orientation to bedding.

- 27 - EnCana Colton S.S. Sample#2 25000

20000 Core was cut to minimize grinding. Fracture is man-made.

15000 UCS (psi) UCS 10000

5000

0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Length (ft)

- 28 - EnCana Colton S.S. Sample#3 16000

14000

12000

10000

8000 UCS (psi)

6000

4000

2000

0 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Length (ft)

- 29 -

EnCana Mancos Shale

45000

40000

35000

30000

25000

UCS (psi) 20000

15000

10000

5000

0 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 Length (ft)

- 30 -

Stage I Report

Section 1: Perforations

Perforations Top Top Bottom Bottom Shot Number Diameter MD TVD MD TVD Density (ft) (ft) (ft) (ft) (shot/ft) (in) 6700.0 6700.0 6830.0 6830.0 0.00 0 0.32 6830.0 6830.0 6911.0 6911.0 4.00 324 0.32 6911.0 6911.0 6937.0 6937.0 0.00 0 0.32 6937.0 6937.0 6960.0 6960.0 4.00 92 0.32 6960.0 6960.0 7038.0 7038.0 0.00 0 0.32 7038.0 7038.0 7058.0 7058.0 4.00 80 0.32 7058.0 7058.0 7158.0 7158.0 0.00 0 0.32

Section 2: Wellbore Configuration

Tubing OD Weight ID Depth (in) (lb/ft) (in) (ft) 2.875 6.5 2.441 6747.0

Casing OD Weight ID Depth (in) (lb/ft) (in) (ft) 4.500 11.6 4.000 7061.0

Section 3: As Measured Pump Schedule

As Measured Pump Schedule Stg Stage Slurry Slurry Pump Fluid Fluid Proppant Max Prop Prop Prop # Name Volume Rate Time Name Volume Name Conc Conc Mass (bbl) (bbl/min) (min) (gal) (PPA) (PPA) (lb) 1 Pad 119.05 9.92 12.000 WF120 4993 None 0.000 0.000 0 [CO2 2 0.5 PPA 87.10 12.01 7.250 WF120 3601 Sand 0.883 0.354 1275 [CO2 3 1.0 PPA 155.78 12.15 12.817 WF120 6015 Sand 2.208 1.939 11663 [CO2 4 1.5 PPA 270.39 10.24 26.400 WF120 9865 Sand 4.380 3.346 33005 [CO2 5 2.0 PPA 281.16 10.82 25.983 WF120 10053 Sand 4.797 3.863 38831 [CO2 6 2.5 PPA 66.44 10.83 6.133 WF120 2335 Sand 5.261 4.309 10062 [CO2 7 Flush 40.12 13.83 2.900 WF120 1652 None 1.965 0.534 882

Appendix C: Stage I treatement report, 1/3.

130

Stage Pressures & Rates Stg Stage Average Maximum Average Maximum # Name Slurry Slurry Rate Treating Treating Rate (bbl/min) Pressure Pressure (bbl/min) (psi) (psi) 1 Pad 9.92 13.04 5645.5 8446.9 2 0.5 PPA 12.01 12.57 6692.5 7325.9 3 1.0 PPA 12.15 13.98 6971.5 7326.2 4 1.5 PPA 10.24 12.23 6696.3 7391.4 5 2.0 PPA 10.82 11.29 6692.0 7283.0 6 2.5 PPA 10.83 10.89 6497.9 6802.9 7 Flush 13.83 19.89 4399.8 6192.9

As Measured CO2 Pump Schedule Stg Stage CO2 CO2 Injection # Name Volume Rate Rate (bbl) (bbl/min) (bbl/min) 1 Pad 84.74 7.06 16.96 2 0.5 PPA 75.65 10.43 22.46 3 1.0 PPA 131.45 10.26 22.40 4 1.5 PPA 216.96 8.22 18.46 5 2.0 PPA 213.19 8.21 19.03 6 2.5 PPA 49.94 8.14 18.95 7 Flush 24.20 8.34 22.24

As Measured Totals Slurry Pump Time Clean Fluid Proppant CO2 Liquid 1 (bbl) (min) (gal) (lb) (bbl) (gal) 1020.07 93.483 38515 95718 796.12 0.0

Average Treating Pressure: 6502.0 psi Maximum Treating Pressure: 8446.9 psi Average Injection Rate: 19.42 bbl/min Maximum Injection Rate: 28.24 bbl/min Average Horsepower: 1758.0 hhp Maximum Horsepower 3007.8 hhp Maximum Prop Concentration 5.261 PPA

Section 4: Material Balance Per Stage

Stage Stage Water CO2 WF120 [CO2 WF120 Sand # Name (gal) (bbl) (gal) (gal) (lb)

1 Pad 4993 84.74 4993 0 0 2 0.5 PPA 3601 75.65 3601 0 1275 3 1.0 PPA 6015 131.45 6015 0 11663 4 1.5 PPA 9865 216.96 9865 0 33005 5 2.0 PPA 10053 213.19 10053 0 38831 6 2.5 PPA 2335 49.94 2335 0 10062 7 Flush 1652 24.20 0 1652 882

Appendix C: Stage I treatement report, 2/3.

131

Stage Stage B221 F104 L064 J218 J475 # Name (gal) (gal) (gal) (lb) (lb)

1 Pad 23.0 24.3 0.0 0 17 2 0.5 PPA 15.5 16.5 0.0 0 14 3 1.0 PPA 26.9 28.0 0.0 0 51 4 1.5 PPA 35.1 48.2 0.0 127 180 5 2.0 PPA 30.5 59.0 0.0 103 149 6 2.5 PPA 0.0 16.5 0.0 22 18 7 Flush 0.0 2.3 0.0 0 0

Section 5: Material Balance

Material Quantity Name Pumped Water 38515 gal CO2 128 Tons WF120 [CO2 36863 gal WF120 1652 gal Sand 104560 lb B221 163.8 gal F104 191.4 gal L064 0.0 gal J218 33 lb J475 110 lb

EnCana Oil and Gas Tr. Press AN_PRESS CFLD_RATE Hells Hole #9139X 05-31-2006

Slurry Rate Prop Con BH_PROP_CON 9000 20 6 19 8000 18 17 5 7000 16 15 Pressure - psi 14 Concentration - PPA 6000 4 13 Rate - bbl/min 12 5000 11 10 3 4000 9 8 3000 7 2 6 5 2000 4 3 1 1000 2 1 0 0 0 09:22:16 10:12:16 11:02:16 11:52:16 12:42:16 Time - hh:mm:ss

© Schlumberger 1994-2003

Appendix C: Stage I treatement report, 3/3.

132

Stage II Report

Section 1: Perforations

Perforations Top Top Bottom Bottom Shot Number Diameter MD TVD MD TVD Density (ft) (ft) (ft) (ft) (shot/ft) (in) 5800.0 5800.0 5994.0 5994.0 0.00 0 0.32 5994.0 5994.0 6019.0 6019.0 4.00 100 0.32 6019.0 6019.0 6035.0 6035.0 0.00 0 0.32 6035.0 6035.0 6085.0 6085.0 4.00 200 0.32 6085.0 6085.0 6164.0 6164.0 0.00 0 0.32 6164.0 6164.0 6243.0 6243.0 4.00 316 0.32 6243.0 6243.0 6343.0 6343.0 0.00 0 0.32

Section 2: Wellbore Configuration

Deviated Hole: NO Treating Down: TUBING Flush Volume to 5891.1 ft is 35.0 bbl

Tubing OD Weight ID Depth (in) (lb/ft) (in) (ft) 2.875 6.5 2.441 5800.0

Casing OD Weight ID Depth (in) (lb/ft) (in) (ft) 4.500 11.6 4.000 7061.0

Appendix C: Stage II treatement report, 1/6.

133

Section 3: As Measured Pump Schedule

As Measured Pump Schedule Stg Stage Slurry Slurry Pump Fluid Fluid Proppant Max Prop Prop Prop # Name Volume Rate Time Name Volume Name Conc Conc Mass (bbl) (bbl/min) (min) (gal) (PPA) (PPA) (lb) 1 Pad 131.0 5.5 23.8 WF120 5491 None 0.00 0.00 0 [CO2 2 0.5 PPA 87.1 12.8 6.8 WF120 3506 Sand 1.30 0.96 3354 [CO2 3 1.0 PPA 181.7 13.0 14.0 WF120 7001 Sand 2.55 2.00 13985 [CO2 4 1.5 PPA 297.4 12.1 24.6 WF120 10888 Sand 27.32 3.28 35690 [CO2 5 2.0 PPA 323.3 13.5 23.9 WF120 11504 Sand 4.20 3.99 45904 [CO2 6 2.5 PPA 104.9 13.7 7.7 WF120 3628 Sand 5.31 4.74 17206 [CO2 7 Flush 34.6 9.2 3.8 WF120 1466 None 0.00 0.00 0

Stage Pressures & Rates Stg Stage Average Maximum Average Maximum # Name Slurry Slurry Rate Treating Treating Rate (bbl/min) Pressure Pressure (bbl/min) (psi) (psi) 1 Pad 5.5 12.8 2740 6690 2 0.5 PPA 12.8 14.2 6002 6323 3 1.0 PPA 13.0 13.5 5975 6262 4 1.5 PPA 12.1 13.8 5828 6938 5 2.0 PPA 13.5 13.9 6151 6510 6 2.5 PPA 13.7 14.3 6430 6824 7 Flush 9.2 19.6 2406 4723

As Measured CO2 Pump Schedule Stg Stage CO2 CO2 Injection # Name Volume Rate Rate (bbl) (bbl/min) (bbl/min) 1 Pad 87.3 3.7 9.2 2 0.5 PPA 70.6 10.4 23.2 3 1.0 PPA 143.0 10.2 23.2 4 1.5 PPA 234.0 9.5 21.6 5 2.0 PPA 233.3 9.8 23.3 6 2.5 PPA 71.3 9.3 23.0 7 Flush 0.0 0.0 9.3

As Measured Totals Slurry Pump Time Clean Fluid Proppant CO2 Liquid 1 (bbl) (min) (gal) (lb) (bbl) (gal) 1160.0 104.5 43484 116139 839.5 0.0

Appendix C: Stage II treatement report, 2/6.

134

Average Treating Pressure: 5156 psi Maximum Treating Pressure: 6938 psi Average Injection Rate: 19.4 bbl/min Maximum Injection Rate: 24.5 bbl/min Average Horsepower: 1615.3 hhp Maximum Horsepower 2269.5 hhp Maximum Prop Concentration 27.32 PPA

Section 4: Material Balance

Material Quantity Name Pumped Water 43484 gal CO2 839.5 bbl WF120 [CO2 42019 gal WF120 1466 gal Sand 116139 lb J218 22 lb J475 27.5 lb F104 209 gal B221 210 gal

Section 5: Message Log

# Time Message Treating Annulus Total Slurry Prop. Pressure Pressure Slurry Rate Conc. (psi) (psi) (bbl) (bbl/min) (PPA) 1 10:28:25 Pressure Test Lines 7395 1 0.0 0.0 0.00 2 11:02:11 Started Pad 46 818 0.0 0.0 0.00 3 11:05:47 Stage at Perfs: Pad 5055 790 27.3 11.9 0.00 4 11:25:54 Started Pumping Prop 6097 760 130.5 12.8 0.00 5 11:25:57 Started 0.5 PPA Automatically 6066 763 131.1 12.8 0.00 6 11:27:40 Stage at Perfs: 0.5 PPA 6005 756 153.1 12.8 0.97 7 11:32:44 Started 1.0 PPA Automatically 6262 761 218.1 13.4 0.97 8 11:34:27 Stage at Perfs: 1.0 PPA 5991 750 240.4 13.0 1.96 9 11:46:43 Started 1.5 PPA Automatically 6068 736 399.9 13.4 2.55 10 11:48:27 Stage at Perfs: 1.5 PPA 5923 730 423.0 13.3 3.01 11 12:11:20 Started 2.0 PPA Automatically 5812 698 697.3 13.4 3.07 12 12:13:04 Stage at Perfs: 2.0 PPA 6009 700 720.7 13.4 3.99 13 12:35:13 Started 2.5 PPA Automatically 6344 675 1020.7 13.8 4.03 14 12:35:17 Activated Extend Stage 6322 671 1021.6 13.8 4.30 15 12:36:57 Stage at Perfs: 2.5 PPA 6494 678 1044.6 13.7 5.23 16 12:42:53 Deactivated Extend Stage 3806 562 1125.4 14.0 -0.09 17 12:42:53 Started Flush Manually 3806 562 1125.4 14.0 -0.09 18 12:43:34 Stopped Pumping Prop 4619 641 1136.7 19.4 -0.08

Appendix C: Stage II treatement report, 3/6.

135

Section 6: Designed Pump Schedule Ramp

Designed Pump Schedule Ramp Stg Stage Slurry Pump Pump Fluid Fluid Prop Name Prop Conc Prop # Name Volume Rate Time Name Volume (PPA) Mass (bbl) (bbl/min (min) (gal) (lb) ) 1. Pad 131.0 12.5 10.5 WF120 [CO2 5500 0.00 0 2. 0.5 PPA 87.1 12.8 6.8 WF120 [CO2 3500 Sand 1.00 3500 3. 1.0 PPA 181.7 13.0 13.9 WF120 [CO2 7000 Sand 2.00 14000 4. 1.5 PPA 297.4 13.3 22.4 WF120 [CO2 11000 Sand 3.00 33000 5. 2.0 PPA 323.3 13.5 23.9 WF120 [CO2 11500 Sand 4.00 46000 6. 2.5 PPA 102.2 13.8 7.4 WF120 [CO2 3500 Sand 5.00 17500 7. Flush 35.0 25.0 1.4 WF120 1469 0.00 0

Designed CO2 Pump Schedule Ramp Stg Stage CO2 # Name Rate

1. Pad 10.6 2. 0.5 PPA 10.4 3. 1.0 PPA 10.2 4. 1.5 PPA 9.9 5. 2.0 PPA 9.7 6. 2.5 PPA 9.5 7. Flush 0.0

Designed Ramp Totals Slurry Pump Time Clean Fluid Proppant CO2 (bbl) (min) (gal) (lb) (bbl)

1157.7 86.3 43469 114000 848.9

Appendix C: Stage II treatement report, 4/6.

136

Section 7: Fluid Description

Fluid Name: 2% KCL Water Additive Type Additive Name Solid Conc Liquid Conc (lb/mgal) (gal/mgal) MISC M117, POTASSIUM CHLORIDE 166.01

Fluid Name: WF120 Additive Type Additive Name Solid Conc Liquid Conc (lb/mgal) (gal/mgal) GELLINGAGT B221, PSG Polymer Slurry 4.50

Fluid Name: WF120 [CO2 50Q] - F104(5) Additive Type Additive Name Solid Conc Liquid Conc (lb/mgal) (gal/mgal) Surfactant F104, Foaming Agent 5.00 GELLINGAGT B221, PSG Polymer Slurry 4.50 ClayStabiliz L064, Clay Stabilizer 2.00 Biocide M275, Microbiocide 0.50

Section 8 Proppant Description

Proppant Proppant Mesh Spec Prop Fric Prop Fric Name Size Gravity Coeff. Coeff. Laminar Turbulent Sand 20/40 2.65 1.00 0.70 20/40 Jordan-Unimin 20/40 2.65 1.00 0.70

Appendix C: Stage II treatement report, 5/6.

137

Hells Hole #9139X 05-31-2006 Tr. Press. & Slr. Rate Vs Slr. Vol.

10000 50 10 Treating Pressure Slurry Rate 9000 BH Injection Rate 45 9 Prop Con 8000 BH Prop Con 40 8

7000 35 bbl/min Slurry Rate - 7 rpCn-PPA Prop Con - 6000 30 6

5000 25 5

4000 20 4

Treating Pressure - Pressure Treating psi 3000 15 3

2000 10 2

1000 5 1

0 0 0 0 200 400 600 800 1000 1200 Slurry Volume - bbl

Appendix C: Stage II treatement report, 6/6.

138

Appendix D: Tracer log of slickwater fracture treatment showing log header, 1/4.

140

Appendix D: Tracer log of slickwater fracture treatment showing two upper sets of perforations of Stage 2, blue represents the pad tracer, red represents the proppant tracer, 2/4.

141

Appendix D: Tracer log of slickwater fracture treatment showing the lower set of perforations of Stage 2, blue represents the pad tracer, red represents the proppant tracer, 3/4.

142 .

Appendix D: Tracer log of slickwater fracture treatment showing the upper and middle (partial) sets of perforations of Stage 1, blue represents the pad tracer, yellow represents the proppant tracer, 4/4.

143