MECHANICAL STRATIGRAPHY OF THE VACA MUERTA FORMATION, NEUQUÉN BASIN,

ARGENTINA

by

Kyla Bishop
A thesis submitted to the Faculty and the Board of Trustees of the Colorado School of Mines in
partial fulfillment of the requirements for the degree of Master of Science (Geology)

Golden, Colorado

Date ____________________

Signed___________________

Kyla Bishop

Signed__________________

Dr. Steve Sonnenberg

Thesis Advisor

Signed__________________

Dr. Thomas L. Davis

Thesis Advisor

Golden, Colorado

Date ____________________

Signed__________________

Dr. Paul Santi

Professor and Head

Department of Geology

ii
 ABSTRACT

The Vaca Muerta offers an excellent opportunity to study field development at an early stage.
The particularities of how best to characterize this play are developing, and various techniques
and processes must be tested. Even in parts of the basin where wells are being drilled in rapid
succession, production history is relatively short, and the dynamics of how the production may
change with time are still to be anticipated (Rimedio et al., 2015).

With limited core data for calibration, a comparison of several predictive methods was made to
estimate rock properties for the Vaca Muerta Formation in a study area consisting of 5 wells.
Borehole image logs were interpreted to study pre-existing natural fractures, and a stability
assessment was conducted to estimate the reactivation potential of the identified natural
fractures.

The comparison results in the characterization of an interval with clear vertical trends between
an upper, middle, and lower section. In the lower section, organic and siliciclastic contents are
highest. In the upper, organic content is lowest, and carbonate content is highest. The middle
zone between them is a transitional interval in which the organic matter continuously decreases
upwards in depth, in relative proportion with the increase in carbonate content.

Borehole image log data were interpreted for the presence of natural fractures, horizontal
discontinuity surfaces, concretions, and other heterogeneities. Documented is the presence of a
naturally fractured zone in the middle Vaca Muerta and abundant calcareous concretions in the
Lower Vaca Muerta. The identified natural fractures are likely candidates for reactivation during
hydraulic stimulation based on a stability assessment.

The middle of the interval is suitably brittle, organic rich, and has now been documented to
contain multiple natural fractures. Although this conclusion was reached independently, it is
corroborated by regional literature (Fantin et al., 2014; Williams et al., 2014). In addition,
geochemical analysis on the quality of the liquid hydrocarbons in the Vaca Muerta indicates that
the middle interval has high volumes, with a moderate gas-to-oil ratio, and moderate viscosities
(Williams et al., 2014).

iii
 TABLE OF CONTENTS

ABSTRACT ........................................................................................................................................ iii

LIST OF FIGURES ............................................................................................................................. vii

LIST OF TABLES .............................................................................................................................. xiv

LIST OF SYMBOLS ........................................................................................................................... xv

ACKNOWLEDGEMENTS ................................................................................................................. xvi

CHAPTER 1 INTRODUCTION ............................................................................................................ 1

1.1 Project Goals and Available Data .............................................................................................. 2

CHAPTER 2 GEOLOGIC BACKGROUND ............................................................................................ 4

2.1 Structural Development of the Neuquén Basin ............................................................ 5

2.2 The Vaca Muerta Formation ......................................................................................... 7

2.3 The Quintuco Formation ............................................................................................... 9

2.4 The Vaca Muerta Petroleum System .......................................................................... 10

CHAPTER 3 GEOMECHANICAL CHARACTERIZATION..................................................................... 12

3.1 Geomechanical Reservoir Characterization................................................................ 12

3.12 Elastic Properties ...................................................................................................... 14

3.13 Pore Pressure ............................................................................................................ 16

3.2 Fractures and Rock Strength ....................................................................................... 17

3.21 Natural Fractures ...................................................................................................... 20

iv
 3.22 Drilling Induced Fractures ......................................................................................... 20

3.23 Stimulation Fractures ................................................................................................ 22

3.3 Fractures in the Vaca Muerta ................................................................................................. 24

CHAPTER 4 WORK FLOW............................................................................................................... 29

4.1 Petrophysical Pre-Conditioning .................................................................................. 29

4.11 Lithology and TOC Determinations ........................................................................... 32

4.2 Borehole Image Theory............................................................................................... 38

4.21 Electric Borehole Imaging Physics............................................................................. 39

4.22 Acoustic Borehole Imaging Physics ........................................................................... 41

4.23 Data Handling............................................................................................................ 42

4.24 Geomechanical Analysis using Borehole Images ...................................................... 44

4.3 Cluster Analysis Principles........................................................................................... 45

4.31 Cluster Analysis Theory ............................................................................................. 46

CHAPTER 5 IMAGE LOG INTERPRETATION ................................................................................... 51

5.1 Image Interpretation Goals ......................................................................................... 51

5.2 Image Interpretation Process ..................................................................................... 51

5.21 OBMI Interpretation for Wells B and C ..................................................................... 55

5.22 FMI Interpretation for Well G ................................................................................... 57

5.3 Stress Determination .................................................................................................. 61

v
 5.4 Natural Fracture Interpretation .................................................................................. 61

5.41 Stability Assessment ................................................................................................. 65

5.5 Horizontal Feature Interpretation .............................................................................. 69

5.6 Integration with Sonic Data ........................................................................................ 72

CHAPTER 6 CLUSTER ANALYSIS ..................................................................................................... 77

6.1 Cluster Analysis Goal and Workflow ........................................................................... 77

6.2 Cluster Model Iterations ............................................................................................. 79

6.3 Interpreting the Cluster Results .................................................................................. 85

6.4 Testing the Cluster Model Sensitivity ......................................................................... 91

CHAPTER 7 FINAL INTERPRETATION ............................................................................................. 95

7.1 Conclusions of the Borehole Image Analysis .............................................................. 95

7.2 Conclusions from Cluster Modelling and Combined Interpretation .......................... 98

CHAPTER 8 CONCLUSIONS .......................................................................................................... 108

8.1 Applicability of Analysis and Implications for Completion Quality ........................... 108

8.2 Recommendations for Further Analysis ................................................................... 109

REFERENCES ................................................................................................................................ 113

APPENDIX CLUSTER PROPERTY TABLE ........................................................................................ 121

vi
 LIST OF FIGURES

Figure 1.01 Location Map of the Neuquén Basin with insert showing major
structural features in the vicinity (Modified from Zeller et al., 2015) ................... 1

Figure 1.11 Data Locations for the study area including wells and seismic survey,
as well as a relevant data audit. Provided by Wintershall and Modified .............. 3

Figure 2.01 Structural domains and large-scale structural features of the

Neuquén Basin (Modified from Sagasti et al., 2014) ............................................. 4

Figure 2.11 Schematic depicting the three tectonic phases key to the evolution of
the Neuquén basin with relevant captions to highlight important features.
(Howell et al., 2005) ............................................................................................... 6

Figure 2.12 Stratigraphic column for the Neuquén basin for the region surrounding
the Wintershall Block of Interest (Simplified from Howell et al., 2005)................. 7

Figure 2.21 The Vaca Muerta interval in the area of interest: a) Representative
gamma ray, results from the ECS tool, resistivity, and dynamic You g s
Modulus a d Poisso s atio, b) Representative seismic geometry flattened
on top Quintuco reflector and compressed laterally for emphasis ....................... 9

Figure 2.22 Schematic cross section of the Vaca Muerta and Quintuco system in
the Central Neuquén Basin sector which is representative of the geometry
in the area of study (Modified from Kietzmann et al., 2014) ............................... 10

Figure 3.11 Stress fields and their relative magnitudes (Fox et al., 2013) .............................. 13

Figure 3.12 Block diagrams depicting layering that creates a VTI medium, or
vertical fracturing that creates an HTI medium, and how these may be
combined in mudstones (modified from Hart et al., 2013).................................. 16

Figure 3.21a Block diagram depicting the relationship between the orientations of
the principal stresses and resultant failure. (Bratton et al., 2006) ...................... 18

Figure 3.21b Alternative block diagram for tensile and shear failure orientations relative
to the maximum principal stress controlling their occurrence
(Fjaer et al., 2008)................................................................................................. 19

Figure 3.22 Components of a Mohr-Coulomb plot (Meissner, 1984) ..................................... 19

Figure 3.23 View of borehole failure in a vertical borehole from above with
horizontal stress orientations depicted (Modified from Schlumberger) ............. 21

vii
Figure 3.24 A Mohr-Coulomb plot showing the lowering of effective stress in
relation to increased pressure during hydraulic stimulation. The
decrease in effective stress forces the Mohr circle towards the failure
envelope (Davey, 2012) ........................................................................................ 23

Figure 3.25 Differing fracture geometries created by hydraulic

stimulation (Warpinski et al.,2009) ...................................................................... 24

Figure 3.31 Stress regimes in the Vaca Muerta showing a transition from normal
in the far east of the basin to full strike slip to the west, where thrust
and reverse faulting may also occur (Garcia et al., 2013) .................................... 25

Figure 3.32 Representative core photos for one of the study wells indicating
multiple fracture forms and a propensity for horizontal discontinuities,
and an image of the characteristic beef from (Zanella and Cobbold, 2012) ....... 26

Figure 3.33 CT scan data for one of the study wells showing the presence of a
possible beef fracture in the core, as well as complex vertical to
sub-vertical fractures, and the presence of fossil fragments............................... 26

Figure 4.11 Results of the Normalization of the gamma ray log for wells A, B, C, G,
and H. which is representative of the normalization that was performed
for the utilized logs ............................................................................................... 30

Figure 4.12 Petrographic slides of the Vaca Muerta with calcite colored
red with Alizarin Red S. and all of the same scale. A) Upper Vaca Muerta
micritic calcite matrix B) Middle Vaca Muerta calcareous/argillaceous
mudstone C) Volcanic ash bed from the middle Vaca D) Lower Vaca
Muerta (Williams et al., 2014) .............................................................................. 31

Figure 4.13 Estimates of clay volume from the XRD (both in table form and as a
Data point on the log), Steiber estimate, and lithoscanner. The
Lithoscanner and XRD are in relatively close agreement, and the Steiber
method tends to frequently overestimate shale volume .................................... 33

Figure 4.14 Hydrogen Index and Oxygen Index with the overlay for a modified Van
Krevelin diagram showing the location of cuttings from Well G plotting as
type II kerogen, indicating an algal origin (Provided by Wintershall) .................. 36

Figure 4.15 Comparison of the preferred Schmoker and [U] methods with TOC
from cuttings in Well G ......................................................................................... 37

viii
Figure 4.16 Log responses over the Vaca Muerta interval for gamma ray, bulk
density, resistivity, o p essio al so i , d a i You g s Modulus Ed),
a d d a i Poisso s ‘atio d). The upper Vaca Muerta is indicated
with a grey color, and the lower Vaca Muerta with blue..................................... 38

Figure 4.21 Tool Design of Schlumberger Oil Based MicroImager (a) and Ultra
Sonic Borehole Imager (b) tools used for acquisition of borehole images
in Wells B and C (Schlumberger) .......................................................................... 40

Figure 4.22 Table listing differing tool configurations, with respect to number of
Arms and buttons, for several commercial electrical borehole-imaging
tools. The number of measurements directly correlates to the resolution of
the image (Lagraba et al., 2010) ........................................................................... 41

Figure 4.23 Schematic depicting the conversion of 3D wellbore imaging data

as acquired from the pads of the micro imager tool to the azimuthal
image which is interpreted based on the characteristic sinusoids
(Hurley, 2004) ....................................................................................................... 43

Figure 4.24 Examples of wellbore failure features in image logs from the data set:
a) Natural fracture observed in FMI showing enlargement at the peak
b) drilling induced tensile fractures in FMI, c) natural fracture observed in
OBMI and acoustic amplitude but not two-way travel time, and
d)borehole breakout as observed in acoustic log, and with a chaotic
response in the OBMI ........................................................................................... 45

Figure 4.31 Mapping of data points in color along with the position of the
e t oids desig ated ased o the use s defi itio of the u e of
desired classes ...................................................................................................... 47

Figure 4.32 Fall Off plot for the cluster analysis showing decreasing
cumulative distance with subsequent model run ................................................ 48

Figure 4.33 Silhouette plot showing the distance ratio between the data in
a given class and the adjoining class. Negative values are indicative of
data which may equally have been assigned to another class ............................ 48

Figure 5.21 Location map depicting the locations of the 4 wells with borehole
images, next to a cross sectional view of how much of the Vaca
Muerta was imaged .............................................................................................. 53

Figure 5.22 Image Processing and Interpretation Workflow as sequentially

performed for Wells B, C, and G........................................................................... 54

ix
Figure 5.23 Natural Fracture interpretation for Well C showing resistive
Fractures (dark blue) as well as high confidence discontinuities in light
blue and mid confidence discontinuities in red ................................................... 56

Figure 5.24 Natural Fracture Orientation for Well B. The dark blue is a single
resistive fracture, and the red and light blue represents the high
and mid confidence discontinuities ..................................................................... 57

Figure 5.25 Example Formation Snaps. Representative sections from the 3

formations Imaged in Well G and their individual trace in the FMI image .......... 58

Figure 5.26 Concretion in the Lower Vaca Muerta which has caused deformation
in the underlying sediment .................................................................................. 59

Figure 5.27 Rotation and distortion of the drilling induced fractures as they
Interact with individual bedding planes in the Upper Vaca Muerta, with
adjacent diagram for emphasis ............................................................................ 59

Figure 5.28 Stereonet plots of the strike of the fractures interpreted in

Well G: a) Resistive fractures, b) High- (dark green) and Mid-
(light green) confidence discontinuities, c) Conductive fractures ....................... 60

Figure 5.31 Wellbore Failure results indicating the principal stress

orientations: a) Induced fractures from Well G, b) Borehole breakout seen
in Well C, c) in contrast, the induced fracture interpretation
from Well B shows no alignment and reflects the insufficient quality
of the interpretation............................................................................................. 62

Figure 5.41 Stereonet plots for resistive and conductive fractures interpreted
in the borehole image alongside image interpretation layout. The
stereonets for each of the three identified fracture zones indicate that
all three zones have the same orientations present zones ................................. 63

Figure 5.42 Conductive and resistive bed terminations of single fractures in

the middle Vaca Muerta ....................................................................................... 63

Figure 5.43 Drilling enhancement at the fracture tip and abrupt fracture
termination at bedding interface from the middle Vaca Muerta ........................ 64

Figure 5.44 Suggested interpretation for the relative timing of four natural
fractures in the middle Vaca Muerta based on cross cutting. The black, low
angle fractures interpreted in Time 4 consist of two conductive fracture
and two features which appear to be conductive fractures but for
which the trace is less certain .............................................................................. 64

x
Figure 5.45 Stereogram and Mohr failure analysis for the identified natural
Fractures with the in-situ stress field estimations showing that the
fractures are stable. The natural fractures are plotted as small red
diamonds on the Mohr diagram, and as small black diamonds
on the stereogram ................................................................................................ 66

Figure 5.46 Stereogram and Mohr failure diagram for increased pore pressure
and minimum horizontal stress showing tensile failure initiation once
Pressure was increased by 2000 psi but other stress and rock strength
values were held constant.................................................................................... 67

Figure 5.47 Shear failure envelope reached in Trap Tester model with
Unconfined Cohesive Strength reduced to 400 psi .............................................. 68

Figure 5.51 Comparison of a fracture trace observed in the FMI with CT scan
photos from the whole core, inner core, and outer core .................................... 70

Figure 5.52 Comparison of the FMI with a likely beef identified using the CT
scan pictures, showing a highly resistive bed in FMI that is likely slightly
offset due to an imperfect depth match between the image log and the
core depths ........................................................................................................... 71

Figure 5.53 Comparison showing the lack of agreement between nodular trace
feature count in the FMI interpretation with the count of possible beef in
the CT scan data. High beef density in the upper CT section does not
correspond to an increase in nodular feature count in the count log ................. 72

Figure 5.61 The highest fracture intensity zone corresponds to a very brief preferred
fast shear orientation, as well as a small local maximum in the time and
slowness anisotropy ............................................................................................. 74

Figure 5.62 The zone of fracture occurrence as indicated by the F.I. log
corresponds qualitatively to an increase in Stoneley transit time ....................... 75

Figure 5.63 Stoneley Slowness plotted with kerogen estimates from log and
cuttings data, depicting a better match between Stoneley response and
kerogen content rather than the presence of healed fractures .......................... 76

Figure 6.11 Cluster analysis workflow ..................................................................................... 78

Figure 6.21 Example from Well G for all logs which were used as input for the
three cluster models............................................................................................. 79

xi
Figure 6.22 Compositionally based log model, as Well G had not yet been
logged, only Wells A, B, and C were modeled ...................................................... 80

Figure 6.23 XRD for Wells A, B, C, with red line indicating approximate
location of 30% clay cutoff, showing that the clay content in this area
remains nearly constantly below 30% (Wintershall) ........................................... 81

Figure 6.24 Fabric based log model, with data from Well G also included............................. 83

Figure 6.25 The mechanical model which involves both texturally and
compositionally sensitive logs and was built to identify mechanically
similar zones, and to attempt to resolve mechanically significant thin layers .... 84

Figure 6.26 Input logs to the mechanical model from Well G next to two
versions of the model, built from Well G logs, or built from Well B logs
but tagged to Well G ............................................................................................ 86

Figure 6.31 Mechanical cluster model with the frequency and type logs used to
describe the cluster class occurrences, with fracture intensity, and the
difference between the cluster frequency (C.F.) and cluster/interval
type (I.T.) log for Well G ....................................................................................... 88

Figure 6.32 Cluster by occurrence depth for Well G. Scale is arbitrary and was
Chosen for visualization purposes. For the same reason, class 3 was
colored white as opposed to light blue ................................................................ 89

Figure 6.33 Final cluster models, with an interpretation of six zones (Zx) of similar
stacking patterns. The discriminations based on stacking pattern are
consistent in the compositional and fabric models for Well B ............................ 90

Figure 6.34 The 6 cluster zones identified in Well B compared to well log and XRD data ...... 90

Figure 6.35 Correlation of the cluster zones for the additional wells in the study area ........ 91

Figure 6.41 Comparison of the different gamma curves used to test the cluster
model and the resultant output. The model initiates randomly with
each rerun, but the stacking patterns are visibly similar ..................................... 93

Figure 6.42 Correlation between anomalous feature identified in image log and
well log response and subsequent cluster class discrimination .......................... 94

Figure 7.11 Comparison of natural fracture orientations recognized in the study area
analysis and an independent study conducted by YPF and Chevron in
another part of the basin (Rimedio et al., 2015) .................................................. 96

xii
Figure 7.12 Geometric representation of hydraulic fracture growth in the
Barnett where natural fracture reactivation is contributing to
fracture complexity, and a larger stimulated area as seen in the
microseismic inset (b) (Gale et al., 2007) ............................................................. 97

Figure 7.21 Correlation of the cluster class zones between the study wells ........................ 100

Figure 7.22 Calcareous concretions in the image log that appear as bright white
spots which corresponds to the black cluster group recognized
by the mechanical cluster model. The micro-resistivity log is also
displayed and shows a response to the presence of the clusters...................... 101

Figure 7.23 High degree of lamination in the light green cluster class seen in
two examples on the borehole image (scales are same) and a
comparison of the light green cluster occurrence with the C.F. log .................. 102

Figure 7.24 Cluster Output alongside sonic logs used to characterize

Formation anisotropy ......................................................................................... 103

Figure 7.25 XRD data for Wells A, B, and C split into Lower and Upper Vaca
Muerta depicting the higher proportion of carbonate in the Upper Vaca
Muerta relative to the Lower ............................................................................. 104

Figure 7.26 Cluster class zones in Well G and corresponding mineralogy and
TOC estimates, along resistivity drop that occurs at the limit between
Zone 2 and Zone 4 .............................................................................................. 105

Figure 7.27 Subset of the seismic data showing the bright, relatively
planar and parallel nature of the Base Vaca Muerta when compared to
the middle Vaca Muerta. Cyan reflector is approximation of top Vaca
Muerta ................................................................................................................ 105

Figure 7.28 The location of Well G, Well A, and Well C on a SW-NE transect of
which Well B is off trend .................................................................................... 106

Figure 7.29 Possible explanation for the additional zone is Well B is that it was
Drilled closer to position y, including more aggradational topsets,
relative to position x ........................................................................................... 106

Figure 8.21 Casing restriction related to shear movement recorded in the

Loma Campana Field (Rimedio et al., 2015)....................................................... 112

xiii
 LIST OF TABLES

Table 4.1 Data table for image log and cluster analysis, where a * symbol indicates
data of questionable quality, but which were used for the analysis
he e eeded. Ŧ i di ates data available, but of too poor quality to be of
use to the analysis ................................................................................................ 30

Table 5.1 Different types of borehole image available in the study area, and
the extent of imaging. Well B had an addition 600 m run above the
reservoir interval in the same imaging run, however, gas in the wellbore
caused image quality issues above the Vaca Muerta level .................................. 53

Table 5.2 Depth correcting performed on core photos in order to compare

between the core photo, CT scan images, image log, and wireline logs ............. 55

Table 5.3 Stress and rock strength values input to calculate fracture stability .................. 65

Table 7.1 Evaluation parameters for an unconventional reservoir ................................... 100

Table 7.2 Zones defined by the mechanical cluster model ............................................... 105

Table A.1 Statistics from cluster output organized by cluster class from mechanical
model .................................................................................................................. 123

xiv
 LIST OF SYMBOLS

DTCO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Compressional Sonic Slowness

DTSM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shear Sonic Slowness

Ed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Dynamic You g s Modulus

Es. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Static You g s Modulus

GR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Gamma Ray

K. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Permeability or Bulk Modulus

NPOR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Neutron Porosity

Pp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Pore Pressure

RHOZ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bulk Density

[U] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concentration of Uranium

Vp. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Compressional Sonic Velocity

Vs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Shear Sonic Velocity

Vp/Vs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ratio of Compressional to Shear Sonic Velocities

vd. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Poisso s ‘atio

vs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Static Poisso s ‘atio

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zone Color for Upper Vaca Muerta

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zone Color for Lower Vaca Muerta

xv
 ACKNOWLEDGEMENTS

To my co-advisors, Steve Sonnenberg and Tom Davis, I am so grateful for the support and
input that I received over the course of my time at Mines. The Geology Department provided me
with classes that allowed me to improve my technical skills, and to do it inside well equipped
labs, or in phenomenal field locations. I will be proud to be a graduate of this Department.

Wintershall have supported me in my work, by providing data, technical advice and a myriad of
opportunities. Harald Karg acted on my committee and provided feedback that was welcome and
insightful. In addition, Heather Davey and Christian Hanitzsch were champions of this project and
helped me to spend 6 months of my time working at the office in Kassel, and benefitting from
the expertise of the team there. They will serve as examples to me of how to comport myself in
a professional environment for the rest of my career. To top it off, Heather was a fantastic travel
companion on several adventures.

I had high hopes about coming down to Golden and they have been exceeded time and time
again thanks, in large part, to the incredible RCP group that Tom has built over the last 30 years.
Tom himself is the best of us, and I will forever be appreciative of his acceptance of me into the
group, his indulgence of my ambitions, and his humor throughout everything. Ma e it s
something about having shared Prairie roots, I am honored to have had Tom as a mentor.

The RCP Sponsors who support the consortium, and who make the trek to the biannual meeting
deserve acknowledgement. The lead up to the meeting was always stressful, but the experience
was worth it. I feel personally motivated in my work by imagining the questions that I might be
asked up on that stage. I also want to thank the administrators and RCP staff who helped to
coordinate my special requests, software needs, and travel plans.

Another Tom who is worthy of mention is Tom Bratton. RCP is privileged to have him as a
pa ti ipa t a d the e is t a stude t i the o so tiu ho has t e efited f o his e pe tise,
and his endless willingness to talk science. I am thankful for the many emails, meetings, and
phone calls from Germany that he participated in. He inspires students to explore further areas
of science and to be appropriately critical of the data and the processes.

xvi
Sue Jackson was an ear to me throughout this experience. Her critical eye, her organizational
skills, and her willingness to engage were such valuable assets to me. Not only that, but I enjoyed
our conversations and her input will guide me in my day-to-day life as much as for presentations
and reports in the future.

My colleagues in RCP and in Geology were a varied bunch. The time that I spent chatting with
them, sometimes late at night in the office, sometimes over a beer, were the best times that I
had at Mines. Wherever we all end up, we have shared this experience together and I am happy
to have had that with each of you.

This thesis was frustrating and fun in equal measures, though more fun now that it is said and
done. My family listened to me complain or exalt whenever I needed a chat, and I want to thank
them for always picking up the phone, whether I was calling from Golden, Buenos Aires, or Kassel.
It does t atte he e I a , I k o that I ha e ou suppo t, a d that e a les e to pu sue
goals and to live the life that I want to live. I love you all.

xvii
 CHAPTER 1

INTRODUCTION

The Neuquén Basin is located at the base of the Andes in west-central Argentina (Figure 1), with
a western margin that passes just over the border into central Chile. It occupies an area of
approximately 120 000 km2 (75 000 mi2) between 30- ˚“ Ho ell et al., 2005; Monti et al., 2013).
The Neu ué is the ost p olifi of A ge ti a s pet olife ous asi s Lega eta a d Villa ,
a d the Va a Mue ta is the i hest sou e ithi the asi , f o hi h up to % of A ge ti a s
current produced hydrocarbons are sourced (Urien et al., 1995; Kietzmann et al., 2011). There
have been 220 wells drilled in the Vaca Muerta as of May 2014, 90% of which are vertical
completions with an average of 4-5 hydraulic fracture stages (Ejofodomi et al., 2014). As
Argentina becomes the focus of foreign unconventional exploration interest, the Vaca Muerta
pla is of p i e i po ta e to A ge ti a s e e g pote tial, a d u de sta di g the
heterogeneities inherent to the reservoir will be essential to creating effective development
strategies (Suarez-Rivera et al., 2005).

Figure 1.01 Location Map of the Neuquén Basin with insert showing major structural features in
the vicinity (Modified from Zavala et al., 2006 and Zeller et al., 2015).

1
Natural fractures present within the reservoir interval are likely candidates for reactivation during
stimulation, and are the focus of this research.

The analysis includes a discussion of available log data, a review of the geomechanical principles
which describe fracture occurrence, the analysis of borehole image logs for fracture
characterization, and the construction of a mechanical stratigraphy model which is related to the
fracture stratigraphy.

1.1 Project Goals and Available Data

The data for this project were provided by Wintershall Holding, GmbH as part of an agreement
with the Reservoir Characterization Project Research Consortium. The data set available is from
4 blocks in the Neuquén basin, and consists of 5 wells and a 600 km 2 narrow azimuth 3D seismic
survey (Figure 1.11). There are an additional number of wells further offset from the primary area
of interest for which a limited data set was provided and for which the data are strictly
confidential, but which has been made available to the RCP consortium for internal uncertainty
analysis.

Shale plays are heterogeneous in their nature, and in order to properly be characterized, both in
terms of lateral and vertical heterogeneity, there must be a large, diverse, and complete data set
(Passey et al., 2010). Although such a data set is ideal, it is not always feasible, and therefore
analysis requires the incorporation of rigorous workflows, and diverse approaches to properly
analyze the data, and to integrate across disciplines for optimal understanding. The collaboration
between Wintershall and the RCP group is designed to incorporate aspects of geology,
geophysics, and geomechanics in order to create a holistic reservoir characterization. Using
electric and borehole image logs for the geologic research allows for a quantification of the
vertical heterogeneities in the reservoir, and a spatial component to the heterogeneity analysis
is introduced by a coincident geophysical analysis.

Of the five wells that are fully available to the RCP consortium, the data suite consists of wireline
logs (gamma ray, bulk density, resistivity, photo electric, etc.), enhanced logs such as image logs,
nuclear magnetic resonance logs, and lithology logs such as the elemental capture spectroscopy

2
sonde. Additional information about these wells, such as core photos, x-ray diffraction data, daily
drilling reports, fluid analysis, and pyrolysis data was provided to the RCP group where available.
Four of the wells (A, B, C, and H)) were drilled from 2013 to 2014 in cooperation with a partner
company, and the fifth well (G) was drilled in 2015 in the block operated by Wintershall.

Because well numbers and production history in the Vaca Muerta are relatively limited compared
to analogous North American plays, there are still significant uncertainties with regards to
optimized completion strategy. The significant thickness of the interval with high organic content
suggests that it may be possible to develop the play using either horizontal or vertical wells. The
heterogeneous nature of the interval also suggests that depending on the depth of the
stimulation stage or landing zone, the most effective completion design may vary and this has
been the subject of recent investigations in the Neuquén Basin (Ejofodomi et al., 2014). The
ultimate goal of this study is to examine how to achieve an applicable model of the variable rock
properties within the Vaca Muerta interval, to examine the heterogeneity of rock properties as
they relate to fracture analysis, and to combine these concepts in the framework of stimulation
efficacy, particularly as it relates to the development of a complex fracture system.

Figure 1.11 Data Locations for the study area including wells and seismic survey, as well as a
relevant data audit. Provided by Wintershall and altered for confidentiality.

3
 CHAPTER 2

GEOLOGY OF THE NEUQUÉN BASIN

The Neuquén basin is a foreland basin with a thick package of sediment fill which includes
several organic rich potential source units in a series of transgressive-regressive cycles (Urien and
Zambrano, 1994). The basin is bounded by the Andes Mountains to the west, the North
Patagonian Massif igneous platform to the south, and by the Sierra Pintada Massif to the north-
east (Howell et al., 2005). Directly adjacent to the Andes, the Neuquén Basin is highly deformed
with several thrust related features towards its western margin, but relatively un-deformed to
the east on the gently dipping Neuquén embayment (Figure 2.1) (Howell et al., 2005). It is in the
embayment where basin fill is the thickest and where the majority of hydrocarbon fields occur,
with the sediment package reaching thicknesses of over 7000 m (22000 ft) (Garcia et al., 2013).

Figure 2.01 Structural domains and large-scale structural features of the Neuquén Basin
(Modified from Sagasti Et al., 2014).

4
2.1 Structural Development

Three phases led to the development of the present day physical expression of the
Neuquén basin (Figure 2.11):

1) In the late-Triassic Synrift Phase, intraplate rifting related to the collapse of the
Gondwana Orogen led to the formation of a series of rift basins down the center of the modern
day South American continent (Hogg, 1993). The basin fill at this time consists of the primarily
continental deposits of the Precuyo Group, whose distribution was controlled by the geometry
of large, transcurrent basement faults (Kietzmann et al., 2014).

2) During the Early Jurassic to Early Cretaceous Post-rift Phase, subduction of the Nazca
plate below the South American plate began in response to convergence caused by rifting in the
proto-Atlantic and the breakup of the Gondwana supercontinent (Howell et al., 2005). This
convergence led to the development of a volcanic arc alo g the asi s este edge a d a k-
arc subsidence. Flooding of the basin occurred, although water level was highly variable and the
basin was only periodically connected with the proto-Pacific as the Andean volcanic arc evolved.
The thermal subsidence in the back-arc occurred slowly at a fairly constant rate through the early
Cretaceous with only minor influence from localized tectonic activity (Mitchum and Uliana, 1985;
Kietzmann et al., 2014).

3) At the end of the Early Cretaceous the angle of subduction of the Nazca plate shallowed
significantly, prompted by a decrease in Atlantic spreading rate, and reorganization of tectonic
plates in the Pacific. This change in angle resulted in a switch to a compressional tectonic regime
and the inversion of several pre-existing extensional features, converting the Neuquén into a
foreland basin (Foreland Phase) (Howell et al., 2005; Zeller et al., 2015). Several stages of
compression and relaxation have been ongoing since the Cretaceous through the Cenozoic and
Tertiary, as continuing growth of the Andes Mountains has led to uplift of the western portion of
the basin (Kietzmann et al., 2014; Reijenstein et al., 2015).

The thermal subsidence sag (Post rift) phase of the Late Jurassic to Early Cretaceous is
currently the most actively pursued oil and gas target interval in the Neuquén basin (Howell et

5
al., 2005; Urien et al., 1995). This section of the fill is characterized by five major transgressive-
regressive cycles bounded by unconformities, which can be further subdivided into several
higher-order cycles (Sagasti et al., 2014; Licitra et al., 2015). During the third of the major
transgressive-regressive cycles – the A di le – the Mendoza Group, including the
continental sandstones of the Tordillo Formation, the deep marine calcareous shales of the Vaca
Muerta Formation, and the limestones of the Quintuco Formation, was deposited (Figure 2.12).
Since the rates of thermal subsidence and sediment influx were relatively constant through the
mid-Mesozoic, the transgressive-regressive sequences are attributed primarily to changes in
eustatic sea-level or occasional pulses of tectonic inversion (Mitchum and Uliana, 1985; Garcia et
al., 2013; Sagasti et al., 2014).

Figure 2.11 Schematic depicting the three tectonic phases key to the evolution of the Neuquén
basin with relevant captions. The dark grey represents deep marine deposits, the light grey zones
are shallow marine deposits, and the black is continental material. The Postrift Phase is the period
of Vaca Muerta and Quintuco deposition (Howell et al., 2005).

6
Figure 2.12 Stratigraphic column for the Neuquén basin for the region surrounding the
Wintershall Block of Interest (Simplified from Howell et al., 2005).

2.2 The Vaca Muerta Formation

The Vaca Muerta Formation consists of dark grey, organic rich shales deposited in a deep
water, low energy, and anoxic, marine environment between the Tithonian to Valanginian
periods, approximately 150 -135 million years ago (Garcia et al., 2013: Licitra et al., 2015). Depth

7
to top of the Formation ranges between outcrop to 3000 m (Garcia et al., 2013), and averages
2700 m in the study area. The Vaca Muerta is a time-transgressive litho-unit that was deposited
as part of a prograding sequence system that moved north-northwestward along a shallow ramp
(>2.5°), so that the Formation is younger to the West.

The formation can be subdivided based on mineralogy and interpreted depositional

environment (Garcia et al., 2013). The marls and shales of the lower Vaca Muerta represent the
most distal deposits, and correspond with a high GR response, low bulk density, and relatively
high sonic log response. The relation of these log profiles to kerogen content is discussed in
further detail in chapter 4. The seismic reflectors seen in the lower Vaca Muerta are sub-
horizontal and parallel.

The middle portion of the interval has increasingly lower organic content, and increasingly
higher carbonate to clastic ratios. The uppermost Vaca Muerta reflects carbonate deposition and
growth in an increasingly more proximal and less restricted marine environment (Garcia et al.,
2013). The GR log decreases upwards, as does the compressional sonic log, and the resistivity
and bulk density logs increase. When the seismic is flattened on an overlying surface, the
reflectors in the upper Vaca Muerta begin to exhibit low angle clinoform geometries (Figure
2.21).

The depositional model is therefore of a low angle, mixed clastic-carbonate ramp, with
the upper portion being primarily carbonate, and the lower portion primarily clastic (Figure 2.22)
(Mitchum and Uliana, 1985; Kietzmann et al., 2014). The siliciclastic material in the system was
sourced from the Pre-Andean magmatic arc and the carbonate was primarily produced in-situ
(Simo et al., 2013). The relative proportion of the source materials was largely controlled by
accommodation space and the resulting transport capacity of offshore currents (Zeller et al.,
2015).

Both the Vaca Muerta and Quintuco are over-pressured, with pressure gradients in the
lower Vaca Muerta ranging between 0.65 psi/ft up to 1.02 psi/ft (Garcia et al., 2013; Fantin et al.,
2014). In terms of brittleness as defined by either Jarvie et al. o Wa g a d Gale s

8
brittleness indices, the Vaca Muerta is sufficiently brittle for effective hydraulic stimulation, with
clay content consistently below 30% (Garcia et al., 2013).

Figure 2.21 The Vaca Muerta interval in the area of interest: a) Representative gamma ray, results
for clastic, carbonate, and clay weight fractions from the lithoscanner, resistivity, and dynamic
You g s Modulus a d Poisso s atio. The g e , uppe i te al is the Uppe Va a Mue ta as
defi ed fo the pu pose of this a al sis, a d the lue lo e po tio is the Lo e Va a Mue ta.
b) Representative seismic geometry flattened on top Quintuco reflector and compressed laterally
for emphasis.

The Vaca Muerta is naturally fractured, and frequently observed are characteristic
fi ous, eddi g pa allel, al ite filled ei s alled BEEF“ Co old et al., 2013). Calcareous
nodules can be observed in core and in borehole images which are evidence of the diagenetic
processes which have altered the Formation since deposition, and an additional layer of
complexity is added by the presence of nearby volcanics, which have introduced ash beds and
intrusive igneous sills to the Neuquén basin (Fantin et al., 2014). The igneous intrusions are not
observed in the data, although ash beds, nodules, and BEEFs are. These features may have
implications for hydraulic fracture propagation and are important considerations for completion
design (Sagasti et al., 2014).

2.3 The Quintuco Formation

The Quintuco overlies the Vaca Muerta in the RCP study area, and consists of marls and
limestones which were deposited in a proximal ramp or open shelf environment, coevally with

9
the distal shales of the Vaca Muerta (Zeller et al., 2015). As such, the contact between the Vaca
Muerta and Quintuco is diachronous and can be difficult to define (Sagasti et al., 2014). Because
of the lithostratigraphic nature of the definition of the Vaca Muerta and Quintuco Formations,
the Formations should be considered as part of larger-scale system. If the Vaca Muerta
represents the bottom sets of the prograding clinoform system, where the organic content is the
highest, and would have been the distal ramp setting, then the Quintuco is the organic-poor,
carbonate rich, shallow marine equivalent in the topsets (Sagasti et al., 2014; Reijenstein et al.,
2015). Although it is not the focus of the RCP – Wintershall collaboration, the Quintuco is a
hydrocarbon reservoir and produces in several areas of the Neuquén Basin (Licitra et al., 2015).

Figure 2.22 Schematic cross section of the Vaca Muerta and Quintuco system in the Central
Neuquén Basin sector which is representative of the geometry in the area of study (Modified
from Kietzmann et al., 2014).

2.4 The Vaca Muerta Petroleum System

Reservoir quality in unconventional plays is commonly evaluated based on thickness,

organic richness and the thermal maturity of the organic content, mineralogy, geomechanical
properties that are favorable for hydraulic stimulation, porosity, and permeability (Fantin et al.,
2014; Sagasti et al., 2014). In all of these regards, the Vaca Muerta makes an excellent
unconventional reservoir.

10
 As a result of deposition in a quiescent, anoxic environment, organic matter preservation
in the Vaca Muerta is exceptionally high, with total organic carbon (TOC) content averaging 4%
and reaching a maximum of 14% (Rodrigues et al., 2009; Garcia et al., 2013; Fantin et al., 2014).
The TOC is highest at the base of the Vaca Muerta, and decreases upward through the interval in
parallel with the decreasing gamma ray log signature. The organic matter present is primarily
type II kerogen, derived from amorphous marine algae (pyrolysis data for study area is shown in
Chapter 4), and the oil found is generally light (35-45°) and low in paraffin and ashphaltene
content (Garcia et al., 2013).

Maturation of the Vaca Muerta organic matter was primarily due to burial, although
additional heat input from nearby volcanics may have contributed (Rodrigues et al., 2009). The
majority of the embayment area is in the oil window, with vitrinite reflectance values between
0.6-1.5, and moving westwards transitions through the condensate to dry gas windows further
into the foothills of the Andes (Rodrigues et al., 2009).

11
 CHAPTER 3

GEOMECHANICAL CHARACTERIZATION

In unconventional reservoirs where permeabilities are low, some type of additional

stimulation is required to allow for economic production (Warpinski et al., 2009). In this case, the
mechanical behavior of the reservoir as it relates to the ability of natural and induced fractures
to sustain hydraulic conductivity pathways may be the primary factor controlling play economics.
In order to assess the optimal completions design for such a reservoir, a thorough
characterization of the mechanical properties of the formation, the stress state at the reservoir
level, and the natural discontinuities already present at depth, is essential (Gale et al., 2009).

Such a characterization must begin with a discussion of the effects of stress and strain,
he e st ess des i es the fo es a ti g i the su su fa e, a d st ai des i es the
deformation of a given material in response to a stress (Fox et al., 2013). The relationship
between these concepts, the means by which they are evaluated and described, the concepts of
formation failure in response to stress, and the specific properties of the Vaca Muerta are
discussed herein.

3.1 Geomechanical Reservoir Characterization

A geomechanical model describes the magnitude and orientations of the principal

stresses, as well as pore pressure, the elastic properties of a rock, and the rock strength over an
area of interest (Barton and Moos, 2010). The estimation of these parameters can be done using
several methods and incorporating data from core tests, well logs, and seismic surveys.

The three principal stresses define the reservoir stress depending on their relative
magnitudes and orientations (Figure 3.1). There are three defined stress states which are
commonly described, and which will be a primary control on the resulting reservoir failure. These
stress regimes can be regionally extensive, but locally variable, as they are affected by all scales
of geologic heterogeneity (Fox et al., 2013).

To estimate stress magnitudes, there are several possible methods. Micro-fracturing

tests, such as leak-off tests or diagnostic fracture injection tests may be used to obtain a

12
magnitude for the minimum horizontal stress (Barton and Moos, 2010). Vertical stress is assumed
to be a principal stress in rock formations that are buried at a sufficient depth, although that is
not always the case and was previously a common pitfall in geomechanical modelling. Vertical
stress is derived primarily from density and sonic well logs (Fjaer et al., 2008; Osorio and Muzzio,
2013). The maximum horizontal stress magnitude is obtained from wellbore failure analysis and
forward modelling (Barton and Moos, 2010).

The stress field immediately around the wellbore is altered by the drilling process and
then transition to the far field stress state away from the wellbore, and both may be relevant to
a geomechanical study. It should be noted that there is no consensus within the geomechanics
community on notation, and that different symbols are frequently used to describe the same
parameters, or the same symbols may be used to describe different parameters. Effort has been
made to remain consistent throughout this thesis, but in the figures sourced from external
papers, there may be some discrepancy. In particular, the notation for the principal stresses
commonly oscillates between Sv, SHmax, Shmin, a d σ1, σ2, σ3. The sigma notation makes no
reference to the orientation of the stresses and simply describes the relative magnitudes,
whereas the S notation describes orientations and requires that the relative magnitudes be
specified.

Figure 3.11 Stress fields and their relative magnitudes (Fox et al., 2013).

13
3.12 Elastic Properties

Elastic properties which relate to the deformation and failure of a rock material are
estimated from well logs or measured on core samples in a laboratory. These are parameters
which have been derived in order to quantify the relationship between stress and strain, using
linear elastic assumptions (Fjaer et al., . You g s odulus E des i es the atio of applied
stress in one direction, to the resulting strain in the same direction. It is frequently used as a
proxy for a rock s stiff ess , ith the assu ptio that a high E ill o espo d to a g eate
required applied stress to induce a given strain (Fox et al., 201 . Poisso s ‘atio des i es the
ratio of lateral expansion to axial contraction, bulk modulus (K) relates volumetric strain to the
mean hydrostatic stress, and shear modulus (G) is defined as the ability of a material to resist
shear deformation (Fjaer et al., . Togethe ith the La é o sta t λ , the shea odulus
defi es You g s Modulus, Poisso s atio, a d the ulk odulus, ea i g that if a t o of the
moduli are known, then the others can be calculated (Equations 3.1 through 3.3).

(3.1)

(3.2)

(3.3)

The elastic moduli as estimated from well logs represent dynamic estimates of the
parameters, with the values being derived from the velocities of elastic waves and the density of
the rock (Fjaer et al., 2008). Static moduli are measured in rock mechanical tests, such as uniaxial

14
or triaxial tests in a controlled laboratory setting. The static estimates of elastic moduli are
frequently much smaller than the dynamic moduli, and this is commonly attributed to the
differences in how the properties are measured. Static moduli are measured using stress-strain
tests on core plugs in a laboratory, and such tests inherently involve large strain magnitudes, but
relatively low strain rates. Conversely, dynamic moduli are estimated using high frequency sonic
measurements (from either laboratory tests or from the sonic wireline logs), where there is a
relatively small strain but a large strain rate (Crain and Holgate, 2014). The differing values for
the static and dynamic properties also relates to fluid saturations, the external stress, borehole
conditions during logging, fractures, and rock fabric (Barree et al., 2009; Fox et al., 2013).

Depositional and diagenetic processes cause a layered nature that is characteristic of

mudrocks, although the tendency of clay minerals to align in a preferred orientation may be
inhibited in some shales by the presence of silica and carbonate minerals (Hart et al., 2013). This
layering creates rock properties that vary much more rapidly vertically relative to laterally (Hart
et al., 2013) The tools used to measure rock properties cannot account for this difference and
the result is that the measurements are very dependent on orientation– in particular, that
measured properties will be different in the orientation perpendicular to the bedding plane
relative to the orientation parallel to bedding (Sullivan Glaser et al., 2014). The magnitude of the
difference between the perpendicular rock properties to parallel rock properties is related to the
linear anisotropic behavior of the rock. In horizontally layered media the anisotropic behavior is
commonly transversely isotropic about a vertical axis of symmetry, (VTI – vertical transverse
isotropy, Figure 3.12) (Ejofodomi et al., 2014). In a VTI medium, linear elastic rock behavior is
described the e ti al a d ho izo tal You g s odulus, the e ti al a d ho izo tal Poisso s
ration, and the vertical shear modulus (Ejofodomi et al., 2014).

The laminated nature of shales is also significant to the hydraulic stimulation process as
the interface between layers of differing rock properties is a surface of stress concentration
where failure will preferentially occur (Davey, 2012).

15
Figure 3.12 Block diagrams depicting layering that creates a VTI medium, or vertical fracturing
that creates an HTI medium, and how these may be combined in mudstones (modified from Hart
et al., 2013).

3.13 Pore Pressure

The oid spa e et ee i e al g ai s ill ha e a sig ifi a t i pa t o a o k s failu e

behavior and elastic response depending on the fluid that is occupying the pore space, and the
resultant fluid pressure which opposes the three principal stresses (Fjaer et al., 2008; Fox et al.,
2013). Pore pressure may be measured directly or predicted from well logs or seismic data using
previously established effective stress models (Eaton, 1969; Bowers, 1994). However, pore
pressure prediction is complicated for unconventional reservoirs, as the majority of techniques
were developed for relatively low stress conventional systems, and fail to account for the

16
complicated tectonic effects which may have a significant impact on the local stress regime (Fox
et al., 2013). This is true for the stress profiles as well, and the assumptions that underlie the
estimation of stresses in conventional reservoirs may not apply at all in unconventional reservoirs
and be misleading (Barree et al., 2009). Ideally, when modelling pore pressures, the prediction
method would be chosen based on the pressure generation mechanism, and would be calibrated
to local pressure test results (Ejofodomi et al., 2014).

Defining a reasonable pore pressure for a formation is important for the geomechanical
model because it is needed in order to determine the effective stress. The effective stress is a
more accurate way to describe reservoir stresses, and involves the total stress but also accounts
for the amount of that total stress which is accommodated by the fluid pressure (Fjaer et al.,
2008).

3.2 Fractures and Rock Strength

Having established that rocks in the reservoir are subject to a variety of stresses, there
are circumstances in which the o k s st e gth a ot ithsta d the fo es a ti g upo it.
Eventually, for a given set of conditions, a rock body may surpass its strength limit and fail, either
in compression or tension (Fox et al., 2013). The plane along which the failure occurs is a
f a tu e , hi h ill eithe e shea o te sile depe di g o the st ess hi h i du ed the failu e
(Figure 3.21a and 3.21b).

Depending on the stress, failure can be initiated in the reservoir at any point. In addition
to the existing in-situ stresses causing natural failures, the drilling of a well causes a concentration
of stress directly around the wellbore and commonly causes borehole failure (Barton and Moos,
2010), and during hydraulic stimulation, failure is being very purposely initiated to create
permeability. Therefore, the three forms of failure that are of interest to this analysis are; failure
that has occurred in the past and created natural fractures which are present in the reservoir,
failure that has been induced by drilling activity, and failure that is likely to occur during
stimulation.

17
 In addition to the stress and pressure regime present at depth, rock failure also depends
o i he e t o k p ope ties that o t ol the ate ial s a ilit to esist the force acting upon it,
which depe d o the o k s i he e t st e gth. In laboratory experiments, rock samples are
stressed to the point of failure in order to determine their strength (Fox et al., 2013). The failure
point, or breaking point, in tension is the tensile strength, and in compression, it is the
compressive strength (Nygaard and Hareland, 2007). Because core data is frequently missing,
empirical relationships have been derived to relate rock strength to the response of sonic or
neutron logs and such models are frequently calibrated with nearby core data where available
(Nygaard and Hareland, 2007). Rock failure is very commonly discussed using Mohr-Coulomb
failure analysis, which allows for simple and straightforward calculations and modelling of the
rock failure criteria (Figure 3.22).

Figure 3.21a. Block diagram depicting the relationship between the orientations of the principal
stresses and resultant failure (Bratton et al., 2006).

18
Figure 3.21b Alternative block diagram for tensile and shear failure orientations relative to the
maximum principal stress controlling their occurrence (Fjaer et al., 2008).

Figure 3.22 Components of a Mohr-Coulomb plot (Meissner, 1984).

19
3.21 Natural Fractures

Natural fractures may be tectonic or non-tectonic, vertical to sub-vertical, or fully

horizontal and bedding parallel (Gale et al., 2014). Fracture analysis is generally concerned with
the geometry of the fracture (length, aperture, azimuth), relative distributions, which may
involve examining fracture occurrence intensity, or other fracture features such as fill material,
level of connectivity, and fluid conductivity (Fox et al., 2013). Advanced seismic interpretation
may detect fracture systems, but the resolution of seismic data is not suitably high for the analysis
of individual fracture orientations, the relative extent of those individual fractures within a
system, or the spacing between them (Wennberg et al., 2006).

Early analysis of whether natural fractures are present and in what form, orientation, and
distribution they occur in will be vital to understanding later production. In particular, for low
permeability reservoirs, the key is how these fractures are likely to react to stimulation and
possible interactions with any fractures created by the hydraulic stimulation treatment. Studies
in North American unconventional reservoirs suggest that there is no guarantee that the
presence of natural fractures will be a benefit to production, as it is in the Bakken, and that
instead, they may be detrimental to well production, as appears to be true in the Barnett (Garcia
et al., 2014).

Critically stressed natural fractures are the subset of fractures which are active under a
fo atio s p ese t da st ess field, and are the most likely to contribute to reservoir
permeability (Barton and Zoback, 2002). Understanding the location of the natural fractures, and
which of these may be critically stressed is therefore essential to creating an effective stimulation
treatment. In areas where core data is missing or lacking, borehole images may provide a high
resolution alternative to study natural fractures.

3.22 Drilling Induced Fractures

During drilling, as material is removed, the surrounding rock must account for the stress that was
previously supported by the missing material and the introduction of additional fluid pressure
from the drilling mud (Fjaer et al., 2008). This results in a concentration of stress around the

20
borehole and, frequently, failure (Tingay et al., 2008). Drilling induced fractures (DIFs) occur as
the result of tensile rock failure around the wellbore, but this concentration can also result in
compressive failure in the form of borehole breakouts (Figure 3.23).

Figure 3.23 View of borehole failure in a vertical borehole with stress orientations depicted
(Modified from Wikel, 2011).

Drilling induced fractures preferentially initiate at a point of geologic heterogeneity,

where the rock strength is reduced, and will grow until they are limited by the stress/pressure
conditions, the presence of natural fractures, shear failure, bed slip, or if the mechanical
properties become sufficiently different so as to cause a switch to plastic, rather than elastic,
deformation (Fox et al., 2013). One of the most common factors in fracture containment is the
stress contrast at a boundary between two differing rock bodies (Fox et al., 2013). The larger the
stress contrast, the higher the likelihood that the fracture will cease to propagate and will instead
terminate at the boundary.

21
3.23 Stimulation Fractures

During hydraulic stimulation, a large volume of fluids are forced into the reservoir, in
order to increase the pore pressure, and to reduce the effective stress to the point where the
fluid pressure exceeds the minimum principal stress and the tensile strength of the rock, and in
doing so induces failure (Figure 3.24) (Fjaer et al., 2008). The stress concentration will be greatest
in the orientation parallel to the minimum stress, and so it is the magnitude of that minimum
stress which the pressure must overcome in order to initiate fracturing (Nygaard and Hareland,
2007).

Once fractures have been initiated, pumping is maintained to encourage fracture growth
which will continue in the direction of least resistance (Fjaer et al., 2008). Small solids may be
included in the stimulation fluid i o de to p op ope the eated f a tu es afte pu pi g
ceases and maintain the fracture conductivity. There are three components to fracture growth
and the ability of fractures to contribute to production: the fracture must be kept open in the
direction of the minimum stress to be maintained, the pressure gradient must allow for fluid to
flow through the fracture, and the pressure at the fracture tip must be kept high enough to
o e o e the o k s inherent resistance to breaking (Fjaer et al., 2008).

During stimulation, the reactivation of large-scale features can occur due to any number
of mechanisms; pressure induced slip, stress induced dilation, the creation of new fractures that
may link to the natural fractures (Fox et al., 2013). This possible reactivation may cause the pre-
existing fracture planes to act as a sink for energy and fluids, or as a baffle for fluid conduit if the
re-activation occurs in multiple orientations (Williams-Stroud et al., 2012). The creation of
complex fracture networks should preferentially depend on the critically stressed fractures, since
these will be the first to reactivate and, as stated, are the best contributors to permeability
(Figure 3.24)

The reactivation of the fracture, or opening of a new fracture, may only be maintained
while pumping is continuous. Once pumping has stopped, the pressure will begin to decrease
and the fracture may relax and close (Davey, 2012). Repeated shearing of previous failure planes
may result in the production of fine grained material, sheared off of the failure surface, and that

22
material could in time fill the fracture and cut-off the flow channel (Fjaer et al., 2008). The
creation or reactivation of failure planes during hydraulic stimulation is therefore not a simple
subject and must be approached carefully.

Figure 3.24. A Mohr-Coulomb plot showing the lowering of effective stress in relation to
increased pressure during hydraulic stimulation. The decrease in effective stress forces the Mohr
circle towards the failure envelope. If there are features oriented such that they have already
passed into the failure envelope given the initial stress conditions, those features are deemed to
be critically stressed (Davey, 2012).

The fracture networks created by hydraulic stimulation may be simple, linear features, or
complex, interconnected networks (Figure 3.25) (Warpinski et al., 2009). Complex fracture
networks will result in wells that perform better relative to continuous, but simple geometries,
and are the ultimate goal of the stimulation design (Baihly et al., 2010). The stress concentration
at the fracture tip controls fracture propagation, and this can be influenced by discontinuities,
changes in the material properties of the rock, or deviations from a perfectly isotropic material
(Fox et al., 2013). High-level models are used in order to best try and match reservoir conditions
to predict hydraulic fracture development and optimize completions design. Increasingly, these
models are able to handle hetereogeneous descriptions of reservoirs which better match real
conditions (Hryb et al., 2014). Whether or not we can create complex induced fracture networks

23
by preferentially targeting zones with pre-existing natural fractures is ambiguous and only
experience and time can really tell.

Figure 3.25 Differing fracture geometries created by hydraulic stimulation (Warpinski et al.,
2009).

3.3 Fractures in the Vaca Muerta

Natural, drilling induced, and stimulation fractures may be examined using a variety of
direct or indirect methods, and the examination of the natural fractures and drilling induced
fractures in the Vaca Muerta using borehole imaging data is discussed in Chapter 6.

The stress regime throughout the majority of the Neuquén basin is strike-slip but
transitions to normal, and decreases in stress anisotropy to the East away from the influence of
the Andes Mountains and into the thick sediment package of the embayment (Figure 3.31)
(Garcia et al., 2013). The vertical stress in this instance is the intermediary stress, and the
maximum horizontal stress has the highest magnitude and is generally oriented W-E in response
to the compressive force created by on-going subduction and uplift of the Andes.

24
Figure 3.31 Stress regimes in the Vaca Muerta showing a transition from normal in the far east of
the basin to full strike slip to the west, where thrust and reverse faulting may also occur (Garcia
et al., 2013).

The strike slip regime is an important consideration for fracture propagation, as the
difference in magnitude between Sv and SHmax tends to be relatively small in strike slip regimes,
increasing the risk of developing long, horizontal fractures instead of the desired complex
fracture networks (Garcia et al., 2013).

The Vaca Muerta and Quintuco Formations are consistently overpressured with pore
pressure gradients, as stated in Chapter 2, ranging between 0.65 to 1.02 psi/ft , with an average
of 0.83 psi/ft, and this will have implications when it comes to drilling and completions safety, as
well as to the modeling of reservoir stress and fracture propagation.

In the Vaca Muerta, available core photos show that there are several types of fractures
present, and that they may be encountered as single, unique features, or that they may interact
to form complex fracture swarms spa i g se e al s of Figu e . . E a i atio of the
core available from offset wells in the near vicinity to the Wintershall Block of Interest indicates
that there are fractures in complex orientations, filled with primarily carbonate cements present
throughout the cored intervals – which represent primarily the lowermost Vaca Muerta. CT scan
data corroborates this and further indicates that there are several varieties of fossils present,

25
which may further obfuscate log readings, or be confused for healed fractures when using core
photos for comparison (Figure 3.33). Literature from the producing fields in the basin suggests
that a high number of vertical to sub-vertical fractures has previously been observed in the
uppermost Vaca Muerta and in to the Quintuco (Badessich and Krasuk, 2014).

Figure 3.32. Representative core photos for one of the study wells indicating multiple fracture
forms and a propensity for horizontal discontinuities, and an image of the characteristic beef
(Zanella and Cobbold, 2012).

Figure 3.33 CT scan data for Well C showing the presence of a possible beef fracture in the core,
as well as complex vertical to sub-vertical fractures, and the presence of fossil fragments, which
may be difficult to distinguish in core photos from healed fractures.

26
 In addition to any vertical or sub-vertical fractures present, two types of significant
horizontal discontinuities are present in the study area. On-going volcanism since the
development of the Andean arc in the Jurassic has resulted in thin layers of reworked volcanic
ash in the Neuquén basin fill (Williams et al., 2014). The ash is composed of siliciclastic and clay
materials and has been cemented by silica or carbonate (see Chapter 4 for petrographic sample).
The base of the ash beds can be sharp or transitional, as can the top contact. The ash beds are
described as having an extreme effect on the wireline log response, and causing localized
washouts depending on their thickness (Williams et al., 2014).

Bedding parallel veins of fibrous material (beefs) are one of the most commonly
recognized features in the Vaca Muerta, and have been known to characterize sedimentary
basins with high volumes of organic matter and carbonate content. Beefs consist of horizontal
expansion seams created by tensile fracturing and vertical dilation during diagenesis, and then
slowly infilled by fibrous minerals that precipitate from supersaturated solutions in an orientation
perpendicular to the fracture surface (Rodrigues et al., 2009). Horizontal fractures theoretically
cannot form in normal stress regime, and require that the vertical stress be the weakest (thrust
faulting regime), that the tensile strength of the rock is very weak in the horizontal orientation,
or that an overpressure is generated that is so high that it is stronger than the overburden weight
(Zanella and Cobbold, 2009). The opening of the beefs in the Vaca Muerta is attributed to the
presence of high pressures, and observations of hydrocarbon fluid inclusions in the beef suggests
that the maturation of kerogen may have been coeval with beef formation, and may have been
the mechanism for generating the high fluid pressure (Rodrigues et al., 2009).

The distribution and extent of the beef are relevant to completions, since the reactivation
of large-scale linear features will limit the surface area of the created hydraulic fractures, relative
to if a complex pre-existing fracture network is reactivated (Williams-Stroud et al., 2012). The
relative distributions, and possible interactions, between any vertical to sub-vertical natural
fractures and the beef is an as-of-yet poorly defined topic.

Finally, to add complexity on top of complexity, there are numerous, small concretions
frequently observed in core and borehole images, and there may be intrusive igneous sills

27
encountered in the play (Fantin et al., 2014). Although no sills have as of yet been documented
in the study area, their presence is a possibility and must be kept in mind during characterization
of the play.

28
 CHAPTER 4

PROJECT WORK FLOW

This p oje t ep ese ts the fi st stage i the ‘ese oi Cha a te izatio P oje t s
collaboration with Wintershall concerning the Vaca Muerta Formation. The data was compiled,
audited, and primary geophysical and geological analysis of the available data was completed
simultaneously.

For the geologic analysis the focus was to understand the distribution and nature of
observed fractures within the context of the geomechanical properties of the reservoir on a well
log scale. This was done using image log data, as well as petrophysical analysis using a clustering
algorithm. Where available, images of core and CT scan data were used to provide validation to
the interpretation of the petrophysical and image logs.

Of primary interest to the analysis is the variability in carbonate to siliciclastic content

within the studied interval, and the changing concentrations of total organic carbon. In particular,
the relation of these features to the mechanical properties of the rock as observed and as
predicted by the mechanical stratigraphy model discussed in chapter 6. Relating the variability in
rock properties to parameters such as the ratio of carbonate to silica, and weight percent TOC,
to the distribution of natural fractures, and drilling induced fractures as observed in borehole
image logs, an estimation of the zones most likely to develop complex fracture networks is
attempted.

4.1 Petrophysical Pre-Conditioning

The logs that were utilized in this analysis are listed in Table 4.1. The first process in the
overall workflow was to normalize the logs to account for environmental differences, and in
particular to account for the different drilling company used for Well A. The gamma ray, bulk
density, neutron porosity, and photoelectric logs were normalized for Wells A, B, C, G, and H
(Figure 4.11).

29
Table 4.1 Data table for image log and cluster analysis, where a * symbol indicates data of
questionable quality, but which were used fo the a al sis he e eeded. Ŧ i di ates data
available, but of too poor quality to be of use to the analysis.
Well GR RHOZ PE NPOR Ed vd Vp/Vs Electric Acoustic
Image Image

Well A x x x x x x x Ŧ

Well B x x x x x x x x x*

Well C x x x x x x x x x*

Well G x x x x x x x x

Well H x x x x x x x

Well A Well B Well C Well G Well H

Figure 4.11 Results of the Normalization of the gamma ray log for wells A, B, C, G, and H. which
is representative of the normalization that was performed for the utilized logs.

XRD analysis indicates that the clay that is present is predominantly illite with some illite-
smectite, and trace amounts of chlorite or kaolinite, and this is corroborated by literature from
other parts of the embayment (Sagasti et al., 2014). Petrographic data is currently being collected

30
but was not available for this analysis, as a result, there is uncertainty related to the form of the
estimated minerals. Petrographic studies from nearby areas indicate that the carbonate is in the
form of skeletal debris and carbonate cement, and the siliciclastic minerals are silt-sized grains
(Figure 4.12) (Reijenstein et al., 2015; Williams et al., 2014).

Figure 4.12 Petrographic slides of the Vaca Muerta with calcite colored red with Alizarin Red S.
and all of the same scale. A) Upper Vaca Muerta micritic calcite matrix with fossil fragments. B)
Middle Vaca Muerta calcareous/argillaceous mudstone with increasing clay content and
increasing organic material, which results in darkening color. C) Volcanic ash bed from the middle
Vaca Muerta with high clay content, mica and quartz. D) Lower Vaca Muerta silty, argillaceous
mudstone with calcite-filled horizontal vein (Williams et al., 2014)

31
 Although the overburden and underburden are significant to the geomechanical behavior
of the reservoir interval, the nature of logging for Wells A, B, and C is such that the reservoir
interval was logged over one continuous run, without significant intervals measured in the
overburden or underburden. To avoid splicing the logs from various runs together, modelling was
done using the log run for the reservoir interval and without additional input from above or below
that run. As the new wells being drilled are logged including some amount of underburden and
overburden on the same run as the reservoir interval, the adjustment of the model to the new
wells with the overburden and underburden included is one recommendation for future work.
The top of the Vaca Muerta is gradational with the overlying Quintuco Formation, for the sake of
this analysis, the placement of the upper cutoff was based on a correlated surface picked using
the gamma ray and compressional sonic logs.

4.11 Lithology and TOC Determinations

A petrophysical determination of lithology was performed using the gamma ray to

estimate shale volume, X-ray diffraction data for calibration, and compared to the lithoscanner
estimate.

The shale volume was computed using the gamma ray index and a Steiber approximation
(Equation 4.1, Steiber, 1970). The index gamma ray is computed using the GR values that would
be representative of a clean sand, and a clean shale (4.2).

(4.1)

(4.2)

The Steiber method proved to overestimate the volume of shale relative to the
lithoscanner prediction and the XRD results (Figure 4.13). The Steiber estimate is also quite noisy

32
relative to the other estimates, due to the influence of the GR log. The presence of uranium in
the interval, as a result of the primarily type II kerogen present (Figure 4.14) means that the
estimate of clay volume from gamma ray must be adjusted to reflect that the uranium from
ke oge is o t i uti g to the o k s adioa ti it is ot just due to the tho iu a d potassiu
present in the clay minerals (Crain and Holgate, 2014). Because the lithoscanner estimate
provided a good match for the XRD data without further adjustment, the lithoscanner estimate
was used instead of a gamma derived estimate.

Figure 4.13 Estimates of clay volume from the XRD (both in table form and as a data point on the
log), Steiber estimate, and lithoscanner. The lithoscanner and XRD are in relatively close
agreement, and the Steiber method overestimates shale volume.

33
 Mineralogy and lithology estimates were therefore primarily interpreted using the
lithoscanner data as corroborated by the XRD data available for wells A, B, C, and H. At present,
there is no XRD data available for well G, but any future study could incorporate this data.

I o de to e aluate total o ga i a o , th ee ethods e e tested. The Passe , o Δ

Log ‘ , ethod uses so i a d esisti it logs i o de to esti ate the total o ganic carbon. This
is based on the understanding that the sonic log will preferentially be sensitive to kerogen with
respect to the resistivity log (Passey et al., 1990). Kerogen has a much higher volume percent
than would be expected due to the low grain density of organic matter (Passey et al., 2010),

The Passey method relates the quantity of separation between the compressional sonic
log and the resistivity curve using equation 4.3, where the value of the measured resistivity (RT)
is related to the sonic transit time (t). The baseline values are established in a non-reservoir
interval where the sonic log and resistivity log should overlie (Passey et al., 1990)

 log R = log10 (RT / RTbaseline) + 0.02 (t / tbaseline)

(4.3)

The quantified  log R value is then related to the total organic carbon weight percent by
equation 4.4, where the LOM parameter describes the level of organic maturity and is empirically
derived from vitrinite reflectance data (Passey et al., 1990). The vitrinite reflectance values
provided by Wintershall for Wells B and C range between 0.74 and 1.36, with averages of
approximately 0.9, which corresponds to a LOM value between 10 and 11 (Passey et al., 1990).

TOC = ( log R)*10(0.297 – 0.1688*LOM)

(4.4)

The Schmoker method uses the density log and the assumption that the formation
generally consists of four components: the rock matrix, the interstitial pores, pyrite, and the
organic matter. The porosity is assumed to be a fixed property in low porosity rocks, since any
density difference between the pore fluids would be negligible to the log reading when there is
only a small volume of that fluid present (Schmoker and Hester, 1983). Once the assumptions
have been made, the equation takes the general form of:

34
 TOC = A / ρ - B
(4.5)

Whe e ρ is the fo atio de sit a d A a d B a e the “ h oke pa a ete s fo hi h the

defaults are A = 154.497 and B = 57.261 (Schmoker and Hester, 1983).

The final TOC estimation method was a simple plot of the concentration of uranium.
Assuming that the uranium present would be associated with the type II marine kerogen in the
Vaca Muerta allowed for the use of the uranium log from the spectral gamma ray tool to be used
as a proxy for the relative concentration of TOC (Passey et al., 2010). This relationship could be
quantified in order to establish an estimate for the weight % of TOC, but in the instance of this
study, the relative concentration was deemed sufficient. In unconventional reservoirs, the
organic-rich intervals may be hundreds of meters thick, but vertical variability can be significant.
The [U] method allows for a relative estimate of this vertical variability that is simple and does
not involve scaling or introducing location specific parameters. Apatite in the reservoir would
require additional consideration when using the [U] (Bust Et al., 2013), but the assumption of no
apatite in the system is made.

The Schmoker and [U] methods matched the laboratory results for TOC from cuttings and
core where it was available (Figure 4.15).

The dynamic mechanical moduli were calculated using the shear and compressional sonic
logs using the equations stated in Chapter 3. Empirical relationships have been developed to
estimate static moduli from dynamic. However, the use of such relationships is not generally
advised, and the determination of the static moduli in laboratory testing is always preferred
(Barree et al., 2009). The lack of core data available and the uncertainty related to the empirical
relationships means that no static moduli were incorporated in this analysis.

The end result of the petrophysical pre-conditioning is that the logs have been
standardized and corrections for environmental variability have been made. Lithologies and TOC
have been estimated using several compared methods, and the dynamic moduli have been

35
calculated. The character of the normalized and calculated logs is consistent with the
i te p etatio of the Va a Mue ta s depositio as des i ed i Chapte Figu e . .

Figure 4.14 Hydrogen Index and Oxygen Index with the overlay for a modified Van Krevelin
diagram showing the location of cuttings from Well G plotting as type II kerogen, indicating an
algal origin. It should be noted that this analysis makes no distinction between the Upper and
Lower Vaca Muerta despite there being considerable difference between them (provided by
Wintershall).

36
Figure 4.15 Comparison of the preferred Schmoker and [U] methods with TOC from cuttings in
Well G.

37
Figure 4.16 Log response over the Vaca Muerta interval for gamma ray, resistivity, dynamic
You g s Modulus Ed , a d d a i Poisso s ‘atio d). The upper Vaca Muerta is indicated with
a grey color, and the lower Vaca Muerta with blue, these colors are consequently used to
differentiate the two zones.

4.2 Borehole Image Theory

Borehole images provide a high-resolution alternative data source when physical rock, in
the form of core or representative outcrop sections, is not available. Details of the
sedimentological features of the reservoir, fractures present at depth, and the borehole

38
conditions are viewable in image logs, and can be used for reservoir characterization or for
development planning (Pöppelreiter et al., 2010). The configuration of modern micro-imaging
tools generates images that capture features at the centimeter scale, which is significant in
unconventional reservoirs where small heterogeneities may be critical to successful stimulation
(Gamero-Diaz et al., 2013; Hurley, 2004). For geomechanical considerations, borehole images
may provide valuable information about in-situ natural fractures and borehole breakout, and
therefore provide insight into stress orientations and relative magnitudes (Tingay et al., 2008).

4.21 Electric Borehole Imaging Physics

The borehole-imaging tool records either an electric (conductivity) or acoustic signal using
a o figu atio of pad su fa es o e ed ith a a a of ele t ode se so s, o utto s Hu le ,
2004). During logging, the pads of the electric imaging tool press against the borehole all, an
alternating current is emitted, and the variation between the emitted current and the received
response is measured and then related to the resistivity contrast of the formation (Figure 4.21)
(Schlumberger FMI Brochure). The measurement of the incoming current between the pads at
the base of the tool and the electrode towards the top of the tool is used to establish the baseline
resistivity of the formation. Simultaneously, the incoming current is measured at each button on
the pads, and the relative magnitude of the variation at each button compared with the baseline
formation resistivity is used to produce the scaled image. It is the array configuration and high
density of the buttons, along with a high sample rate (400 samples/m), that makes for the high-
resolution of the final image (Hurley, 2004, Schlumberger FMI Brochure).

The number of arms and buttons is tool dependent and varies by Service Company (see
Figu e . . “ hlu e ge s Oil Based Mi o I age tool OBMI , hi h is a adaptatio of the
Formation MicroImager Tool (FMI) for wells drilled with oil based mud, was the tool used for
Wells B and C in this study. An FMI was run in Well G. The quality of the image depends on the
mud system for drilling, the borehole conditions (since the recording depends on the pads being
properly pressed to the borehole wall), and logging speed (Lagraba et al., 2010).

39
Figure 4.21. Tool Design of Schlumberger Oil Based MicroImager (a) and Ultra Sonic Borehole
Imager (b) tools used for acquisition of borehole images in Wells B and C (Schlumberger).

40
Figure 4.22 Table listing differing tool configurations, with respect to number of arms and
buttons, for several commercial electrical borehole-imaging tools. The number of measurements
directly correlates to the resolution of the image (Lagraba et al., 2010).

4.22 Acoustic Borehole Imaging Physics

The acoustic tool is generally configured in a similar fashion to the electrical tool. There is
a transducer towards the base that both emits and records an acoustic signal, creating an
ultrasonic scan of the borehole wall (Lagraba et al., 2010). The acoustic tool rotates as it moves
upward in the borehole and emits ultrasonic pulses, which are then processed into the reflected
amplitude and the interval transit time, it is essential that the tool be correctly centered in the
borehole (Hurley, 2004). The final images used for interpretation are of the acoustic amplitude
and this two-way travel time between the emission of the pulse and the recording of the
response. Whereas the resistivity imager is capable of recording data up to a few cm into the
formation, the acoustic image is only capable of recording data at the borehole wall and as such
is commonly used for analysis of borehole stability issues. The acoustic image does, however,
provide full 100% borehole wall coverage, as opposed to a maximum of 94% with electrical
imaging tools (Lagraba et al., 2010).

The lack of depth of penetration by the acoustic tools means that they are preferentially
suited to analysis of features that affect borehole condition – fractures, faults, and borehole
breakout. Because the oil-based mud will create a resistive trace in an open fracture, the OBMI
is not able to resolve the difference between an open fracture and the resistive cement in a
healed fracture. As such, an acoustic tool is frequently run in tandem with OBMI, since the
acoustic image will preferentially record open fractures, particularly as the edges of such

41
fractures become increasingly rugose during the drilling activity (Schlumberger UBI Brochure). A
healed fracture filled with a material, which has an acoustic amplitude that differs from that of
the greater rock matrix will be seen on the acoustic amplitude image, but not necessarily the
travel time image, and this can act as a further distinguishing feature between fracture types
(Hurley, 2004; Lagraba et al., 2010). In theory, the resolution of the electrical imaging tool and
acoustic imaging tool are equal, however, in reality, the range of resistivities in oil reservoirs is
much higher than the range of acoustic impedance values and therefore the electrical imaging
tool will capture much smaller variations in the formation relative to the acoustic (Lagraba et al.,
2010).

“ hlu e ge s Ult aso i Bo ehole I agi g UBI tool as used i Wells B a d C, a d

Bake Hughes Ci u fe e tial Bo ehole I agi g Log CBIL as u i Well A. No a ousti
imaging tool was run in Well G.

4.23 Data Handling

Borehole image data is acquired at very high sample rates (400 samples/m), and the raw
data consists of the readings from each electrode and the x-, y-, and z-axis accelerometer and
magnetometer information, to allow for accurate tool orientation (Hurley, 2004). The hole
deviation and tool orientation are determined from the magnetometers, with pad 1 orientation
being used as the standard reference point for the majority of tools (Tingay et al., 2008; Hurley,
2004).

Processing of the data file includes the orientation of the image in relation to the borehole
using the accelerometer and magnetometer data, correction for emitter excitement, and depth
matching of the button readings (Lagraba et al., 2010; Hansen and Buczak, 2010). The individual
readings must be adjusted to be representative of the same depth, by correcting for the physical
separation of the buttons – which is specific to the tool dimensions, as well as the constant
acceleration and deceleration of the tool during logging in response to the tension in the tool
cable and the effects of friction between the borehole wall and the tool (Hurley, 2004; Hansen
and Buczak, 2010).

42
 It is not possible to create an image directly from the raw data recorded by the electrical
imaging tool, but instead the image is constructed using the relative distributions of the resistivity
alues Ha se a d Bu zak, . The fi al esisti it i ages ge e ated a e stati a d d a i
versions of the constructed image. The static image has been scaled broadly across the entire
logged interval, whereas the dynamic image has a variable contrast that can be applied over any
defined interval. The dynamic image is used for the study of subtle features, particularly in rocks
that are either very lowly resistive, or highly resistive and where contrasts are not as extreme.
The static image, by contrast, provides a more generalized understanding of resistivity variability
across the formation (Hurley, 2004).

For interpretation of the image, human or automatic feature picking can be employed,
with the caveat that any automatic interpretation must be carefully quality controlled. The log
interpreter will fit sinusoidal waves to any observed feature, such as bed boundaries,
sedimentary features, or fractures, depending on the goal of the interpretation (Figure 4.23). The
interpreter must be careful to avoid treating the image as if it were a picture, and to remain
cognizant of the possibility of tool or processing artifacts distorting the image (Hansen and
Buczak, 2010).

Figure 4.23 Schematic depicting the conversion of 3D wellbore imaging data as acquired from the
pads of the micro imager tool to the azimuthal image which is interpreted based on the
characteristic sinusoids generated by any resistivity or acoustic anomaly which is inclined relative
to the wellbore (Hurley, 2004).

43
4.24 Geomechanical Analysis using Borehole Images

The relationship between the occurrence of stress-induced wellbore failure features in

the borehole image, and the corresponding stresses that induced such a feature is a powerful
tool for improved understanding of the stress state around a wellbore (Pöppelreiter et al., 2010).

The orientation and character of fractures as interpreted in image logs can be used to
make determinations about the magnitudes and orientations of the principle stresses in a
formation, and can increasingly accurately be used to make determinations about optimal
wellbore placement, and optimal hydraulic stimulation design (Pöppelreiter et al., 2010).
Generally, borehole breakout will occur in the minimum horizontal stress orientation in vertical
wells that have less than 10° of inclination (Tingay et al., 2008; Ejofodomi et al., 2014), and
induced fractures will be oriented in the direction of the maximum horizontal stress (Zoback et
al., 1985). Natural fractures, induced fractures, and borehole breakout are all resolvable in
borehole image logs, provided that the quality of the log is sufficiently high (Figure 4.24), and the
resolution of these features may be especially significant in low-permeability unconventional
reservoirs where enhanced permeability is essential to production performance (Barton and
Moos, 2010).

Distinguishing between natural and induced fractures is only sometimes possible in

images, as the differences may be non-existent or subtle (Barton and Zoback, 2002). Natural
fractures are more likely to occur as one or more sets, each with a distinct orientation. In contrast,
the drilling induced fractures are most likely to be vertical or near-vertical and to have a well-
defined strike azimuth (SHmax). In general, natural fractures can be fit with a sinusoid which
reflects that they are planar features intersecting the borehole, whereas drilling induced
fractures will propagate asymmetrically from their point of initiation and may be non-planar
(Barton and Moos, 2010). At the same time, there is frequently an enhancement of natural
fractures by drilling activity, due the creation of a mechanically weakened zone at the
intersection point between the natural fracture and the wellbore. This may be seen in the image
as an enlargement in aperture of the natural fracture at its peak and its trough, and is another
diagnostic tool for distinguishing between enhanced natural fractures and drilling-induced

44
fractures (Barton and Moos, 2010). Best practice is to establish strict criteria by which to evaluate
the likelihood of a feature being natural or drilling-induced, and to ensure that image quality is
sufficiently high to allow for such an interpretation (Barton and Moos, 2010).

Figure 4.24 Examples of wellbore failure features in image logs from the data set: a) Natural
fracture observed in FMI showing enlargement at the peak and a faulty flap, b) drilling induced
tensile fractures in FMI, c) natural fracture observed in OBMI and acoustic amplitude but not two-
way travel time, indicating perhaps a cemented fracture, and d) borehole breakout as observed
in the acoustic log, and with a chaotic response in the OBMI.

4.3 Cluster Analysis Principles

The material properties of the given rock volume will have several implications for the
likelihood of developing fractures during stimulation, the nature of the developed fractures, and
the ability of the material to maintain those fractures post-stimulation (Barree, 2002). Analysis
of the geologic factors that affect log response, and the relation of the logs to the material
properties which may affect the hydraulic stimulation, is facilitated using a clustering technique

45
(Williams et al., 2014), particularly when large rock volumes are being considered, as in the Vaca
Muerta.

4.31 Cluster Analysis Theory

Mudrocks are by their nature heterogeneous, and must be discussed in terms of not only
composition but also fabric and texture, which can be controlled by any number of factors such
as tectonics, source, water dynamics, climate etc. The differences between different portions of
a mudrock may be subtle, and the transitions and variations may be difficult to identify, even
when core or samples are available (Al Ibrahim, 2013).

There are several methods for clustering which can be applied using readily available
commercial software (Al Ibrahim, 2013) and in this instance, the Heterogeneous Rock Analysis
odule i “ hlu e ge s Te hLog® software was chosen. The method consists of the
construction of a fully unsupervised model and is heavily influenced by the user input
parameters. The model is unsupervised such that there is not a defined objective, and there are
no training parameters inherent to the initial construction (Duenas, 2013). The HRA module uses
the k-means algorithm in order to group like data, and allows for the user to set the number of
cluster, or sub groups, by which to divide the data based on a series of diagnostic tools that are
generated during the model building process. The user choses input logs, and then will set an
approximate number of cluster groupings (x) to subdivide the data into. The program reads the
data from the logs and converts the data points in to principal component space. A centroid
corresponding to each principal component is then mapped and the data points are assigned to
the nearest centroid. The cumulative distance between each point and the corresponding
centroid is calculated using equation 4.6, where d is the distance between point x and the
e t oid . P efe e es the u e of di e sio s i the data, hi h o espo ds to the
number of cluster groups assigned by the user (Al Ibrahim, 2013). The position of the centroid is
adjusted with each log run, in order to minimize this distance, and to maximize the distance
between each cluster (Duenas, 2013). The model is run for the number of iterations set by the
user, at which point diagnostic plots are generated in order to inform the user of whether or not
they have achieved the goal of creating sufficiently unique clusters with a minimized cumulative

46
distance. If the model does not fit the critical criteria as defined by the user, then the number of
runs, the number of clusters, or the input log suite may be adjusted until an acceptable model
has been achieved.

d2 , = Σpi=1(xi-ci)2
(4.6)

The diagnostic plots include a map of the data in principal component space with the
cluster assignments indicated by color scheme (Figure 4.31), a fall off plot which shows the
change in the cumulative distance for each run of the model (Figure 4.32), and a silhouette plot
which gives an indication of cluster uniqueness by depicting how many points of a given cluster
may have been assigned to an adjoining class (Figure 4.33) (Willis, 2013; Duenas, 2013).

Figure 4.31 Mapping of data points in color along with the position of the centroids designated
ased o the use s defi itio of the u e of desi ed lasses.

47
Figure 4.32 Fall Off plot for the cluster analysis showing decreasing cumulative distance with
subsequent model runs. A flattening of the cumulative distance within the last 10% of model runs
indicates that a minimum distance has been achieved and therefore the model is satisfactory.

Figure 4.33 Silhouette plot showing the distance ratio between the data in a given class and the
adjoining class. Negative values are indicative of data which may equally have been assigned to
another class.

48
 The logs that may be input into the model must first be normalized to preclude any
variability resulting from differences in external factors such as the service company used for the
logging, the tool specifications, or problems with the logging, as was discussed earlier in this
chapter. The HRA module initiates the model randomly, so that running the model multiple times
may result in somewhat different final results (Al Ibrahim, 2013). The unsupervised nature of the
initial model means that there is no bias towards either class uniqueness or class size. The model
is purely partitioning the data based on the true variance and will not create classes which are
non-unique, but nor will it include too narrow a data distribution in order to achieve uniqueness
(Williams et al., 2014; TechLog Manual).

The input logs for the cluster model are fully at the discretion of the user, and can be
manipulated depending on the end goal of the modelling. Because the model is solely
constructed using log data, it is recommended to relate the resulting classification scheme to
other data forms where available, such as core or cuttings, in order to interpret the definitive
rock properties (Williams et al., 2014).

Once the model has been assessed to be suitable by the interpreter, the model can be
applied to additional wells using the same log suite in order to become a predictive tool. Here,
the e a e also diag osti tools i o de to assess the ualit of the tag of the odel to the e
well(s).

The tagging uses the initially unsupervised model from one well to train the models of the
subsequent wells. When tagging, a compliance distance (a Mahalanobis distance), which
describes the distance between a data point in the new well and the centroid position defined
by the original model, is used to assess the quality of the tag. This is a quantification of the
amount of similarity between the data points in the original model and the data points grouped
in the same classes, but from a new data source. In general, a compliance distance of 20 or less
is considered acceptable (Duenas, 2013).

49
 Clustering appears to be increasingly popular for use in unconventional plays where well
productivity may be heavily reliant on optimized completions design (Duenas, 2013; Willis, 2013;
Suarez-Rivera et al., 2011; Williams et al., 2014). Because of the high degree of user dependence
of the HRA module, the cluster model output can be highly tailored to the interpretive goal. The
cluster models used in this analysis are discussed in detail in Chapter 6.

50
 CHAPTER 5

IMAGE LOG INTERPRETATION

Access to core was limited for the study wells and borehole image logs were chosen to act as the
primary means of studying borehole scale natural fractures present through the reservoir
interval.

5.1 Image Interpretation Goals

High fracture gradients are present within the Vaca Muerta (Ejofodomi et al., 2014),
meaning that it may be very difficult to induce tensile fractures during stimulation. Documented
previously in the Vaca Muerta are vertical, sub-vertical, and bedding parallel fractures (Zanella
and Cobbold, 2009; Gale et al., 2009; Williams et al., 2014; Fantin et al., 2014), and during
stimulation, these pre-existing weak points may act as initiation points, or be reactivated to
provide fracture permeability and transport liquids to the wellbore for production.

The borehole image data available in the study area were examined in order to
characterize the presence and form of natural fractures present in the wells. The data were
acquired during different time periods and under different logging conditions, and provide the
opportunity for a case-study on how drilling methods may impact the determination of stress
and fracture interpretations from borehole images.

5.2 Image Interpretation Process

The four wells for which the borehole image data were available include different imaging
tools and cover different depths within the reservoir (Figure 5.11 and Table 5.1). The image logs
e e p o essed a d a al zed usi g “ hlu e ge s Te hLog soft a e suite in order to study
depositional and diagenetic features such as bedding, sedimentary structures, and natural
fractures, with a particular emphasis on the fractures. The results from the image interpretation
were compared to core photo and CT scan data where available.

For Wells B, C, and G, the original Digital Log Information Standard (DLIS) were available
to be processed and interpreted. The Well A data set included only the array processed amplitude
image available and not the original data file. As a result, it is not possible to interpret Well A at

51
the same resolution as the other wells, and for this reason, the interpretation of Well A was done
only for general comparison purposes, and not for quantitative analysis, and the results are not
discussed beyond confirming that the bedding which could be observed is consistent with what
is seen in the other wells. Processing begins with a quality control check of the accelerometer
and magnetometer data, essential to ensuring that the dips recorded in the image interpretation
are accurate and represent true dips (Figure 5.12). First, the azimuth and direction data are
adjusted for magnetic declination and magnetic inclination. Then the accelerometer is corrected
to ensure that the curve is on depth with the resistivity traces recorded by the tool, and to correct
for any sticking of the tool, variation in logging speed, and any resonant vibrations as the tool
moved up-hole. In the case of Well B, the magnetometer data showed anomalous readings near
the top of the image during processing, and further examination indicated that the uppermost
portion had been run too close to the casing for reliable measurement. In this instance, the first
10 m of the image were cropped from the processing and interpretation.

The mud weights used for drilling in Wells B and C would normally preclude the
acquisition of an acoustic image, because the densities are too high and the signal to noise ratio
is diminished by the density of the mud. However, it appears that the image quality is sufficient
to be used for interpretation of geologic features.

The quality of the FMI run for Well G is very high, and except for a logging error which
rendered Flap 3 unusable, allows for the most precise and accurate interpretation of any of the
images in the area.

For validation, the interpretation of the image logs was compared to available core photos
for wells B and C. In both wells, significant adjustments to the depth profiles had to be made to
the core photos using the image gamma ray and the core gamma, where the core gamma had
previously been corrected to match the wireline gamma. This correction was performed using
the most significant peaks in the gamma log without splicing or stretching, so that one shift was
performed per core. The depth shifts performed are listed in Table 5.2.

52
 Figure 5.21 Location map depicting the locations of the 4 wells with borehole images, next to a
cross sectional view of how much of the Vaca Muerta was imaged.

Table 5.1 Different types of borehole image available in the study area, and the extent of imaging.
Well B had an addition 600 m run above the reservoir interval in the same imaging run, however,
gas in the wellbore caused image quality issues above the Vaca Muerta level.
Well G Well A Well B Well C

Image Type FMI Acoustic OBMI, UBI OBMI, UBI

Amplitude

Extent 382 m 202 m 258 m 947 m

Total Recorded Thickness 240 + 170 m 268 + 611 m 260 + 638 m 270 + 677 m
(Vaca Muerta + Quintuco)

Mud Weight 1.82 g/cm3 1.55 g/cm3 1.78 g/cm3 1.79 g/cm3

Quality Good Bad Poor Poor

53
 Load DLIS file in TechLog

Run Processing Wizard

Image Interpretation

Quality control check of

Workflow

magnetic and
accelerometry Data

Create final images for

interpretation
Drilling Induced Fractures
and Borehole Breakout for
Stress Determination
Manually interpret relevant
features
Natural Fractures for
Fracture Stratigraphy
Incorporate other data
forms for integrated final
interpretation

Figure 5.22 Image processing and interpretation workflow as sequentially performed for Wells B,
C, and G.

The depth shift for Well G was performed by Schlumberger and provided by Wintershall.
In lieu of a comparison with core photos, which were not available at the time of the analysis, CT
scan images were used for the interpretation validation.

Table 5.2 Depth correcting performed on core photos in order to compare between the core
photo, CT scan images, image log, and wireline logs.
DEPTH CORRECTIONS FOR CORE PHOTOS AND CORE SPECTRAL GAMMA

WELL CORE 1 CORE 2 CORE 3

Well B -8.35m -8.42m -8.66m

Well C -5.68m -6.89m -4.74m

54
5.21 OBMI Interpretation for Wells B and C

The imaging tools were run over the Quintuco for both Wells B and C, however, the only
features interpreted in the Quintuco were borehole breakout, and the interpretation was
otherwise confined to the Vaca Muerta.

The majority of wells in the Neuquén Basin have been drilled using oil-based drilling fluids,
which may be more cost effective in this region and allow for broader borehole stability windows.
However, the use of oil-based muds reduces the quality of the image log acquired, as the tools
developed for use in oil-based muds (“ hlu e ge s Oil-Base MicroImager, OBMI, in Wells B and
C, he eas Bake Hughes Ci u fe e tial Bo ehole I age Log, CBIL, as used i Well A do ot
offer the same quality of data as those developed for use in water based muds (Garcia et al.,
2013). The resolution of the OBMI is such that for an 8 inch diameter wellbore, approximately
32% of the wellbore surface is imaged, as opposed to upwards of 80% when using the water
based Formation MicroImager (FMI) tool (Schlumberger).

The interpretation for Wells B and C was generally inconclusive. Although some fractures
were visible, the number was very low and prevented any robust analysis. The wellbore failure
features observed in the image were inconsistently oriented (Figures 5.23 and 5.24), and this is
attributed to image quality and a lack of sufficient borehole coverage. The interpretation was
further complicated by a lack of sufficiently high agreement between the core photos and the
image trace, even after rigorous depth matching. A confidence qualification was added to the
analysis in order to rank the relative degree of certainty for each interpreted feature. The scheme
a ked possi le f a tu es as eithe high o fide e , id o fide e , o lo o fide e
dis o ti uities, o as a esisti e f a tu e o o du ti e f a tu e he e a defi iti e o lusio
could be made. This ranking was relative and depended on how clearly a feature could be viewed
in all versions of the processed image data (the static OBMI, dynamic OBMI, acoustic amplitude,
and two-way travel time) and a qualitative assessment of how significant the contrast between
the surrounding rock matrix and the feature trace appeared on the resistivity or acoustic scale.

Acoustic images are preferentially sensitive to the borehole wall conditions and are very
useful for identifying borehole breakout and facilitating stress determinations. The same cannot

55
be said of OBMI, and the interpretation of wellbore failures in OBMI images is further made
difficult by chemical interactions between the drilling mud and the reservoir rock which can
appear to be failure. In particular, the development of syneresis cracks can appear as drilling
induced fractures (DIFs), which can invalidate any stress determinations since the controlling
mechanism for the syneresis crack formation is not stress related (Personal Communication S.
Ganguly).

Figure 5.23 Natural fracture orientations for Well C showing resistive fractures (dark blue) as well
as high confidence discontinuities in light blue and mid confidence discontinuities in red.

Bedding is easily visible in all the images and makes up the primary structure for all units.
Bedding indicates a relatively thinly layered rock matrix, with alternations on the centimeter
scale. The bedding dip is very low angle, reaching a maximum of 8° with an average of 3° in all
wells.

56
Figure 5.24 Natural fracture orientations for Well B. The dark blue is a single resistive fracture,
and the red and light blue represents the high and mid confidence discontinuities.

5.22 FMI Interpretation for Well G

Well G is the sole to have been logged using an FMI tool, as it was the only well in the
study to have been drilled with a water based mud. The image quality for the FMI was high
relative to the OBMI images in the offset wells, which provided for the most robust interpretation
of any of the images available for study (Figure 5.25).

The interpreted interval covers the uppermost 10 m of the Tordillo Formation, the whole
of the Vaca Muerta, and the lowermost 10 m of the Quintuco Formation. Within the Tordillo, the
bedding dips are the highest, and crossbedding is observed. In the Vaca Muerta and the Quintuco
the bedding is much more shallowly dipping, with a maximum of 3° of dip. The lower Vaca Muerta
has abundant concretions, or nodules, that are clearly seen in the image which have, in places,
caused soft sediment deformation (Figure 5.26). These concretions are both resistive and
conductive, indicating that they may be composed of calcite and clay, although there does not
appear to be a particular control on the relative location of each type. In the middle to upper
Vaca Muerta the concretions are still present, though in much lower quantities, and fractures
begin to be observed. The middle Vaca Muerta is where the highest degree of natural fracturing

57
takes place, showing isolated resistive or conductive fractures, or else small swarms of several
cross-cutting fractures. The upper Vaca Muerta is where the drilling induced fractures are located
and it would appear that the orientation of the DIFs is generally consistent but with slight rotation
in heavily laminated areas across bed boundaries (Figure 5.27). Finally, the Quintuco has a
mottled texture in the image which is characteristic of carbonates, and many natural fractures.
Drilling induced fractures are difficult to resolve in the Quintuco, but borehole breakout was
observed.

Figure 5.25 Representative dynamic FMI sections from the formations imaged in Well G.

In Well G, two preferred fracture orientations corresponding to azimuths of N 50-230°

and N 145-325° can clearly be seen in the natural fractures (Figure 5.28). These fractures are
preferentially observed in the middle of the interval and are absent in the uppermost or
lowermost portions of the formation. That the lowermost section would be lacking in fractures
is perhaps intuitively understood given the high proportion of organic material there results in

58
 e ha i al oduli that a e ot fa o a le to f a tu e i itiatio elati el highe Poisso s ‘atio,
a d lo e You g s Modulus . Ho e e , the uppermost Vaca Muerta is where the mechanical
moduli would indicate the most likely fracture occurrence and the relative lack of fractures
observed in this area is a first indication of the lack of correlation between the fracture
stratigraphy and mechanical stratigraphy in the Vaca Muerta.

Figure 5.26 Concretion in the Lower Vaca Muerta which has caused deformation in the underlying
sediment.

Figure 5.27 Rotation and distortion of the drilling induced fractures as they interact with
individual bedding planes in the Upper Vaca Muerta, with adjacent diagram for emphasis.

59
 There are still discontinuities that could not be characterized in Well G, although these
are distinguished from the discontinuities in Well B and Well C because of the better recognition
of fractures in Well G relative to the images from Wells B and C. In Wells B and C, the
discontinuities possibly represent natural fractures, whereas in Well G the natural fractures tend
to be recognizable and the discontinuities are less likely to be fractures, and may instead be
related to bedding, the presence of concretions or nodules, or some other rock anomaly.

Figure 5.28 Stereonet plots of the strike of the fractures interpreted in Well G. a) Resistive
fractures, b) High- (dark green) and Mid- (light green) confidence discontinuities, c) Conductive
fractures.

60
5.3 Stress Determination

The orientation and character of wellbore failures as interpreted in image logs can be
used to make determinations about the horizontal stress directions in a formation (Figure 5.31).
Borehole breakout occurs as the result of compressive shear failure as stresses become
concentrated around the borehole as the original material is removed to clear the way for further
operations. As reviewed in Chapter 4, breakout will elongate the borehole in the orientation of
the minimum horizontal stress in vertical wells with less than 10° of inclination (Tinguay et al.,
2008). The drilling induced fractures will propagate in the direction of SHmax and open in the
direction of Shmin, since the borehole stress concentration is minimized in the direction of S Hmax
(Tingay et al., 2008).

Borehole breakout is very clear in the acoustic logs, because of the sensitivity of the
acoustic tool to the borehole wall conditions. In the wells where the acoustic was run, breakout
is the easiest to define, and frequently corresponds to areas that appear to be heavily fractured
(see Figure 4.24 for example).

Drilling induced fractures appear as conductive lineations in the Well G image and
although there is a slight rotation in their orientations, selective interpretation of the more
common orientation trends indicates that the maximum horizontal stress is oriented between
N95-100°. The induced fractures are observed in the middle to upper Vaca Muerta and are
completely absent from the lower intervals of the image. In contrast, borehole breakout is
uniquely observed in the Quintuco. The borehole breakout orientation corresponds to a
minimum horizontal stress oriented N20-200°. Comparing these orientations to other wells in
the areas corroborates that these values are in line.

5.4 Natural Fracture Interpretation

There were 24 resistive fractures and 4 conductive fractures observed in Well G. The
fractures occurred primarily in three zones through the middle Vaca Muerta, and are oriented in
two directions. One set of more highly dipping fractures strikes N50 - 230°, and a second set of
more shallowly dipping fractures strikes orthogonally to that at N145 - 325° (Figure 5.41). The
fractures terminate abruptly at bed boundaries but are not confined to a single bed, and the

61
 nature of the bed at which they terminate varies (Figure 5.42). The resistive fractures, in
particular, have a smeared appearance, or halo, at the fracture peaks on the image sinusoids,
indicating a degree of enhancement during the drilling process (Figure 5.43).

Figure 5.31 Wellbore Failure results indicating the principal stress orientations: a) Induced
fractures from Well G indicate a maximum horizontal stress oriented approximately E-W, b)
Borehole breakout seen in Well C indicating a minimum horizontal stress oriented approximately
N-S, c) in contrast, the induced fracture interpretation from Well B shows no alignment and reflects
the insufficient quality of the interpretation.

62
Figure 5.41 Stereonet plots for resistive and conductive fractures interpreted in the borehole
image alongside image interpretation layout. Feature count log generated by TechLog for the dip
interpretation, a fracture intensity (F.I.) log for 5m intervals, the FMI images, and the tadpole plot
of the dip interpretation.

Figure 5.42 Conductive and resistive bed terminations of single fractures in the middle Vaca
Muerta.

63
Figure 5.43 Drilling enhancement at the fracture tip and abrupt fracture termination at bedding
interface from the middle Vaca Muerta.

There are locations in Well G where the fracture sets cross-cut which might allow for a
determination of relative timing. Once such interpretation is presented in Figure 5.44, which
infers that the fracture set oriented N145 – 325° formed first, and was followed by the secondary
set at N50 – 230°.

Figure 5.44 Suggested interpretation for the relative timing of four natural fractures in the middle
Vaca Muerta based on cross cutting. The black, low angle fractures interpreted in Time 4 consist
of two conductive fracture and two features which appear to be conductive fractures but for
hi h the t a e is less e tai , a d hi h e e i te p eted as high o fide e dis o ti uities
rather than as conductive fractures.

64
5.41 Stability Assessment

The stability of the natural fractures was determined using the Trap Tester software from
Badley Geoscience with the goal of determining how many of the identified fractures are critically
stressed. The azimuth and dip for each of the resistive and conductive fractures in Well G was
loaded in to the software and subjected to the in-situ reservoir stresses at a chosen
representative depth. The values for the stress field were taken from literature (Garcia et al.,
2013) and rock strength estimates were chosen through consultation with Wintershall (Personal
Communication with P. Pateti and H. Davey). The chosen values are listed in Table 5.3, and
represent the strength of the rock matrix devoid of any consideration of the pre-exiting failure
surfaces.

Table 5.3 Stress and rock strength values input to calculate fracture stability.
The result is that all of the fractures appear stable given the current reservoir conditions and that,
actually, none of the identified fractures are critically stressed (Figure 5.45).
Gradient Value

Shmin 0.92 psi/ft 8263 psi

SHmax 1.14 psi/ft 10234 psi

Sv 1.06 psi/ft 9521 psi

PP 0.83 psi/ft 7544 psi

Friction Angle 37°

Coefficient of internal friction 0.75

Cohesive Strength 1000 psi

The pressure was then increased in the model incrementally (increments of 1000 psi) with
a corresponding 2/3 increase in Shmin in order to simulate the reservoir conditions during a
theoretical, very simple, stimulation. In practice, the pressure increase would likely be a sudden
jump of a large magnitude, but the incremental increase was done to assess the point of failure
initiation.

65
Figure 5.45 Stereogram and Mohr failure analysis for the identified natural fractures with the in-
situ stress field estimations showing that the fractures are stable. The natural fractures are
plotted as small red diamonds on the Mohr diagram, and as small black diamonds on the
stereogram.

A pressure increase of 2000 psi, and a corresponding increase of 600 psi in Shmin resulted
in tensile failure occurring in the Trap Tester model (Figure 5.46). Shear failure cannot be
achieved in this model without lowering the cohesive strength input significantly. In order to see
failure in the Trap Tester model with the in-situ stress conditions, the cohesive strength of the
rock had to be lowered to 400 psi to begin to induce shear failure (Figure 5.47). Altindag and
Guney (2010) compiled a list of UCS values referenced in literature for various rock types and
particular formations around the world. There is no value listed for a calcite filled fracture, or for
a marl, but their range of values for limestones or similar carbonate rocks ranges between 20 –

66
140 MPa/ 2900 – 20300 psi, which is considerably higher than the 400 psi at which shear failure
is observed in this model.

Figure 5.46 Stereogram and Mohr failure diagram for increased pore pressure and minimum
horizontal stress showing tensile failure initiation once pressure was increased by 2000 psi but
other stress and rock strength values were held constant.

Healed fractures will be somewhere between the original rock strength and the non-
existent strength of an open void (0 psi), we expect healed fractures to be more compliant than
the background matrix, but the difference may be subtle (Personal Communication T. Bratton).
Experimental results indicate that the healing process and the re-strengthening of the rock after
initial failure can occur over short time periods (Tenthorey and Cox, 2006), and so it would be
inaccurate to assume that the natural fractures observed in the borehole image data are
necessarily very weak. Assessing the strength is also complicated by the fact that the fractures

67
may not be consistently healed, and may still be partially open based on changing trace intensity
in the image data. Regardless, the stability assessment shows that the rock will fail in tension
with an increase of reservoir pressure by 2000 psi, and it is reasonable to assume that shear
failure could be initiated in the fractures before that, or else concurrently, particularly because
in this model the coefficient of internal friction was held high.

Figure 5.47 Shear failure envelope reached in Trap Tester model with Unconfined Cohesive
Strength reduced to 400 psi.

The stability assessment set the upper bound of possible fracture fill strength before the
fractures would be critically stressed at 400 psi. Because this value is much higher than the
conventional use of zero strength, I propose that the identified fractures are likely candidates for
reactivation upon stimulation. Further refinement of the strength of the fracture fill would
improve the accuracy of this analysis.

68
5.5 Horizontal Feature Interpretation

The presence of horizontal features in the Vaca Muerta has been previously documented
(see Chapter 3). In particular, the presence of cemented ash beds, and the fibrous beefs may be
relevant to reservoir stimulation. The ash beds may be particularly difficult as their composition,
and the resulting mechanical contrast between the ash bed and surrounding rock matrix, might
result in fracture termination or in reservoir compartmentalization, if the fractures are able to
propagate through the ash bed but are not able to be maintained as open.

In addition to the natural horizontal fractures present, shales are commonly damaged by
tensile failure during the coring process (Fjaer et al., 2008) and methods to distinguish between
coring induced failure and natural fractures may be difficult to determine in a reservoir with a
high frequency of horizontal natural fractures such as the Vaca Muerta. Cementation on the
inside of any void space would indicate that a fracture had existed for long enough to have
reservoir fluids circulate and for mineral precipitation to occur, but this cannot always be seen
from core photos alone. In this instance, the choice was made to examine whether a
quantification of the number of beef and ash beds present in the wells is possible using the
integration of available core photos, CT scan photos, and the borehole image logs.

There is not a single well for which all three data sources are available. The lack of image
quality for Wells B and C means that, even after carefully correcting the core photo depths, it is
not advisable to try and correlate between the core features and the image log to discern what
the trace of the beef or ash bed look like. As such, Well G is a better candidate, but the core data
is not available for Well G, and so, only the CT scan photos can be used for comparison.
Distinguishing between BEEF structures and other resistive fractures in the CT scan data is
qualitative at best, and involves a high degree of uncertainty. The CT scan provides a density
based imaging of the unslabbed core, and will be best suited to features with highly contrasting
density fill relative to the rock matrix. The CT scan pictures are very high resolution and are a
good match for the FMI in terms of the relative scale of the resolvable features. Where the CT
scan pictures were loaded, there is generally very good agreement with the image log
interpretation (Figure 5.51).

69
Figure 5.51 Comparison of fracture trace observed in FMI with CT scan photos from the whole
core, inner core, and outer core.

Because the ash beds are highly variable, any idea of attempting to identify them without
the core photos was immediately discarded. The ash beds can be silica- or calcite-cemented, and
are in both sharp and gradational contact with the surrounding rock (Personal Communication,
LCV Core Lab, Buenos Aires, Argentina). Despite the challenges, a feasibility test for the
determination of the beef using borehole images was performed.

Comparing the CT scan photos to the image log, it appears that the features that could be
beef correspond to a bright, resistive layer in the FMI which may or may not be associated with
a planar, discontinuous layer which sometimes appeared nodular (Figure 5.52). The FMI was
interpreted for the most resistive layers, and for a corresponding nodular trace – ignoring
concretions and concentrating instead solely on planar features with a discontinuous trace which
might be related to the fibrous nature of the beef. Separately, the CT scan photos were flagged
in locations which might correspond to the beef, where high density layers with an outer halo
were interpreted to potentially be beef, and that the halo might be related to the two phases of
beef fill that has been proposed in the Neuquén basin (Rodrigues et al., 2009).

70
Figure 5.52 Comparison of the FMI with a likely beef identified using the CT scan pictures,
showing a highly resistive bed in FMI that is likely slightly offset due to an imperfect depth match
between the image log and the core depths.

The result was that there was not a match between the independent interpretation of
the bright, nodular trace in the FMI and the high density layers in the CT scan photos (Figure
5.53). The lack of agreement is inconclusive. Rather than disproving that the beef have a nodular
trace in the FMI, the only conclusions that can be made is that not all of the bright, nodular traces

71
in the FMI must correspond to beef. Without the core data to validate the interpretation, it is not
possible to assess whether the FMI or CT scan photo features truly correspond to beef.

Figure 5.53 Comparison showing the lack of agreement between nodular trace feature count in
the FMI interpretation with the count of possible beef in the CT scan data. High beef density in
the upper CT section does not correspond to an increase in nodular feature count in the count
log.

5.6 Integration with Sonic Data

For Well G a sonic scanner log was run, which provides information about the acoustic
anisotropy seen in the rock which may be related to stress or to the presence of fractures (Prioul
et al., 2007). The raw data was processed by Schlumberger and the interpretation provided to
RCP by Wintershall, and this was compared with the image interpretation for an additional
validation.

72
 The sonic scanner tool defines the near-field and far-field slownesses from axial,
azimuthal, and radial data collected using monopole and cross-dipole measurements
(Schlumberger). In this case, the Stoneley wave arrival and amplitude, the fast shear azimuth, as
well as the difference in arrival times and amplitude between the fast shear and slow shear could
be indicative of the presence of natural fractures. These logs have been used to characterize open
fractures (Prioul et al., 2007; Donald and Bratton, 2006) and since there is a degree of drilling
enhancement at the fracture margins evidenced in the BHI data, there may be a signal even
though the fractures are predominantly healed.

The difference between the fast and slow shear arrival times is at a maximum in the upper
Vaca Muerta, and in general, the arrival time anisotropy for the Vaca Muerta interval is relatively
low, with a maximum of 6%. This could imply that there is either relatively low stress sensitivity
over the Vaca Muerta interval in the area, or that the difference in magnitude between the
minimum and maximum horizontal stresses is very low.

The fast shear azimuth is not consistent over the Vaca Muerta, and only indicates a
preferred orientation at a few, thin intervals in the Quintuco, otherwise the fast shear azimuth
rotates with the tool orientation around the borehole (Figure 5.61). Where the fast shear azimuth
does lock on to a consistent orientation, the azimuth corresponds with SHmax as interpreted by
the DIFs in the Upper Vaca Muerta. This could be a further indication that the fractures are healed
with a material that is similar to the rock cement and therefore the fractures are not creating a
response in the shear arrival times. In rock where the fractures are open, there should be a
preferred fast shear orientation that is aligned with fracture strike (Prioul et al., 2007).

That the entire fractured zone does not display high anisotropy values in the sonic scanner
is a further confirmation of the healed nature of the natural fractures that were observed in the
BHI data. What the local maximum does correlate with is a small cluster of conductive fractures,
so, the sonic data is sensitive to the presence of fractures, but the cemented fractures are not
differentiated from the surrounding matrix.

73
Figure 5.61 The highest fracture intensity (F.I.) zone corresponds to a very brief preferred fast
shear orientation, as well as a small local maximum in the time and slowness anisotropy.

The Stoneley wave slowness does correspond better to the fractured zone (Figure 5.62).
The Stoneley wave has been proven to be sensitive to open fractures (Hornby et al., 1989) and it
is possible that in this case the wave is responding to the enhancement of the natural fractures
by drilling. The sonic anisotropy may be missing the fracture presence because that signal is
masked by the combination of the presence of the natural fractures, the effects of the horizontal
stress anisotropy – and the fact that the current orientation of the maximum horizontal stress
does not correspond with the fracture orientations, or the laminated nature of the formation.
The upper and lower Vaca Muerta tend to be more highly laminated than the middle, transitional
zone (Garcia et al., 2013), and this may be the cause of the relatively higher acoustic anisotropy
o se ed i the uppe a d lo e i te als a o di g to “ hlu e ge s p o essi g.

74
Figure 5.62 The zone of fracture occurrence as indicated by the F.I. log corresponds qualitatively
to an increase in Stoneley transit time.

Having observed that the Stoneley slowness corresponds roughly to the zone of highest
fracture occurrence, a comparison in Figure 5.20 between the Stoneley Slowness and the organic
content of the interval, and that comparison is in much better agreement. In Figure 5.63, the
fractured zone has been colored red to distinguish it, but the Stoneley slowness begins to
increase ahead of the fractured zone, in perfect timing with the increase in weight percent of
organic matter as predicted by the lithoscanner. For this reason, I propose that the sonic log
response is controlled by the degree of lamination and concentration of organic matter in the
rock, and not to the presence of the healed fractures.

75
Figure 5.63 Stoneley Slowness plotted with kerogen estimates from log and cuttings data,
depicting a better match between Stoneley response and kerogen content rather than the
presence of healed fractures.

76
 Chapter 6

Cluster Analysis

Previous hydraulic fracture modelling has suggested that, in order to effectively produce
in the Vaca Muerta, the completion may need to be varied depending on the depth. In particular,
that the Upper Vaca Muerta requires greater fracture conductivity to be productive, whereas in
the lower Vaca Muerta production is more dependent on having large fracture surface areas
(Ejofodomi et al., 2014). The formation is variable from top to base, but the lithological
differences are subtle and may be difficult to distinguish in wireline logs, which can make
predicting reservoir behavior challenging. The cluster analysis was used to try and resolve the
subtle differences that may be relevant to the mechanical characterization of the reservoir at the
log scale.

6.1 Cluster Analysis Workflow

Earlier work in the RCP group has examined the differing components of rock behavior in
terms of composition and fabric (Duenas, 2013; Davey, 2012). In unconventional reservoirs, the
behavior of the rock matrix when stimulated depends on a variety of factors, meaning that a
ese oi s ualit does t solel elate to po osit a d pe ea ilit , ut also to the deg ee of
lamination, the mineralogy, and the stresses present at depth. If, in unconventional reservoirs,
the traditional parameters used to evaluate reservoir quality are not likely to be the criteria that
determine success, then it is essential to develop unique criterion that are subject to the
particular sensitivities of a given reservoir.

Standard logs are not capable of resolving the thin-bedded nature of some
unconventional reservoirs, as the sample rates at which they acquire data do not capture the
layering of shale formations (Sagasti et al., 2014; Lagraba et al., 2010). In the Vaca Muerta, the
productive interval is thick and not at risk of being missed by the log sampling rates, but since the
lithological variations within it are relatively minor, the clustering technique is used to group
similar rocks in order to look for patterns in their occurrence, vertically, and in adjacent wells. In
order to separate out the different influences on the log response in the modeled interval, the

77
final model was built iteratively, with added components in each step by incorporating logs which
are sensitive to a differing range of rock properties (Figure 6.11)

Log Preconditioning for

Quality Control

First Model: GR, RHOZ,

Choose Input Suite
Cluster Analysis Workflow

PE, NPOR

Run Cluster Model on

Training Well

Assess Model Outputs 9 cluster classes suitable

Apply Training Model to

Offset Wells

Evaluate appropriate
model usage
Second Model: GR, RHOZ,
VPVS, YM
Alter Input Suite as
Necessary
Final Model: GR, RHOZ,
YM, VPVS, Res

Figure 6.11 Cluster analysis workflow.

78
6.2 Cluster Model Iterations

The cluster models use a mix of recorded and calculated logs which are commonly run,
and were chosen in part to maximize the number of wells on which the model can be applied
(Figure 6.21). All of the models were first built with Well B and then tagged to the other wells.

Figure 6.21 Example from Well G for all logs which were used as input for the three cluster
models.

The first cluster model was built to preferentially examine compositionally based
behavior, and input logs were chosen that would represent bulk volume rock properties without
any effects from texture or fabric (Figure 6.22). Although overburden and underburden rock are

79
contributors to the mechanical behavior of the reservoir (Fjaer et al., 2008), the decision was
made to perform the cluster analysis over the Vaca Muerta interval alone, leaving out any log
response recorded in the Quintuco or Tordillo Formations in order to avoid the need to splice
together logs from different runs and to obtain the maximum amount of consistency between
logging runs from the different wells. The first model consisted of bulk density, gamma ray,
neutron porosity, and photo electric factor.

Figure 6.22 Compositionally based log model, as Well G had not yet been logged, only Wells A, B,
and C were modelled.

80
 The compositional model displays orange and green cluster classes grouped towards the
top of the interval, transitioning to pink, red and purple through the middle. The Lower Vaca
Muerta plots quite differently from well to well, with a dark blue trace in Well B in which the
model was built, compared to a homogeneous purple trace in Well C, and a highly variable purple
to blue to red trace in Well A. For reference, Well A and Well C are each located approximately
10 km away from Well B, and 20 km from each other and so a degree of inhomogeneity is
expected between them. Well B has the highest gamma log signature of any of the wells in the
study area, and it is possible that the uniqueness of the blue class dominating the compositional
cluster in the Lower Vaca Muerta is related to the gamma response in that well. The alternations
between the pink, red, and purple classes through the middle interval may be related to changing
carbonate and silica ratios. Figure 6.23 displays the different XRD data for Wells A, B, and C, and
demonstrates a relatively constant amount of clay content with high frequency variability in the
silica to carbonate relative amounts.

Figure 6.23 XRD for Wells A, B, C, with red line indicating approximate location of 30% clay cutoff,
showing that the clay content in this area remains nearly constantly below 30% (Wintershall).

81
 After running the bulk volume properties model, a second model was run which
incorporated rock fabric by adding sonic logs (Figure 6.24). The compressional and shear sonic
logs were added to the model, not directly, but indirectly by replacing the neutron porosity and
photo ele t i logs ith d a i You g s Modulus, a d the atio of o p essio al elo it to
shear elo it VpVs . The d a i You g s odulus is al ulated f o so i logs, ut usi g
You g s Modulus i stead of the o p essio al o shea slo ess, the e i a i he e t li k to the
language used to describe mechanical behavior. Similarly with Vp/Vs, the goal was to incorporate
parameters in the model which would create electrofacies which could intuitively be related to
other forms of analysis, such as seismic inversion. By including texture and composition in the
second log run, it is possible to begin to determine zones of similar material properties (Williams
et al., 2014), and the output will be related both to deposition and diagenesis. This intermediate
odel is alled the fa i odel.

The fabric model incorporates both composition and texture, which are equally important
in terms of mechanical rock behavior. A rock body that is compositionally homogeneous, but
which varies in terms of texture or fabric will behave differently when stressed (Montaut, 2012).

These four logs were chosen for input to the intermediate model; however, there are
other logs that describe the mechanical behavior of the rock that could have been included. In
pa ti ula , the e ha i al ualit o f a ka ilit of rocks is frequently described in terms of
ittle ess as des i ed You g s Modulus a d Poisso s ‘atio ‘i k a et al., . Poisso s
Ratio, as defined in Chapter 3, describes the lateral deformation of a material in response to axial
loading (Fox et al., . The alues of Poisso s ‘atio ge e all fall et ee a d . , he e
is a material that exhibits no lateral strain when the axial stress is applied, and 0.5 is a plastic
material that will expand equally laterally and axially. The value ranges in the Vaca Muerta are
p i a il et ee . a d . , hi h ould o espo d to a ha d a o ate ate ial Ba ee,
. The Poisso s ‘atio o e the Va a Mue ta i the ells a aila le fo this stud a ies e
little (variance ranges 0.2 - 0.09 x 10-3) and would not add value as a discriminator for the cluster
model.

82
Figure 6.24 Fabric based log model, with data from Well G also included.

The fabric model, like the compositional model, has three predominant zones obvious in
terms of cluster occurrence. The uppermost Vaca Muerta plots as green, blue, and yellow cluster
classes, whereas the middle transitional to orange and pink, with a pink and/or purple Lower
Vaca Muerta. The depth at which the transition out of the upper clusters to the middle clusters
occurs is much more variable per well relative to the compositional model, with a much thinner
blue/green/yellow zone observed in Well G relative to Well C, for example. Similar to the

83
compositional model, the Lower Vaca Muerta is most unique in Well B, although Well A and Well
C are more similar in the fabric model than in the compositional.

Finally, a purpose driven model was created (Figure 6.25). The purpose driven model was
designed specifically to try and address the mechanical behavior of the rock as it relates to
composition and texture, which was already inherent in the fabric model, but with the added
goal of identification of ash beds, which are known to exist in the area and may be mechanically
significant (Williams et al., 2014).

Figure 6.25 The mechanical model which involves both texturally and compositionally sensitive
logs and was built to identify mechanically similar zones, and to attempt to resolve mechanically
significant thin layers.

84
 Because the ash beds have a higher abundance of clay material relative to the rock matrix,
they are most readily visible in the bulk density and resistivity logs – where the low grain density
of clay, and the fluid captured by the clays should have the greatest effect on the log response
(Passey et al., 2010). The density log was already in the first mechanical model, and with the
addition of resistivity, the fit-to-purpose model was created. The ash beds should result in a
unique log response depending on their thickness and have previously resulted in the appearance
of wholly unique and thin cluster classes (Williams et al., 2014), but it was not definitively clear
which of the thin beds were associated with the ash beds, as many of the classes are thin. The
effect of the ash bed will be amplified when water based drilling muds are used, and it is possible
that the oil based mud that was used in Wells A, B, and C is not picking up the signal as well as if
they had been drilled with water based mud. To test this, the same input logs were used, but
with the model being constructed with the data from Well G as opposed to Well B (Figure 6.26).
The result is that the stacking pattern of the clusters is not changed if all the other parameters
are kept the same. For that reason, and because there is no core data to validate the
interpretation with, it is not possible to tell which of the thin cluster classes represent the ash
beds.

6.3 Interpreting the Cluster Results

When considering geomechanical properties, relatively thin features can be significant to

the mechanical behavior of the system as it relates to fracture propagation (Chapter 3). Stress
contrasts between different layers mean that a relatively thin bed, which is significantly different
in terms of its material properties from the surrounding rock matrix, will have a significant impact
on fracture propagation or containment (Fox et al., 2013). Since the stress profiles from logs are
derived using averages (because of the sampling rate of acoustic logs), it is difficult to accurately
predict discrete layer properties – there is an inherent smoothing that has already taken place.
These discontinuities may be significant as shear planes, or points of stress concentration and by
setting a high number of cluster classes the thinnest unique zones can be identified. However,
the thinner the cluster group is, the higher the uncertainty related to that cluster will be, since
there may be insufficient sample volume for a statistically significant interpretation (Sagasti et

85
al., 2014). It is a fine line that the interpreter must navigate to balance the need to resolve thin,
but significant intervals, with the uncertainty that modeling thin beds may introduce.

Figure 6.26 Input logs to the mechanical model from Well G next to two versions of the model,
built from Well G logs, or built from Well B logs but tagged to Well G.

For the sake of resolving thin beds, and to minimize losing model resolution early in the
process, the number of clusters was purposely set high. For the bulk composition model, the
fabric model, and the mechanical model, 9 cluster classes were chosen. The fall-off plots
indicated a better minimum was achieved with the bulk composition and fabric models, but the
mechanical model was still deemed to be acceptable.

The initial interpretation served simply to define the unique attributes of each cluster.
The average thicknesses, frequency of occurrence, and the typical property ranges are the basis

86
for the initial interpretation. The property ranges for each cluster class in the mechanical model
are defined in Table 6.1 in the Appendix. In general, the property variability between unique
cluster classes is low. For a property that is generally variable over the interval, such as GR, the
arithmetic mean of the clusters varies between 62 API and 182 API. For a property that is
generally constant over the interval, such as VPVS, the arithmetic mean varies only between 1.67
and 1.93.

The frequency of cluster variation was calculated over 1 m and 5 m intervals and is called
luste f e ue , as ell as a s ale of ho a luste lasses a e p ese t o e the sa e
interval, hi h is alled i te al t pe Figu e . . The diffe e e et ee these t o
parameters is that the i te als t pe o l diffe e tiates et ee diffe e t luste lasses ei g
present but does not account for whether it is in one package or several, split occurrences, where
the luste f e ue dis i i ates et ee epetitio s of the sa e lass i a gi e i te al i
order to distinguish how many discrete clusters are present. The difference between these two
parameters is therefore a measure of whether a given interval consists of simply alternating
classes or if several unique clusters are involved. Differentiating one from the other is relevant
because of the similarity between some of the cluster classes. Although the model incorporates
all inputs equally in order to assign appropriate classes, using 9 classes in the model means that
there is some overlap in the material properties described by adjacent clusters.

Figure 6.31 displays the difference between cluster frequency logs normalized by 1 m or
5 m intervals, and smoothed in both instances, with the fracture intensity log from the image log
interpretation. The cluster frequency is highest in the upper half of the well, and decreases with
depth. The difference between the two measures, the count of unique clusters per interval and
the ou t of luste lasses pe i te al is the Δ I.T. a d C.F. log, hi h spikes ut is elati el
constant with depth, indicating that no particular zone is more prone to alternation between
similar cluster types.

Garcia et al. (2013) describes the upper and lower Vaca Muerta as being the most
laminated, with a relatively homogeneous middle interval. Although the cluster model predicts a
higher degree of lamination towards the Upper Vaca Muerta, it does not suggest a

87
correspondingly high degree of lamination in the lower interval. It is possible that the presence
of concretions, and beef are obscuring the log response in the lower zone, along with the
presence of a significant amount of organic material by weight %.

Figure 6.31 Mechanical cluster model with the frequency and type logs used to describe the
cluster class occurrences, with fracture intensity, and the difference between the cluster
frequency (C.F.) and cluster/interval type (I.T.) log for Well G.

Each cluster was also plotted on its own log in order to document more clearly how each
occurrence is distributed within the interval (Figure 6.32). Some classes (1 in black, and 7 in
orange) are confined to particular depth zones, but class 5 (green) occurs in much of the middle
of the interval and also appears in the uppermost and Lower Vaca Muerta. Class 4 is not present
in Well G, which is not a problem, as it was defined in Well B and may be absent in Well G, due
to any number of geologic reasons, such as a slightly different depositional position.

88
Figure 6.32 Cluster by occurrence depth for Well G. Scale is arbitrary and was chosen for
visualization purposes. For the same reason, class 3 was colored white as opposed to light blue.

How the clusters are distributed and how they alternate depending on their individual
class is a starting point for the cluster interpretation, but what is of greater interest to this analysis
are the stacking patterns of the cluster classes. Figure 6.33 shows the 3 cluster output models
side by side, and then attempts to separate out the gross zones which appear to exhibit similar
cluster patterns. There are 3 large scale zones in the Vaca Muerta that are clear from a cursory
glance at the logs, and were described in Chapter 1. With the cluster analysis, there are 6 zones
that exhibit similar clustering patterns in Well B. The top three zones correspond to a clear shift
between the silica and carbonate estimates from the XRD (Figure 6.34), where the uppermost
zone is more siliceous, then the next zone is more carbonate rich, and then there is an additional
influx of silica.

89
Figure 6.33 Final cluster models with an interpretation of six zones (Zx) of similar stacking
patterns. The discriminations based on stacking pattern are consistent in the compositional and
fabric models for Well B, for which all models were built.

Figure 6.34 The six cluster zones identified in Well B compared to well log and XRD data.

90
 Matching the cluster zones across all wells is not perfectly consistent, although Z1, Z5,
and Z6 are very constant across all four wells. The middle zones are the most variable, with
thicknesses that range, and some ambiguity as to the definitions of Z2-Z4 (Figure 6.35). However,
this could be related to depositional or diagenetic differences, and there is no requirement for
the clusters to be consistent from well to well, rather, the power of the cluster model is to allow
for the prediction of similar rock properties in the context of a model which is preferentially
sensitive to whatever parameter the interpreter choses. In this instance, between 4 and 6 similar
zones were recognized over a distance of approximately 25 km of offset. It appears that in the
upper interval, the stacking patterns correspond more closely with mineralogy, whereas towards
the base of the Formation, the stacking patterns are influenced heavily by the concentration of
organic material.

Figure 6.35 Correlation of the cluster zones for the additional wells in the study area.

91
6.4 Testing the Cluster Model Sensitivity

The number of assigned clusters must be chosen in order to ensure that the thin-bedded
variations are captured without creating a model that includes too much noise relative to
information. The diagnostic plots that are generated by the HRA module in TechLog allow for the
user to determine whether their model inputs are acceptable, but it is difficult to quantify what
the difference between an acceptable model and a robust model is. To that end, some additional
sensitivity analysis on the cluster models was performed, primarily on the fabric model.

The sensitivity analysis consisted of running the model several times over while changing
si gle i put pa a ete s. Be ause of the u supe ised atu e of the odel s i itial construction,
there should be a degree of variability allowed for between separate trials. The class assignation
also has no physical meaning, and so the classification number does not relate to any particular
feature of the model and is not considered when comparing output models. These factors make
the quantified testing of model similarity challenging, and so, a qualitative analysis was
performed looking simply for reasonable match between the class stacking patterns.

The first test was to run the bulk composition and fabric models while removing a single
log each time. This test should alter the model to a certain degree because each log should be
equally weighted i the odel s o st u tio , ut as do e to e su e that o si gle log as
controlling the model to an undue degree. The total gamma ray was also swapped with the
spectral gamma minus uranium, since the concentration of uranium has a large effect on the
gamma character in the Vaca Muerta.

The second test involved altering the logs to simulate the effects of logging noise. This
was designed to determine how sensitive the model is to noise, and whether there may be an
over-interpretation of the variability in the classes. A random noise mask was added to the
gamma ray log in one run, and in the other, Gaussian smoothing was applied to reduce variability.
These were separately swapped in for the actual gamma log. Neither the artificially noisy nor the
artificially smooth gamma altered the model in a significant way. This indicates to the interpreter
that there is not an overabundance of noise contained in the model and that it is sufficiently
robust.

92
Figure 6.41 Comparison of the different gamma curves used to test the cluster model and the
resultant output. The model initiates randomly with each rerun, but the stacking patterns are
visibly similar.

The cluster output was also compared in a few places to the image log for Well G for
assessment of the validity of thin, anomalous beds. There are spikes in the well logs that may be
geologically valid or else may be logging artifacts. No de-spiking was done to the well logs, and
this comparison provides an indication of whether or not the spikes are causing an alteration in
the cluster output that may skew the model. Not every spike can clearly be correlated to a
geologic anomaly in the image log, but in a few instances, it was clear that the log spikes, and the
corresponding cluster output, corresponded to real geologic variability. Figure 6.42 shows an
anomalous trace in the FMI where the sediment has been deformed by the presence of a feature
in the center of the image, perhaps a calcareous concretions which has some internal structure
that can still be seen in the image. The location of this anomaly corresponds in depth to a spike
on the GR, RHOZ, and lithoscanner logs, and also to the additional logs that have subsequently

93
been calculated. This seems to further confirm that the cluster model is recording true signal,
and that there is not a significant overweighting of log noise.

Figure 6.42 Correlation between anomalous feature identified in image log and well log response
and subsequent cluster class discrimination.

94
 CHAPTER 7

Final Interpretation

The Vaca Muerta offers an excellent opportunity to study field development at an early
stage, as drilling in the Neuquén basin has not yet reached the levels of the Barnett, Eagle Ford
or Niobrara. Even where wells are being drilled in rapid succession, as in the 174 wells drilled in
YPF “ Lo a Ca pa a field i , p odu tio histo is elati el sho t, a d the d a i s of
how the production may change with time is still to be anticipated (Rimedio et al., 2015). The
particularities of how best to characterize and develop this play are still very much being
developed, and this presents a fantastic opportunity to test various techniques and processes.

7.1 Conclusions of the Borehole Image Analysis

The interpretation of natural fractures in the borehole image data available for the study
area indicates the presence of two fracture sets, oriented orthogonally to each other at azimuths
of N50 - 230°, and N145 - 325° which can clearly be identified in Well G. In Well B two resistive
fractures could be definitively recognized, and in Well C there was one. Although these fractures
are not numerous enough to be statistically significant on their own, they are oriented in the
same direction as the two sets identified in Well G. Beyond the study area, similar natural fracture
orientations have been identified in other areas of the Neuquén embayment (Figure 7.11), and
have been used to identify sweet spots, with the caveat that their presence requires a thoughtful
development strategy to mitigate against well interference (Rimedio et al., 2015). As such, this
appears to be a relatively regionally consistent set of fractures which may either be related to
large-scale tectonics or rock properties.

Generally, tectonic fractures would be inconsistent regionally relative to non-tectonic

fractures (Fox et al., 2013). Tectonic fracture orientations can be related to paleo tectonic
stresses and the non-tectonic orientations are governed by rock properties. Compression and
resulting deformation in the Neuquén basin has been ongoing since the Triassic, broken up by
pulses of relaxation. Mosquera and Ramos (2006) describe a rotation in the dominant
compressional orientation in the Neuquén basin from NW-SE to W-E in the late Mesozoic to the
Cenozoic, which is in line with the orientation of one of the fracture sets observed in these wells.

95
Other studies have related fracturing in the Vaca Muerta to the volume expansion of organic
matter during catagenesis (Rodrigues et al., 2009), the process by which organic matter becomes
oil. This is a perfectly reasonable theory and might explain why the natural fractures
preferentially occur in the middle of the Vaca Muerta, close to the richest source interval, but in
lithologies more prone to fracturing.

Figure 7.11 Comparison of natural fracture orientations recognized in the study area analysis and
an independent study conducted by YPF and Chevron in another part of the basin (Rimedio et al.,
2015).

I propose that the natural fractures observed in this analysis were created initially due to
the formation of hydrocarbons, but that their orientation was related to the paleo-stress
orientation at the time, and the timing of their formation was during the Cenozoic while the
stress rotatio as o goi g. This is o siste t ith Mit hu a d Ulia a s asse tio that
generation of oil in the Vaca Muerta began at the Cretaceous – Paleogene boundary.

96
 Previously, natural fractures have been described in the Upper Vaca Muerta and Quintuco
Formations (Badessich and Krasuk, 2014; Licitra et al., 2015), and the presence of natural
fractures has been shown to result in the formation of complex fracture networks in the play
(Badessich and Berrios, 2012; Ejofodomi et al., 2014), and to sometimes be a controlling factor
in initial production rates (Collins et al., 2015; Rimedio et al., 2015). If it is possible to reactivate
those natural fractures, as well as create new tensile fractures, then a mixed mechanism
stimulation could be achieved (McClure, 2013), and production could be maximized (Figure 7.12).
The stability assessment performed indicated that there is a high possibility of reactivating the
natural fracture network documented in the middle Vaca Muerta and I suggest that future
development plans should incorporate the presence of this regionally extensive system.

Figure 7.12 Geometric representation of hydraulic fracture growth in the Barnett where natural
fracture reactivation is contributing to fracture complexity, and a larger stimulated volume as
seen in the microseismic inset (b) (Gale et al., 2007).

97
7.2 Conclusions from Cluster Modelling and Combined Interpretation

Predicting the likely mechanical behavior of any reservoir is challenging under the best of
conditions given the dynamic processes that are constantly at work in geologic reservoirs, and
the variety of time scale over which these processes are active. In this instance, an analysis was
performed in order to try and constrain log scale variability within the Vaca Muerta using cluster
analysis, so as to group zones of similar material properties. This workflow is a relatively simple
way of discriminating the reservoir into like zones at a given time, without considering the
dynamic nature of the reservoir. This is not a true prediction of the mechanical behavior of the
rock because no static or rock strength information was available from core tests. However, as a
high level discriminator over the reservoir interval, a thoughtful discussion of the cluster model,
even where the uncertainty is known to be high, is suitable.

There are several criteria that are frequently used to assess the potential of
unconventional reservoirs (Table 7.1). The organic matter in the Vaca Muerta is sufficiently high
and though maturity ranges, the Wintershall study area is located within the oil window. The
conditions are favorable in the reservoir for completions, in the sense that the clay content is
low, and that the mineralogy consists primarily of carbonate and silica, Pressure is high to
facilitate completions, but there is still uncertainty with regards to the stress and pressure
relations and the rock strength (Ejofodomi et al., 2014; Garcia et al., 2014). In terms of
infrastructure, the Neuquén basin is relatively remote, and this certainly contributes to the
relatively high well costs. Despite the infrastructure challenge, it should be clear, that the Vaca
Muerta makes for an excellent unconventional reservoir. In fact, the Vaca Muerta makes for such
an excellent unconventional reservoir that choosing the best landing zone is not straightforward,
as there are several possible targets which may each be unique in their challenges. Some kind of
reservoir discrimination scheme is needed in order to evaluate the relative quality of the different
zones. Mechanical property variations across layer interfaces are particularly interesting because
of the role these variations play in fracture growth or containment (Sagasti et al., 2015). The
mechanical property variations are associated with facies variations, and by examining the
distributions of similar mechanical zones the impact that facies changes are likely to have on
reservoir performance are examined.

98
Table 7.1 Evaluation parameters for an unconventional reservoir.
Category Criteria Vaca Muerta Evaluation

Thickness Excellent

Volume Concentration of Organic Matter Excellent

Maturity of Organic Matter Excellent

Mineralogy Excellent

Completions Pressure Good but with uncertainty

Natural Fractures Present

Porosity Good
Reservoir Quality
Permeability Good

Access to infrastructure Challenging

Development
Well costs High

Discriminating the reservoir for mechanical properties can be done in any number of
ways, using sequence stratigraphic interpretations from seismic, core analysis, comparing
production results between wells, or looking at representative outcrops. In this case, the decision
as ade to use Te hLog s Hete oge eous ‘o k A al sis odule to uild a luste odel. The
cluster model was built iteratively to incorporate additional levels of complexity in the properties
which were being modelled. The final result predicts 5 zones of similar cluster class patterns
consistently observed between all of the modelled wells (Figure 7.21).

The lithological variations in the clusters are relatively minor and some of the clusters
describe similar ranges of material properties, but pattern recognition can be used to correlate
between wells and to provide the geologic insight without risking over-interpretation. The zones

99
are described in terms of the cluster class distributions and rock properties in Table 7.2. Well B is
unique for having 6 zones, with an additional breakout in the middle of the interval which was
not observed in the other wells. Z1 is the uppermost zone recognized in all wells and has thicker
alternations of the yellow and red and orange clusters. These are the clusters with the highest
esisti it , lo est ga a, highest You g s Modulus, a d high de sit , hi h ep ese t the ost
carbonate rich rocks that are poorest in organic material. Moving in to the middle zones (2 and
3) there is an increase in the gamma response, and the amount of organic material. From Zone 2
to Zone 4 there is an abrupt change in the resistivity log (Zone 3 to Zone 4 in Well B); this could
likely be related to a change in porosity that corresponds with the increasing siliciclastic content,
and the increase in organic matter. At the base of the interval, the green clusters are seen just
above the top of the Lower Vaca Muerta, and then the kitchen itself is relatively homogeneous
and models as the black cluster class.

Figure 7.21 Correlation of the cluster class zones between the study wells.

Correlation of the cluster occurrence with the image log offers valuable insight into the
discrimination. The black cluster group begins to appear in Well G at the top of the Lower Vaca

100
Muerta and very rarely before that. This is coincident with the first appearance of the calcareous
concretions in relatively high concentrations; previously they appeared as scattered features. In
the Lower Vaca Muerta there are decimeter thick intervals that are nodule rich that correspond
with assignation of the black cluster class (Figure 7.22). Similarly, the light green cluster group
corresponds to a high degree of lamination (Figure 7.23), a finding which is not consistent with
the interpretation that the cluster frequency log is able to recognize laminations, as the green
group occurs over the whole interval and is not limited to the upper Vaca Muerta where the C.F.
log is highest. It does confirm, however, that the mechanical model is resolving zones of similar
texture.

Figure 7.22 Calcareous concretions in the image log that appear as bright white spots which,
where multiple concretions appear, corresponds to the black cluster group recognized by the
mechanical cluster model. The micro-resistivity log is also displayed on the two left hand images
and shows a response to the presence of the clusters.

101
Figure 7.23 High degree of lamination in the light green cluster class seen in two examples on the
borehole image (scales are same) and a comparison of the light green cluster occurrence with
the C.F. log.

This result suggests that there is actually a higher degree of lamination towards the center
of the Vaca Muerta than the C.F. alone suggested and should serve as a reminder to be mindful
of the incorporation of the cluster model. The higher frequency alternation of clusters in the
uppermost Vaca Muerta relative to the middle, where one cluster class defines a highly laminated
unit, may suggest that the compositional differences between layers in the upper portion are
more significant, and that is what is driving the model to define thin and rapidly changing classes
as opposed to a single, highly laminated class such as is defined in the middle.

The designation of the green cluster class as being highly laminated, and being dominant
through the middle of the interval, is important when considering fracture containment. The
lamination planes will act as barriers to fracture growth as energy is lost along the lamination
surface (McCounaughy and Engelder, 1999), and differences in elastic properties between layers
will arrest growth. Because the drilling induced fractures are able to continuously propagate
through the layers in the middle section with minimal rotation of their orientation, it is
reasonable to assume that the elastic properties vary relatively little from layer to layer, which
may confirm that the green cluster class represents zones of high lamination where the adjacent
laminae are relatively similar. The elastic property contrast will affect the propagation from one
layer to another, but the contrast in stress is also capable of halting fracture growth.

102
 The lack of sonic anisotropy over the middle interval, where the presence of natural
fractures has been observed, is puzzling (Figure 7.24). The sonic anisotropy is dependent on the
presence of fractures, the stress sensitivity of the rocks, and the magnitude of difference
between the horizontal stresses. If the rocks are not stress sensitive or if the magnitudes of the
horizontal stresses are approximately equal then the rock will not display much anisotropy
(Personal Communication T. Bratton).

The magnitude of the difference between the horizontal stresses has been approximated
to be 1.4 in areas around the study site in the Neuquén basin, although it decreases with depth
(Garcia et al., 2013), and the rocks should be stress sensitive based on their material properties.
I suggest that the lack of anisotropy seen in the sonic log is because of the competing influence
of fractures, bedding, and stress over the middle Vaca Muerta interval, and that this competition
is creating a series of preferred orientations which make the interval appear relatively isotropic
in the log response. The direction of anisotropy relative to the orientation of the largest principal
stress affects the orientation of any initiated failure (Fjaer et al., 2008), and the sonic scanner
response should be studied in more detail in order to refine this interpretation. If, instead, it can
be shown that the sonic log response is due to a lack of stress anisotropy or stress sensitivity,
then this would suggest a relative lack of barriers to fracture growth through the middle Vaca
Muerta.

Figure 7.24 Cluster Output alongside sonic logs used to characterize formation anisotropy.

103
Table 7.2 Zones defined by the mechanical cluster model.

Zone 1 Red, orange, and yellow clusters with highest carbonate content, lowest TOC.
Classes are relatively thick.

Zone 2 Predominantly red, orange, and yellow clusters with light blue. Classes are
relatively thin and alternate rapidly.

Zone 3 Only present in Well B, light blue cluster alternating with yellow.

Zone 4 Dark blue and green clusters – which represent a high degree of lamination.
Sudden drop in resistivity which could be related to change in pore structure,
particularly as gamma is also rising along with TOC increase through this
interval.

Zone 5 Thin layer of green clusters directly above the Lower Vaca Muerta.

Zone 6 Predominantly consists of the black cluster group, has abundant calcareous
nodules, and the highest TOC.

Figure 7.25 XRD data for Wells A, B, and C split into Lower and Upper Vaca Muerta depicting the
higher proportion of carbonate in the Upper Vaca Muerta relative to the Lower.

104
 In a regional context, the lowermost zones (5 and 6) are most likely representative of a
transgressive, aggradational system. This is the lowermost Vaca Muerta which appears as bright,
parallel reflectors in the seismic (Figure 7.27). Above the Lower Vaca Muerta, the system is
progradational and the direction of progradation is known to have been from south-east to
north-west (Licitra et al., 2015).

Figure 7.26 Cluster class zones in Well G and corresponding mineralogy and TOC estimates, along
with the resistivity drop that occurs at the limit between Zone 2 and Zone 4.

Figure 7.27 Flattened seismic data showing the bright, relatively planar and parallel nature of the
Base Vaca Muerta when compared to the middle Vaca Muerta. Cyan reflector is approximation
of top Vaca Muerta.

105
 The direction of progradation is significant, as it may explain why Well B displays an
additional clustered zone. Looking at the well location map (Figure 7.28), Well B sits further to
the SE relative to the other wells, which all follow a roughly SW-NE transect. A simple explanation
for the additional zone (zone 3) could be that Well B represents a position in the system that is
further back and higher up in the clinoform and that the additional zone recognized by the cluster
model is related to an additional aggradational or progradational package (Figure 7.29).

Figure 7.28 The location of Well G, Well A, and Well C on a SW-NE transect of which Well B is off
trend.

Figure 7.29 Possible explanation for the additional zone is Well B is that it was drilled closer to
position y, including more aggradational topsets, relative to position x.

106
 The result of this integration is a characterization of the formation as consisting of an
uppermost Vaca Muerta that is relatively unfractured and where sonic anisotropy is relatively
high due possibly to the reservoir stress. Moving through the middle Vaca Muerta into a more
thinly bedded zone, there are natural fractures present which are filled with calcite cement, and
in the Lower Vaca Muerta, organic content is the highest and the presence of multiple calcareous
concretions is noted. These interpretations are consistent with regional literature while also
reflecting the unique aspects of the data set.

107
 CHAPTER 8

Conclusions

In Argentina, well costs may be much higher than in analogous North American shale
plays, and so predictive models are needed to ensure that development can proceed
economically (Fantin et al., 2014). During early development, the goal must be to systematically
characterize the differing aspects of the Vaca Muerta which may contribute to production, and
in doing so, to eventually move beyond purely observational characterization to building the
necessary predictive models.

8.1 Applicability of Analysis and Implications for Completion Quality

Reservoir stress and points of pre-existing weakness related to earlier failure and rock
fabric are major influences on the development of hydraulic fractures (Garcia et al., 2013).
Knowing about these factors will allow for the most informed asset management decisions to be
made (Hryb et al., 2014), and the motivation for this thesis was to perform a systematic of the
natural fractures present in the reservoir, and the changing rock properties in the study wells.
This is just one aspect of the systematic characterization of the Vaca Muerta, and continued
analysis will further build the predictive ability of the RCP groups work.

Rickman et al. (2008) defined the four criteria required to successfully design a hydraulic
fracture stimulation; formation brittleness, closure pressure, proppant design, and likely fracture
initiation points. Because fracture initiation and propagation are highly dependent on
mineralogy, stress state, and rock fabric, the different mineralogical distributions across the Vaca
Muerta may have a significant effect on what constitutes the most effective completion method
for a given depth in the interval (Garcia et al., 2013; Ejofodomi et al., 2014).

There is no way to definitively link mechanical properties to fracture characterization;

both are dynamic properties and change with time (Laubach et al., 2009). Gale et al. (2009) rightly
notes that even where the mechanical properties of a rock are the same, the presence of natural
fractures is not guaranteed, and may vary based on other factors such as structural position, layer
thickness, or for any number of ambiguous factors. This thesis is not suggesting that the presence

108
of the natural fractures in Zone 4 in Well G means that natural fractures will be found in Zone 4
in all other wells. Rather, it is suggesting that based on an evaluation of the quality of Zone 4, and
complimented by the observed presence of natural fractures, the middle Vaca Muerta has been
identified as a highly prospective zone.

The middle of the interval is suitably brittle, organic rich, and has now been documented
to contain multiple natural fractures which are likely to be reactivated during stimulation.
Although this conclusion was reached independently, it is corroborated by regional literature
(Fantin et al., 2014; Williams et al., 2014). In addition, geochemical analysis on the quality of the
liquid hydrocarbons in the Vaca Muerta indicates that the middle interval has high volumes, with
a moderate gas-to-oil ratio, and moderate viscosities (Williams et al., 2014).

The upper most Vaca Muerta and Lower Vaca Muerta represent end members, each with
both benefits and disadvantages in regards to their reservoir quality. The middle Vaca Muerta
represents a transitional zone where the positive reservoir qualities are present, but the negative
reservoir qualities are minimized. This zone is upwards of 50+ m thick and has been identified as
a high-graded interval. Because other studies have identified the uppermost Vaca Muerta as
being the best quality zone, I suggest that continued study of vertical wells would allow for a
comparative analysis, with perforations being placed in both the upper and middle intervals, to
test whether there is indeed any production enhancement from the presence of natural
fractures. Downhole microseismic, if acquired in the future, would be a powerful tool for
validating the interpretation made in this thesis.

8.2 Recommendations for Further Analysis

The scale of fractures discussed in this analysis is sub-seismic by design. The seismic data
should be used to resolve seismic scale features, but anything that is below that requires differing
data sets for analysis, and the goal in this study was to determine whether or not sub-seismic
features were present in abundance, and if so, what their character might be. In addition, a
feasibility analysis about using image log data to resolve horizontal features that could be
significant to the hydraulic stimulation treatment was performed.

109
 What was not discussed, or considered, in this analysis, was the presence of features which
would be smaller still than the log scale. In particular, the volume increase that results when
kerogen is converted into oil and gas is significant and will create a pressure increase which must
be accommodated – this accommodation frequently takes the form of micro scale fractures
which can be resolved only with petrographic data.

1. Representative samples from differing zones in the reservoir should be examined for the
presence of microfractures and the implications of such features should be considered.

Petrographic data could also help to resolve the question of the diagenetic history of the
formation. The mineral form is not considered by the wireline logs, and although petrographic
data was available from other studies, it is not necessarily accurate for the Wintershall area. Log
based studies are valuable and highly applicable for field scale development, but there is an
inherent averaging that takes place due to the resolution of wireline logs. I suggest that core tests
and petrographic study will be highly beneficial to refining the porosity, permeability, and
cementation uncertainty that is inherent in the log analysis.

2. Incorporate petrographic data and perform a diagenetic study on the local rocks.

The presence of calcareous concretions is noted in this analysis, but the implications of their
presence are not discussed in any great length. A better understanding of the concretions may
be of interest to developing an understanding of the diagenetic evolution of this formation. I
would recommend that the implications of the concretions be considered, and that the
distribution of concretions could be further quantified using a textural analysis of the borehole
image data. The textural analysis could also be compared to the cluster model, and perhaps could
offer greater insight on how the textural partitioning of the reservoir affects the cluster output
in a more quantified way that what was described here in section 7.2

3. A textural description from the image log and quantification of the relationship between the
observed textures and the cluster modelling.

The limitation of using the clustering algorithm to define likely mechanical behavior is that
the e is t u e tl a data to alidate su h a i te p etatio . Willia s et al. (2014) chose

110
 samples for core tests from each of the defined cluster classes in order to discriminate more
precisely the differences between each class. This may be challenging for a couple of the classes,
which manifest as very thin zones but would allow for a more robust interpretation and core
tests in general are recommended in order to determine the rock strength and static elastic
properties.

Because of the general lack of calibration data available for this study, the stress, pressure,
and rock strength values that were used in this study are either log derived, representing purely
the dynamic measurements and horizontal orientations, and appropriate values for fracture
stability assessment were chosen in consultation with Wintershall based on internal modelling
for the area. The resulting stability analysis described in Chapter 5 showed that the natural
fractures are stable given the current stress estimate, but the uncertainty involved with those
figures is very high and the values have not been properly correlated to local data. Refining the
stress magnitudes and the rock strength parameters is essential in order to ascertain whether or
not the fractures are indeed stable or are, in fact, critically stressed.

The logs used in this study are solely capable of resolving rock properties in the vertical
direction. In addition, owing to the narrow azimuth configuration of the acquisition survey for
the available seismic volume, anisotropy analysis is not possible with the existing seismic data
set. Triaxial core tests from nearby studies record vertical to horizontal anisotropy of up to 40%,
and the establishment of the relevant local anisotropy will be useful in order to understand likely
fracture growth or possible containment.

4. Core test results should be incorporated into future analysis.

The resolution of the horizontal features was one aim of this analysis which could not be
successfully completed and which would have been a valuable insight to complete the natural
fracture characterization. The need for core to corroborate the interpretation of the image
limited the ability to characterize the distribution of the horizontal discontinuities. The ash beds,
beef, and possible intrusive sills have been identified to be the most likely barrier to fracture
growth (Fantin et al., 2014) and the mapping of their locations should be highly prioritized. Once
core data is available, the image log trace of the beef can be refined, and the possibility that some

111
of the thin-bedded cluster classes represent ash beds can be confirmed or disproved. This will
be vital to further operations as both features may have significant impacts during completions
and stimulation. Casing shear has been recorded in nearby producing fields, and it is very possible
that the location of these horizontal shear events could be directly related to the presence of the
beef (Garcia et al., 2013) (Figure 8.21). Williams et al. (2014) also observed relatively fewer ash
beds through the middle section as compared to the lower Vaca Muerta, and this would be a
further data point to confirm that the middle of the Vaca Muerta represents the best target for
horizontal landings.

Figure 8.21 Casing restriction related to shear movement recorded in the Loma Campana Field
(Rimedio et al., 2015).

112
 REFERENCES

Al Ibrahim, M. (2013) Multi-Scale Sequence Stratigraphy, Cyclostratigraphy, and Depositional

Environment of Carbonate Mudrocks in the Tuwaiq Mountain and Hanifa Formations, Saudi
A a ia, Maste s Thesis, Colo ado “ hool of Mi es

Altindag, R., and Guney, A. (2010) Predicting the relationships between brittleness and
mechanical properties (UCS, TS and SH) of rocks, Scientific Research and Essays, Vol. 5, No. 16,
pp 2107-2118

Badessich, M.F., and Berrios, V. (2012) Integrated Dynamic Flow Analysis to Characterize an
Unconventional Reservoir in Argentina: The Loma La Lata Case, SPE 156163, Prepared for
Presentation at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 8-10
October, 2012

Badessich, M.F., and Krasuk, R. (2014) Reservoir and Drilling Engineering Combined Effort to
Develop a Naturally Fractured Over Pressured Shale Formation: Understanding and Operational
Specific Procedures, SPE 171587, Prepared for Presentation at the SPE/CSUR Unconventional
Resources Conference, Calgary, Alberta, 30 September – 2 October, 2014

Baihly, J.D., Malpani, R., Edwards, C., Han, S.Y., Kok, J.C.L., Tollefsen, E.M., Wheeler, C.W. (2010)
Unlocking the Shale Mystery: How Lateral Measurements and Well Placement Impact
Completions and Resultant Production, SPE 138427, Prepared for Presentation at the SPE Tight
Gas Conference, San Antonio, Texas, 2-3 November, 2010

Barree, R.D. (2002) Introduction to Gohfer and Stimulation Design, Internal Publication. Barree
and Associates

Barree, R.D, Gilbert, J.V., and Conway, M.W. (2009) Stress and Rock Property Profiling for
Unconventional Reservoir Stimulation

Barton, C.A., and Zoback, M.D. (2002) Discrimination of Natural Fractures From Drilling-Induced
Wellbore Failures in Wellbore Image Data – Implications for Reservoir Permeability, SPE 78599

Barton, C.A., and Moos, D. (2010) Geomechanical Wellbore Imaging: Key to Managing the Asset
Life Cycle, in Pöppelreiter, M., Garcia-Carballido, C., Kraaijveld, M., eds., Dipmeter and Borehole
Image Log Technology, The American Association of Petroleum Geologists Memoir 92, 1-32

Bowers, G.L. (1994) Pore pressure estimation from velocity data: accounting from overpressure
mechanisms besides undercompaction, International Journal of Rock Mechanics and Mining
Sciences & Geomechanics, Vol. 31, No. 6, pp 515-530

113
Bratton, T., Viet Canh, D., Van Que, N., Duc, N.V., Gillespie, P., Hunt, D., Li, B., Marcinew, R., Ray,
S., Montaron, B., Nelson, R., Schoderbek, D., Sonneland, L. (2006) The Nature of Naturally
Fractured Reservoirs, Schlumberger Oilfield Review, Accessed at www.schlumberger.com

Bust, V.K., Majid, A.A., Oletu, J.U., Worthington, P.F. (2014) The petrophysics of shale gas
reservoirs: Technical challenges and pragmatic solutions, Petroleum Geoscience, Vol. 19, pp 91-
103

Cobbold, P.R., Zanella, A., Rodrigues, N., Loseth, H. (2013) Bedding-parallel fibrous veins (beef
and cone-in-cone): worldwide occurrence and possible significance in terms of fluid
overpressure, hydrocarbon generation and mineralization, Marine and Petroleum Geology, Vol.
43, pp 1-20

Collins, P.W., Fernandez Badessich, M., Ilk, D. (2015) Addressing Forecasting Non-Uniqueness and
Uncertainty in Unconventional Reservoir Systems Using Experimental Design, SPE 175139,
Prepared for Presentation at the SPE Annual Conference and Exhibition, Houston, Texas, 28-30
September, 2015

Crain, E.R. and Holgate, D. (2014) Digital Log Data to Mechanical Rock Properties for Stimulation
Design, Prepared for Presentation at GeoConvention 2014: Focus Conference in Calgary, Alberta,
12-14 May, 2014

Crain, E.R. and Holdate, D. (2014) A 12-step Program to Reduce Uncertainty in Kerogen-rich
Reservoirs, Prepared for Presentation at GeoConvention 2014: Focus Conference in Calgary,
Alberta, 12-14 May, 2014

Davey, H. (2012) Geomechanical Characterization of the Montney Shale, Northwest Alberta and
No theast B itish Colu ia, Ca ada, Maste s Thesis, Colo ado “ hool of Mi es

Donald, J.A., and Bratton, T.R. (2006) Advancements in acoustic techniques for evaluating open
natural fractures, SPWLA 47th Annual Logging Symposium, Society of Petrophysicists and Well-
Log Analysts

Duenas, C. (2013) Understanding rock quality heterogeneity of Montney shale reservoir, Pouce
Coupe Field, Al e ta, Ca ada, Maste s Thesis, Colo ado “ hool of Mi es

Eaton, B.A. (1969) Fracture gradient prediction and its application in oilfield operations, Journal
of Petroleum Technology, Vol. 21, No. 10, pp. 1-353

Ejofodomi, E.A., Varela, R.A., Cavazzoli, G., Velez, E.I., Peano, J. (2014) Development of an
Optimized Completion Strategy in the Vaca Muerta Shale with an Anisotropic Geomechanical
Model, SPE 167806, Prepared for Presentation at the SPE/EAGE European Unconventional
Conference and Exhibition, Vienna, Austria, 25-27 February, 2014

114
Fantin, M., Crousse, L., Cuervo, S., Vallejo, D., Gonzalez Tomassini, F., Reijenstein, H., Lipinski, C.
(2014) Vaca Muerta Stratigraphy in Central Neuquén Basin: Impact on Emergent Unconventional
Project, URTeC 1923793, Prepared for Presentation at the Unconventional Resources
Technology Conferences held in Denver, Colorado, 25-27 August, 2014

Fjaer, E., Holt, R.M., Horsrud, P., Raaen, A.M., Risnes, R. (2008) Petroleum Related Rock
Mechanics 2nd Edition, Developments in Petroleum Science Vol. 53, Elsevier

Fox, A., Snelling, P., McKenna, J., Neale, C., Neuhaus, C., Miskimmins, J. (2013) Geomechanical
Principles for Unconventional Resources, www.microseismic.com

Gale, J.F.W., Reed, R.M., and Holder, J. (2007) Natural Fractures in the Barnett Shale and their
importance for hydraulic fracture treatments, AAPG Bulletin, V. 91, No. 4, pp 603-622

Gale, J.F.W. and Laubach, S.E. (2009) Natural fractures in the New Albany Shale and their
importance for shale-gas production, Proceedings, 2009 International Coalbed & Shale Gas
Symposium

Gale, J.F.W., Laubach, S.E., Olson, J.E., Eichhubl, P., Fall, A. (2014) Natural Fractures in Shale: A
Review and New Observations, The American Association of Petroleum Geologists Bulletin, V. 98,
No. 11, pp. 2165-2216

Gamero-Diaz, H., Miller, C.K., Lewis, R. (2013) sCore: A Mineralogy Based Classification Scheme
for Organic Mudstones, SPE 166284, Prepared for Presentation at the SPE Annual Technical
Conference and Exhibition, New Orleans, Louisiana, 30 September – 2 October, 2013

Garcia, M.N., Sorenson, F., Bonapace, J.C., Motta, F., Bajuk, C., Stockman, H. (2013) Vaca Muerta
Shale Reservoir Characterization and Description: The Starting Point for Development of a Shale
Play with Very Good Possibilities for a Successful Project, SPE 168666, Presented at the
Unconventional Resources Technology Conference, Denver, Colorado, 12-14 August, 2013

Garcia, M.N., Sorenson, F., Stockman, H., Zavala, C. (2014) Shale and Tights Reservoirs:
Geomechanical Control in the Success of Producing Wells, Neuquén Basin, Argentina, SPE
167707, Prepared for Presentation at the SPE/EAGE European Unconventional Conference and
Exhibition, Vienna, Austria, 25-27 February, 2014

Hansen, B. and Buczak, J. (2010) Making Interpretable Images from Image Logs, in Pöppelreiter,
M., Garcia-Carballido, C., and Kraaijveld, M., eds., Dipmeter and borehole image log technology,
The American Association of Petroleum Geologists Memoir 92, pp. 15-38

Ha t, B. “., Ma uake , J. H., & Ta lo , K. G. Mudsto e shale depositio al and

diagenetic processes: Implications for seismic analyses of source-rock reservoirs. Interpretation,
Vol. 1, No. 1, pp 7-26

115
Hogg, S.L. (1993) Geology and hydrocarbon potential of the Neuquén Basin, Journal of Petroleum
Geology, Vol. 15, No. 4, pp 383-396

Hornby, B.E., Johnson, D.L., Winkler, K.W., Plumb, R.A. (1989) Fracture Evaluation using reflected
Stoneley-wave arrivals, Geophysics, vol 54, no. 10, 99 1274-1288

Howell, J.A., Schwarz, E., Spalletti, L.A., Veiga, G.D. (2005) The Neuquén Basin: and overview,
Geological Society of London Special Publication, Vol. 252, No. 1, pp 1-14

Hryb, D., Archimio, A., Badessich, M., Ejofodomi, E., Diaz, A., Cavazzoli, G., Zalazar, F., Lagarrigue,
E., Pichon, S. (2014) Unlocking the True Potential of the Vaca Muerta Shale via an Integrated
Completion Optimization Approach, SPE 170580, Prepared for Presentation at the SPE Annual
Technical Conference and Exhibition, Amsterdam, The Netherlands, 27-29 October, 2014

Hurley, N. (2004) AAPG Methods in Exploration, No. 16, Chapter 9: Borehole Images

Jarvie, D.M., Hill, R.J., Ruble, T.E., Pollastro, R.M. (2007) Unconventional shale-gas systems: The
Mississippian Barnett Shale of north-central Texas as one model for thermogenic shale-gas
assessment, AAPG Bulletin, V. 91, No. 4, pp 475-499

Kietzmann, D.A., Martin-Chivelet, J., Palma, R.A., Lopez-Gomez, J., Lescano, M., Concheyro, A.
(2011) Evidence of precessional and eccentricity orbital cycles in a Tithonian source rock: the mid-
outer carbonate ramp of the Vaca Muerta Formation, northern Neuquén Basin, Argentina, AAPG
Bulletin, Vol. 95, No. 9, pp 1459-1474

Kietzmann, D.A., Palma, R.M., Riccardi, A.C., Martin-Chivelet, J., Lopez-Gomez, J. (2014)
Sedimentology and sequence stratigraphy of a Tithonian-Valanginian carbonate ramp (Vaca
Muerta Formation): A misunderstood exceptional source rock in the Southern Mendoza area of
the Neuquén Basin, Argentina, Sedimentary Geology, Vol. 302, pp 64-86

Lagraba P., J.O., Hansen, S.M., Spalburg, M., Helmy, M. (2010) Borehole image tool design, value
of information, and tool selection, in Pöppelreiter, M., Garcia-Carballido, C., and Kraaijveld, M.,
eds., Dipmeter and borehole image log technology, The American Association of Petroleum
Geologists Memoir 92, pp. 15-38

Laubach, S.E., Olson, J.E., and Gross, M.R. (2009) Mechanical and fracture stratigraphy, AAPG
Bulletin, Vol. 93, No. 11, pp 1413-1426

Legarreta, L., and Villar, H.J. (2011) Geological and Geochemical Keys of the Potential Shale
Resources, Argentina Basins, Search and Discovery 80196

Licitra, D., Lovrincevich, E., Vittore, F., Quiroga, J., Oviedo, P., Montoya, V., Shannon, C., Monti,
L. (2015) Sweet Spots in Vaca Muerta: Integration of Subsurface and Production Data in Loma
Campana Shale Development, Argentina, SPE 178563 ,Presented at the Unconventional
Resources Technology Conference in San Antonio, Texas, 20-22 July 2015

116
McClure, M.W. (2013) Understanding, diagnosing, and modeling the causes of fracture network
complexity in unconventional reservoirs, The Leading Edge, Vol. 32, No. 12, pp 1494-1500

McCounaughy, D.T., and Engelder, T. (1999) Joint Interaction with embedded concretions: joint
loading configurations inferred from propagation paths, Journal of Structural Geology, Vol. 21,
No. 11, pp 1637-1652

Meissner, F.F. (1984) Petroleum Geology of the Bakken Formation Williston Basin, North Dakota
and Montana, AAPG Memoir 35: Petroleum Geochemistry and Basin Evaluation, pp 159-179.

Mitchum, R.M. and Uliana, M.A. (1985) Seismic stratigraphy of carbonate depositional
sequences, Upper Jurassic-Lower Cretaceous, Neuquén Basin, Argentina, Seismic Stratigraphy,
Vol. 2, pp 255-283

Montaut, A.M.M. (2012) Detection and quantification of rock physics properties for improved
h d auli f a tu i g i h d o a o ea i g shales, Maste s Thesis, U i e sit of Te as at Austi

Monti, L.J., Suarez, M.I., Thompson, A.R., Vega, L. (2013) Modeling Vertical Multifractured Wells
in Vaca Muerta Shale Oil Play, Argentina, SPE 164537, Prepared for Presentation at the
Unconventional Resources Conference, The Woodlands, Texas, 10-12 April, 2013

Mosquera, A., and Ramos, V.A. (2006) Intraplate deformation in the Neuquén Embayment,
Geological Society of America Special Paper 407, 99 97-123

Nygaard, R., and Hareland, G. (2007) Application of rock strength in drilling evaluation, SPE
106573, Prepared for Presentation at the 2007 SPE Latin American and Caribbean Petroleum
Engineering Conference, Buenos Aires, Argentina, 15-18 April, 2007

Osorio, J.G., and Muzzio, E. (2013) Correlation between Microseismicity and Geomechanics
Factors Affecting the Hydraulic Fracturing Performance in Unconventional Reservoirs – A Field
Case in Neuquén Argentina, Prepared for Presentation at the 47 th US Rock
Mechanics/Geomechanics Symposium, American Rock Mechanics Association, 2013

Passey, Q.R., Creaney, S., Kulla, J.B., Moretti, F.J., Stroud, J.D. (1990) A Practical Model for Organic
Richness from Porosity and Resistivity Logs, The American Association of Petroleum Geologists
Bulletin V. 74, No.12, pp 1777-1794

Passey, Q.R., Bohacs, K.M., Esch, W.L., Klimentidis, R., Sinha, S. (2010) From Oil-Prone Source
Rock to Gas-Producing Shale Reservoir – Geologic and Petrophysical Characterization of
Unconventional Shale-Gas Reservoirs, SPE 131350, Prepared for Presentation at the CPS/SPE
International Oil & Gas Conference and Exhibition, Beijing, China, 8-10 June, 2010

117
Pöppelreiter, M., Garcia-Carballido, C., and Kraaijveld, M., eds., Dipmeter and borehole image log
technology, The American Association of Petroleum Geologists Memoir 92, pp. 15-38

Prioul, R., Donald, A., Koepsell, R., El Marzouki, Z., Bratton, T. (2007) Forward modeling of
fracture-induced sonic anisotropy using a combination of borehole image and sonic logs,
Geophysics, Vol 72, No. 4, pp 135-147

Reijenstein, H.M., Lipinski, C., Fantin, M., Cuervo, S., Vallejo, D., Crousse, L., Gonzalez Tomassini,
F., and Kietzmann, D. (2015) Where is the Vaca Muerta Sweet Spot? The Importance of Regional
Facies Trends, Thickness, and Maturity in Generating Play Concepts, Prepared for Presentation
at the Unconventional Resource Technology Conference in San Antonio, Texas, 20-22 July 2015

Rickman, R., Mullen, M., Petre, E., Grieser, B., Kundert, D. (2008) A Practical use of Shale
Petrophysics for Stimulation Design Optimization: All Shale Plays are Not Clones of the Barnett
Shale, SPE 115258, Prepared for Presentation at the SPE Annual Technical Conference and
Exhibition, Denver, Colorado, 21-24 September, 2008

Rimedio, M., Shannon, C., Monti, L., Lerza, A., Roberts, M., Quiroga, J. (2015) Interference
Behavior Analysis in Vaca Muerta Shale Oil Development, Loma Campana Field, Argentina, SPE
178620, Prepared for Presentation at the Unconventional Resources Technology Conference, San
Antonio, Texas, 20-22 July, 2015

Rodrigues, N., Cobbold, P. R., Loseth, H., & Ruffet, G. (2009) Widespread bedding-parallel veins
of fi ous al ite eef' i a atu e sou e o k Va a Mue ta F , Neu ué Basin, Argentina):
evidence for overpressure and horizontal compression. Journal of the Geological Society, Vol.
166, No. 4, pp 695-709

Sagasti, G., Ortiz, A., Hryb, D., Foster, M., Lazzari, V. (2014) Understanding Geological
Heterogeneity to Customize Field Development: An Example From the Vaca Muerta
Unconventional Play, Argentina, URTeC 1923357, Prepared for Presentation at the
Unconventional Resources Technology Conference, Denver, Colorado, 25-27 August, 2014

Schlumberger FMI brochure www.schlumberger.com

Schlumberger UBI brochure www.schlumberger.com

Schmoker, J.W. and Hester, T.C. (1983) Organic Carbon in Bakken Formation, United States
Portion of Williston Basin, The American Association of Petroleum Geologists Bulletin V. 67, No.
12, pp. 2165-2174

Simo, T., Ottman, J.D., Bohacs, K.M., Macquaker, J.H. (2013) Unconventional Reservoirs in Back
Arc Basins, Comparison Between the Vaca Muerta (Argentina) and Indonesian Projects,
Proceedings, Indonesian Petroleum Association Thirty-Seventh Annual Convention and
Exhibition, May 2013

Steiber, S. J. (1970) Pulsed neutron capture log evaluation in the Louisiana Gulf Coast: Society of
Petroleum Engineers. In 45th Annual Meeting, paper SPE-2961

118
Suarez-Rivera, R., Handwerger, D., Kieschnick, J., Martin, W., Green, S. (2005) Accounting for
Heterogeneity Provides a New Perspective for Completions in Tight Gas Shales, ARMA 05-758,
Prepared for Presentation at Alaska Rocks, the 40th US Symposium on Rock Mechanics,
Anchorage, Alaska, 25-29 June, 2005

Suarez- Rivera, R., Deenadayalu, C., Chertov, M., Novalo Hartanto, R., Gathogo, P., Kunjir, R.
(2011) Improving Horizontal Completions on Heterogeneous Tight Shales, SPE 146998, Prepared
for Presentation at the Canadian Unconventional Resources Conference, Calgary, Alberta, 15-17
November, 2011

Sullivan Glaser, K., Miller, C.K., Johnson, G.M., Toelle, B., Miller, P., Pennington, W.D. (2014)
Seeking the Sweet Spot: Reservoir and Completion Quality in Organic Shales, Schlumberger
Oilfield Review Winter 2013/2014, Vol. 25, No. 4, pp 16-29

Tenthorey, E. and Cox, S.F. (2006) Cohesive strengthening of fault zones during the interseismic
period: An experimental study, Journal of Geophysical Research, Vol. 111, pp 1-14

Tingay, M., Reinecker, J., & Müller, B. (2008) Borehole breakout and drilling-induced fracture
analysis from image logs.

Urien, C.M., and Zambrano, J.J. (1994) Petroleum Systems in the Neuquen Basin, Argentina, in
Magoon, L.B. and Dow, W.G. The Petroleum System: From Source to Trap, AAPG Memoir 60, pp
513-534

Villar, H. J., Laffite, G.A., Legarreta, L. (1998) The source rocks of the Mesozoic petroleum systems
of Argentina: A comparative overview on their geochemistry, paleoenvironments and
hydrocarbon generation patterns, International Congress and Exhibition of the AAPG and the
Brazilian Association of Petroleum Geologists, p. 186–187.

Urien, C.M., Zambrano, J.J, Yrigoyen, M.R. (1995) Petroleum basins of southern
South America: an overview, in A. J. Tankard, R. Suárez S., and H. J. Welsink, Petroleum
basins of South America: AAPG Memoir 62, pp 63–77

Wang, F.P., and Gale, J.F.W. (2009) Screening Criteria for Shale-Gas Systems, Gulf Coast
Association of Geological Societies Transactions, Vol. 59, pp 779-793

Warpinski, N.R., Mayerhoffer, M.J., Vincent, M.C., Cipolla, C.L., Lolon, E.P. (2009) Stimulating
Unconventional Reservoirs: Maximinzing Network Growth While Optimizing Fracture
Conductivity, Journal of Canadian Petroleum Technology, Vol 48, No. 10, pp 39-51

Wennberg, O.P., Svana, T., Azizzadeh, M., Aqrawi, A.M.M., Brockbank, P., Lyslo, K.B., Ogilvie, S.
(2006) Fracture intensity vs. mechanical stratigraphy in platform top carbonates: the Aquitanian

119
of the Asmari Formation, Khaviz Anticline, Zagros, SW Iran, Petroleum Geoscience, Vol. 12, pp
235-246

Wikel, K. (2011) Geomechanics: Bridging the Gap from Geophysics to Engineering in

Unconventional Reservoirs, CSEG Recorder, Vol. 36, No. 05

Williams-Stroud, S., Barker, W., and Smith, K. (2012) Distinguishing Natural Reactivated Fractures
from Hydraulic Induced Fractures using Microseismic Event Analysis, Prepared for Presentation
at Geoconvention 2012, Calgary, Alberta, 14-18 May, 2012

Williams, R.D., Willberg, D.M., Handwerger, D., Ekart, D., Petriello, J., Suarez-Rivera, R. (2014)
Advanced Core Analysis Methodologies Quantify and Characterize Prolific Liquid Hydrocarbon
Quantities in the Vaca Mueta Shale, URTeC 1922938, Prepared for Presentation at the
Unconventional Resources Technology Conference, Denver, Colorado, 25-27 August, 2014

Willis, M. (2013) Upscaling Anisotropic Geomechanical Properties Using Backus Averaging and
Pet oph si al Cluste s i the Va a Mue ta Fo atio , Maste s Thesis, Colo ado “ hool of Mi es

Zanella, A., Cobbold, P. R., and Lemrabott, A. (2010). Physical models of chemical compaction,
overpressure development and hydraulic fracturing in various tectonic contexts. In GeoMod 2010

Zanella, A., and Cobbold, P. R. (2012) 'Beef': Evidence for fluid overpressure and hydraulic
fracturing in source rocks during hydrocarbon generation and tectonic events: Field studies and
physical modelling. Prepared for Presentation at the Geofluids VII International Conference

Zavala, C., Ponce, J.J., Arcuri, M., Drittanti, D., Freije, H., Asensio, M. (2006) Ancient Lacustrine
Hyperpycnites: A Depositional Model from a Case Study In the Rayoso Formation (Cretaceous) of
West-Central Argentina, Journal of Sedimentary Research, Vol. 76, pp 41-59

Zeller, M., Verwer, K., Eberli, G.P., Massaferro, J.L., Schwarz, E., Spallett, L. (2015) Depositional
Controls On Mixed Carbonate-Siliciclastic Cycles And Sequences On Gently Inclined Shelf Profiles,
Sedimentology, doi: 10.1111/sed.12215

Zoback, M.D., Moos, D., Mastin, L., Anderson, R.N. (1985) Well bore breakouts and in situ stress,
Accessed at www.digitalcommons.uni.edu

120
 APPENDIX A

Table A.1 Statistics from cluster output organized by cluster class from the mechanical model

Arithmetic Standard
Variable Number of values Minimum Maximum
mean Deviation
1 Deep_Res 120 9.98 252.47 43.50 37.69

1 GR 146.83 237.21 182.53 21.18

1 NPHI 0.2315 0.3282 0.2780 0.01963

1 PEFZ 3.347 4.353 3.824 0.210

1 PR_DYN 0.2238 0.2884 0.2486 0.0136

1 RHOZ 2.232 2.3873 2.2845 0.0293

1 VPVS 1.6768 1.8284 1.7327 0.0294

1 YM_DYN 14.40 21.2820 16.67 1.34

2 Deep_Res 232 4.04 16.24 8.41 2.48

2 GR 82.17 169.39 124.96 15.67

2 NPHI 0.1858 0.3267 0.2564 0.0242

2 PEFZ 3.326 5.783 3.986 0.397

2 PR_DYN 0.19097 0.2766 0.2319 0.0154

2 RHOZ 2.2690 2.5212 2.3742 0.0407

2 VPVS 1.6229 1.7976 1.6952 0.0310

2 YM_DYN 15.22 22.50 19.21 1.48

3 Deep_Res 264 6.66 414.29 61.01 65.25

3 GR 75.99 179.35 122.24 21.96

3 NPHI 0.1590 0.3260 0.2454 0.0274

3 PEFZ 3.539 5.124 4.124 0.273

3 PR_DYN 0.1551 0.2583 0.2197 0.0179

3 RHOZ 2.2176 2.5250 2.3916 0.0558

3 VPVS 1.5726 1.7531 1.6730 0.0318

3 YM_DYN 16.22 28.09 21.36 2.04

121
Table A.1 Continued
5 Deep_Res 504 3.47 122.67 12.99 12.95

5 GR 63.30 175.37 113.75 21.31

5 NPHI 0.1431 0.3239 0.2320 0.0275

5 PEFZ 3.183 5.990 4.189 0.444

5 PR_DYN 0.2242 0.3229 0.2629 0.0176

5 RHOZ 2.2520 2.5864 2.4270 0.0577

5 VPVS 1.6863 1.9395 1.7659 0.0445

5 YM_DYN 15.78 31.78 22.47 2.74

6 Deep_Res 67 5.55 77.08 24.01 18.85

6 GR 63.95 154.98 97.97 17.24

6 NPHI 0.1398 0.3113 0.2160 0.03037

6 PEFZ 3.239 5.955 4.376 0.505

6 PR_DYN 0.1962 0.2517 0.2249 0.01141

6 RHOZ 2.3144 2.5662 2.4451 0.04884

6 VPVS 1.6399 1.7172 1.6830 0.01846

6 YM_DYN 21.30 30.49 25.77 2.00

7 Deep_Res 212 6.11 469.00 91.22 76.43

7 GR 53.69 161.68 83.66 15.32

7 NPHI 0.1166 0.2661 0.1934 0.02778

7 PEFZ 3.558 5.096 4.365 0.228

7 PR_DYN 0.1399 0.2769 0.2359 0.02098

7 RHOZ 2.38198 2.7175 2.5064 0.06120

7 VPVS 1.5612 1.7933 1.7066 0.04075

7 YM_DYN 22.32 38.46 29.42 3.317

8 Deep_Res 113 6.74 309.35 46.55 44.98

8 GR 34.47 107.35 62.25 15.73

122
Table A.1 Continued
8 NPHI 0.0546 0.2207 0.1414 0.04233

8 PEFZ 3.791 6.187 4.779 0.466

8 PR_DYN 0.2181 0.2904 0.26238 0.0150

8 RHOZ 2.4142 2.7789 2.6018 0.0554
8 VPVS 1.6808 1.8500 1.7728 0.03495
8 YM_DYN 31.44 63.89 41.04 6.53
9 Deep_Res 30 2.83 45.63 22.55 12.30
9 GR 36.89 163.83 83.72 33.68
9 NPHI 0.052 0.2685 0.1613 0.0660
9 PEFZ 3.632 6.238 4.890 0.639
9 PR_DYN 0.2817 0.3836 0.3132 0.02550
9 RHOZ 2.4171 2.7112 2.5658 0.08336
9 VPVS 1.8140 2.2920 1.9319 0.1155
9 YM_DYN 19.64 57.16 35.07 9.33

123