Está en la página 1de 132



Catalina Acua
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 (Geophysics).
Golden, Colorado
Date ___________________
Signed: ___________________________
Catalina Acua
Approved: __________________________
Dr. Thomas L. Davis
Professor of Geophysics
Thesis Advisor
Golden, Colorado
Dr. Terence K. Young
Professor and Head
Department of Geophysics
A 4D-3C seismic volume has been interpreted to characterize the Glorieta
Paddock reservoir at Vacuum Field, New Mexico. This reservoir has been under
water flood since 1992. Seismic velocity, amplitude and anisotropy attributes
have been analyzed in a time-lapse manner.
Compressional and shear waves have been able to detect changes within the
reservoir due to water flooding and the results are supported by production data.
The results achieved in this study suggest that it is possible to monitor water
flood in a carbonate reservoir using multicomponent data.
1.1 Vacuum Field 1
1.2 Data Set. 2
1.3 Seismic Data Processing 6
1.4 Characterization Methodology.... 10
1.4.1 Shear wave method.. 11
1.4.2 4D or seismic monitoring method... 13
2.1 Tectonic history of Permian Basin and Delaware Basin 16
2.2 Origin of Delaware Basin: 19
2.3 Local geology and rock characteristics of the main producers.. 20
3.1 Introduction 27
3.2 Calibration P-waves. 27
3.3 Calibration S-waves. 32
4.1 Introduction.. 37
4.2 Structural interpretation. Coherence cube.. 41
5.1 Production History.. 53
5.2 Reservoir performance. 54
5.3 Reservoir fluid data 58
5.4 Data available. 58
5.5 Production maps. 60
6.1 Coherence cube P-waves..... 68
6.2 P-waves.. 73
6.2.1 Structure maps in time. 73
6.2.2 Time Delay analysis. 73
6.3 Multicomponent S
and S
... 78
6.3.1 Time Delay analysis. 79
6.3.2 Anisotropy analysis 82
6.3.3 Amplitude analysis. 91
6.4 Gassmann modeling.. 97
6.5 Reliability of the data. 103
7.1 Conclusions. 110
7.2 Recommendations. 110
1.1 Regional map showing geological location of Vacuum Field 3
1.2 Location of Vacuum Glorieta Unit.. 4
1.3 Index map showing relative location of seismic survey areas... 5
1.4 Shear wave splitting.. 12
2.5 Subsidence curves from the Permian Basin. 19
2.6 Stratigraphy column of the Permian Basin 21
2.7 Type log for the well Bridges State #95. 22
2.8 Core photos of the Paddock formation.. 24
2.9 Depositional setting for the Paddock interval.. 25
2.10 Core photo of the Glorieta interval 26
3.11 Location of the wells with synthetic seismograms 28
3.12 Synthetic seismogram for the well WS-13 (P-waves). 30
3.13 Synthetic seismogram for the well VGWU-127 (P-waves).. 31
3.14 Comparison between Trace 49 and corridor stack for S
-wave . 36
3.15 Time depth table for P and S waves.. 36
4.16 Tie between VSP-127 and seismic data from Maljamar. 38
4.17 Inline from Maljamar survey .. 39
4.18 The Glorieta map in time 40
4.19 Volumetric visualization of the Glorieta map in time 41
4.20 Time slice 934 ms from coherence cube.. 44
4.21 Faults interpreted in TS 934 using faults from seismic sections .. 45
4.22 Block diagram showing linked faults and relay ramps 46
4.23 Block diagram of the possible spatial distribution of relay ramps .. 47
4.24 Time slice from the coherence cube at 918 ms 48
4.25 Interpretation of faults using just coherence analysis . 49
4.26 Time slice from the coherence cube at 1046 ms.. 50
4.27 Time slice showing how discontinuities disappear in depth 51
4.28 Diagram. Transference zone between two overlapping faults 52
5.29 Historical production of the field before unitization ..................................55
5,30 Historical production of the field after unitization ......................................57
5.31 Water injection wells with water injection profiles ....................................61
5.32 Total oil production maps (1997 and 1998) ...............................................62
5.33 Total gas production maps (1997 and 1998) ............................................63
5.34 Total water production maps (1997 and 1998) .........................................63
5.35 Water cut maps (1997-1998) .......................................................................64
5.36 Total water injection maps (1997 and 1998) .............................................65
5.37 Average water injection pressure maps (1997 and 1998) ......................65
5.38 FVF vs reservoir pressure (oil and gas) .....................................................66
6.39 Time slices from the Vacuum survey coherence cube ............................70
6.40 Structural model for the Paddock dolomite ...............................................71
6.41 The Maljamar and Vacuum coherence cubes comparison ....................72
6.42 Line extracted from P-wave and S-wave volumes ...................................74
6.43 The Glorieta and the Paddock structural maps in time ...........................74
6.44 P-wave relative velocity differences ...........................................................75
6.45 P-wave relative velocity differences & structural model ..........................77
6.46 Relative S
and S
velocity increase ............................................................80
6.47 Relative S
and S
velocity decrease ..........................................................80
6.48 Inline 66 extracted from S
and S
, pre and post, volumes .....................84
6.49 Anisotropy maps (pre and post) .................................................................85
6.50 Anisotropy difference map ..........................................................................87
6.51 Inline 55. Intervals used in anisotropy computations ...............................89
6.52 Anisotropy maps computed for two different windows ............................90
6.53 Interpretation of a "middle" event in all S-volumes ..................................92
6.54 Time delay in S
. Relationship with "middle" event .................................... 93
6.55 RMS amplitude difference maps (S
and S
wave) .................................... 94
6.56 RMS amplitude ratio (S
) pre and post .................................................... 95
6.57 RMS amplitude ratio difference ..................................................................... 96
6.58 Anisotropy maps computed in the overburden (800-1500ms) ............... 106
6.59 Anisotropy maps computed in the overburden (10000-1500ms) ........... 107
6.60 Anisotropy maps computed in the overburden (800-1000ms) ............... 108
6.61 Crossline 80. base and repeat S-wave traces side by side .................... 109
1.1. Phase VII compressional wave data processing flow ................................. 7
1.2. Phase VII shear wave final processing sequences .................................... 9
1.3. Phase VII shear wave final statics values ..................................................10
5.4. Oil properties and initial reservoir characteristics ......................................59
6.5 Bulk modulus and density for oil and water..................................................99
6.6 Bulk and shear modulus of the mineral for dolomite and limestone .......... 99
6.7 Parameters used in the Vp and Vs calculations ......................................... 100
6.8 Compressional and shear velocities calculated with Gassmann ............. 102
I wish to express my appreciation to the following faculty members at Colorado
School of mines: Dr. Thomas Davis for his advice, support, guidance and his
enormous encouragement. Without his foresight and expertise I would not have
appreciated the significance of this dataset and the results of this study. Thanks
I want to thank the members of my committee: Dr. Michael Batzle and Dr.
Robert Benson. Especially for Mike I have the deepest gratitude feeling. I thank
him as a scientific, for all the hours he dedicated to explain me several aspects of
the fascinating geophysics as a science in his, no less fascinating, lab. I also
thank him as a person, for his unconditional support and understanding of many
other students and me, more as persons than as students.
Much appreciation to TEXACOs people: Mike Raines, Robert Martin and
Kevin Hickey for their support and expertise.
My very special thanks to the members of RCP Phase VII and Phase VIII.
Especially people like Gwenola, Luca, Raul, Reynaldo and David were crucial to
the success of my research, and were totally necessary to make my life happier
during these two years.
I want to express my sincere thanks to Lucy and Ed Jenner for their
unconditional and valuable help reading and correcting my thesis. I could not ask
for better people to do that annoying job: professionally perfectly capable, and
personally just great. Thanks for being an example of a wonderful couple.
The daily support of my friend Ronny made my life at Colorado School of
mines something special and beautiful. Thanks Ronny for helping me,
understand me and for always being there when I needed you.
I thank my parents for their education, their encouragement and their love.
Without them none of my goals could have been reached.
Finally, and above all, I thank Maite, my daughter, and Werner, my love;
without their love, understanding and unbelievable patience, the successful
culmination of this enormous task would have not been possible.
Vacuum Field is a mature operating field located 20 miles west of the town of
Hobbs in Lea County, New Mexico. This field is located on the northwestern shelf
of the Delaware Basin, on a large east-northeast anticlinal structure formed by
drape of sediments over a basement controlled fault block. Regionally, the study
area forms part of the Northwestern Shelf of the Permian Basin region. The
Permian Basin is a segmented foreland basin composed of two structural
depressions known as the Delaware and Midland basins, separated by the north-
south trending uplift, the Central Basin Platform (Figure 1). The data come
specifically from Central Vacuum Unit, which includes approximate 7 sections (7
square miles) and is operated by Texaco.
The Glorieta Paddock interval is a carbonate reservoir with an extended
production history (Chapters 2 and 5).
RCP Phase VII was originally conceived in an attempt to develop seismic
techniques to monitor associated with CO
flooding of the San Andres Formation
the primary producing unit in the Permian Basin. The data available imaged
deeper intervals so that it could be used to determine the reservoir performance
of Glorieta Paddock interval as well.
The Glorieta Paddock interval is being produced by Texaco and Phillips, in
two production units called Vacuum Glorieta West and Vacuum Glorieta East
respectively (Figure 2). This study is focused on the southeastern portion of
Vacuum Glorieta West Unit, since the seismic data available was recorded in the
overlapping of both the Central Vacuum and the Glorieta Units (Figure 2).
RCP Phase VII started July 1, 1997 and concluded December 30, 1999. The
project involved a multidisciplinary study of a CO
injection program for San
Andres reservoir. This project was extended to the study of deeper intervals of
Glorieta - Paddock and Drinkard - Abo in August 1999.
Several data sets have been made available to RCP Phase VII study in
Vacuum Field, specifically for the Glorieta Paddock interval. Beside the seismic
and well data provided for Central Vacuum Unit, a complete set of production
data including well logs have been provided for Vacuum Glorieta West Unit in the
interval of interest.
The seismic data available includes a large seismic survey and a subset of
two surveys recorded by RCP in Phase VII to monitor the reservoir (Figure 3).
Figure 1. Regional map showing geological location of Vacuum Field and
structural provinces of the Permian Basin. Black spots are oil and gas fields
producing from the different units in the Permian Basin (after Mattocks, 1998).
Figure 2. Location of Vacuum Glorieta Unit and the respective sub-division of Vacuum Glorieta West Unit
operated by Texaco since the unitization in 1992. The 4D area corresponds to the overlapping of Central
Vacuum Unit and Vacuum Glorieta West Unit.
Figure 3. Index map showing relative location of seismic survey areas and well location for Phase VII San
Andres and the Glorieta Paddock reservoir characterization.
This subset consists of two 3D-3C (3-dimensional multicomponent) seismic
surveys. The baseline survey was acquired in December 1997, prior to the CO
injection program in San Andres. The monitoring survey took place on December
In conjunction with each seismic survey, a downhole 3D-3C seismic survey, a
VSP and a walk-away VSP were acquired in the well VGWU-127 (Figure 3). A
passive seismic recording system was also used to detect induced seismicity in
the reservoir interval prior to the second 3D seismic survey.
Since the main objective of this study is to pursue a reliable 4D interpretation
of the Glorieta - Paddock, the processing flows applied to the baseline and
repeat survey are key in the development of the dynamic interpretation.
The main objective was to enhance the shear wave quality. Compressional
waves were processed with the same parameters shown in Table 1. The primary
difference for Phase VII was a new approach in the calculation of residual statics.
Shallow evaporite dissolution beds formed playa lakes at the surface
influencing the position of the seismic events at depth (Mndez, 1999). New
statics were calculated for both baseline and repeat volumes in P-wave and S-
wave. The new static values applied to P-wave volume are listed in Table 3. To
preserve the repeatability of the data during the 4D interpretation the seismic
Geometry description
Source/Receiver trace edit
Refraction static estimation
Minimum phase conversion
Surface consistent deconvolution (source/ receiver domain)
Cross-correlation of common-trace pairs (P-wave source / vertical
Apply source and receiver phase corrections to match surveys
Residual statics / Velocity estimations
Individual trace edit
Surface consistent amplitude corrections
CMP stack
FX Filter (spatial deconvolution: Inline Crossline)
Spectral shaping
Bandpass filter
3D migration (phase shift algorithm with single velocity model from stacking
velocities and VSP survey)
Table 1. Phase VII Compressional wave data processing schematic flow.
processing flows for both P-wave and S-wave volumes (Table 2) were the same
for the baseline and the repeated surveys. It is important to state that the
deconvolution operator used was the same for both the baseline and the repeat
survey. With this it is anticipated that the changes in frequency and phase that
could be observed in the wavelet would be due to changes intrinsic to the
reservoir through time.
The final shear wave processing sequences were determined through
multiple processing tests and even the complete reprocessing of the pre-injection
volumes. It was determined that the playa-lakes were also affecting the shear
wave image locally. As a consequence, new processing tests and the
reprocessing of the multicomponent pre-injection data set was planned to
reevaluate the effects of playa-lakes on seismic signal. A new strategy, involving
increased trace editing as well as new deconvolution operators and velocity
analysis was employed to improve the shear wave image. From the 650,000
traces, approximately 10,000 were killed in the vicinity of the playa-lakes during
the initial pre-injection data processing.
After reprocessing the pre-injection survey, the new data showed more
continuous reflections and more reliable time delays for anisotropy and time-
lapse analysis (Mndez, 1999; Cabrera, 1999). The same polarization angle of
118 degrees used in Phase VI was determined for Phase VII using VSP
polarization information at well VGWU-17 to separate the S
and S
Pre Post
Phase VI
- Reformat SEGD field tapes to internal format
* * * *
- Build identical geometries S1H1, S1H2, S2H1, S2H2
* * * *
- Reverse traces
* * * * *
- Rotate 118 degrees S1H1, S1H2, S2H1, S2H2 data
* * * *
- True amplitude correction, time raised power .8
* * * *
- Surface consistent shot and receiver amplitude
* * * *
- Trace kills
* * * *
- Shift to processing datum, based on flattening statics
from Phase VI
- Surface consistent shot and receiver flattening statics
* *
- Residual statics, pass 1 and pass 2
* * *
- NMO. 2
pass Phase VII
* * * *
- Shot domain TX dip filtering
* * * *
- Receiver domain TX dip filtering
* * * *
- Inverse NMO, 2
pass Phase VII
* * * *
- Bandpass Butterworth filter 6-18 Db, 60-72 Db
* * * *
- Minimum phase conversion
* *
- Notch filter 30 Hz noise for area receiver
* * * *
- Convolve receiver deconvolution filters
* *
- Convolve shot deconvolution filters
* *
- Residual statics, 3
* * *
- NMO final Phase VII
* * *
- Top mute
* * *
- CDP stack
* * * *
- FXY filter 8-50 Hz, 50% addback
* * * *
- TXY filter reject mode, pass real data
* * * *
- CDP fold taper
* * * *
- Finite difference migration, VSP velocity field
* * * *
Table 2. Phase VII Shear wave final data processing sequences.
A more extensive discussion about the processing algorithms applied to each
survey, S
and S
, pre-CO
and post CO
, can be found in Mndez, 1999.
Table 2 shows the final processing sequences for both shear volumes and
where each processing parameter was determined. The new statics values
applied are listed in Table 3 together with P-wave statics values for Phase VII.
Phase VII total refraction statics -75 to 105 ms
Phase VII total statics -75 to 109 ms
Phase VI flattening statics +50 to 85 ms
Phase VII additional flattening statics +20 to 27 ms
Phase VII residual statics +14 to 14 ms
Phase VII total statics +60 to 120 ms
Table 3. Phase VII P-wave and S-wave statics values, after Mndez (1999).
In order to characterize the Glorieta Paddock interval with the data available
several methods were applied. Well log data were used to create synthetic
seismograms to tie depth with time for the compressional and multicomponent
volumes. Production data were analyzed to explain the seismic results obtained
from the 4D point of view and shear wave anisotropy, shear and compressional
velocity and shear amplitude analyses were made in a time-lapse sense.
1.4.1 Shear wave method
In an anisotropic medium, shear waves are split into fast (S
) and slow (S
models. Particle motion within the S
component is parallel to the orientation of
the features creating anisotropy (e.g. fractures), with the S
particle motion
oriented perpendicularly (Figure 4). The time difference between corresponding
events on the two principal time series directions and a slow component,
perpendicular to the fracture direction, provides a measure of average anisotropy
over extended depth intervals (from Martin and Davis, 1987).
Shear wave anisotropy estimates are computed by normalizing processed
fast (S
) and slow (S
) shear wave isochron differences as follows:
2 1
tS tS

= (1)
Where tS
is the isochron of the interval of interest in the S
volume and tS
the correspondent isochron in the S
volume. The subtraction is normalized by
and multiplied by 100 so that the result is given in percentage.
Figure 4. Shear wave splitting. When a shear wave enters an azimuthally
anisotropic medium (the anisotropy is here represented by a set of aligned
fractures), it is effectively split into a fast (S
) component parallel to fracture
orientation, and a slow (S2) component, polarized perpendicular to the fracture
trend. The time delay between S1and S2 component is illustrated (Martin and
Davis, 1989).
1.4.2 4D seismic monitoring method
Time-lapse seismic is the link between seismic observables and reservoir
variables such as fluids, pressure or temperatures through time.
Most of the information regarding the 4D method was taken from Jack (1997),
and it is extensively discussed in his book Time Lapse in Reservoir
A reservoir under a secondary recovery program (water flood) may change it
properties through time. The properties that potentially altered include:
1. Pore pressure: It can fall as fluids are extracted from the reservoir, or it can
increase as water is injected.
2. Reservoir pore fluids: may change as fluids are extracted. Pore fluid
properties could be very sensitive to pressure changes. Density and
saturation could vary.
3. Temperature: is usually fairly constant, but injection of cold water into a
reservoir causes a reduced temperature behind the flood front. Cooling of the
reservoir rock may change rock properties. The effects of the cold water bank
may show up on the time-lapse seismic in some cases.
4. Reservoir medium properties: there will be a change of density, porosity, and
compressibility with pressure due to deformation of the rock frame and the
pore fluids.
5. Microearthquakes and induced fractures: can occur during production due to
temperature reduction or effective stress changes caused by water injection
or compaction of the reservoir rock. The direction of the fracturing will be
controlled by local stress.
6. Reservoir rock properties: could change as a result of chemical interaction
between an injection fluid and the rock.
The seismic observables that need to be analyzed and related with the
changes within the reservoir include:
1. Times on surface seismic data: Changes in event time due to a change in
seismic velocity due to extraction (production or injection) might be visible.
Seismic features are analyzed in terms of time differences (t or isochron) for
one specific interval (i.e. the reservoir interval) between the baseline and the
repeat surveys.
2. Amplitudes: If we assume consistent data processing, amplitudes may vary
due to changes in the reservoir. Amplitude analysis in a time-lapse sense was
not meaningful for compressional waves, nonetheless amplitude analysis for
shear waves revealed some important information. This analysis was pursued
taking the ratios between S
and S
RMS amplitudes for each survey.
3. Anisotropy measurements: anisotropy measured in a time-lapse sense is
nothing but the difference in anisotropy measured between baseline and
repeat surveys. Just for convention, an anisotropy difference is given by:
) ( ) ( post Anisotropy pre Anisotropy Anisotropy = (6)
Therefore, a negative difference of anisotropy means an increase in
anisotropy through time. On the other hand, a positive difference of anisotropy
means a decrease in anisotropy through time.
4. Frequency and phase: changes in frequency and phase could arise due to
changes in rock absorption (inelastic attenuation). They might also occur due
to effects resulting from changes in velocity within the rock and variable
conditions during the acquisition.
The tectonic evolution of the Permian Basin has been examined by many
authors including Horak (1985); Galley (1958); and Hills (1988) and it can be
outlined as follows:
Passive margin (late Precambrian to Mississippian, 850-310 ma.): the
ancestral Permian Basin occupied a passive margin characterized by weak
crustal extension and low rate subsidence (Horak, 1985). A broad, shallow,
gently dipping depression known as the Tobosa Basin developed (Galley, 1958).
Relatively uniform and widespread shelf carbonates and thin shales occupied the
basin (Hills, 1988).
Collision phase (Late Mississippian through Pennsylvanian, 310-265 Ma): The
significant structural features that characterize the gross geometry of the
Permian Basin resulted during this period, as a result of the collision of North
America and Gondwana Land. This collision gave rise to the Oachita Marathon
fold belt. The ancestral Tobosa Basin was intensively deformed along high
angle basement faults and pre-existing zones of weaknesses (Horak, 1985).
High heat flow, rapid basin subsidence and sedimentary filling took place during
this phase. Given its equatorial location at this period, broad carbonate shelves
developed along the western, northern, and eastern margins of the Permian
Basin. By Late Paleozoic time the original Tobosa Basin was divided into the
Delaware Basin, Central Basin Platform and Midland Basin.
Permian Basin phase (265-230 Ma.): during this phase, rapid filling of the
basin with fine to coarse-grained clastic and the development of extensive reef-
fringed carbonate/evaporite platforms and shelves proceeded until only the
Delaware Basin remained as a small depocenter (Horak, 1985). During this
period the Glorieta Paddock interval (with clastic and carbonate character) was
Stable Platform phase (Mesozoic, 230-80 Ma.): Mobility rates were low and
structural deformation was limited (Horak, 1985).
Laramide deformation (Late Cretaceous through Early Eocene, 80-50 Ma.):
The western side of the Permian Basin was elevated approximately 4000 ft.
Volcanic phase (Early Eocene through Middle Oligocene, 50-30 Ma.):
Extension and crustal thinning followed the Laramide, which resulted in
widespread volcanic activity.
Basin and Range tectonism (recent, 24-0 Ma.): Rifting, crustal thinning, and
high heat flow characterized the region from the western Delaware Basin across
the southwestern U.S. to California. Figure 5 illustrates subsidence curves from
different areas of Permian Basin and highlights the major tectonic phases and
periods of most active basin development.
The Delaware Basin of western Texas and south eastern New Mexico is a
major petroleum producing province about 200 miles long and 100 miles wide.
Late Mississippian: In excess of 7000 ft of pre-Pennsylvanian sediments
were deposited in the central portion of the basin. Upthrusting of the Central
Basin Platform began at the end of the Mississippian.
Pennsylvanian: Rapid subsidence of the basin occurred. Clastics including
turbidites deposition dominated the early Pennsylvanian.
Permian: Rapid subsidence permitted the accumulation of up to 8000 mts of
During the Early Permian considerable areas in the northern, north western,
and northeastern Delaware basin were shallow enough and free enough from
clastics for limestone shelves to develop. During the Middle Permian, subsidence
continued although not as rapidly as during the preceding Early Permian. The
amount of clastic influx decreased and carbonates formed. Evaporite deposits
were also common in restricted lagoons. The increase of clastic source in Late
Leonardian time occurred due to the renewed uplifts on the northwest. Late
Permian is characterized by tectonic calm.
Figure 5. Basement mobility profiles for each of the major provinces illustrate the
time and duration of successive tectonic phases which have affected the
Permian Basin. Major tectonic phases and periods of most active basin
development are displayed (From Horak, 1985)
Deposition at this time consists predominately of limestone, although clastics
and evaporites are also common. During this period fault reactivation may have
occurred. Reef and other sedimentary features often formed on these structures
and were important for oil and gas traps. At the end of Permian time, stress
relaxation continued and local subsidence formed many salt pans on the floor of
the receding sea. In the northeastern part of Delaware basin, evaporation of
lagoonal brines proceeded to completely fill depressions, leaving large deposits
of potassium salts (Hill, 1970).
Post Permian: lower clastic deposition rate. Compressional stage
accompanied the uplift of Delaware Mountains. Minor volcanism and intrusions,
as well as considerable faulting were the effects of this uplift.
The Vacuum Glorieta Field is located on the Vacuum structural High. Although
designated the Vacuum Glorieta Pool by the New Mexico Oil Conservation
Division, production is primarily from the Leonardian Paddock zone (Figure 6).
The upper seal of the reservoir is the overlaying Glorieta interval, and the lower is
the Blinebry interval. These tops are shown in a type-log for the well Bridges
State #95 (Figure 7).
The Paddock formation is subdivided into the Upper and Lower Paddock. The
Upper Paddock is subdivided into Upper Paddock Limestone and Upper
Era Period Epoch Formation
Dewey Lake
Rustler Halite 100
Salado Halite/Anhydrite 1000
Tansill Anhydrite /
Ss Sh /
Seven Rivers
Dolomite /
Sandy Dolomite/

Dolomite / Anhy. /
Shale / Sandstone
San Andres
Dolomite /
Glorieta Sandy Dolomite 100
Limestone /
Lower Dolomite
Dolomite /
Sandy Dolomite
Dolomite /
Sandy interval
Drinkard Dolomite 300
Dolomite / Anhy. /
Wolfcampian Wolfcamp
Limestone /
Figure 6. Stratigraphy column of the Permian Basin in the vicinity of the study
area on the Northwest Shelf for the formations deposited during Permian period.
The age, general lithology, and approximate thickness of each formation are
listed. The main producer in Vacuum Glorieta West Unit is shaded. Modified from
Pranter (1999); Broadhead (1993), and Talley (1997).
Figure 7. Type log for the well Bridges State #95. The top of the reservoir (The
Glorieta), Upper and Lower Paddock, as well as Blinebry (the bottom of the
reservoir, are picked).
Paddock Dolomite. The Upper Paddock Limestone has about 0-150 ft of
thickness and is the main producer. The Upper Paddock Dolomite is very tight
and is the permeable barrier between the limestone and Lower Paddock, which
is another producer interval. The Lower Paddock is a dolomite and it is extremely
fractured (Martin, 2000, Pers. Comm.). The production interval of the Main
Paddock is primarily within the upper 100 ft of the formation.
The Main Paddock as seen in core studies is predominantly dolomite,
nonetheless an irregular limestone interval is also observed (Upper Paddock
Limestone). The dolomite is tan to light brown, fine to medium crystalline, with
vuggy and intercrystalline porosity. Core photos (core intervals and thinsections)
of the Paddock interval are displayed in Figure 8. From observed cores, core
descriptions, mapping and petrophysical log analysis, the productive interval of
the Main Paddock is interpreted as having been deposited as an oolitic shoal
environment along an east-west trending shelf edge (Figure 9).
The Glorieta section overlying the Paddock consists of dolomitic sandstones.
It is Late Leonardian in age and is regarded as lowstand deposit, restricted to a
shallow shelf environment. The Glorieta Formation is composed of shelf
dolomites and minor interbedded sands in Vacuum study area. It also contains
isolated, lenticular porous zones. The Paddock and Glorieta is naturally fractured
(Broadhead, 1993)
Figure 8. Core photos and thin sections of the Paddock Formation. Figure (A)
shows a core from the Lower Paddock Dolomite, which is extremely fractured.
Figure (B) is a thin section of Upper Paddock porous limestone, vuggy and/or
moldic porosity can be seen. Figure (C) is a thin section of Upper Paddock non-
porous dolomite, which is the permeable barrier between the Upper Paddock
porous limestone and Lower Paddock fractured dolomite.
Figure 9. Schematic figure showing the depositional setting for the Paddock
interval (Burman, 1991).
The Glorieta Formation is composed of cyclically deposited siliciclastics,
carbonate, and carbonate-evaporite units. Environments of deposition range from
supratidal, through shallow sub-tidal to open-marine conditions. The siliciclastics
are dominantly eolian-derived sediments that were deposited on the shelf.
Although the Glorieta interval is not considered as a producer, three wells
within the Vacuum Glorieta West Unit are producing hydrocarbons from this
interval (Hickey, 2000, Pers. Comm.). Porosity of the productive Glorieta
reservoirs in south east New Mexico generally ranges from 5% to 12%, with
permeabilities from 2 to 9 md. (Broadhead, 1993). A core photo from the
Glorieta interval is displayed in Figure 10.
Figure 10. Thin section of the dolomitized sandstone of the Glorieta interval. Low
porosity can be appreciated.
As a first step, P-wave and S-wave seismic surface data were tied to the well
data was. Five wells with sonic logs (one of these with accompanying density
log), and one multicomponent vertical seismic profile (VSP) data set was used
(well VGWU-127), although two of them were available (wells CVU-200 and
Calibration data for P-wave was extensively studied by Baylock (1999) for the
San Andres interval. His results were assumed to be reliable and were taken as
a base or datum for the calibration within Glorieta - Paddock interval.
Synthetic seismograms were constructed from the sonic logs for the 5 wells
within the seismic area. In addition others synthetic seismograms were used from
2 wells located on the edge of the study area. Only one well had a density log, for
the remainder, a constant density of 2.76 g/cm
was assumed. Figure 11 shows
the map of the wells used for P-wave calibration.
Figure 11. Location of the wells with synthetic seismograms. The small square is
the area of the 4D study.
A synthetic created for the Warren State-13 well in the south central part of
the area is displayed in Figure 12. This well shows the whole section, which
2000 ft
includes the San Andres interval and Glorieta Paddock interval. It is possible to
observe a good tie at the San Andres level, and a relatively good tie at the
GLORIETA - PADDOCK level, with a correlation coefficient of 61. After Baylocks
work it is important to note that although there are successive changes in
processing flows (see Chapter 6.5), the polarity of the data is still the same:
normal polarity SEG standard (peak = negative impedance contrast), or reverse
polarity Landmark - Syntool (peak = positive impedance contrast) convention.
The correlation also suggests that the data is approximately zero phase.
All the synthetic seismograms were constructed with the following common
parameters: the wavelet convoluted with the reflectivity model was a trapezoid of
8-14-60-80 or 8-14-40-60 frequency bandwidth, zero phase, reverse polarity
Syntool convention or normal polarity SEG convention, and 100 ms operator
length. The datum for the seismic is 4000 ft., the velocity for the overburden (the
rock layers between the well elevation and the datum was taken from the VSP
VGWU-127, and is of 11,500 ft/sec.
Figure 13, displays the synthetic constructed for the VSP VGWU-127
correlated with the surface seismic and the corridor stack from the VSP.
Figure 12. Synthetic seismogram calculated for the well WS-13, which display a
good tie for both San Andres and Glorieta - Paddock intervals. Panel (a) displays
the sonic log used in the synthetic seismogram construction; panel (b) is the
impedance curve calculated overlain the reflection coefficient series; panel (c) is
the synthetic seismogram, and panel (d) is the synthetic overlain the trace # 74
from the surface seismic P-wave volume.
(A) (D) (C) (B)
Figure 13. Panel (a) displays is the synthetic seismogram constructed for the well
VGWU-127; panel (b) is the correlation coefficient between the seismic trace #49
and the synthetic; panel (c) is the synthetic overlain the trace # 79 from the
surface seismic P-wave volume; panel (d) is the correlation coefficient between
the corridor stack and the synthetic; and panel (f) is the same synthetic overlain
the VSP corridor stack for P-waves. Note the good correlation synthetic-seismic
in contrast with the poor correlation synthetic-corridor stack.
(A) (D) (C) (B)
Correlation between synthetic and surface seismic is good, with a correlation
coefficient of 60.
The VSP corridor stack was reprocessed in order to get the same resolution
as in the surface seismic data as well as time-shifted. Several shifts were applied
and 76 ms appeared to be the best choice. The wavelet shape or seismic
signature of the corridor is, in general, the same as in the synthetic for the
intervals between San Andres top and the Paddock, nonetheless differences can
be observed. The correlation coefficient is 45.
From all the synthetics made it can be concluded that Glorietas top is a
trough and can be found between 840 to 860 msecs. The Paddock top is
immediately below the Glorieta, and it is picked between 850 to 870 msecs.
A similar investigation by Li (2000) over the New Mexico State #18 well, which
mainly focussed on deeper intervals, has shown the same time picks for the
Glorieta and Paddock in compressional waves.
Since there was no shear wave sonic within the Glorieta Paddock interval
some assumptions and extrapolations were taken into account. The deepest
sonic dipole belongs to the New Mexico State #18 well, and it reaches 5200 ft in
depth. The shear synthetic seismogram for this well was constructed with the
following parameters: the wavelet convoluted with the reflectivity model was a
trapezoid of 8-12-20-30 frequency bandwidth, zero phase, normal polarity
Syntool convention or reverse polarity SEG convention, and 100 ms operator
The S-wave corridor stack for the VSP in well VGWU-127 was available but
the Glorieta - Paddock interval was seen as part of the reflected wave field.
Direct waves end at about 5600 ft. Two different procedures were made to
address the problem of no shear wave data within Glorieta - Paddock interval.
First, a visual correlation between the S-wave corridor stack and the closest
seismic section to the VGWU-127 well was established. Figure 14 shows in one
panel the trace # 49 and in the other the corridor stack for S-waves from the VSP
well. This correlation was possible through the extrapolation of the time-depth
table available for this well and calculated by Michaud (1999) from the first
arrivals of the direct shear wave in the borehole (Figure 15). The extrapolation
was made assuming that below the Queen (3600 ft) the behavior of the curve is
linear. Using this new Time-Depth curve, the corridor stack for S-waves and the
surface seismic can be relatively well correlated. For this synthetic it can be seen
that the Glorieta top can be picked around 1950 ms, and the top of the Paddock
would be the trough immediately below the Glorieta. This agrees with Baylock
(1999) where he established that shear wave volume has reverse polarity with
respect to the compressional volume. The reason for that is the way that shear
waves were acquired in the field for both, baseline and repeat survey.
The second approach was to use the deepest dipole sonic available (New
Mexico State #18 well) and construct a synthetic seismogram for it. This
synthetic was later compared with the VSP corridor stack for S-waves (well
VGWU-127) and with the surface seismic in the vicinity of this well. (Figure 14).
The objective of this comparison was to confirm that the Time-Depth table for the
-wave was valid at least till 5200 ft in depth. Another way to prove the validity
of the time-depth table was to calculate the Vp/Vs average for the interval
between San Andres top and the end of the curve, which is of 1.8, and it is agree
with published Vp/Vs values for carbonates.
Figure 14. Visual comparison between the trace # 49 extracted from the S1-wave
volume, and corridor stack for S1-wave. Panel (a) displays the sonic log used in
the synthetic seismogram construction; panel (b) is the synthetic seismogram
calculated for the well New Mexico State #18; panel (c) is the synthetic overlain
the trace # 79 from the surface seismic S1-wave volume; panel (d) is the
synthetic overlain the VSP corridor stack for S1-wave.
(C) (B) (F)
Figure 15. Time-Depth table calculated by Michaud (1999) using the first arrivals
of the VSP for compressional and shear wavelet. Notice the linear relationship
that exists between S1 and S2 below Queen.
The Maljamar P-wave seismic survey is a 30 mi
extraction taken from a
larger data set (about 120 mi
) and was donated for RCP use by Philips,
Chevron, Marathon, Mobil, Shell, and Texaco (Figure 3). The Maljamar survey
represents a regional overview of the Permian shelf margin at Vacuum. The
Maljamar survey was interpreted in order to determine the regional structure and
major stratigraphic patterns for the Glorieta Paddock reservoir. These data
were mainly used in the determination of the regional structural framework for the
Glorieta Paddock interval.
The VSP-127 synthetic seismogram (Figure 16) plus all the others calculated
and tied in the previous studies were used to initiate the current interpretation,
focussing the interpretative work on the definition of the structure of the Glorieta-
Paddock interval (Figures 17, 18 and 19).
Structural interpretation in the Glorieta Paddock interval is key to the
prediction fractures within the interval. This work was pursued using the seismic
coherence cube.
Figure 16. Tie between the VSP-127 and the seismic data for Maljamar survey.
The Glorieta is a trough and Paddock is a zero crossing.
Figure 17. Crossline #360 taken from the Maljamar volume, showing the
interpretation of Glorieta and Paddock seismic events. In the lower left corner the
relative location in the Maljamar survey is displayed.
Figure 18. The Glorieta map in time. The black lines represent fault systems: one
in the shelf break with east-west trend and the other in the slope with a north-
south trend. The blue square is the 4D area.
Figure 19. Visualization of the Glorieta time structure map.
As a first step, faults were interpreted in vertical sections, following obvious
discontinuities in the horizons. From this first step, a system of two major normal
faults, accompanied by one minor normal fault in the shelf break is clearly
observable (Figures 18 and 19).
The next step was the utilization of the coherence cube tool. With this attribute
seismic discontinuities were assumed to be related to faults if the general
tectonic history matched with them, otherwise they could be related with
stratigraphic features. Previous work in the structural interpretation of the
Maljamar survey using coherence, has been done (Galarraga, 1999), although it
was focussed in the San Andres interval.
The coherency cube used in this study was the same as used by Galarraga
(1999). The parameters and general considerations in the construction of the
cube were the following:
- Landmark PostStack Continuity from Correlation algorithm was used: this
option measures data continuity via crosscorrelation. For each trace, data
within a sliding window is crosscorrelated with data from 2, 4 or 8 adjacent
traces. The correlation coefficients are analyzed to determine a continuity
attribute which is assigned to the central sample.
- The minimum correlation was used within the correlation option in order to
emphasize the largest local discontinuities.
- A vertical window size of 16 ms was used since the lateral resolution of the
faults appears to increase as the vertical window size is decreased. At the
same time the signal to noise ratio can be increased as the size of the vertical
window is also increased.
- In the tests made by Galarraga (1999) coherency algorithm was shown to be
more robust as more traces were used in the crosscorrelation and lineaments
of low coherency are better defined, therefore the maximum number possible
was used (8 traces).
- FK filtering was applied: seismic discontinuities due to faulting can be
separated from other discontinuities related to noise and dipping seismic
reflections through the definition of the dip for each one of them. The dips of
the faults in the area are nearly vertical, while the seismic discontinuities from
stratigraphic reflection range from flat to 25
. A filter was created in the t-x
domain to eliminate events with dips of zero to five milliseconds per trace
(zero to 25
). This filter was then transformed into the f-k domain and applied
to the f-k before being transformed back to the (t,x) domain with the undesired
events removed.
In order to start the interpretative process of this attribute, the faults
interpreted in sections were overlain onto the time slices extracted from the
coherence cube, and a first interpretation was made for the Glorieta - Paddock
time interval (900-1200 ms). The fault interpretation in time slices was made in
time slices every 16 ms, and it followed the faults previously interpreted in
sections (Figures 20 and 21).
At this time the faults were interpreted as segments, not as large continuous
curvilinear planes. Justification for this interpretation comes from advances in
descriptive structural geology at outcrops and the better resolution in seismic
data. This new approach proposes that faults are presented as a system of
interlinked faults.
Figure 20. Time slice from the coherence cube (934 ms). The marks show where
the faults interpreted in section cut the time slice: each color (green, blue, pink) is
a different fault, each one interpreted as a continuous fault.
4D Area
Figure 21. Faults interpreted as a system of linked segments using the faults
interpreted in section as a guide. Observe the overlapping between faults.
Much recent work involving detailed mapping of fault traces using 3D seismic
data, has demonstrated that faults are usually made up of many overstepping
segments, linked by areas of complex deformation, termed transfer zones or
relay ramps. These areas are difficult to interpret from limited subsurface data,
yet are often sites of hydrocarbon traps and their evolution is important to the
understanding of the formation of many oil and gas fields. (Peacock and
4D Area
Sanderson, 1994). See Figures 22 and 23 for graphic visualization of the
Figure 22. Block diagram showing the main features of linked faults and relay
ramps. Bedding is reoriented in the relay ramp to accommodate displacement
transfer between the overstepping segments. (Peacock, et. al., 1994)
Figure 23. Block diagram of the possible spatial distribution of relay ramps.
Different levels have different displacement and different stages of relay
evolution. (Peacock, et. al., 1994). The evolution of relay ramps is directly related
with the evolution of fractures in carbonate settings.
Using this approach, coherence data were used to interpret fault segments
that afterward are used to locate the relative position of the relay ramps. The
second step is to use the coherence data alone, not following the faults
interpreted in sections but following the discontinuities of the data itself (Figures
24 and 25).
Figure 24. Time slice from the coherence cube at 918 ms., near the top of the
Glorieta. The discontinuities were associated with faults and based on the
coincidence of the faults interpreted in section with those discontinuities.
4D Area
Figure 25. Interpretation of faults using just coherence analysis. The faults in
green were interpreted as a separate group of faults because they never
intersected the Glorieta interval (they disappeared in depth). In yellow the top
Glorieta Paddock seismic interpretation at 918 ms.
The interpretation through the Glorieta Paddock interval shows that the fault
zones are composed of arrays of overstepping and linked segments (Figures 26
and 27).
4D Area
Figure 26. Time slice of the coherence cube at 1046 ms. Notice how the strong
anomalies or discontinuities do not persist due to the decrease in seismic
resolution with depth in some cases and to the end of the faults in some other.
Figure 27. At this time (1046 ms) is not possible to distinguish the character of
many fault segments. The south faults (green color in figure 28) are no longer
visible. However the character of the main faults in the shelf break is preserved, it
is just weaker. The set of north-south trending faults (blue) is now observable.
The possible presence of relay ramps in these arrays has a particularly
important role in the definition of the fracture pattern for the carbonates of the
Glorieta Paddock interval.
4D Area
Studies made in other areas have shown that depending on the evolution of
relay ramp, transfer faults can be developed generating fractures in between the
main faults (Figure 28). However, the Vacuum survey P-wave data has a higher
resolution, so that a more accurate interpretation of the linked fault system
pattern is expected, with a better description of the relay ramps role in the
fracture distribution and hydrocarbon trapping mechanism.
Figure 28. Some of the features common in the transference zone between two
overlapped faults. Fractures are generated in between the two faults and they
can have parallel as well as perpendicular direction respect to the main fault
trend (Trudgill and Cartwright, 1994).
The initial discovery of the field was in 1929 by the Socony Vacuum Oil
Company with the completion of Bridges State Well #1 in the San Andres. The
Vacuum Glorieta Pool was discovered on January 11, 1963 by Texacos #12
New Mexico O State NCT-1. The completed interval is in the dolomitized
Paddock member of the Upper Yeso Formation.
The Vacuum Glorieta Pool was established in 1963 after a meeting that
resolved the vertical limits of each pay zone within the reservoir. The type log
was designated as the Mobil State Bridges #95 (Figure 7, Chapter 3). The upper
pool was designated the Vacuum Glorieta Pool, starting at the top of Glorieta (log
depth 5838 ft) and ending at the top of the Blinebry. The Geologic Subcommittee
designated the top of the Paddock at 5950 ft. The major production in the
Vacuum Glorieta Pool is from the Paddock and a minor amount from the
Glorieta. After the initial discovery of the Vacuum Glorieta Pool, rapid
development extended the field to the north and east. These wells were drilled on
state-wide 40-acre spacing with a total of 185 wells having produced 61,816
MMBO plus 75,778 MMCFG and 35,837 MBW (as of January 1, 1990).
The resultant OOIP (Original Oil In Place) of 171 MMSTBO is calculated for
the Main Paddock using a 6% porosity cutoff and a 50% gamma ray, along with
capillary pressure provided by Shell and a geological oil/water contact map.
The predominant producing mechanism for the Vacuum Glorieta Field started
been solution gas drive with some pressure support from the surrounding aquifer.
Although the field has produced a significant amount of water, especially near the
edges, aquifer activity can best be described as encroachment rather than active
influx providing any significant pressure support.
The historical production plot for the field before unitization and water flooding
(Figure 29) shows that the initial production began in February 1963 with full field
development essentially completed in early 1967. Most wells in the field
produced at the maximum allowable rate of 107 BOPD until early 1974, when the
field began to decline as a whole. Water influx and pressure support from the
surrounding aquifer were limited until early 1974, as evidenced by the constant
water cut of 25% (once full field development had taken place in early 1967) and
constantly increasing gas-oil ratio from initial field discovery in early 1963. After
early 1974, the field had undergone significant fluid and pressure depletion and
Figure 29. Historical production of the field before unitization. Production from both Vacuum Glorieta West
and Vacuum Glorieta East are included.
some support. Generally water cut and reservoir pressure are higher and GOR's
lower on all flanks of the reservoir.
After the analysis of historical GOR, water cut and bottom hole pressure
(BHP) maps, it was determined that the Vacuum Glorieta West Unit has large
amounts of solution gas drive coming higher GOR's, lower reservoir pressure
and lower water production. This indicated that the west unit holds a great
percentage of secondary oil that needed to be waterflooded for an extended
period of time before any tertiary recovery (CO
injection) could be applied.
Logs suggested that the Glorieta is not connected with the Paddock and may
not be a candidate for unit wide water flooding or CO
injection. Primary reserves
from the Glorieta as determined by decline curve analysis were included in the
total projected recovery.
The previous analysis led to the formation of two units of production: Vacuum
Glorieta West Unit (VGWU) operated by Texaco and Vacuum Glorieta East Unit
(VGEU) operated by Philips. This process was accomplished in 1992 and water
flooding was initiated in VGWU later that same year after unitization.
A production history plot shows the behavior of the reservoir after the initiation
of the water flood program (Figure 30). From this plot it can be observed that gas
production decreased while oil production increased uniformly after June 1994.
Figure 30. Historical production of the field after unitization and water flooding program started. (Hickey,
In June 1996 oil production experienced an increase which was maintained
during the whole of 1997. Flushing fractured dolomite in the Lower Paddock with
water appears to be the reason for the increase (see reasons in the next
section). Shortly after, 1997, a new increase in oil production occurred due to the
implementation of a horizontal-drilling program.
Bottomhole fluid samples were obtained from different wells soon after the
reservoir discovery in 1964. PVT analysis were run on the samples by three
different labs. Table 4 summarizes the main characteristics of the oil samples
taken at that time. From the composition, it was demonstrated that the reservoir
fluid is black oil.
Texaco provided production information from 62 wells from the Vacuum
Glorieta West Unit, as well as water injection information (amount of water
injected and pressure of injection) for 54 injector wells. Production information
includes water, oil and gas production per month from 1995 to 1999.
The data include production information for 9 horizontal wells. These wells are
part of a horizontal-drilling program started in 1997 in the Upper Limestone. The
objective of this program is to recover the oil bypassed within the porous
Initial reservoir pressure 2260 psi
Bubble point pressure 1331 psi
Oil FVF @ BPP 1.306 RB/STB
Oil FVF @ 14.7 1.019 RB/STB
Solution GOR @ BPP 552 SCF/STB
Oil viscosity @ BBP 0.662 cp
Oil viscosity @ 14.7 2.08 cp
Oil density @ BBP 0.7278 gm/cc (62.9
Oil density @ 14.7 psia 0.811 gm/cc (43.0
Avg Reservoir Temperature 118.7
Table 4. Oil properties and initial reservoir characteristics. Information displayed
corresponds to samples taken at the beginning of the field production history.
limestone. There is evidence that proves that the water originally being injected
to the Upper Paddock was no longer being absorbed by the formation. Over
saturation of this interval induced the water to flow into the Lower Paddock
fractured dolomite (Martin and Hickey, 2000, Per. Com.).
Water injection profiles (Figure 31) logged in water injector wells showed
unequivocally that the water injected within the Upper Paddock Limestone since
1994 was being repulsed and going to the next open interval, the Lower Paddock
Dolomite during 1997. Reservoir engineers suspected that the Upper Paddock
Limestone was definitely not drained appropriately and the water injected there
created an over saturation of the formation. Thus the water started to go to the
Lower Paddock Dolomite and was rapidly canalized due to the high fracture
porosity in that formation.
Several production maps were constructed for each year in order to determine
production trends that could help to determine the behavior of the fluids in the
reservoir during the time-lapse period (1997-1998). The maps constructed
involved total oil, gas and water production (Figures 32, 33 and 34), as well as
water cut, water injection and surface tubing injection pressure maps (Figures 35,
36 and 37). All maps were generated based on a monthly report of fluid extracted
and fluid injected, and their associated injection pressures.
Analysis of the production data maps for the seismic time-lapse interval,
provides useful information about the reservoir behavior over time.
Figure 31. Water injection wells with water injection profiles shown on the right of
each log. Note the change in water injection profiles from 1994 to 1997 (most of
the water was initially being injected into the Upper Paddock limestone and now
is being injected into the Lower Paddock fractured dolomite.
Figure 32. Total oil production bubble map for 1997 (left) and 1998 (right) in
barrels. The size of the bubbles is relative to the amount of oil produced for each
well. Injectors are green and producers are red. Main wells from the CO
program in San Andres are labeled as a reference. The big square is the area
that corresponds with the seismic survey Phase VI. The red square is the 4D
seismic area for Phase VII. The whole southern line of wells have been shut off
because of high water production and mechanical problems (Hickey, 2000).
VGWU 103
VGWU 103
VGWU 116
VGWU 116
VGWU 118
VGWU 118
Figure 33. Total gas production bubble map for 1997 (left) and 1998 (right) in
Figure 34. Total water production bubble map for 1997 (left) and 1998 (right) in
VGWU 103
VGWU 103
VGWU 103 VGWU 103
VGWU 116
VGWU 116
VGWU 116
VGWU 118 VGWU 118
VGWU 118 VGWU 118
VGWU 114
VGWU 114
VGWU 126 VGWU 126
Figure 35. Water cut map for 1997 (left) and 1998 (right). Notice how the oil
production moved from the southeastern corner to the northwestern corner from
one year to the other.
Production and water injection volumes are given in stock tank barrels (STB).
In the case of gas production they were given in MMSCF and converted to
reservoir barrels (RB) using PVT data provided by TEXACO (Figure 39). The
bottom hole pressure in the reservoir varies greatly but generally ranges between
1500 and 2500 psi. An average reservoir pressure is 2000 psi (Hickey, 2000). At
2000 psi., a GOR of 350, temperature of 118
F and 40 API oil, the FVI (Bo) is
about 1.2 RB/MSCF. This was the gas volume factor used in the calculation of
gas in reservoir barrels.
Figure 36. Total water injection bubble map for 1997 (left) and 1998 (right) in
Figure 37. Average water injection pressure bubble map for 1997 (left) and 1998
(right) in psi.
Figure 38. Variation of the Formation Volume Factor (FVF) versus reservoir
pressure for oil and gas (reservoir fluids).
The analysis of the total oil production maps shows that oil moved toward the
northwestern portion of the study area from 1997 to 1998. The VGWU-103 well
shows a considerable increase in oil production while well VGWU-119 in the
eastern portion of the area, and well VGWU-116 south of 103, exhibited a
noticeable decrease. It is important to note that the whole bottom line of producer
wells in the area, including the VSP well VGWU-127, have been shut off since
1995, because of low production (mostly due to flooding) and mechanical
problems in the two most eastern wells in the line.
Gas production maps show that increasing gas production toward the
northwestern portion of the area has occurred. Wells VGWU 118 and 116
experienced a gas production drop while VGWU-103 had an increase in gas
Water production maps for the two years indicates that the aquifer is active at
the edges of the field . As previously stated, the bottom line of producer wells has
been shut off because they produced mostly water. One piece of evidence for
this is that well VGWU-126 still had some production, however, it was all water
(see Figure 35). In general water production behavior remained constant from
1997 to 1998, nevertheless, well VGWU-116 had an obvious increase in water
production whereas well VGWU-114 to the west, relatively close to well #116,
experienced a dramatic decrease.
Water cut maps reveal important information about the spatial movement of
the oil between 1997 and 1998. From the 1997 water cut map it is possible to
observe that the majority of oil per barrel of total fluid produced was concentrated
in the southeastern portion of the area. During 1998 oil appeared to move toward
the northwestern corner. The largest amount of oil per water produced are now
concentrated in that portion of the study area. A brief look over the water injection
maps show that a great amount of water has been injected in the west side of the
area, and almost no water has been injected in the east portion.
A coherence cube was constructed for the compressional waves in the
baseline survey of the 4D seismic area. The main objective was to determine
possible structural patterns within the Glorieta Paddock interval using the
higher resolution Vacuum seismic survey. Correlation of the structural pattern
inside the 4D area within the Maljamar structural seismic interpretation was also
The parameters used in the computation of this new coherence cube were
slightly different from those used by Galarraga (1999) in the Maljamar survey,
and they can roughly be listed as follows:
- Continuity/Coherency algorithm used (Poststack/PAL Landmark):
Correlation pattern of 4 traces,12 msec window and 0 dip search: 4 traces
pattern filtrate the information for more clarity, 12 msec window and 0 dip
because the area is relatively flat and the continuity changes expected are not
- From Promax: trace/scalar, multiply by a scalar value of 200 (increases the
amplitude values and changes the polarity, for easier display)
- F-K Dip Filter from Poststack/PAL: F-K fan filter
- Promax: Trace DC removal
The last two process are related with the enhance of the data display.
The interpretation of the coherency cube was made in the conventional way
with time-slices every 8 ms, starting with the stratigraphic top of the reservoir and
finishing with the bottom of the Paddock.
Figure 39 shows a sequence of time-slices within the reservoir. For the time-
slices near the top and within the Lower Paddock fractured dolomite (time-slices
904, 912 and 920ms) patterns of low coherence have northwest-southeast and
northeast-southwest trends, which were interpreted as faults (Figure 40).
Comparisons between coherence time-slices from both the Maljamar and
Vacuum surveys were made. The main fault trends observed in the Vacuum
coherency cube coincide with the regional structural trends interpreted in
Maljamar survey (Figure 41).
Considering the coherence time-slice from Maljamar, it is possible to identify a
set of faults near the Vacuum survey, one to the west with a NW-SE trend,
another to the south with a E-W trend and some others further south with NE-SW
trend. Coincidentally, these are the same trends interpreted for the coherence
cube in the Vacuum survey at the dolomite level. Furthermore, time-lapse and
production data seem to support the existence of the two faults interpreted in the
Vacuum survey (see sections 6.2 and 6.3.2).
Figure 39. Time-slices every 8 ms showing the coherent (red) or non-coherent (black) character of Glorieta
Paddock interval. Notice two trends of non-coherent patterns.
Figure 40. Structural interpretation of the lower part of the Upper Paddock dolomite and the Lower Paddock
fractured dolomite at Vacuum survey. Two sets of faults, one NW-SE and the other NE-SW, with juxtaposed
fractures is proposed.
Figure 41. Comparison of Maljamar (left) and Vacuum (right) surveys coherence
cube interpretation. The absence of structural features within the Vacuum area
for Maljamar coherence cube is due to the poor resolution of this survey. Label 1.
indicates the presence of faults with NW-SE trend in Maljamar survey. Label 2.
indicates the faults with NE-SE trend. Both trends are observed in the Vacuum
coherence cube as non-coherent patterns.
6.2.1 Structure maps in time
The seismic events for the tops of the Glorieta and Paddock were clear and
continuous for both baseline and repeat compressional wave volumes (Figure
42). Time structural maps show flexural trends that have similar directions on
those mentioned previously (Figures 43).
A horizon located at about 920 ms was interpreted as the approximate bottom
of the Paddock. The upper limit of the interval studied was a horizon located
above the Glorieta (a trough). Top Paddock was not considered for the
calculations at any time (see section 6.3.1 for reasons).
6.2.2 Time delay analysis (relative velocity change)
Compressional seismic velocity maps in a time-lapse sense were computed
over an interval between one cycle above the Glorieta event and the bottom of
the Paddock, approximately a 55 ms window. Figure 44 displays the velocity
difference maps for decreasing (negative values in percentage) and increasing
(positive values in percentage) P-wave velocity.
There is a general tendency for the P-wave velocities to decrease (about -
4%). This decrease is larger in the areas surrounding the injector well VGWU
122. Nonetheless there are two other localized tendencies: About the south
central portion of the area velocities are increasing (5%), and following this area
Figure 42. Comparison between the same line (Line 66) extracted from
compressional volume baseline (left) and repeat (right). The main horizons
picked are displayed.
Figure 43. Time structural maps of Glorieta top (left) and Paddock top (right).
Figure 44. Compressional velocity differences Pre-Post (baseline repeat
survey). Left map displays the relative velocity decrease, the right map displays
the relative velocity increase. The scales for both are percentages (See text for
to the north west there is a narrower area where the velocities seem to be
unchanged. Almost no variation in velocity is displayed in the dark shaded area (-
1 to 1%).
An increase in pore pressure (Pp) within the reservoir, hence a decrease in
effective pressure, is proposed as an explanation for the general decrease in
compressional velocity. The bulk modulus is strongly affected by pressure
changes (Mavko, et al, 1998). When pore pressure increases K and decrease,
therefore Vp decreases:


Vp (7)
If the structural interpretation made using coherency analysis is overlain by
both velocity decreasing and velocity increasing maps an interesting correlation
is observed. Figure 45 displays the velocity decreasing map with the structural
trends proposed. The fault with NW-SE trend seems to fit the narrower area
where the relative velocities did not change with time. On the other hand, at the
junction of the two faults, there is an increase in velocity through time.
Since the water injection within the Lower Paddock dolomite can be affecting
the reservoir, the results in velocity changes for P-wave can be explained as
In the NW-SE fault: there could be an increase in pore pressure (total
decrease in effective pressure) then, as a consequence, the compressional
velocity decreases. The same effect is observed in general for the whole area. At
the same time, it is possible to have a fluid substitution effect. Water is displacing
the small amounts of oil remaining in the fractured area. The water bulk modulus
is larger than the oil bulk modulus. On the other hand, water is denser than oil.
Figure 45. Compressional velocity difference decrease with structural
interpretation proposed overlaid. Notice how the NW-SE fault matches with the
non-variability in Vp through time.
However, the bulk modulus increases more significantly than the density due to
water substitution. Hence, an increase in Vp is more likely to be expected. Both
effects, fluid substitution (Vp increase) and pore pressure increase (Vp decrease)
could cancel each other and the combined effect would be no change in Vp. This
is what is being observed.
In the junction of the two faults: Production information (Chapter 5) indicates
that in the southern portion of the area the bottom line of wells are shut in
because by 1996 they were producing mostly water. This fact suggests the
existence of better permeability patterns in the areas close to the southern fault.
This is evidence that in some wells the water started to flow to the Lower
Paddock dolomite in 1996. A total flush of the fractures associated to the
southern fault can be expected. Hence, assuming better permeability for that
area, especially for the junction of these two faults, a simple pore pressure effect
is unlikely to occur. Fluid substitution of hydrocarbon to brine can be expected
and therefore Vp would increase. The results are in agreement with this
interpretation: Vp increases in the junction of the two faults.
Multicomponent analysis was done using time differences for the fast shear
wave (S
) and the slow shear wave (S
), as well as anisotropy measurements
and amplitude extraction.
6.3.1 Time delay analysis (relative velocity change)
Relative velocity difference maps (post - pre) were calculated for S
and S
Figure 46 and Figure 47 display velocity increase and velocity decrease maps,
respectively. The scale for both is given as a percentage, positive for increase in
velocity and negative for decrease. The velocity difference maps were calculated
with the same methodology used for P-wave. The computations were made over
an isochron between a horizon above the Glorieta top and approximately the
bottom of the Paddock. The Glorieta top was not taken as the top of the
computation window since the first interpretation attempt showed that the
Glorieta seismic event was changing in time for S
and S
volumes (see Figure
49). The Glorieta top was a horizon chosen as a non-variant, since there is no
water flooding going in the Glorieta. Nevertheless, it was evidently changing. For
that reason, a horizon one cycle above Glorieta (~1900 ms) was chosen as a
static horizon, since it was discernibly more stable. This horizon was flattened to
1900 ms and the calculations were recomputed between this horizon and the
bottom of Paddock (~2010 ms). The total window was then ~110 ms.
The top of Paddock was not used either since there was a remarkable
variability or non-continuity of the event from one volume to another. Moreover
the seismic window between the top of Paddock and the bottom where it could
be defined, is extremely small resulting in unstable computations.
Figure 46. Relative velocity increase (pre post) for S1 (left) and S2 (right)
Figure 47. Relative velocity decrease (pre post) for S1 (left) and S2 (right)
The resultant velocity maps show that there is an increase in S
(about 25%) in an area with NW-SE trend. Coincidentally, in the same area it is
possible to see that S
velocities decrease by about 23 to 26%. With the
maximum values located toward the NW corner of the survey. Isolated spots of
small decreases in S
and increases in S
are located to the sides of the main
trend. The high percentage values of velocity variation for both shear waves are
due to the small window taken in the calculations. In general, the reservoir is
seismically thin, it means that the reservoir is defined in a small time window and
therefore has limited seismic response, but the interpretation was also limited by
the fact that no event could be picked below the bottom of Paddock in a reliable
The fact that an event one cycle above Glorieta is being used as the top of the
interval is some what dangerous since the San Andres immediately overlies the
Paddock reservoir, and the San Andres is changing through time. There is a risk
of confusing responses if events within the San Andres are taken as a top of the
Paddock interval. However, one cycle above Glorieta top is still considered more
The trend of increasing S
and decrease S
coincides with the NW-SE fault
proposed in the structural model, as well as with the junction of that fault and the
NE-SW fault.
An increase in pore pressure, resulting in a decrease in effective pressure, is
interpreted to cause of the decrease in S
velocity in the fractured area. The fluid
being injected opens up the fractures with NW-SE direction associated with the
fault of the same trend. The aperture of the fractures in a perpendicular direction
with respect to S
could cause the shear modulus () to decrease and
consequently VS
will decrease. Furthermore, if there is an effect of fluid
substitution in the fracture area, which is known to be small due to the lack of
porosity within the fractures, that change will also affect VS
. As discussed in the
previous chapter, the density of the water is larger that the density of the oil, so
from equation (8) it can be seen that VS
will decrease.

= Vs (8)
Where is the shear modulus and the total density.
Talley (1997) in his work on the San Andres showed that the anisotropy
computations over the San Andres interval were due to an increase in S
He proposed that the shear wave traveling parallel to the open fracture direction
) responds to changes in viscosity within the fractures, with less attenuation
and higher velocity. This is just a hypothesis and barely applicable in this study.
6.3.2 Anisotropy analysis
Anisotropy measurements were the basic measurements made for the
Glorieta-Paddock reservoir. It is well known that these measurements are very
sensitive to horizon time interpretation errors. Although careful attention was
made into the picking process, the fact that this interval is deeper than the San
Andres interval means the resolution is lower. In addition, the interval is varying
in time due to the water injection and production, which made the interpretation
unusually complicated. Picked horizons and how they vary for each volume are
shown in Figure 48.
One of the first complications observed was the fact that it was not possible to
pick the top Paddock on either the S
post or both S
pre and post volumes (as
explained in the previous section). A first attempt to calculate the anisotropy
between the Glorieta horizon and the bottom of the Paddock horizon, resulted in
anisotropy values in the order of 50 to 50%. Although it is known that small
intervals of highly fractured rock can have high anisotropy change, a test was
conducted to verify the validity of the interpretation.
An event one cycle above the Glorieta was chosen as the top of the interval,
and the results at this time were more reliable and stable. The result of pre and
post anisotropy obtained was in the order of 25 to 25% (Figure 49). It is a high
anisotropy value compared with the San Andres interval (Cabrera, 2000) but the
size of the window is a factor to take into account.
Figure 48. Line 66 for each shear wave volume, S
and S
(pre and post) the Glorieta and Paddock bottom
picks are displayed.
Figure 49. Anisotropy maps baseline (left) and repeat (right).
f we analyze both the baseline and repeat anisotropy maps, a dramatic
change is observable for the NW corner of the seismic area. Calculating an
anisotropy difference map between pre and post (Figure 50) resulted in the
anomaly becoming more evident. This anomaly has negative values and an
elongated shape with a NW-SE trend.
The baseline anisotropy map for the Glorieta Paddock interval shows the
NW-SE trend to have negative anisotropy. It means VS
is faster than VS
delay for S
is smaller than the time delay for S
). By definition or nomenclature
has to be slower than VS
, if the opposite happens an incorrect rotation of
the volumes could be the problem. At this point it is necessary to clarify that the
rotation of the volumes used in this study was constant, using an angle of 118
for the fast shear waves. The anisotropy values changed radically in the post
anisotropy map. Here VS
is bigger than VS
, as was expected. A possible
explanation could be that there was a set of open fractures with a NE-SW
orientation created by the main or regional stresses in the field.
By the time of the water flooding in the dolomite the NW-SE fractures
associated with the fault with the same trend could have been opened. Different
fracture sets have different aspect ratios, the response of each one of them to
pressure and fluid effects due to water injection varies greatly.
Figure 50. Baseline repeat anisotropy map difference. Faults are overlaid. See
text for explanations.
The southeastern portion of the NE-SE fault does not have significant
anisotropy. It could be interpreted as either an absence of fractures or the
existence of a set of open conjugate fractures. The assumption of better
permeability paths for this fault based on production data makes us to think that
the second interpretation is the best alternative. In the other extreme the fault has
negative anisotropy values in both the baseline and repeat maps, indicating that
is greater than VS
. Since the fractures interpreted here have a NE-SW
direction the chosen S
direction would be the natural S
direction. Moreover, for
the post map we observe that the southeastern portion of the NE-SW fault has
negative anisotropy values, indicating that a possible increase in water flooding
in that area possible caused the NE-SW fractures to open more than the NW-SE
fractures (increasing compliance in that direction).
Since the anisotropy values are quite high, a test was conducted to prove the
validity of these results: a horizon at about 2750 ms was picked. This horizon
presented good continuity conditions and its time map for both S
and S
(pre and
post) showed small changes. A window between the horizon picked and flattened
(above Glorieta) and the 2750 ms horizon was taken. Anisotropy calculations
were made over that interval, a total of 850 ms, and the results were visually
similar to the ones obtained for the 110 ms interval (trough above Glorieta and
bottom of Paddock) (Figure 51). Figure 52 show a comparison between the
anisotropy difference maps for both intervals. The same anisotropy features but
with smaller anisotropy values (from 5 to 5%), can be observed.
Figure 51. Seismic line (L66) showing the top and the bottom of the two intervals
used in the anisotropy calculations
Figure 52. Anisotropy maps computed over an interval of 110 ms (left) and over an interval of 850 ms (right).
0 1000
The anomalies are there and the high values obtained for the Glorieta-
Paddock interval are possibly a consequence of the size of the interval used in
the calculations. A better multicomponent data quality (by reprocessing, for
instance) could assure better definition of the events below the bottom of the
Paddock so that an interpretation of deeper events and new anisotropy
calculations can be conducted.
6.3.3 Amplitude analysis
During the interpretation of the S
and S
volumes (pre and post) a change in
the character of the reflectivity of the seismic response within the reservoir
interval was observed between the two surveys. A seismic event (a pick) barely
observable in S
baseline had greater amplitude in S
repeat and even more
evident in S
baseline. In the case of S
repeat, the event appears as a
continuous horizon. Here it had a strong clear signature specifically for an area
where it was not observable in any of the other volumes. A map of this event was
derived for each volume. Observe that in S
post, the reflector appears in an area
where it does not appear in any of the other volumes (Figure 53). Coincidentally,
this area coincides in some lines with the increase in travel time for S
in velocity) (Figure 54).
RMS amplitude difference maps were computed for both S
and S
amplitude pre minus RMS amplitude post), and they reveal a significant
Figure 53. Time structure maps of the middle event between the top and the
bottom of the Paddock for both S
(upper maps) and S
(lower maps) shear wave
volumes, for baseline (left maps) and repeat (right maps) surveys. Notice how
the horizon covers the entire map in the repeat maps, especially in the
interpretation made on the S
Baseline Repeat
Figure 54. Line 66 for both S
baseline (left) and repeat (right) surveys displaying
the Glorieta top and the Paddock bottom. Note the time delay (decrease in S2
velocity) at the bottom of the Paddock in the same area where there is an
amplitude anomaly or an event that exists in the repeat survey but not in the
baseline. The interpretation of this middle event is also displayed (two first
maps). The third map is the RMS amplitude ratio difference map (explained
afterward). Note how the anomaly coincides with an amplitude increase.
Baseline Repeat
difference in amplitude range for S
with respect to S
(Figure 55). To avoid
interpretation problems due to different amplitude ranges the amplitude ratio
between S1 and S
was computed.
Figure 55. RMS amplitude difference maps (base repeat) computed for S
and S
(right). Notice the large scale difference between each one of the shear
waves. The range of S
amplitude values is roughly 3 times larger.
Baseline Repeat
Figure 56. RMS amplitude ratio (RMS amplitude S
/ RMS amplitude S
) baseline
(left) and repeat (right). The range of amplitude values is now normalized.
For both surveys RMS amplitude ratio maps were computed (RMS amplitude
/RMS amplitude S
) (Figure 56). The maps themselves do not show particular
differences or anomalies that could be related to either anisotropy or velocities,
however if the difference between both ratios is taken this result exhibits a trend
similar to the one obtained for anisotropy (Figure 57). Therefore it follows the
trend of the non-coherent pattern observed in the Vacuum coherency cube
analyzed previously. Although the amplitude trend does not seem to be in the
same position, there is an amplitude anomaly that appears to be related to the
shear wave splitting and the time-lapse results.
Baseline Repeat
A possible explanation for the slightly different location of this feature with
respect to all the others (anisotropy, velocity and coherence) could be given by
Cabrera (2000) in the analysis of the influence of processing parameters on
amplitude extractions (see section 6.4)
Figure 57. RMS amplitude ratio difference (S
pre S
post) (left) and
anisotropy difference (right) maps. It seems that the high positive amplitude
difference values are related with the negative anisotropy differences, or at least
they display similar trends.
Modeling was performed in order to explain quantitatively the anisotropy and
velocity difference anomaly. In this regard Gassmanns equations were applied
using well log information from five wells. The low frequency assumption of
Gassmans equations (strictly valid at zero frequency) predict the resulting
increase in effective bulk modulus, K
, of the saturated rock through the
following equation:
) (
0 0 0 fl


dry sat
= (9)
Where K
= effective bulk modulus of the dry rock
= effective bulk modulus of the rock with pore fluid
= bulk modulus of the mineral material making up the rock
= effective bulk modulus of pore fluid
= porosity

= effective shear modulus of dry rock

= effective shear modulus of rock saturated with pore fluid
The parameters necessary for the input into the model were extracted either
from well logs or calculated. The total reservoir was subdivided in the Glorieta
(considered as a dolomite in the calculations), the Upper Paddock limestone and
the Lower Paddock dolomite. Average porosity was obtained from well logs for
each of these intervals. Vp and Vs were calculated for a dry rock using porosity-
Vp and porosity-Vs relationships available in the literature (Mavko et. al. 1998).
Similar relationships were used to calculate densities. Dry bulk modulus (K
) and
dry shear modulus (
) were calculated from equations (7) and (8) for
compressional and shear velocities respectively.
Table 5 summarizes the values for the bulk and the shear moduli of the
mineral, extracted from published values (Mavko, et. al., 1998). The wells used in
the calculations were VGWU-127, WS-13, SR-9, SR-10 and SR-12. Since no
partial saturation data were available, two extreme conditions were considered:
1. Initially the rock was 100% saturated with oil, 2. After water flooding, the rock
was 100% saturated with water or brine.
The density and bulk modulus for brine and oil under the reservoir conditions
(given in the Chapter 5.5), were calculated with the program FLAG (Fluid
Acoustic for Geophysics) designed by Batzle and De hua Han (2000). The
reservoir conditions for the calculations of those fluid properties were the same
as the ones described in the production data chapter. However, some
parameters necessary for the calculations, such as salinity and gas gravity, were
assumed. It is important to recall that, after the unitization and the water flooding
started, the oil from this reservoir is above bubble point with no gas cap and the
gas is in solution.
Table 6 summarizes fluid properties used in the calculations. Table 7 shows
the porosity values extracted and all the other calculated values for each interval.
Fluid K
(GPA) (gr/cm
Oil 0.159 0.54
Water 2.68 1.03
Table 5. Bulk modulus and density for each fluid considered in the calculations.
From Batzle and Han (2000).
Mineral k
Dolomite 76.4 49.7
Limestone 70.2 29
Table 6. Bulk and shear modulus of the mineral for each rock type used in the
calculations. From Mavko et. al. (1998).
Well Porosity (%) Vp dry (Km/sec) Vs dry (Km/sec)
GLR U/Pdk L/Pdk GLR U/Pdk L/Pdk GLR U/Pdk L/Pdk
VGWU127 11.14 14.83 10.89 5.5617 4.6388 5.5852 3.0560 2.4797 3.0678
WS-13 12.08 14.63 10.36 5.4739 4.6519 5.6351 3.0118 2.4874 3.0929
SR-9 6.41 8.37 8.49 6.0057 5.0684 5.8103 3.2794 2.7295 3.1810
SR-12 9.07 14.71 6.48 5.7556 4.6463 5.9989 3.1535 2.4841 3.2759
SR-10 12.20 13.77 9.46 5.4618 4.7091 5.7191 3.0057 2.5206 3.1352
Density dry (gr/cm
) Density 100% oil (gr/cm
) Density 100% water (gr/cm
VGWU127 2.611 2.483 2.614 2.671 2.563 2.673 2.725 2.636 2.726
WS-13 2.599 2.485 2.621 2.664 2.564 2.677 2.723 2.636 2.727
SR-9 2.672 2.543 2.645 2.706 2.588 2.691 2.738 2.629 2.732
SR-12 2.637 2.484 2.671 2.686 2.564 2.706 2.731 2.636 2.738
SR-10 2.597 2.493 2.632 2.663 2.567 2.683 2.722 2.635 2.730
Kdry (GPA) K 100% oil (GPA) K 100% water (GPA)
VGWU127 48.25 33.08 48.74 48.44 33.38 48.93 51.27 37.68 51.72
WS-13 46.43 33.28 49.79 46.63 33.58 49.98 49.60 37.89 52.69
SR-9 58.06 40.06 53.61 58.20 40.41 53.77 60.26 45.16 56.19
SR-12 52.40 33.19 57.90 52.57 33.49 58.04 55.08 37.80 60.11
SR-10 46.19 34.17 51.60 46.39 34.47 51.78 49.37 38.81 54.35
Table 7. Extracted (from well logs) and calculated values used in the calculations of Vp and Vs using
Gassmanns equations.
The results summarized in Table 8 show the modeled behavior that is
expected for water flooding in a carbonate reservoir, using Gassmanns
approach. The differences expected are small, nonetheless they give an idea of
the general response. Compressional velocity appears to increase slightly when
water substitute oil in the dolomite, the increase is even greater for the limestone.
On the other hand, shear velocities seem to decrease, and again the greatest
change is observable in the limestone rather than in the dolomite.
For the case of shear waves, the model agrees with the results analyzed
previously in the seismic data, specifically for the slow shear wave S
. It does not
agree with the behavior of the fast shear wave, S
, observed in the seismic
(Chapter 6.2.2). The results for compressional velocities are not in agreement
with the results of the model either.
Here we have to take in account all the limitations or assumptions that this
model works with. Moreover, it is necessary to consider that pressure effect was
not taken into account in any stage of the calculations, and, as it is well known,
velocity is largely dependent on pressure, particularly for fractured rocks.
One of the basic assumptions of Gassmann model is that the medium is
isotropic and homogeneous (Mavko, et al, 1998). This assumption is totally
violated when there is evidently anisotropy in the medium, as the shear wave
analysis showed. The fact that the interval being considered is composed of
Vp 100% oil (Km/sec) Vp 100% water (Km/sec)
VGWU127 5.5053 4.5785 5.5298 5.5444 4.6924 5.5683
WS-13 5.4134 4.5924 5.5821 5.4550 4.7067 5.6192
SR-9 5.9716 5.0372 5.7659 6.0003 5.1756 5.7987
SR-12 5.7085 4.5864 5.9644 5.7426 4.7006 5.9932
SR-10 5.4008 4.6530 5.6702 5.4428 4.7692 5.7051
Vs 100% oil (Km/sec) Vs 100% water (Km/sec)
VGWU127 3.0214 2.4407 3.0338 2.9910 2.4069 3.0041
WS-13 2.9747 2.4487 3.0604 2.9422 2.4153 3.0318
SR-9 3.2583 2.7056 3.1538 3.2396 2.6844 3.1298
SR-12 3.1246 2.4453 3.2547 3.0991 2.4117 3.2358
SR-10 2.9683 2.4838 3.1052 2.9355 2.4518 3.0787
Vp difference (water sat -
oil sat)(%) (Km/sec)
Vs difference (water sat -
oil sat(%)(Km/sec)
VGWU127 0.71 2.43 0.69 -1.02 -1.41 -0.99
WS-13 0.76 2.43 0.66 -1.10 -1.39 -0.94
SR-9 0.48 2.67 0.57 -0.58 -0.79 -0.77
SR-12 0.59 2.43 0.48 -0.82 -1.39 -0.58
SR-10 0.77 2.44 0.61 -1.12 -1.30 -0.86
Table 8. Compressional and shear velocities calculated with the parameters
given in Tables 5, 6 and 7. The difference in percentage between the velocities
calculated with 100% water saturation and the ones calculated with 100% oil
saturation are also summarized.
three sub-intervals with different lithologies makes the reservoir anything but
Considering all these facts, the Gassmanns equations, although very useful
and precise in many cases (particularly for clastics), are not the best way to
model the Glorieta Paddock reservoir, with the information available to date.
Several questions arise about the reliability of the seismic data available in
this study: Are the anisotropy and velocity anomalies real? Are there processing
effects of or is the response totally geological and reservoir related? This section
attempts to answer these questions, specifically for the Glorieta-Paddock, and to
provide a larger degree of confidence in the data and the process.
Processing applied to the data is the first aspect of discussion. An extensive
and well developed analysis of processing variables and their effect on the time-
lapse results, has been conducted by Cabrera (2000). Cabrera made
interpretations of the main horizons related to the San Andres reservoir in each
set of volumes with a different processing flow applied. He also computed
anisotropy for each of them. The results showed that no matter which processing
was applied, including changes in migration algorithms, the anisotropy anomalies
were similar. Cabrera determined that migration (pre or post stack) moved the
amplitude anomalies at the San Andres level, in comparison with stack data.
The same analysis could be applied to the Glorieta Paddock interval to
determine the direction of amplitude movement. In the amplitude analysis made
for this interval (Chapter 6.3.3) it was found that there is a similar trend in
amplitude differences when compared with anisotropy differences. However, the
amplitude trend was moved to the N-NE with respect to the anisotropy trend. The
hypothesis that the migration algorithms applied are moving the amplitudes to N-
NE in the Glorieta Paddock interval is proposed as an explanation.
Cabrera also calculated the anisotropy values for the static overburden. The
interval above San Andres is supposed to remain unchanged in time, since no
significant production or injection program is occurring there. To that end he used
three different windows, one from 800 to 1500 ms, another from 1000 to 1500ms
and the last from 800 to 1000 ms. for the anisotropy calculations. Figures 58, 59
and 60 display his results in this experiment.
The anisotropy maps for pre and post CO
injection in the San Andres and the
anisotropy difference map, show that the anisotropy values for the static interval
are in a range of -2% to 2%, which is already the level of anisotropy that every
rock has (Crampin, 1994) or it could also be associated with the associated noise
level. Hence, anisotropy changes within the San Andres reservoir are believed to
be due to the CO
and water injection currently underway.
A parallel study on the processing confidence was done by Karr (2000) from
Golden Geophysical, the company that processed all the volumes for RCP
Phase VI and VII. Figure 61 displays a comparison trace by trace of the pre and
post shear wave volumes. Beside one trace from the pre volume there is a trace
from the post volume, and no noticeable difference can be seen. This figure
suggests that the definition of the wavelet parameters for each volume was the
Figure 58. Anisotropy maps computed for the static overburden in baseline and
repeat surveys, and the difference between them. Notice that the anisotropy
values range between 2 to 2%, which is the natural anisotropy typical of many
rock (From Cabrera, 2000)
Figure 59. Anisotropy computations of the static overburden within a window
from 1000 to 1500 ms. The last anisotropy map indicates the few areas where
the anisotropy is out of the background 2 to 2% range due to the static
differences between the two surveys (From Cabrera, 2000)
Figure 60. Anisotropy map of the static overburden computed within a window
from 800 to 1000 ms. Notice the range of anisotropy values increased due to
decrease in window size used in the computations (From Cabrera, 2000)
Figure 61. Crossline 80 displaying Pre and Post CO
shear wave traces side by
side in an interval between 700 and 1400 ms, which is the static overburden. No
appreciable change in the character of the wavelet is observed. (From Karr,
1. At CVU the Glorieta Paddock interval is faulted and fractured. The faults are
NW and NE oriented.
2. Water flooding program within the Glorieta Paddock reservoir can be
seismically monitored.
3. Anisotropy measurements calculated from shear-wave isochron differences
over the Glorieta Paddock interval identified zones associated with
fractures. Due to the effects of the water flooding the slow shear wave
velocities indicate a pressure effect due to water injection and a change in
anisotropy trends, indicative of changes in open fractures. NEE-SWW
fractures were open at the time the baseline survey was acquired, and NW-
SE fractures were opened at the moment the repeat survey was acquired.
1. The major contribution of this study was the correlation of seismic character
change due to the fracture pattern directly affected by the water flooding:
water channeling is more likely to occur in the fractured zones. In extremely
fractured areas there may be no flushing of the matrix oil because the water
displaces oil only through fractures and bypasses the matrix. Considerable
amounts of oil then remain in the non-fractured areas. Therefore, horizontal
drilling is recommended for the zones in the east and west sides of the NW-
SE fault and for the north side of the NE-SW fault.
2. Re-processing of shear-waves looking for the accurate definition of the
anisotropy directions in the Glorieta Paddock reservoir, which are going to
validate the anisotropy results obtained in this study:
Layer stripping in shear waves: It has been expected that S-wave
polarization directions remain constant with depth, but several data
analysis have shown that they did not. To assure the right polarization of
S-waves through depth layer stripping is a method that has been proved
to be useful. Layer stripping involves simply subtracting anisotropy effects
in a layer in order to analyze anisotropy effects in the layer immediately
below. S-wave splitting is cumulative, so that if anisotropy changes with
depth, the effects of anisotropy above the change, unless removed, will
persist in the changed region and confuse the analysis there (Winterstein
and Meadow, 1991). By removing the anisotropy of the overburden and
the San Andres interval, and analyzing the anisotropy per se of the
Glorieta Paddock interval, the anisotropy calculation derived from S-
waves will be more accurate and realistic.
Alford rotation by CMP: The Alford rotation process (Alford, 1986) was
applied to the original field data, since the orientation of the lines, and
therefore the orientation of the sources, did not correspond with the
preferred directions for the azimuthally anisotropic subsurface. A global
rotation of the shear wave data using anisotropy measurements made in
the VSP CVU-200 was conducted. However, there is another alternative
approach using the same technique to correct for horizontal anisotropy
variations within the same area: application of Alford rotation every CMP
has been conducted successfully in other fields. In every CMP the
anisotropy of the medium surrounding that point is corrected for by
applying the right rotation of shear wave energy.
3. 3-dimensional modeling of an anisotropic fractured media should be pursued
in order to have a better understanding of the structure and the process
occurring in the reservoir. Simple models as the penny-shape aligned
cracks (Hudson, 1988) can be considered as a first approach. This model is
the first one to consider variations in the conditions within the crack, from
completely dry, to partially saturated with fluid and gas, to solid. Thomsen
(1995) proposes the same model of saturated aligned cracks but with less
restrictions that allow the model to perform better in upper crustal conditions
(with stiffer crack fluids, equant porosity often substantial, and lower seismic
frequency band). This model is insensitive, at low frequency, to the aspect
ratio of the cracks, frequently difficult to measure; only the crack ratio is
critical. One of the last models developed by Hudson (1999) is most
applicable for this case. This model includes two different cases depending
upon the crack density: in the first, the fault is modeled as a plane distribution
of approximately circular cracks, while elsewhere the faces of the fault are
held together by the ambient pressure and friction. In this model the
distribution of the crack is sparse. The second model consists in plane
distribution of approximately circular stuck regions, within an area where the
faces are separated as for a crack. This model works for high density
fractured media. Commonly, the mathematical expressions of these models
involve rock and crack parameters and/or properties that should be measured
in real samples from the field we wish to model, beside they are pressure
dependent. For that reason some studies should be conducted before a 4D
Rock physics study: as an indispensable part of a multicomponent seismic
monitoring study over a field under a secondary recovery program (in this
case water injection), a complete set of compressional and shear velocities
must be measured on core plugs, using different pressure and fluid saturation
conditions. Although ultrasonic measurements are commonly the most used,
a set of low frequency measurements would be the most useful in order to
make direct comparisons with seismic velocities. With these measurements it
is possible to establish the link between what it is observed in seismic and
what actually happens in the reservoir.
Fracture orientation and fracture density analysis: The crucial parameters in
the definition of a model in fractured media are related to fracture
characteristics. It becomes important to establish a fracture study in order to
have the parameters taken from real fractures in the reservoir. This kind of
analysis is commonly accomplished using borehole images or FMI


and/or cores. Fracture intensity, which is measured as the mean horizontal

spacing of fractures, fracture aperture and fracture orientation (dip and strike)
relative to the borehole position are the parameters obtained from borehole
image analysis (Hurley et. al. 1994). With the definition of these parameters, a
more accurate modeling can be expected, as well as a more precise
multicomponent interpretation, relatively free of bias due to processing
concerns in the seismic data.
Adams, J. E., 1965, Stratigraphic tectonic development of Delaware Basin:
AAPG Bulletin, 49, no.11, p. 2140-2148
Alford, R. M., 1986, Shear data in the presence of azimuthal anisotropy: Dilley,
Texas: Annual Meeting Abstracts, Society Of Exploration Geophysicists,
Batzle, M. L., 2000, Colorado School of Mines, Department of Geophysics
Professor of Geophysics. Personal communication.
Batzle, M., Huan, D., 2000, Software FLAG: Fluid Acoustic for Geophysics.
Bahorich M. and Farmer S., 1995, The coherence cube: The Leading Edge, 14,
no. 10, p. 1053-1058
Benson, R., 2000, Colorado School of Mines, Department of Geophysics
Professor of Geophysics. Personal communication
Blaylock, J. J., 1999, Interpretation of a baseline 4-D multicomponent seismic
survey at Vacuum Field, Lea County, New Mexico: M.Sc. thesis, Colorado
School of Mines, Golden, Co.
Broadhead, R. F., 1993, Glorieta and Upper Yeso shelf. in: Hjellming, C.A., ed.,
Atlas of Major Rocky Mountain Gas Reservoirs: New Mexico Bureau of Mines
and Mineral Resources, p. 146-147.
Cabrera, R., 2000, Analysis and use of seismic attributes for Static and Dynamic
Reservoir Characterization at Vacuum Field, New Mexico. Ph.D. Thesis.
Colorado School of Mines, Golden, Colorado.
Cardona, R., 2000, Colorado School of Mines, Department of Geophysics, Ph. D.
Candidate. Personal communication
Crampin, S., 1985, Evaluation of anisotropy by shear wave splitting: Geophysics,
50, no. 1, p. 142-152.
Davidson, W. T., 1986, depositional and diagenetic aspects of siliciclastic and
carbonate reservoirs in Glorieta Formation (Permian), Northern Midland Basin,
Texas, Association Round Table, AAPG Bulletin, p. 248
Davis, T. L., 2000. Colorado School of Mines, Department of Geophysics
Professor of Geophysics. Personal communication.
Duranti, L., 2000, Colorado School of Mines, Department of Geophysics, Ph. D.
Candidate. Personal communication
Galarraga, M., 1999, 3D interpretation in Vacuum Field area, Permian Basin, Lea
County, New Mexico. MS Thesis. Colorado School of Mines, Golden, Colorado.
Grechka, V., and Tsvankin, I., 1998, 3D desciption of normal moveout in
anisotropic inhomogeneous media: Geophysics, 63, no. 3, p. 1079-1092.
Grechka, V., Tsvankin, I., and Cohen, J., 1999, Generalized Dix equation and
analytic treatment of normal-moveout velocity for anisotropic media: Geophysical
Prospecting, 147, p. 117-148.
Grechka, V., 2000, Colorado School of Mines, Department of Geophysics,
Professor of Geophysics. Personal communication
Haack, R. C., Jacka, A. D., 1985, Deposition, diagenesis, and porosity
relationships in the Glorieta Formation, Keystone Field, Winkler County, Texas,
Association Round Table, AAPG Bulletin, 69, p. 248
Hickey, K., 2000, TEXACO, Vacuum Glorieta West Unit. Personal
Hills, J.M., 1984, Sedimentation, tectonism, and hydrocarbon generation in
Delaware Basin, West Texas and southeastern New Mexico: AAPG Bulletin, 68,
no.3, p. 250-267
Hofmann, R., 2000, Colorado School of Mines, Department of Geophysics, Ph.
D. Candidate. Personal communication
Hudson, J. A., 1988, Seismic wave propagation through material containing
partially saturated cracks: Geophysical Journal, 92, p. 33-37.
Hudson, J. A., 1999, Effective elastic properties of heavily faulted structures:
Geophysics, 64, n. 2, p. 479-485.
Hurley, N. F., 1994, Using borehole images for target-zone evaluation in
horizontal wells: AAPG Bulletin, 78, n. 2, p. 238-246.
Jack, I., 1998, Time Lapse seismic in reservoir management: Distinguish short
course, Society of Exploration Geophysicist.
Jinsong, L., 2000, 3D-3C interpretation of the Drinkard and Abo: in Reservoir
Characterization Project Meeting Report April 13-14, 2000.
Martin, R., 2000, TEXACO, Vacuum Glorieta West Unit. Personal communication
Mattocks, B. W., 1998, Borehole geophysical investigation of seismic anisotropy
at Vacuum Field, Lea County, New Mexico: Ph.D. Thesis. Colorado School of
Mavko, G., Mukerji, T. and Dvorkin, J., 1998, The rock physics handbook: Tools
for seismic analysis in porous media: Cambridge University Press, USA.
Mndez - Hernandez, E., 1999, Influence of shallow heterogeneities on
multicomponent 4D seismic data at Vacuum Field, New Mexico. MS Thesis.
Colorado School of Mines, Golden, Colorado.
Michaud, G., 2000, Colorado School of Mines, Department of Geophysics, Ph. D.
Candidate. Personal communication
Pantoja, D., 2000, Colorado School of Mines, Department of Geophysics, Ph. D.
Candidate. Personal communication
Peacock, C. P., Sanderson, D. J., 1994, Geometry and Development of Relay
Ramps in Normal Fault System: AAPG Bulletin, 78, p. 147-165
Pranter, M.J., 1999, Use of a petrophysical-based reservoir zonation and
multicomponent seismic attributes for improved geologic modeling: Ph.D. thesis,
Colorado School of Mines, Golden, CO
Purves, W. J., 1990, Reservoir description of the Mobil Oil Bridges State Leases
(Upper San Andres Reservoir), Vacuum Field, Lea County, New Mexico, in D.G.
Bebout and P.M. Harris, eds. Geologic and engineering approaches in evaluation
of San Andres/Grayburg hydrocarbon reservoirs-Permian Basin: Bureau of
Economic Geology, University of Texas at Austin, p. 87-112
Raines, M., 2000, TEXACO, Central Vacuum Unit. Personal communication
Reservoir Characterization Project Meeting Report April 13-14, 2000
Talley, D. J., 1997, Characterization of a San Andres carbonate reservoir using
four dimensional, multicomponent attribute analysis. MS Thesis. Colorado School
of Mines, Golden, Colorado.
Tatham, R. H., and McCormack, M. D., 1991, Investigations in geophysics no. 6
Multicomponent seismology in petroleum exploration: Society of Exploration
TEXACO PHILLIPS, 1990, Proposed Vacuum Glorieta Unit, Lea County, New
Mexico: Engineer geological technical committee report.
Thomsen, L., 1988, Reflection seismology over azimuthally anisotropic media:
Geophysics, 53, no. 3, p. 304-313.
Thomsen, L., 1995, Elastic anisotropy due to aligned cracks in porous rocks:
Geophysical Prospecting, 43, p. 805-829.
Trudgill, B., Cartwright , J., 1994, Relay-ramp forms and normal fault linkages,
Canyonlands National Park, Utah: Geological Society of America Bulletin, 106, p.
Winterstein, D. F. and Meadows, M. A., 1991, Shear-wave polarizations and
subsurface stress directions at Lost Hills field: Geophysics, 56, no. 09, 1331-
Winterstein, D. F. and Meadows, M. A., 1991, Changes in shear-wave
polarization azimuth with depth in Cymric and Railroad Gap oil fields:
Geophysics, 56, no. 09, 1349-1364.