You are on page 1of 107

3D SEISMIC REFLECTION SURVEY DESIGN AND MODELLING AT THE

BEENYUP WASTE WATER TREATMENT SITE, WESTERN AUSTRALIA












Report No.: GPH 4/07







By Christopher A. Semeniuk
BSc. (Geophysics) Curtin







This report is presented as part of the requirement for the units
Geophysics Honours Dissertation 494 Part A & B, units totaling 50 credit points in the
BSc (Geophysics) Honours from Curtin University of Technology. The work
is the result of supervised research; however, the report has been prepared
by the student who is solely responsible for its contents.




DEPARTMENT OF EXPLORATION GEOPHYSICS
Curtin University of Technology



November 2007

ii
Abstract
In Perth, Western Australia, there has been both an increasing demand for water and
decreasing rainfall over recent years. Managed Aquifer Recharge is a water
recycling method identified as having the potential to reduce pressure on Western
Australias surface and ground resources. For this reason a treated waste water
injection trial is planned for the Beenyup Waste Water Treatment plant. The
treatment plant is located in the northern Perth suburb of Craigie. The trial will
include detailed hydraulic flow and reactive transport modelling of the injected
water.

Accurate modelling requires precise knowledge of the hydrostratigraphy below the
injection site. Consequently a high resolution 3-D seismic reflection survey will be
used to assist in building a detailed groundwater flow model. An optimal 3D survey
geometry has been designed based on a preliminary 2D survey, VSP data and the
resolution required for the target injection zone within the Leederville Formation.
3D survey design was faced with various difficulties as it needed to be designed with
a number of exclusions zones related to topographic mounds, vegetation and existing
infrastructure.

A 26 degree skewed design was selected to be the best design for the site. This
design showed a compromise between acquisition time, fold and azimuth
requirements. However site conditions changed substantially and the survey was
redesigned a day before acquisition commenced. An orthogonal geometry with 20
source lines, 8 receiver lines, and a maximum of 1254 shots was used. Fold, offset
and azimuth distributions were optimised to obtain the best results in the new site
conditions.

iii
The ability of time lapse seismic to detect changes in elastic properties caused by the
injection at Beenyup, was then examined by constructing synthetic seismic models.
The models were constrained using information from borehole logs, 2D seismic,
VSP, and results from core analysis because sonic or density logs were not acquired
near the site. Velocity-pressure relationships presented by Zimmer et al. (2007a & b)
were used to determine changes in seismic velocity as a function of effective
pressure.

Pore pressure is expected to increase during injection, resulting in a reduction in
effective pressure. Both P and S wave velocities decrease as a result of the reduced
grain to grain contact. If an increase in pore pressure between 50 and 100kPa is
expected at Beenyup, then the velocity will decrease by 3 to 6m/s. The amplitude of
reflections within the injection interval increase by a maximum of 3%. These small
changes may be detectable by seismic methods, particularly 3C VSP, if coupling and
near surface conditions are consistent between surveys. This will reduce the
probability of introducing other unwanted time lapse effects.

iv

















Dedication

To the memory of my late grandfather Waclaw Semeniuk, who funded my university
studies

v
Acknowledgments


I would like to acknowledge the Water Corporation for providing funding and data
for this research.

Special thanks to Dr Milovan Urosevic and Dr Brett Harris for their guidance and
technical input. Without their help my thesis will not be as complete.

I would also like to thank Luke Frederick, Aleksander Dzunic, Michael Sykes, and
Putri Wiseman for providing help with software and offering new ideas and
perspectives.

Lastly I would like to acknowledge my parents who have helped fund my university
studies, and put up with me during the more stressful times during the year.

vi
Table of Contents

1 INTRODUCTION 1
1.1 Perths Water Supply and MAR 1
1.2 Proposed research 4
1.3 History and Location of the project area 5

2 PERTH BASIN GEOLOGICAL SETTING AND GEOLOGY 7
2.1 Regional Geology 7
2.2 Stratigraphy 11
2.3 Local geology interpreted from borehole logs 16

3 EARLY SEISMIC MEASUREMENTS AT THE SITE 21
3.1 2D seismic 21
3.2 Check shot survey 25

4 3D SEISMIC SURVEY DESIGN 27
4.1 Access constraints (exclusion zones) 27
4.2 Survey constraints 30
4.3 Survey configuration 30
4.3.1 Shot and receiver interval 31
4.3.2 Line position and spacing 34
4.3.3 Migration aperture 35

5 SURVEY GEOMETRIES 36
5.1 Orthogonal geometry 37
5.2 Skewed geometry 48
5.2.1 Results of the 45 degree skew 48
5.2.2 Results of the 26 degree skew 53
5.3 Final Survey design 57





vii
6 TIME LAPSE SEISMIC THEORY 60
6.1 Pressure effects 61
6.2 Fluid effects 64

7 SYNTHETIC MODEL CREATION 68
7.1 Interval velocities prior to injection 70
7.2 Groundwater temperature, salinity and pressure 73
7.3 Injected water temperature, salinity and pressure 74
7.4 Formation density 74

8 MODELLING RESULTS 77
8.1 Calculation of Interval velocities after injection 77
8.2 Synthetic models 83
8.3 Application to pressure monitoring at Beenyup 90

9 CONCLUSIONS AND RECOMMENDATIONS 92

10 REFERENCES 95




viii
List of figures

FIGURE 1.1: The mar process to be trialled at Beenyup WWTP. 3

FIGURE 1.2: Location of the project 6

FIGURE 2.1: Structural components of the Perth Basin 9

FIGURE 2.2: Mesozoic stratigraphy of the Perth Basin 10

FIGURE 2.3: Cainozoic stratigraphy of the Perth Basin 15

FIGURE 2.4: Location of the boreholes used to interpret local geology 17

FIGURE 2.5: Gamma and resistivity log from the Beenyup borehole. 20

FIGURE 2.6: Gamma and resistivity logs from the Craigie borehole. 20

FIGURE 3.1: location of 2D line 23

FIGURE 3.2: Processed 2D seismic section with the gamma log overlain 24

FIGURE 3.3: Computed average and interval velocity beneath beenyup from the
check shot survey. 26

FIGURE 4.1: Location of exclusion zones 29

FIGURE 4.2: Frequency spectrum of 10 traces (31-40) from the 2d line 33

FIGURE 5.1: Preliminary orthogonal geometry 39

FIGURE 5.2: First template (receiver lines 1-4 and associated shot points) 39

FIGURE 5.3: Fold distribution from the preliminary orthogonal design 40

FIGURE 5.4: Modified orthogonal design with 2m shot intervals to the west, and
4m to the east 44

FIGURE 5.5: Fold distribution resulting from decreasing the shot point interval
in the west 44

FIGURE 5.6: Near offset distribution of the second orthogonal design 45

FIGURE 5.7: Azimuth distribution of the second orthogonal design 45

FIGURE 5.8: Final orthogonal geometry 46

FIGURE 5.9: Fold distribution of the final orthogonal design 46

FIGURE 5.10: Near offsets achieved from the final orthogonal design 47

ix

FIGURE 5.11: Far offsets achieved from the final orthogonal design 47

FIGURE 5.12: 45 degree skewed design 50

FIGURE 5.13: Fold from the 45 degree skewed design 50

FIGURE 5.14: Near offset from the 45 degree skewed design 51

FIGURE 5.15: Far offset from the 45 degree skewed design 51

FIGURE 5.16: Azimuth distribution from the 45 degree skewed design 52

FIGURE 5.17: 26 degree skewed design 54

FIGURE 5.18: Fold distribution of the 26 degree skewed design 54

FIGURE 5.19: Near offset distribution of the 26 degree skewed design 55

FIGURE 5.20: Far offset of the 26 degree skewed design 55

FIGURE 5.21: Azimuth distribution of the 26 degree skewed design 56

FIGURE 5.22: Redesigned survey 59

FIGURE 5.23: Expected fold distribution from the redesigned survey 59

FIGURE 6.1: a) The relationship between differential pressure and velocity can
be approximated by a power law. b) Vp:Vs ratio vs. Effective
pressure. 63

FIGURE 6.2: a) Sonic velocity of pure water as a function of pressure and
temperature; b) density of water as a function of pressure,
temperature, and salinity; c) calculated brine modulus as a function
of pressure, temperature and salinity 65

FIGURE 7.1: MAR injection well in a confined aquifer 69

FIGURE 7.2: Geological model used to forward model the seismic response
before injection. 72

FIGURE 8.1: Average effective pressure vs. Velocity trends calculated from
Zimmer et al. (2007a&b) 79

FIGURE 8.2: Predicted P and S wave velocity change due to the injection. 81

FIGURE 8.3: Plot of Vp:Vs ratio vs. Effective pressure. 82

FIGURE 8.4: Zero offset synthetic traces for each pore pressure change 86



x

FIGURE 8.5: Zero offset sections showing the amplitude difference between
a) 50kpa pore pressure change and; b) 100kpa pore pressure change,
compared to the pre-injection scenario (i.e. t
2
- t
1
). 87

FIGURE 8.6: Zero offset section showing the amplitude difference between
500kpa pore pressure change compared to the pre-injection scenario. 88

FIGURE 8.7: Zero offset sections showing the amplitude difference between
a) 1Mpa pore pressure change; b) 1.5Mpa pore pressure change,
compared to the pre-injection scenario. 89
Perths Water Supply and MAR
1
Chapter 1

1 Introduction

1.1 Perths Water Supply and MAR

Climate modelling by the CSIRO predicts that rainfall will decline in the South West
of Western Australia by up to 20% by 2030 and 60% by 2070, compared to average
rainfall recorded in 1990 (State water plan, 2007). Reduced rainfall over the last 35
years has resulted in a 50% reduction in the amount of water which flows into public
drinking supplies. Decreased aquifer recharge has also occurred in some areas.

Water use has also tripled over the last 25 years, from less than 750GL in 1980 to
2 340 GL in Western Australia (State water plan, 2007) placing extra burden on
water resources. The increase in demand has mostly been compensated by using
greater amounts of ground water from the Perth Basin. The Perth basin is estimated
to contain a sustainable yield of 1 937 GL, the largest fresh groundwater availability
in the state (State water plan, 2007). In the year 2000 it was estimated that only
approximately 908 GL was used.

However groundwater is not an unlimited supply and with decreasing aquifer
recharge the amount of water extracted needs to be monitored. This is where
Managed Aquifer Recharge (MAR) helps. MAR is a method of adding recycled
water to underground aquifers in controlled conditions (CSIRO, n.d., Kresic, 2007).
The purpose of this method is to store excess water for later use, while improving the
quality of the ground water. As the treated water infiltrates the soil, natural
biological, chemical and physical processes remove pathogens, chemicals and
Perths Water Supply and MAR
2
nutrients from the water. The water is further filtered whilst sitting in the aquifer
(CSIRO, nd). The increased water storage makes more water available for use and
preserves water levels. The energy and costs involved in treating reclaimed water
using MAR is less than the desalinisation of seawater (CSIRO, n.d.).

The Government of Western Australia has set a target to recycle 20% of all
wastewater (State water plan, 2007 & Water for life report, 2006) by 2012.
Currently 13% of wastewater is recycled however the long term goal is to exceed
30%. Managed Aquifer Recharge has been identified by the Water Corporation as
an important water recycling method which can help recharge Perth aquifers and
reduce water wastage. Before a full scale MAR facility can be constructed the
method has to be tested on a smaller scale to demonstrate that it is safe. Over $31
million has been put towards such a trial at the Beenyup Wastewater Treatment
Plant, located in the northern Perth suburb of Craigie, over a three year period
commencing in 2009 (Water Corporation, 2006a)

The process of MAR is described in Figure 1.1. Water from the plant will be treated
via micro filtration, reverse osmosis, and advanced oxidisation to produce high
quality water for recycling (Turnbull, M (hon), 2007). The water will then be
injected into the Leederville aquifer at a depth of approximately 200m (Water
Corporation, 2006a and Water for life report, 2006). The injected water will provide
an additional 1.5 billion litres to Perth drinking supplies. Detailed monitoring will be
carried out to demonstrate that the approach is safe.

Figure 1.1: The MAR process to be trialled at Beenyup WWTP.
Treated water will be injected in the Leederville aquifer, where it will be naturally filtered, stored and eventually used as potable water (Australian Government National Water
Commission, n.d)
Proposed research
4
1.2 Proposed research

Aquifer storage potential at the Beenyup site will be assessed during the MAR trial
by analysing changes to the aquifer caused by water movement. Flow and reactive
transport models of the injection process will be created by the Water Corporation to
show how the aquifer reacts to the injection process. In order to predict the
movement of the water front it is necessary to have a very detailed knowledge of the
lithology and subtle structural features beneath the site. This cannot be achieved
through the extrapolation of borehole information, so geophysical methods are
required.

A 2D seismic survey and vertical seismic profile (VSP), using a cemented 3C
geophone string, were completed to assess the ability of the method to map the
shallow structures and lithologies. Upon the completion of this survey a high
resolution 3D seismic reflection survey will be performed to obtain a detailed three
dimensional stratigraphic and structural model. Completion of this survey is vital
because the site will ultimately be covered by buildings, after which there will be no
opportunities to perform any ground surveys.

A further objective of the research program at Beenyup is to model and if possible
measure changes in elastic properties due to the injection process. The aim of this is
to determine whether 4D seismic methods can be designed to detect pressure changes
within the aquifer, resulting from the injection. This research may potentially be
helpful for future ASR projects. To examine these changes 4D seismic surveys will
be carried out during the three year injection trial, and the recovery process. This
Proposed research
5
will extend the seismic method to its limit by using new ways of transmitting and
measuring seismic signals in a shallow subsurface environment.

The purpose of this project was to carefully design the high resolution 3D seismic
survey to map the fine stratigraphic variations and structures, and to help understand
the injection process. Designs were constructed to optimise fold, offset and azimuth
distributions whilst focussing the seismic energy within the target interval. Careful
consideration was also made for exclusion areas over the site which prevent the
placement of seismic equipment. Synthetic modelling was also carried out to assess
the applicability of using time lapse seismic methodology to monitor pressure
changes caused by the injection process.


1.3 History and Location of the project area

The Beenyup Waste Water Treatment Plant is located in the northern Perth suburb of
Craigie, Western Australia (~19km NW of Perth) and treats wastewater from the
northern suburbs to the foothills. Inorganic and organic substances are removed
from the water in the treatment process and the treated water (~34GL/yr) (N. Turner,
personal communication, June 6, 2007) is discharged into the ocean. The Project
area is a 400 by 400m area located to the north of the plant entrance and is where
MAR will be trialled. An injection bore (identified in blue on Figure 1.2) will be
used at the site to inject the treated water, and a series of monitoring bores (indicated
in red) spaced away from the injection will be used to monitor the process.
Understanding the hydrogeology around these bores is of highest importance to the
project. For that reason we carried out a 3D seismic survey.

Figure 1.2: Location of the project
Left: the Beenyup WWTP is located in Craigie on the corner of the Mitchell Freeway and Ocean Reef Road; Right: The Project area is a 400 by 400m area near the plant entrance.
The position of the monitoring and injection bores are indicated
Mitchell
Fwy
Ocean
Reef Rd
400m
400m
Beenyup
WWTP
SURVEY
AREA
Regional geology
7
Chapter 2

2 Perth Basin geological setting and geology

2.1 Regional Geology

The project area is located within the Perth Basin, on the Swan Coastal Plain. The
Perth Basin is between 34 and 23km wide, and bound to the east by the Gingin and
Darling Scarps. Regional seismic data indicates that the sediments are as thick as
12km in the Perth area (Playford et al. cited in Davidson, 1995) and separated from
the Yilgarn Craton by the Darling Fault. Figure 2.1 shows the main structural
components of the Perth Basin. The sediments have been gently folded to form the
Yanchep Syncline in the west, Swan Syncline in the east, and the Pinjar anticline
(Davidson, 1995). The project area is situated on the eastern limb of the Pinjar
Anticline.

The Perth Basin was formed during periods of rifting and sagging along the
continental margin when Australia separated from Gondwana in the Early
Cretaceous. Sedimentation occurred prior and after break up over the Late Jurassic
to Later Tertiary-Quaternary, with sediments ranging from carbonates, shoreline,
coastal dunes, alluvial, and colluvial deposits. The Leederville aquifer and overlying
sediments, known collectively as the superficial formations, is the main area of
interest for this project.

The Leederville Aquifer is a confined aquifer present below the entire Perth area
except near the Swan Estuary, where it has been eroded prior to the deposition of the
Kings Park Formation (Davidson, 1995). It is a multilayer groundwater flow system
consisting of discontinuous interbedded siltstones, sandstones and shales of the
Regional geology
8
Henley Member (Osborne Formation), and the Pinjar, Wanneroo and Mariginiup
Members of the Leederville Formation (see Figure 2.2 for the stratigraphic column).
The Leederville aquifer ranges in thickness from less than 50m across the Pinjar
Anticline, to approximately 550m in the Yanchep Syncline. Throughout most of the
basin the Kardinya Shale Member, of the Osborne Formation, and the South Perth
Shale confine the aquifer above and below.
Regional geology
9

.

Figure 2.1: Structural components of the Perth Basin (from Davidson, 1995)
The project area is located on the eastern flank of the Pinjar Anticline
Project
area
Regional geology
10



Figure 2.2: Mesozoic stratigraphy of the Perth Basin (from Davidson, 1995)
The units of interest for this project are the Warnbro and Coolyena Groups. The
Leederville aquifer (injection interval) is indicated in yellow.
Injection
interval
Stratigraphy
11
2.2 Stratigraphy

The stratigraphy of the Perth Basin has been summarised from Davidson (1995).
Knowledge of the expected sediment types at the site aids in interpreting the
borehole logs to identify the target interval, and in creating time lapse seismic
models of the injection process. The relationships between each of these units can be
observed in the stratigraphic columns presented in Figure 2.2 and Figure 2.3.

The South Perth Shale is identified as a confining bed within the Perth Basin. It was
deposited during the Early Cretaceous when continental to shallow marine
sedimentation dominated. Davidson (1995) describes the South Perth Shale as
predominantly shallow-marine in origin; consisting of thinly interbedded, grey to
black siltstone and shale; and with minor thin, sandy beds and local thin, calcareous
beds. In the northern area it was found that the South Perth Shale is mostly
uncemented to weakly cemented. It is at depths between 140 and 900m below the
surface, and has a maximum thickness of approximately 300m. The South Perth
Shale is overlain, with a conformable and transitional contact, by the Leederville
Formation (Mariginiup and Wanneroo Members).

The Mariginiup Member defines the base of the Leederville Formation. It has been
classically described as consisting of thinly interbedded and discontinuous grey to
black siltstones and shales with minor fine grained sandstone beds ( 1m thick). The
depositional environment is mainly marine and represents the transition between the
South Perth Shale and Wanneroo Member. In the northern area the Mariginiup
Member is slightly more sandy and coarse grained. Glauconite, mica, and pyrite or
calcareous cement is commonly found within this unit. The depth of the Mariginiup
Stratigraphy
12
Member ranges between 20 and 800m below the surface. It is conformably overlain
by the Wanneroo Member or the Pinjar Member.

The Wanneroo Member contains discontinuous, interbedded sandstones, siltstones
and shales originating from both marine and non marine environments. Sandstone
beds are commonly between 12 and 15m thick but can range from less than 10m to
more than 20m. The sandstone is weakly consolidated, predominantly coarse
grained, poorly sorted, angular to sub angular, and slightly silty. Siltstone and shale
beds are grey and micaceous, and are generally as thick as the sandstone beds. The
Wanneroo Member is commonly found at depths between 20 and 450m. It is
conformably overlain by the Pinjar Member and unconformably by the Osborne
Formation, Molecap Greensand, or superficial formations.

The Pinjar member is the uppermost member of the Leederville Formation. Like the
Wanneroo Member it consists of discontinuous interbedded sandstones, siltstones
and shales of both marine and non marine origin. Sandstone beds are approximately
3 to 6m thick, are weakly consolidated and grey in colour. Grains are fine to very
coarse in size, poorly sorted, sub angular to sub rounded, and commonly silty. The
siltstones and shales are dark grey to black in colour, typically micaceous, and thinly
laminated with fine grained sandstone. The Pinjar Member is found at a depth
between 10 to 360m, and unconformably overlain by the Osborne Formation,
Molecap Greensands, or superficial formations.

The Henley Sandstone is part of the Osborne formation and also part of the
Leederville aquifer. It contains both sandstone and siltstone. The sandstone is
weakly consolidated to friable, fine to coarse grained, very poorly sorted, greenish
Stratigraphy
13
brown in colour and glauconitic. Very coarse gravel sized particles with high
sphericity and roundness can also be found within this unit. The Osborne Formation
has been completely eroded over the Pinjar Anticline prior to the deposition of the
superficial formations. Its upper contact with the Kardinya Shale Member is abrupt,
but can be gradational.

The Kardinya Shale (Osborne Formation) is a relatively thick siltstone-shale unit
within the Osborne Formation. It has been identified as a regional confining layer
for the Leederville aquifer. The interbedded siltstones and shales are moderately to
tightly consolidated, and dark green to black in colour. The Kardinya Shale often
contains glauconite and thin interbeds of fine grained sandstone. It is overlain
conformably by the Mirrabooka Member and unconformably by the Molecap
Greensand.

The Mirrabooka Member (Osborne Formation) is the uppermost section of the
Osborne Formation. It consists of sandstone with thin beds of siltstone and shale.
Sandstone is weakly consolidated, fine to very coarse grained, very poorly sorted,
and rich in glauconite. Siltstone and shales are moderately consolidated, glauconitic
and contain spherical, gravel size quartz grains. Onshore the Mirrabooka Member
has a maximum thickness of approximately 160m. The Mirrabooka Member is
overlain either comformably by the Molecap Greensand or unconformably by the
superficial formations.

The Molecap Greensand is of shallow marine origin and can be found in the Swan
Syncline and the Wanneroo area. It consists of mainly unconsolidated fine to
medium grained sandstone which also contains amounts of clay, glauconite and silt.
Stratigraphy
14
The Molecap Greensand overlies the Mirrabooka member with a transitional and
conformable contact. It lies at a depth of 40 to 230m and has a maximum thickness
of approximately 80m. North of Perth, the Molecap Greensand is overlain
unconformably by the Poison Hill Greensand or superficial formations.

The Poison Hill Greensand consists of unconsolidated fine to very coarse grained
sand. It is very glauconitic and contains small amounts of clay and silt. The Poison
Hill Greensand has a maximum thickness of 90m in the Swan Syncline and may
possibly be channel infill. This unit is unconformably overlain by the superficial
formations.

The Ascot Formation is the base of the group of units deposited in the Late Tertiary
to Quaternary (superficial formations). It consists of hard to friable calcarenite
interbedded with thin, very poorly sorted sand containing shell fragments, glauconite,
and phosphatic nodules. This unit is found to conformably overlie the Osborne
Formation or the Molecap and Poison Hill Greensands. It was deposited as a
prograding shoreline and has a maximum thickness of 30m in the northern Perth
area. The Ascot Formation forms part of the Superficial Aquifer.

The Tamala Limestone is also part of the superficial formations and can be found
along the coastal strip of the Perth Region. It is a calcareous eolianite composed of
quartz sand, fine to medium grained shell fragments and minor clay lenses. The
Tamala Limestone has a maximum thickness of 110m, however its thickness varies
along the coast depending on the topography.
Stratigraphy
15









Figure 2.3: Cainozoic stratigraphy of the Perth Basin (from Davidson, 1995)
Whilst these units dont form part of the Leederville aquifer, knowledge of the
shallow geology helps in log interpretation and time lapse seismic model building
Local Geology
16
2.3 Local geology interpreted from borehole logs

Before a successful 3D survey can be designed and time lapse synthetic modelling
performed; the geological setting, and target interval elastic properties (Leederville
Aquifer) need to be evaluated. This was achieved by examining borehole logs from
Water Corporation boreholes AM27, Beenyup WWTP and Craigie Geothermal Bore.
The Beenyup borehole was drilled during April 2007 and is located within the survey
area. The Craigie hole is located approximately 1.2km to the south of the survey
area in the grounds of the Craigie Leisure Centre, and AM27 (Aquifer Monitoring
bore) is located 1km to the north of the survey area. Gamma and resistivity logs
were acquired in each hole, and supported with lithology logs. Core samples were
retrieved from various depths within the Leederville aquifer and overlying
formations, from the Beenyup hole.

Cross plots of resistivity vs. gamma and volume of shale logs, were created to
differentiate lithologies (i.e. sand and shales). The logs were correlated with
lithology logs and geological descriptions from Davidson (1995) to identify the
formations present beneath the project area. The interpreted Beenyup and Craigie
logs are presented in Figure 2.5 and Figure 2.6 and were used to identify the 3D
seismic reflection survey target. The depth and dip of units within the target interval
dictates source-receiver offset and line and station spacing for the survey.

Local Geology
17









Figure 2.4: Location of the Boreholes used to interpret local geology
The flat geology can be easily correlated between boreholes

Craigie
Borehole
AM27
Borehole
Beenyup
Borehole
Survey
area
~1.2 km
~1 km
Local Geology
18
The logging results correlate very well between boreholes, with distinctive units
easily traced between all three holes. The top of the South Perth Shale can be
identified at a depth of approximately 260 to 280 meters below the surface due to its
distinctive high gamma count and low resistivity. Above this, the Mariginiup
Member of the Leederville Formation, can be located at a depth of 220 to 245m by
the high gamma count of the siltstones and shales, and the spiky, high resistivities
due to cementation. The Wanneroo Member is located at depths between 120 and
270m due to its characteristic blocky appearance on the gamma log, caused by the
interbedded sandstone, siltstone and shale beds. There appears to be two main sand
intervals and one shaly interval within the Wanneroo Member, as identified on the
logs in Figure 2.5 & Figure 2.6.

Above the Wanneroo Member is the Pinjar Member, located at a depth between 100
and 120m. This unit can be recognised by the characteristic spiky nature of the
gamma log caused by the interbedding of thin, low radiation sandstones with higher
gamma radiation siltstones and shales. Overlying the Pinjar Member is the Osborne
Formation. This unit is difficult to separate into its distinctive units due to the highly
variable gamma log and ambiguous lithology logs. The sandy unit overlying the
Pinjar member appears to be the Henley Sandstone Member, because of its low
gamma radiation and relatively high resistivity caused by poorly sorted lithology.

The high gamma count recorded in the unit above the Henley Sandstone Member is
probably the Mirrabooka member which also has high electrical resistivity, due to the
presence of Glauconite within pore spaces, and poor sorting. The radioactive decay
of potassium in the Glauconite may also explain the high gamma count. The
Kardinya Shale Member does not appear to be present below the site because there is
Local Geology
19
no prominent decrease in resistivity, which would be caused by the presence of fine
clay particles.

The top 50 metres of the Craigie hole, and the top 70m of the Beenyup hole, were
cased off with steel casing. Gamma ray attenuation increases in this area and the
resistivity tool no longer works. As a result the top of each borehole is difficult to
interpret using the logs, so greater emphasis was put on using the lithological
descriptions to identify the formations. Above the Osborne Formation the lithology
has been described as a grey/green (and sometimes slightly orange) glauconitic
sandstone with minor silt and moderate consolidation. This description fits with that
of the Molecap Greensand described by Davidson (1995). A unit of this type can be
found at depths from 50m and overlies the Osborne Formation.

Above the Molecap Greensand the cavernous limestones of the Tamala Limestone
formation were encountered at depths as shallow as 20m at Beenyup and 6m at
Craigie. Above this is a poor to moderately sorted, yellow quartz superficial sand.
Resistivity (ohm-m) Resistivity (ohm-m)


.
Figure 2.5: Gamma and Resistivity log from the Beenyup borehole.
The sand units have been identified using a yellow colour, and shale units in
green. The Leederville aquifer is at depths between 100 and 250m



























Figure 2.6: Gamma and resistivity logs from the Craigie Borehole.
The Leederville aquifer is present at depths between 115 and 280m

Leederville
Formation
South Perth Shale
Gage Formation
Osborne Formation
Molecap Greensand?
Shale
Sand
Sand
Pinjar
Member
Wanneroo
Member
Mariginiup
Member
Superficial sand
Superficial sand
Tamala Limestone
Molecap Greensand?
Osborne Formation
Wanneroo
Member
Leederville
Formation
Pinjar
Member
Mariginiup
Member
South Perth Shale
Sand
Sand
Shale
Gamma Ray (API)
Gamma Ray (API)
Initial seismic measurements at the site
21
Chapter 3

3 Early seismic measurements at the site

A 2D seismic line and check shot survey were performed at the site at the beginning
of the year. Examining the results of these surveys helped interpret the local geology
(i.e. target depths and the formations that are expected to cause reflections) and
assign velocities for the time lapse modelling.
3.1 2D seismic

A 2D line of seismic data was acquired at the site approximately 4 months before the
acquisition of the 3D survey. This line was positioned, as displayed in Figure 3.1,
with an approximate E-W orientation. The geophone spacing was set at 2m and
shots were positioned 2m apart and halfway between geophones. The first shot was
located on the western side of the line and the last shot was located close to the
eastern end.

The data was processed by Dr Michael Sykes of the Curtin Geophysics Department,
using Landmarks Promax. The processed results are displayed in Figure 3.2 with
the Beenyup gamma log overlain. The tops of each major formation, and some units
within the Wanneroo Formation, can be identified clearly within the data. Major
boundaries between each of these units are marked by red dashed lines and can be
observed to correlate very well with the gamma log. Reflections within the
Wanneroo Member and Osborne Formation can also be observed to correspond to
spikes in the gamma log (indicated by the yellow arrows).

Initial seismic measurements at the site
22
The quality of the data is not ideal for hydraulic modelling because the reflections
appear to be discontinuous in parts (possibly caused by a smaller acoustic impedance
contrast rather than a hole in the aquifer) and lateral variations are not imaged. The
quality of reflections also begins to deteriorate past 220m, with reflections from the
Mariginiup Member and South Perth Shale difficult to identify. This may be due to
the offsets not being large enough, or due to the geology not having a significant
contrast in acoustic impedance. The 3D survey should be an improvement on these
results because it is capable of mapping lateral changes in seismic response, with
greater resolution than a series of 2D lines. This should allow the aerial extent of the
sedimentary units to be determined. The signal to noise ratio will also be improved
over the 2D due to a larger number of offsets and azimuths used.

Initial seismic measurements at the site
23











Figure 3.1: location of 2D line
The seismic line was positioned over the entire width of the site and located near the cemented 3C
geophone string used for the VSP




Figure 3.2: Processed 2D seismic section with the gamma log overlain
Most major reflections correspond to the top of units identified in the logs. Reflections deeper than 200m are difficult to identify
Superficial
Tamala Limestone
Pinjar Member
Wanneroo Member sst
Wanneroo Member sst
Wanneroo Member sltst
Osborne Formation
Molecap Greensand
Past geophysical methods
25
3.2 Check shot survey

A check shot survey, using the geophone string identified in Figure 3.1, was
performed after the 2D seismic data was acquired. A 10lb sledge hammer was used
as a seismic source and the seismic energy was recorded by a geophone string
consisting of 6 triaxial geophones and 10 standard geophones, cemented in the
Beenyup borehole. Geophones were placed at depths between 70 and 280m, with a
30m triaxial spacing and 10m geophone spacing. The data was processed by Dr
Michael Sykes and Dr Brett Harris.

First break picks were used to calculate the average velocities, which were converted
to interval velocities using the Dix equation. The average and interval velocities
calculated from the check shot survey are shown in Figure 3.3, where each point
represents the location of a geophone. The velocity beneath the site ranges between
1800m/s and 2700m/s, with the mean interval velocity being approximately 2200m/s.
Above 70m depth the velocity is not calculated very accurately because no phones
are above 70m. The velocities were used during the design phase of the 3D survey,
and also to constrain the time lapse model. The exact velocities were not used to
define each geological unit due to the coarse sampling of the velocity field.
Past geophysical methods
26








Average velocity Interval Velocity

Figure 3.3: Computed average and interval velocity beneath Beenyup from the check shot
survey. These values were used to constrain the 3D survey design and time lapse model creation
3D seismic survey design
27
Chapter 4

4 3D seismic survey design

Very good results were achieved with the 2D survey. The next phase in the project
was to design the 3D seismic survey. Survey planning was completed using GMGs
MESA Expert v10. An aerial photo was used to aid in positioning receiver lines and
shot points, allowing access constraints (due to topography, vegetation and existing
infrastructure) to be considered. Seismic resolution requirements, target depth,
survey time frame, and location of the proposed injection and monitoring wells were
also taken into account during planning. Two designs, an orthogonal and skewed,
were tested to compare fold, offset, and azimuth distributions. The best survey
geometry is a compromise between uniform fold, offset and azimuth distributions,
ease of acquisition and acquisition time.

4.1 Access constraints (exclusion zones)

The site has many inaccessible zones which prohibit or limit the placement of shots
and receivers. As a result the greatest obstacle in designing a successful survey at
Beenyup is the careful positioning of shots around these areas to minimise
irregularities in the CMP coverage. A weight drop mounted on a small bobcat was
used as a seismic source for the survey. The positioning of source lines has to allow
adequate access for such machinery.

As part of the preliminary survey planning process, exclusion zones were identified
and marked using a handheld GPS. These areas can be divided into four different
categories relating to dense vegetation, mounds/rubble, steep depressions/holes
3D seismic survey design
28
which prohibit the use of a bobcat, and cultural objects such as roads and buildings.
Whilst the number of exclusion zones is large, numerous obstacles, such as the dirt
mounds in the northwest of the area, can be moved by machinery if required.

Figure 4.1 shows the location of the exclusion zones within the survey area. The
orange zones represent obstacles such as sand mounds and limestone dumps. These
mounds can be very steep in places and are between 1 to 3m high. It was deemed
unsafe to use the bobcat in these areas. In addition, poor source coupling was
expected at these locations. Receivers could not be located in the limestone dump
due to its cavernous nature, which prevents coupling of the receivers to the ground.

The green zones in Figure 4.1 represent areas where the vegetation is too dense to
allow bobcat access without requiring significant clearing, leading to flora
destruction. This area has been designated as a bush for life location, prohibiting
any large scale clearing. As a result shots can not be positioned in areas of dense
vegetation due to the size and environmental impact of the machinery. However
receivers can still be easily placed in these zones. Areas where vegetation is
relatively sparse, and where small obstacles can be quickly and easily moved by
hand without costly impacts on the flora and fauna, were not marked as exclusion
zones.

The main access road to the treatment plant, external car park, and shed/power
station in the north east are the only cultural objects which restrict shot placement.
Both source and receivers are restricted within these areas due to potential asphalt
damage from the source, and the inability of receivers to penetrate the ground. If
required, holes can be drilled into the asphalt to allow receivers to couple with the
ground





Figure 4.1: location of exclusion zones
Exclusion zones prohibit the use of some seismic equipment due to thick vegetation, dirt and rubbish mounds.
Difficult for shots
No shots & receivers
No shots on mound
3D seismic survey design
30
4.2 Survey constraints

The availability of equipment and personnel, crew productivity, and the amount of
channels available all need to be factored into the survey planning process to allow
maximum efficiency within the survey timeframe.

The seismic crew for this project was planned to consist of three to five persons. A
maximum of two weeks were allocated for the survey and approximately 250 shots
per day were required to fit within the planned survey period. Consequently up to
3500 shots could be achieved within the timeframe, not allowing for down time
associated with major problems. A larger number of shots will result in increased
CMP fold and hence higher signal to noise ratio.

Currently the Curtin Department of Exploration Geophysics owns a Seistronix
distributed system with 72 Line Tap Units (LTUs). Each unit allows a maximum of
six geophones to be connected, resulting in a maximum of 432 active channels at one
time. The recording geophone line pattern (patch) is required to be designed to allow
the maximum amount of equipment to be used at one time, facilitating maximum
productivity/efficiency during recording. This means that each recording patch
should have at most 432 receivers. However in reality this will be slightly less
because some LTUs will be required as spares in case of equipment failure.

4.3 Survey configuration

The wastewater injection interval has been identified on logs and 2D seismic data to
be between 120 to 225m below the surface, within the Wanneroo Member of the
Leederville Formation. Most of the seismic energy should be focused within this
3D seismic survey design
31
interval so that subtle lithology and structural properties of the injection interval, and
confining layers, can be imaged adequately. The Mariginiup Member is expected to
be the basal seal of the injection interval. It is important to image this surface
accurately using 3D seismic so that its continuity can be assessed to help define
where the injected water will migrate.

Sand/shale sequences often have small normal incidence reflectivity but potentially
greater reflectivity at larger incidence due to the difference in Poissons ratio
between sands and shales (M. Urosevic, 2007, personal communication).
Consequently a desired maximum offset should be approximately two times greater
than target depth. As a result the largest source-receiver offsets have to be at least
225m. The top seal is located at a depth of 100m. This unit also is required to be
imaged accurately by the 3D survey, meaning that the largest minimum offset
(Xmin) should be at most 100 to 120m to adequately sample the shallowest horizon
of direct importance with the injection. If the Xmin is greater than this, the horizon
may be under sampled, resulting in single fold data or areas of no coverage.

4.3.1 Shot and receiver interval

Receiver spacing is selected to provide the required horizontal resolution and to
avoid spatial aliasing of source generated waves. Hydraulic modelling at this site
also requires a horizontal resolution of less than 15m in order to be accurate enough.
Modelling typically requires at least 6 samples per anomaly (rather than 4 as
commonly taken based on the Nyquist criterion), meaning that the bin size for this
survey should be no greater than 2.5m (i.e. 2x2m bins). However before a 2m bin is
used for the survey, the range of appropriate bin sizes need to be determined to
3D seismic survey design
32
ensure that events can be resolved laterally after migration. This depends on the
dominant frequency reflected from the zone of interest, and the interval velocity
above the target. The simple equation presented by Cordsen et al. (2000) can be used
to estimate the range of bin sizes:
int
4
dom
v
B
f
=


where the value of 4 indicates that lateral resolution is one quarter of the dominant
wavelength. If the dominant frequency of the wavelet is 70hz (Figure 4.2) and the
interval velocity above the target is 2100m/s (obtained from the check shot), then the
bin size should be no greater than 7.5m. A 2x2m bin will thus be able to adequately
resolve lateral events.

Spatial aliasing of dipping events also needs to be considered. If the bins are not
small enough then some frequencies may be aliased and the true dip of the event will
not be recorded. The bin size which ensures that the maximum expected frequency
at the target is not aliased can be calculated by the equation presented by Cordsen et
al. (2000):
max
4 sin( )
=
v
B
f

If the interval velocity just above the injection interval is 2100m/s, the maximum
geological dip is 5 degrees (based on the comparison of the Beenyup, Craigie and
AM27 logs), and the maximum expected frequency is 150hz at the target, then the
bin size should be at most 40m to ensure that the maximum frequency at the target is
not aliased with reflector dip. This result shows that spatial aliasing will not be a
problem at the site because the geology is almost flat. The lateral resolution
requirement is much smaller than the anti-aliasing requirement hence a 2m bin size
and 4m shot and receiver spacing is adequate for the survey.
3D seismic survey design
33







Figure 4.2: Frequency spectrum of 10 traces (31-40) from the 2D line
The dominant frequency is approximately 70Hz and the maximum is at most 200hz

3D seismic survey design
34

4.3.2 Line position and spacing

Normally receiver lines in a 2D survey are orientated so that they are in the direction
of maximum dip to minimise the presence of out-of-plane events. However this is
not a problem in a 3D land survey as the target is illuminated from all angles.
Consequently the positioning of source and receiver lines for the Beenyup survey is
dictated by accessibility rather than illumination requirements with respect to the
target complexity. Receiver lines with east-west and north-south orientations were
both considered however, as the majority of the exclusion areas are orientated north-
south, it was decided that receiver lines should be positioned east-west to allow
source lines to be more continuous when using an orthogonal geometry. Receiver
lines can easily be placed perpendicular to his direction because their placement does
not require easy access.

The source and receiver lines have to be spaced close enough to prevent spatial
aliasing. Using the same equation and parameters as in section 4.3.1 the maximum
line spacing before spatial aliasing occurs is approximately 40m. However as 2m
bins are to be used, the line spacing should be much smaller than this to ensure that
there is coverage between lines. The area of interest is considered to include the
whole site, with particular emphasis close to the injection and monitoring wells. In
order to have CMP coverage over this large area, whilst keeping the total number of
shots down, the line spacing should be 20-30m.



3D seismic survey design
35
4.3.3 Migration aperture

During the migration process reflections are moved up dip to their true subsurface
location. As a result coverage may not be as large as expected after migration. The
amount in which the imaged area decreases in size after migration can be calculated
using the equation from Evans (1997):
tan( ) = A Z
If the depth of the deepest reflector within the injection interval is less than 225m,
and the maximum geological dip at the site is 5 degrees, then the maximum
migration aperture is approximately 20m. This result means that the survey should
have at least an extra 20m of coverage on all sides of the survey area to account for
the migrated reflections moving up dip by a maximum of 20m.

3D survey designs
36
Chapter 5

5 Survey Geometries

Two of the most commonly used geometries for 3D land seismic are orthogonal and
skewed designs. Both were tested for the Beenyup survey to determine which would
be most ideal. The advantage of using these geometries, where the source and
receiver line are not parallel, is that many different offsets and azimuths can be
recorded for each CMP. However both configurations have strengths and
weaknesses.

An orthogonal survey geometry has source and receiver lines perpendicular to one
another. This geometry is the easiest for the crew to acquire data because lay out in
the field is easy, extra equipment can be laid ahead of shooting, and roll-along
operations are possible. Keeping track of station numbering is also simple (Corsden
et al., 2000). Offset distribution is also good with an orthogonal geometry but
deteriorates towards the center of the boxes where shorter offsets are absent, and the
azimuth distribution is uniform, as long as a wide recording patch is used. The major
disadvantage using this geometry is that offsets in parallel rows of bins show a
tendency toward distinct patterning for offset limited stacks, which may result in
severe acquisition footprints at shallow levels (Corsden et al., 2000).

A skewed geometry has source lines positioned obliquely to the receiver lines. This
geometry attempts to make the offset distribution more random in nature, compared
to an orthogonal geometry. The offset distribution is well dispersed with few
duplicated offsets, but the position of the largest minimum offset (Xmin) changes
with line angle and interval. Azimuth distribution is also very good but depends on
3D survey designs
37
the number of receiver lines in the patch (Corsden et al., 2000). For this survey two
angles between source and receiver lines were tested: 45 and 26 degrees. Typically
the 26 degree skew should provide a better offset and azimuth distribution, hence
less acquisition footprint.

Simple designs for each geometry were first created to determine the best source and
receiver line positions. These designs were then refined to improve fold, offset and
azimuth distributions to a level which will produce acceptable results. The
commonly used template approach was used to shoot the survey. This shooting
method involves laying out a patch of several receiver lines and firing shots into
them. The receiver patch is then moved to a new template position and the process is
repeated. This shooting method allows for maximum efficiency because all receiver
equipment can be used at one time. The receiver lines were positioned in an east-
west direction.

5.1 Orthogonal geometry

Receivers were positioned East-West and a North-South shot line orientation was
used for the orthogonal design. A bin size of 2m was used with 4m station spacing.
Twelve receiver lines, each with ninety six receivers, spaced 30m apart, were used
for the initial designs. This allowed four receiver lines, each with 16 LTUs, to form
each patch. Ten LTUs were set aside as spares in case of equipment failure. A
receiver line spacing of 20m was initially selected however this resulted in too many
templates, so an interval of 30m was used to reduce the amount of patch moves.

3D survey designs
38
Shot lines were initially positioned as a grid, with the same line interval as the
receivers, and then manipulated so that shot points were not located in exclusion
zones. Nineteen source lines, each with 90 shot points spaced 4m apart, were used.
This survey geometry can be viewed in Figure 5.1. Receiver lines are shown in blue
and source lines in red. Source lines were not manipulated too much through the dirt
and rubbish mounds in the northwest because these can be moved by machinery if
required. Shooting into each template was made simple to reduce errors during
acquisition. Shots within, and half a receiver line interval above and below the
active patch were shot. Once all shots in the template were fired the patch was rolled
to the north by advancing the two trailing lines. Figure 5.2 shows the first template.
A total of 1890 shots were fired in this design.

Whilst the number of shots is well below the maximum allowed, the results do show
a slight acquisition footprint, identified as fold striping (Figure 5.3). The azimuths
are also limited to an east-west direction due to the small size of the template. Fold
distribution is also not ideal, with fold less than 30 around the injection and
monitoring wells. This can be increased by increasing the bin size to 4m; however
this may affect spatial resolution.
3D survey designs
39

Figure 5.1: Preliminary orthogonal geometry
Fourteen source lines and 12 receiver lines are positioned 30m apart with a 4m station spacing.



Figure 5.2: First template (receiver lines 1-4 and associated shot points)
Only shots near the receiver patch were fired. The two trailing receiver lines are rolled to the north for
the next template.
3D survey designs
40













Figure 5.3: Fold distribution from the preliminary orthogonal design
Fold ( 8) near the wells (red=monitoring, green=injection) is not high enough.
3D survey designs
41
To increase fold around the injection and monitoring wells two modifications can be
made: the number of receivers can be increased in the western half of the site or the
number of shots can be increased. It would take longer to set out and move a greater
number of geophones compared to shooting extra shots, which requires no extra
equipment, so the number of shots was increased instead. This was done in two
different ways: increasing the number of source lines by halving the source line
interval, and increasing the shot density by decreasing the shot interval. Both
methods were tested and whilst increasing the number of lines does improve the near
offsets, more earth works is required for equipment access. Increasing the shot point
density would take less time to acquire because earth works are reduced.

The shot point interval on lines 1 to 7 was decreased to 2m to increase the fold. The
results of this modification are displayed in Figure 5.4 to Figure 5.7. Whilst fold
near the wells has increased to 30, compared to ~14 in the original design, fold at
the centre of the site is much larger than required (~60 to 80). This indicates that
some shots in the east may be removed. The largest minimum offset in this design is
typically less than 44m and distributed more evenly than the previous design due to
some careful shot point repositioning around exclusion zones.

The maximum offsets are greater than the depth of the target interval base (as
required) however, like the previous design, azimuths are limited to an inline
direction because of the narrow, rectangular template used. To increase the signal to
noise ratio the azimuth and offset distributions are required to be more uniform (i.e.
include many offset and azimuths), as the random noise is more effectively
attenuated. This was achieved by making the templates larger in size by adding more
3D survey designs
42
sources. Adding more source positions for each template is possible because the
total amount of fired shots in this design is only 2917.

The templates were enlarged by shooting over the entire source line length. To keep
the number of shots below 3500, shot positions were alternated between each patch
to reduce duplicate ray paths. This was achieved by shooting every second shot into
each patch. Source and receivers were also removed from the eastern side of the
survey to help reduce the total shot number. This modification is permitted because
the area is not as most important for modeling, as it is further away from the wells.
Maximum offsets are also not affected because they are already sufficiently large.

Reducing the number of receivers per line from 96 to 60 also allows the number of
receiver lines in the patch to increase to six. A larger patch results in fewer
templates, with only four (each overlapping by three receiver lines) required to
complete the survey. Essentially every shot position is used twice because there are
four templates. Reducing the number of templates helps reduce survey time because
not as many moves are required. The shot point interval on lines 5 to 7 was also
changed back to 2m to further reduce the total amount of shots. The total number of
shots used for this design is 2503, which is less than the previous design. This would
take approximately 10 days to acquire.

The results of this design can be seen in Figure 5.8 to Figure 5.11. Whilst the fold
does not appear to be distributed as well as the previous design, it does appear to
have less striping and fold has increased near the wells by approximately 20 due to
the templates being larger in size (i.e. more shot points). Most of the wells are
covered by at least 30 fold. The near offsets are comparable to the previous design,
3D survey designs
43
and less than 50m. Only near offsets to the north east and south east are greater, due
to a lack of CMP coverage (caused by no patch overlap on the edges). These larger
offsets may mean that the shallower horizons may not be sampled as well as in the
centre of the site; however these areas are far from the injection wells. Far offsets
are also close to 300m over most of the area.
3D survey designs
44

Figure 5.4: Modified orthogonal design with 2m shot intervals to the west, and 4m to the east
The number of shots was increased in the west to attempt to increase fold near the wells






Figure 5.5: Fold distribution resulting from decreasing the shot point interval in the west
Fold has increased to 30 near the wells (red=monitoring, green=injection)
2m shot
point
interval
3D survey designs
45

Figure 5.6: Near offset distribution of the second orthogonal design
The largest minimum offset is approximately 44m, which will allow shallow horizons to be sampled
adequately






Figure 5.7: Azimuth distribution of the second orthogonal design
The azimuths are limited to an east-west direction because of the narrow template used
3D survey designs
46

Figure 5.8: Final orthogonal geometry
Modifications have resulted in less shot points and template moves





Figure 5.9: Fold distribution of the final orthogonal design
Fold is at least 30 near most of the wells.
3D survey designs
47

Figure 5.10: Near offsets achieved from the final orthogonal design
Near offsets are typically less than 46m so shallow horizons should be sampled adequately






Figure 5.11: Far offsets achieved from the final orthogonal design
The South Perth Shale should be imaged accurately because far offsets are at least 300m near the
wells
3D survey designs
48

5.2 Skewed geometry

The skewed survey designs were created in a similar manner to the final orthogonal
design described in section 5.1. The same receiver locations and station spacing was
used. Source lines were positioned 30m apart and striking at 45 and 26 degrees.
Two different source line orientations were tested to see if either has an advantage.
An extra source line along the western side of the road was also included to improve
fold and offsets near the wells.

5.2.1 Results of the 45 degree skew

The 45 degree skewed design is shown in Figure 5.12. It includes 13 source lines
striking at an angle of 45 degrees, and one striking north. The shot point interval on
the western half of the site is 2m, whilst the eastern side is 4m. This is to increase
fold near the wells, as discussed in section 5.1. Most of the shot lines are straight
however some earth works is required in the centre of the site to create a path for the
seismic source. Other source lines have been manipulated to avoid exclusion areas
which cant be easily cleared.

This design has a total of 1966 shots, which is estimated to take 8 days to complete.
The fold distribution shown in Figure 5.13 is very good, with fold greater or equal to
30 near all wells. The maximum fold in the centre of the site is also only 125, which
is better than the orthogonal design with 153 fold. The fold variation between bins is
also very smooth, which will help reduce the acquisition footprint. The near offsets
displayed in Figure 5.14 are mostly less than 48m meaning that the shallow horizons
should be sampled accurately. Only a few of the larger exclusion zones and
locations covered by only one template have minimum offsets between 60 and 80m.
3D survey designs
49
The far offsets are very good with the largest offset close to 300 over most of the
area. The azimuth distribution is good because there are numerous angles.

3D survey designs
50

Figure 5.12: 45 degree skewed design
Shot point spacing is 2m in the East and 4m in the west



Figure 5.13: Fold from the 45 degree skewed design
Fold is at least 30 near all wells
2m shot point
interval
3D survey designs
51

Figure 5.14: Near offset from the 45 degree skewed design
Most near offsets are small enough to ensure that the shallow horizons are properly sampled, except
near the larger exclusion areas






Figure 5.15: Far offset from the 45 degree skewed design
Far offsets are larger than the deepest reflector
3D survey designs
52













Figure 5.16: Azimuth distribution from the 45 degree skewed design
Azimuths are distributed more uniformly, which will improve S:N
3D survey designs
53

5.2.2 Results of the 26 degree skew

The 26 degree skew is very similar to the 45 degree, just with source lines striking at
a different angle. Thirteen source lines are used in this design and a total of 2178
shots were used to shoot the survey. Acquisition time is estimated to be 9 days. The
fold distribution shown in Figure 5.18 is comparable to that produced by the 45
degree skewed design but slightly better. The maximum fold in the centre of the site
is 130 and reflects the larger amount of shots required to cover the survey area. Near
offsets are adequate, with most less than 40m, and slightly better than the 45 degree
skew because more CMPs have offsets of 30m or less. The far offsets appear very
similar to the previous design, with bins having offsets greater than 300m. Azimuth
distribution is better than the 45 degree design because they are distributed more
uniformly.
3D survey designs
54

Figure 5.17: 26 degree skewed design
This design is similar to the 45 degree but uses a different source line azimuth





Figure 5.18: Fold distribution of the 26 degree skewed design
Fold is at least 30 near all wells
2m shot
point
interval
3D survey designs
55

Figure 5.19: Near offset distribution of the 26 degree skewed design
Most offsets are less than 30m






Figure 5.20: Far offset of the 26 degree skewed design
All offsets are greater than the target depth
3D survey designs
56












Figure 5.21: Azimuth distribution of the 26 degree skewed design
Azimuth distribution is more uniform than the 45 degree design
3D survey designs
57

5.3 Final Survey design

A comparison between each survey design is presented below in Table 5.1. All are
very close statistically. Fold near the wells is very similar in all designs, greater than
30, and the offset and azimuth distributions all satisfy the constraints presented in
section 4.3. All three designs should also be able to be completed in less than the 14
day time period. The best design for the Beenyup project would be the 26 degree
skewed design because it shows a compromise between acquisition time, fold and
azimuth requirements. Fold is not very different to the other designs, however Xmin
is smaller, and far offsets are generally greater. There is more earth moving required
before shooting can commence, compared to the orthogonal design, but this can be
achieved before the survey commences thus not impacting on acquisition time.
Whilst a skewed geometry may appear to be harder to lay out and shoot, than an
orthogonal geometry, the lines are kept relatively straight, so shooting should be
straightforward for the crew.

Table 5.1: Comparison between each survey design.
Based on the number of shots, fold, offset, azimuth, and amount of earth works the 26 degree skew is
the best design

Orthogonal 45 Skew 26 Skew
Total shots 2503 (10 days) 1966 (8 days) 2178 (9 days)
Min fold near
wells
40 30 35
Footprint
(fold pattern)
Similar to the 26 but
slightly more
pronounced
Not good good
Xmin 45m 47m 37m
Min far offset
near wells
229m 234m 235m
Azimuth
distribution
best worst good
Est. amt of earth
works
Less than 303m 421m 370m
3D survey designs
58
However site conditions changed and the 26 degree skewed design could not be used
for the 3D survey because large piles of sand were dumped over the site just before
acquisition was to commence. Due to the central area of the site being inaccessible it
was more feasible to focus the survey around the injection wells where knowledge of
the hydrogeology is more important. The area near the location of the proposed
injection wells had already been cleared to allow machinery to dump the sand. This
means that the exclusion zones presented in Figure 4.1 are no longer present, and the
placement of source and receiver lines is less restricted. The smaller target area also
allows a closer line spacing to be used, for the same number of shots as the previous
designs. Together with Dr Aleksander Dzunic I redesigned the survey using an
orthogonal geometry, just a day before the commencement of the survey.

Source lines are positioned in an East-west direction so that shot points can also be
positioned on the western side of the road. A total of 1254 shots were used to
acquire the data, using 20 source lines and 8 receiver lines. Fold near the wells is
between 35 & 40 and the near offsets are less than 50m, however increases
significantly in the east due to undershooting the sand mound. Far offset are at least
240m over most of the site. Azimuths are limited to northwest-southeast and north-
south directions near the wells, however this was the best distribution that could be
achieved with the current site conditions.
3D survey designs
59

Figure 5.22: Redesigned survey
The site conditions changed substantially so the previous designs were no longer the best for the
conditions




Figure 5.23: Expected fold distribution from the redesigned survey
Fold is at least 35 to 40 near the wells
Time lapse seismic theory
60
Chapter 6

6 Time lapse seismic theory

Time lapse seismic is a method where the difference in seismic response between
repeated surveys is used to image changes in elastic properties caused by, for
example, fluid movement and/or pressure changes in the subsurface. Many factors
affect time lapse seismic responses: (i) repeatability errors (positioning, ground
saturation, availability of the same recording equipment), (ii) data processing and
sophistication of algorithms for data differencing, and (iii) the actual changes in
elastic properties caused by fluid injection or extraction from a reservoir rock.
During the injection process at Beenyup, the effective elastic properties of the aquifer
will change as the original pore water is displaced by the injected water. Models of
the MAR injection into the Leederville aquifer were created to examine whether time
lapse seismic can detect these subtle changes.

Before examining the effects injection has on the seismic response it is necessary to
understand which properties change as a consequence of the water injection, and
their effect on elastic properties of the rock. This analysis is required because it is
the change in elastic properties which will affect the propagation of seismic waves.
Major parameters expected to change as a result of the injection are: the formation
pressure, and density (as a result of the fluid injection and potential opening of
microcracks).





Time lapse seismic theory
61
6.1 Pressure effects

Lithostatic stress within an aquifer, caused by the weight of the overlying water and
sediment, is supported by the aquifer skeleton. Assuming that for the shallow
sediments horizontal stresses are equal and small, the lithostatic pressure can be used
to approximate the confining pressure i.e.
( )
1
3
c H h
L = + +
where L is the lithostatic or vertical stress,
H
is the maximum horizontal stress, and
h
is the minimum horizontal stress. An upward stress is also applied by the
pressure of fluid within the pore spaces, and that provided from the rock frame
(Fetter, 1994). This relationship can be expressed as:
C e f
P = +
where
C
is the total stress,
e
is the effective stress, is an effective stress
coefficient whose value is assumed to be one in unconsolidated sediments (Terzaghi,
1943 cited in Zimmer et al., 2007) and
f
P is the pressure from the pore fluid. It is the
effective stress which affects seismic and acoustic wave velocities the most (Sheriff,
1992, Dodds et al., 2007).

In confined aquifers there can be significant changes in pressure with very little
changes in the thickness of the water column, because the overburden is supported
by the sediment (effective stress) and hydraulic pressure from the pore water (Fetter,
1994, Kasenow, 2001). This means that the total stress can be considered constant.
If pore pressure is increased then the effective stress will reduce (i.e.
e
P = ),
because pore spaces will dilate, causing unloading of the rock.
Time lapse seismic theory
62
The displacement of ground water by the injected water will cause an increase in
pore pressure (thus a decrease in effective pressure) and decrease in the effective
rock density around the injection well, altering the elastic properties of the rock.
Furre (2002) (cited in Hofmann et al., 2005) used the spherical packing theory to
show that as differential pressure increases, the contact area between spherical grains
flattens and the influence of pore pressure on velocity decreases. As a result rock
stiffness and rigidity is increased. The opposite is true when differential pressure
decreases, the stiffness of the rock will reduce due to a reduction in grain contact.
This relationship can be viewed in Figure 6.1 which shows results of work done by
Zimmer et al. (2007a & b) on unconsolidated sediments (with porosities between 33
and 42%, fine/medium grained, and an average of 65% quartz, 12% plagioclase, 8%
K feldspar, 1.4% amphibole, and 4.4% clay) with pressures between 0.1 to 20MPa.

The pressure trends of both dry and water saturated shear wave velocities are
essentially the same and can be approximated by a power law relationship
proportional to
1 4
e
. Dry compressional wave velocities are approximated by a
power law with a shallower slope, between
1 4
e
and
1 5
e
(Zimmer et al., 2007a). The
pressure trend for water saturated compressional velocities can be described by a
power law plus a constant. It is also apparent from these results that effective
pressure change at lower pressures, which are in the same range as those at Beenyup,
causes a more pronounced change in velocity (Figure 6.1a). Experiments by
Huffman and Castagna (2001), and Zimmer et al. (2007b) show that the
s p
V V
ratio is
also most sensitive to effective pressure at pressures below 2MPa (Figure 6.1b).
Time lapse seismic theory
63









Figure 6.1: a) The relationship between differential pressure and velocity can be approximated
by a power law. With decreasing effective pressure Vp and Vs both decrease (reproduced from
Zimmer et al., 2007b); b) Vp:Vs ratio vs. effective pressure. The change in ratio is more pronounced
at lower pressures (from Zimmer et al., 2007b)
Time lapse seismic theory
64

6.2 Fluid effects

Water within the formation also affects the bulk elastic properties of the rock. Pore
fluids form a dynamic system where the bulk modulus and density (and thus seismic
velocities) change with pressure, temperature and salinity. The relationship between
sonic velocity, density, and bulk modulus with pressure, temperature and salinity of
water was examined by Batzle & Wang (1992). Their results are displayed in Figure
6.2.

Increasing the temperature of water generally causes the bulk modulus and density to
decrease. However up to a temperature of 100 degrees Celsius, the bulk modulus
increases slightly, whilst density still decreases, leading to an increase in velocity
below 100 degrees. After the temperature reaches 100 degrees Celsius the velocity
decreases due to a reduction in bulk modulus caused by its gaseous state. The
temperature of the pore water within the Leederville aquifer is expected to be much
lower than 100 degrees, so large velocitytemperature changes are not expected.
The injected water is also expected to reach the same temperature as the groundwater
very quickly so no significant temperature change is expected.

When salinity is increased the bulk modulus and density increase due to the extra
dissolved ions within the fluid. During the MAR trial the in situ ground water will
be displaced by the injected, treated water. The injected water is expected to have a
lower salinity than the groundwater, leading to a density change in the pore fluid. A
reduction in salinity will increase the bulk P and S wave velocities of the saturated
rock, however the decrease in grain contain will have a larger and opposite effect,
causing the velocities to decrease overall.
Time lapse seismic theory
65





Figure 6.2: a) Sonic velocity of pure water as a function of pressure and temperature (from
Helgeson and Kirkham, 1974); b) Density of water as a function of pressure, temperature, and
salinity (from Batzle and Wang, 1992); c) Calculated brine modulus as a function of pressure,
temperature and salinity (from Batzle and Wang, 1992)
b.
c.
a.
Time lapse seismic theory
66
Gassmanns equation, which is only valid for low frequencies, i.e. <100hz (Mavko et
al. cited in Han & Batzle, 2004), was used to calculate the rock bulk and shear
modulus after injection (Zou et al., n.d., Li et al., 2006, Wang, 2000). The equations
describing the bulk and shear moduli of saturated rocks are:
2
2
1
1
grain
dry
grain fluid
grain
dry
sat
K
K
K K
K
K
K

=


dry sat
=
Where
sat
K and
sat
are the bulk and shear moduli for the saturated rock,
dry
K and
dry
are the bulk and shear moduli for the dry frame,
grain
K is the bulk modulus for
the mineral grains,
fluid
K is the bulk modulus of the fluid, and is the porosity. The
shear modulus is not affected by fluid saturation hence the saturated shear modulus
equals the dry modulus.

The frame bulk and shear moduli are calculated from the equations:

=
2 2
3
4
s p dry dry
V V K
2
s dry dry
V =
Similarly P and S wave velocities of the saturated rock can be computed using:


sat sat
p
K
V
3
4
+
=

sat
s
V =

These equations were used to calculate the seismic velocities of the sediments within
the aquifer after injection. The Fluid Replacement Modelling module of Hampson
Russell software was used to do this.

Time lapse seismic theory
67
The change in fluid properties are not expected to be the significant factor affecting
the elastic properties of the aquifer because the change in density and salinity is very
small. Instead the change in pore pressure, due to injection, is expected to be the
dominant process. Both effects were considered during the forward modelling.
Synthetic model creation
68
Chapter 7

7 Synthetic model creation

A synthetic model of the local geology was created to simulate the changes in
seismic response caused by the injection process. This was achieved using Hampson
Russell to generate zero offset models prior, during and after injection. Geological
interfaces were constructed using the boundaries identified on the gamma and
resistivity logs, and 2D seismic data. Densities were computed from the results of
core analysis, on some units within the Leederville Aquifer, by ACS Laboratories in
Bassendean, Western Australia. P wave velocities were estimated using the lithology
and check shot data as a constraint, and S waves velocities were estimated using
Castagnas equation.

Figure 7.1 shows the basic principle of MAR/ASR. The injected, higher quality
water is pumped in with large enough volumes so that the pre existing groundwater
is displaced laterally. Confinement has to be tight enough to prevent rapid vertical
migration of the less dense recharged water (Rosenshein and Hickey, 1977 cited in
Kresic, 2007). This basic principle, that ground water is displaced entirely by the
injected water, was used to create models of the injection. The elastic parameters
within the Leederville Aquifer were changed to represent the changing pressures and
fluid densities associated with the process.
Synthetic model creation
69









Figure 7.1: MAR injection well in a confined aquifer
The idealised flushed and mixing zones created by recharge will have different elastic properties to
the zones with native ground water (from Reece, 2002 in Kresic, 2007)
Synthetic model creation
70
The properties required for modelling include:

P & S wave velocities before injection
Temperature of the aquifer
Salinity of the groundwater and injected water
Formation pressure before and after injection
Density of the groundwater and injected water (which are a function of
temperature, salinity and pressure)
Density of the dry rock and in situ rock (which includes the original pore
water).
P & S wave velocities after injection

7.1 Interval velocities prior to injection

As there were no sonic logs acquired down any boreholes near Beenyup, the main
source of velocity information was from the check shot survey presented in section
3.2. These interval velocities were used as a guide to assign velocities to each
sedimentary unit.

Initially the resistivity log was going to be used to estimate acoustic velocities via the
Faust Transform (which relates P-wave velocity, formation resistivity factor, and
depth) however Hacikoylu et al. (2006a & b), have found that the empirical
relationship is applicable only to consolidated cemented sandstones with low clay
content and porosity between 5 and 20%. It should not be used in shale or
unconsolidated and/or uncemented rock. As most of the sediments at the site are
unconsolidated this relationship will not be very accurate. Errors arising from
variations in water salinity, and cementation would also affect the accuracy of the
Synthetic model creation
71
velocity estimate. Sarasty & Stewart (n.d), and K. Gerlitz (personal communication,
September 14, 2007) also pointed out that the relationship only predicts velocities
accurately if the best set of coefficients are determined through comparison with a
sonic log. The VSP data is too sparsely sampled for an accurate comparison. Thus it
was decided not to use the Faust transform.

Instead the range of typical velocities provided in a database within Tesseral
Technologys Tesseral 2D, and those presented by Berkman (2001), were used as a
guide. The lithology log was used to help fine tune the velocities by taking
mineralogy in account, i.e. how much clay content there is, and how consolidated the
sediment is. S wave velocities were estimated using Castagnas equation.

The synthetic model is displayed in Figure 7.2. Velocities range between 600m/s for
the overburden to 2300m/s for the shales. The top and bottom of each major unit is
identified by the bold black lines. Within the Pinjar member a sandy layer, identified
on the lithology and gamma logs, is also included. This is indicated by the thin, pale
yellow layer. Silty/clay layers within the Wanneroo sands are also included. A
velocity gradient was used for the shallow sediments, down to the Tamala
Limestone. This is because there was no reflection associated with the top of the
Tamala Limestone, and the lithology log indicated that calcium carbonate content
increased gradually with depth.

Figure 7.2: Geological model used to forward model the seismic response before injection.
The major sedimentary units above and within the injection interval were modelled using single Vp, Vs and values
Synthetic model creation
73
7.2 Groundwater temperature, salinity and pressure

According to Davidson (1995) water within the Leederville aquifer near Beenyup has
an average salinity of 450 mg/L, and is at a temperature of 26 degrees Celsius.
Groundwater would be more saline within the Henley Sandstone, Pinjar, and
Mariginiup Members because of salt, locked up during deposition, being released,
however an average of 450mg/L will be appropriate for the modelling. Formation
pressure within the aquifer was estimated by calculating the pressure at a depth of
165m (approximately half way through the aquifer) due to the overlying water from
the equation:
gh P =
Where is the density of water (0.997g/cc for water at 20C, atmospheric pressure,
450mg/L salinity), g is the gravitational acceleration (9.8m/s
2
), and h is the depth
below the water table (165m). In reality a confined aquifer may have a higher
pressure than the hydrostatic pressure but this is a reasonable first pass value. The
estimated formation pressure (pore pressure) before injection is 1.6MPa. The
confining pressure is estimated by calculating the weight of the overlying sediment
and water using the equation:
gh gh P
s w c
) 1 ( + =
Where is porosity (an average of 35%),
w
is the density of the water, and
S
is the
density of the overlying sediment (estimated to be 2.65g/cc). This is estimated to be
3.5MPa.



Synthetic model creation
74
7.3 Injected water temperature, salinity and pressure

The treated water is expected to have a salinity of 12mg/L, and would quickly reach
the formation temperature once injected. If a large enough volume of water is
injected into the aquifer then mixing of this water will be minimal and pore pressure
can be expected to increase due to the filling of pore spaces which are already close
to full. The injected water is expected to cause an increase of 50 to 100 kPa over the
injection interval near the well.

7.4 Formation density

The bulk density of the formation includes the density of the matrix and pore fluids
combined. As there was no density log from any boreholes near Beenyup, the
density of each formation had to be estimated using a combination of lab results and
expected densities based on lithology. Fluid density was determined using the Fluid
Substitution Modelling module in Hampson Russell, for a temperature of 26C,
salinity of 450mg/L, and pressure of 1.6MPa. The grain densities of sand units
within the Leederville Aquifer, calculated from core analysis on a selection of
samples, were used as a benchmark in assigning values to other lithologies.
Densities to all units were then estimated based on the amount of clay and cement
present, degree of consolidation, and dominant mineral etc. The database of typical
densities for various sediments, from Tesseral, and Berkman (2001) were also used
to assist.

A total of five core samples were analysed by ACS Laboratories. The first sample
(SST 1) was extracted from within the Osborne Formation, SLTST 4 from the top of
the top Wanneroo Sandstone, SLST 5 and SST 7 from the bottom of the top
Synthetic model creation
75
Wanneroo Sandstone, and SST11 from the second Wanneroo Sandstone. ACS
Laboratories performed tests on 1" diameter plug samples cut from horizontal and
vertical orientations within each dried core. Helium porosity, grain density, air and
water permeability were measured. For these models only the grain density and
helium porosity were used to calculate the bulk density. The results of the core
analysis are displayed in Table 7.1. These samples were mostly fine to medium
grained quartz sand (except for SST7 which is coarse grained), as reflected in the
average grain density of 2.64g/cc. These values were used to estimate the grain
density of other units, and the porosity values were used to calculate the bulk density
using the equation:
( )


=
100 100
100
h f
h
g B


Where
B
is the bulk density,
g
is the grain density,
h
is the helium porosity of the
sample, and
f
is the density of the pore fluid (water). One hundred percent water
saturation was assumed.

Synthetic model creation
76









Table 7.1: Porosity and grain density, calculated by ACS laboratories, of cores within the
Leederville Aquifer. The bulk densities were calculated using the equation presented above.

Bulk density
(inc pore H
2
0)
Sample
Number

Depth Dir
Porosity
Helium
Grain
Density

(m) (percent) (g/cm
3
) (g/cm
3
)

SST 1 88.04 H
x
30.3 2.72
SST 1 88.08 V
z
31.7 2.69
2.18
SLTST 4 115.06 H
x
37.3 2.69
SLTST 4 115.22 V
z
38.0 2.71
2.06
SLTST 5 137.34 H
x
30.9 2.60
SLTST 5 137.54 V
z
32.1 2.62
2.10
SST 7 151.80 H
x
39.2 2.60
SST 7 151.75 H
y
38.7 2.59
SST 7 151.68 V
z
37.9 2.60
1.98
SST 11 202.40 H
x
35.6 2.63
SST 11 202.53 V
z
36.0 2.62
2.04
Average: 35.2% 2.64g/cc


77
Chapter 8

8 Modelling results

8.1 Calculation of Interval velocities after injection

As described in section 6 the velocity of the sediments within the Leederville Aquifer
after injection is dependant on the change in effective pressure and fluid density.
The change in velocity due to effective pressure was calculated using an average of
the results of Zimmer et al. (2007 a & b), which was shown in Figure 6.1a. The
curves fitted to an average of the Vp and Vs data points is displayed in Figure 8.1.
These were used to calculate the change in velocity because velocity-pressure tests
on the Beenyup core samples have not been completed to date.

Five different injection pressures were examined; two which are expected to be
within the range of pressure of the injection at Beenyup (50 and 100kPa), and larger
pressures of 500kPa, 1MPa, and 1.5MPa, for comparison purposes. The effective
pressure for each scenario was calculated by subtracting the pore pressure (original
pressure plus the pressure change caused by injection) from the confining pressure
(i.e. 3.5-1.6=1.9MPa for the pre-injection scenario). P and S wave velocities were
calculated using the power law trends shown in Figure 8.1. The results of these
calculations are displayed in Table 8.1. The actual velocities from these calculations
were not used, because the sediments studied by Zimmer et al. (2007a) may have
slightly different textures and mineralogy; thus different initial velocities. The pore
and confining pressures were also estimated in a simplistic way, rather than being
measured. So instead the difference between the velocity at 1.9MPa (pre-injection)

78
and the velocities at 1.85MPa, 1.8MPa, 1.4MPa, 0.9MPa, and 0.4MPa effective
pressure, was calculated and applied to the velocities used in the model.

Table 8.1: Calculated velocity change caused by a change in effective pressure. (using the results
of Zimmer et al. (2007b))
Injection
pressure
Effective
pressure
Vp Vp Vs Vs
Pre-injection 1.90 2008.51 - 521.09 -
50kPa 1.85 2005.75 -2.76 517.59 -3.50
100kPa 1.80 2002.92 -5.58 514.02 -7.07
500kPa 1.40 1977.17 -31.34 482.41 -38.69
1MPa 0.90 1932.68 -75.82 431.46 -89.63
1.5MPa 0.40 1853.63 -154.87 351.55 -169.55


If the injected water causes a change of 50kPa to the formation pressure then the P
wave velocity can be expected to decrease by almost 3m/s, and the S wave velocity
is expected to decrease by 3.5m/s. With a pressure change closer to 100kPa the
velocity decrease is approximately double that of the 50kPa pressure. The P wave
velocity decreases by approximately 5.5m/s and the S wave velocity decreases by
just over 7m/s. These are not very large changes but if the pressure change is larger
than expected then velocity changes may be as high as 169m/s (for the 1.5MPa case).
Injection pressures above 500kPa are highly unlikely because reducing the effective
pressure to such small values would mean that the structural integrity of the
formation could be compromised.

Reducing the pore fluid density around the injection well (not taking the change in
pressure into account) does not have a very significant effect on seismic velocities or
bulk densities after injection. This is because the difference in density between the
groundwater and injected water is very small, i.e. 0.9967g/cc for the groundwater (at
450mg/L salinity, 26C, 1.6MPa) and 0.9954g/cc for the injected water (at 12mg/L
salinity, 26C, 1.6MPa). This small change results in a reduction of P wave velocity
Velocity vs. effective pressure
y = 1943.2x
0.0515
y = 443.1x
0.2526
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
0 1 2 3 4 5
Effective Pressure (MPa)
V
e
l
o
c
i
t
y

(
m
/
s
)
Vp wet
Vs wet
1.9MPa
1.8MPa
1.4MPa
0.9MPa
0.4MPa

Figure 8.1: Average effective pressure vs. velocity trends calculated from Zimmer et al. (2007a&b) for water saturated sediments. The effective pressures examined in this
study are indicated by the dashed lines.

80
by 0.4m/s and an increase in S wave velocity by 0.013m/s (due to the bulk modulus
of the rock decreasing whilst the shear modulus stays the same).

The effects of effective pressure and fluid density were treated separately and
combined to calculate the total effect on velocity. These results are presented in
Figure 8.2. The expected pressure range of the injection does not show a very
significant change in either P or S wave velocity. A maximum reduction of 6m/s in
Vp and 7m/s for Vs is predicted from the modelling. This is because the pressure
change is not very large. The effect that these small changes have on reflection
amplitudes was examined by creating synthetic models.

Figure 8.3 shows the calculated Vp:Vs ratios for the pre and post-injection models.
The trend fitted to the results follows a power law relationship proportional to
6
1
e
.
The Vp:Vs ratio of the Wanneroo Member for the pre-injection model is
approximately 3.45. With a 50kPa pore pressure increase the ratio increases to 3.46
and for a 100kPa increase the ratio becomes 3.48. Higher pressure changes of
500kPa, 1MPa and 1.5MPa results in Vp:Vs ratios of 3.63, 3.91, and 4.56
respectively. If ground/survey conditions are ideal then these small changes in
velocity may be used to infer pressure changes.


Seismic velocity vs. pore pressure change
0
250
500
750
1000
1250
1500
1750
2000
2250
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
Pressure (MPa)
V
p

(
m
/
s
)
Wanneroo Member
Vp
Vs
Range of expected
pressure changes
due to injection

Figure 8.2: Predicted P and S wave velocity change due to the injection. The expected range of pore pressure change is shaded in red

Pore



Vp:Vs vs. effective pressure
y = 3.8357x
-0.1652
2
2.5
3
3.5
4
4.5
5
5.5
6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Effective Pressure (MPa)
V
p
/
V
s

Figure 8.3: Plot of Vp:Vs ratio vs. effective pressure. The trend is a power law relationship like that presented by Zimmer et al. (2007b)
Pre-injection
50kPa pore
pressure increase
100kPa pore
pressure increase
500kPa pore
pressure increase
1MPa pore
pressure increase
1.5MPa pore
pressure increase
Modelling results
83

8.2 Synthetic models

Synthetic seismic models were created for each effective pressure modelled in
section 8.1. This was done using Zoeppritz equations and an 80hz Ricker wavelet in
Hampson Russell. Transmission losses or geometrical spreading were neglected so
that results replicate true amplitude sections. The Pro4D module in Hampson
Russell was used to subtract the pre-injection result from each dataset to show
amplitude differences, caused by changes in acoustic impedance contrasts, could be
examined. Time shifts were corrected. Figure 8.4 shows a synthetic trace for each
pressure change. The pre-injection scenario is displayed on the left and pore
pressure changes of 50kPa, 100kPa, 500kPa, 1MPa, and 1.5MPa are displayed
adjacent.

The 50 and 100kPa results are very similar in amplitude to the pre-injection results.
There are no noticeable time shifts caused by the velocity changes. These two
pressures represent the expected scenarios for the Beenyup injection. Very small
amplitude changes and time shifts can be identified in the 500kPa case. The 1MPa
and 1.5MPa results also show obvious differences in both amplitude and time. The
amount in which each reflection changes in amplitude is displayed in Table 8.2. An
increase in pore pressure between 50 and 100kPa causes an increase in reflection
amplitude by 0.7 to 2.9%. A 500kPa increase in pore pressure causes amplitudes to
increase by substantially larger amounts: 11%. Similarly a 1 and 1.5MPa increase
results in an amplitude increase of at least 26% within the injection interval.


Modelling results
84
Table 8.2: percentage increase in amplitude of the major reflections within the Leederville
Aquifer
Pore pressure
change
Top Wanneroo
SS1 amplitude
change (%)
Top Wanneroo
SLST amplitude
change (%)
Top Wanneroo
SS2 amplitude
change (%)
Top Mariginiup
amplitude change
(%)
Pre-injection
50kPa 1.10 1.63 1.53 0.77
100kPa 2.07 2.34 2.87 1.39
500kPa 10.90 13.87 10.72 11.23
1MPa 26.38 32.27 25.43 25.91
1.5MPa 54.65 66.70 51.57 52.36

Figure 8.5 to Figure 8.7 show the differences between the pre-injection and post-
injection synthetic sections (i.e. post-injection sections pre-injection section).
Amplitudes are scaled to 1 and time shifts have been corrected, allowing only
differences in amplitude being displayed. An increase in 50kPa results in the
amplitude of each reflection increasing by 0.0025. The largest increase in
amplitude occurs at the sand/shale interfaces i.e. the top of the Mariginiup Member.
Reflections from interfaces within the Wanneroo Member show smaller amplitude
changes, approximately 0.001, due to the smaller acoustic impedance contrasts.

The change in amplitude caused by a 100kPa pressure change (shown in Figure 8.5b)
is very similar to the 50kPa case, with most amplitudes increasing by at most 0.003.
The reflection from the top of the Wanneroo Member has increased in amplitude by
approximately 0.0018 due to the reduction in velocity within the Leederville Aquifer.
The top Mariginiup Member reflection also shows an increase in amplitude by a
similar amount. A 500kPa increase in pore pressure results in slightly larger
amplitude changes within the Leederville Aquifer. Figure 8.6 shows the results of
the subtraction of the pre-injection data from the 500kPa data. Amplitudes have
increased by at most 0.015. If injection pressures are much larger then the amount in
which the reflections from the injection interval increase in amplitude, increases
significantly. A change of 1MPa in pore pressure results in amplitudes increasing by
Modelling results
85
0.04 at the most. Similarly an increase of 1.5MPa results in amplitudes changing by
a maximum of 0.06.

If time shifts in the data are not accounted for, then the differences between each
dataset can be observed to increase with depth. This is due to differences being
caused by both changes in amplitude and time shifts. The deeper the interface the
longer the seismic wave has to travel through the slower velocity injection interval.
Modelling results
86










Figure 8.4: Zero offset synthetic traces for each pore pressure change, with the pre-injection
model in the background. Notice that the amplitude changes and phase shifts for the 50 and 100kPa
traces are subtle. The 500kPa, 1MPa, and 1.5MPa show larger differences.

Modelling results
87














Figure 8.5: Zero offset sections showing the amplitude difference between a) 50kPa pore
pressure change and; b) 100kPa pore pressure change, compared to the pre-injection scenario
(i.e. t
2
- t
1
).

b.
a.
D
i
f
f
e
r
e
n
c
e

i
n

a
m
p
l
i
t
u
d
e

(
t
2
-
t
1
)

D
i
f
f
e
r
e
n
c
e

i
n

a
m
p
l
i
t
u
d
e

(
t
2
-
t
1
)

T
i
m
e

(
m
s
)

T
i
m
e

(
m
s
)

Injection
Interval
Injection
Interval
Modelling results
88
















Figure 8.6: Zero offset section showing the amplitude difference between 500kPa pore pressure
change compared to the pre-injection scenario.




D
i
f
f
e
r
e
n
c
e

i
n

a
m
p
l
i
t
u
d
e

(
t
2
-
t
1
)

T
i
m
e

(
m
s
)

Injection
Interval
Modelling results
89














Figure 8.7: Zero offset sections showing the amplitude difference between a) 1MPa pore
pressure change; b) 1.5MPa pore pressure change, compared to the pre-injection scenario.

a.
D
i
f
f
e
r
e
n
c
e

i
n

a
m
p
l
i
t
u
d
e

(
t
2
-
t
1
)

D
i
f
f
e
r
e
n
c
e

i
n

a
m
p
l
i
t
u
d
e

(
t
2
-
t
1
)

T
i
m
e

(
m
s
)

T
i
m
e

(
m
s
)

Injection
Interval
Injection
Interval
b.
Modelling results
90
8.3 Application to pressure monitoring at Beenyup

Expected pore pressure changes resulting from the injection at Beenyup are very
small (50-100kPa). Thus the change in elastic properties and velocity is also small.
However considering that the repeatability of land seismic surveys (with respect to
positioning and source wavelet) is potentially very high, very small predicted
amplitude changes may be detectable by time lapse seismic. Any 16 bit or better,
seismic system should be able to detect these changes in ideal conditions.

If the injection causes higher than expected pore pressure changes, then the changes
will be easily detected using the seismic method. Pressure changes of 1MPa, and
1.5MPa in the aquifer may be unrealistic though because this may lead to permanent
damage near the well. The aquifer is also expected to be highly transmissive
(>1000m
2
/day) so large build-up in pressure will not occur.

Having the same receiver coupling and near surface conditions will be the deciding
factor determining whether these changes can ultimately be detected. Ideally near
surface conditions should be the same to ensure that no other time lapse effects are
introduced. If these effects are greater than those caused by injection, then the
pressure effect may not be detectable. Time lapse surveys should also be acquired
during the same time of year to ensure that the properties of the very near surface is
identical for each survey (e.g. water saturation).

A cemented 3C geophone string could be used to measure these changes. Such a
system would have better coupling repeatability compared to a surface seismic
survey, thus more likely to detect the amplitude changes. Another advantage of
Modelling results
91
using a 3C geophone string is that shear wave velocities can also be recorded. Figure
8.3 showed that the Vp:Vs ratio increases with decreasing effective pressure. Whilst
the change in ratio is not very large, the comparison of P and S wave interval
velocities from the injection interval, may provide valuable information about the
subsurface conditions. Whilst the geophone string at Beenyup only coarsely samples
the velocity field (because the geophones are spaced far apart) a change in interval
velocity between each geophone can be detected and would be adequate to compare
Vp:Vs ratios to infer pressure changes.

Conclusions and recommendations
92
9 Conclusions and recommendations

The Wanneroo Member of the Leederville Formation is the injection interval at
Beenyup. The Pinjar Member is expected to be the upper seal, located at a depth of
100m, and the Mariginiup Member is expected to be the basal seal, at a depth of
260m. 2D seismic data acquired prior to the 3D survey showed very good results.
Reflections associated with all important sedimentary units (Pinjar Member and
Wanneroo Member sands & siltstones) could be identified in the data.

The 3D seismic survey was designed, using GMGs MESA v10, to focus seismic
energy within the injection interval. This was achieved using near (< 50m) and far
(>225m) offsets which ensure adequate sampling of the Pinjar and Mariginiup
Members. Spatial aliasing and resolution requirements (dictated by hydraulic
modelling requirements) were also satisfied by using 2m bins and a 30m line
spacing. Three different survey geometries (orthogonal, 45 degree skew, and 26
degree skew) were tested to determine which will provide optimum results.

The 26 degree skew design was determined to be the best design for the current site
conditions because it showed a compromise between high fold, offset & azimuth
distributions, ease of acquisition, acquisition time, and complied with survey
objectives. Thirteen source lines and twelve receiver lines, divided into four
templates comprised of six receiver lines, were used in the design. A total of 2178
shots were proposed, which can potentially be achieved in approximately 9 days.

Site conditions changed and the 26 degree skewed design could not be used for the
3D survey. This was due to large piles of sand that were dumped over the site just
Conclusions and recommendations
93
before acquisition was to commence. Changes to the site resulted in it being more
feasible to focus the survey around the injection wells where knowledge of the
hydrogeology is more important. Dr Aleksander Dzunic and I redesigned the survey
one day before the commencement of acquisition. An orthogonal geometry with 20
source lines, 8 receiver lines, and a maximum of 1254 shots was used. Fold, offset
and azimuth distributions were optimised to obtain the best results under the new site
conditions.

The effect of changes in pressure and fluid properties, due to MAR, on seismic
velocities and densities at the site were examined and used to create time lapse
models of the injection. The dominant factor affecting the seismic response is
expected to be the change in pore pressure. An increase in pore pressure will lead to
a reduction in effective pressure, and thus seismic velocity. Density changes in the
pore fluid will be very small. If a pore pressure change between 50 and 100kPa is
assumed, the velocities of sands within the aquifer are expected to reduce by 3 to
6m/s. Very subtle amplitude changes (as large as 2.9%) and small time shifts of
reflections within and underlying the Leederville Aquifer can be observed. These
changes are very small however may be detectable by VSP and continuous source
seismic methods if near surface conditions and receiver coupling are consistent.

The models used to investigate the time lapse seismic response due to injection,
represent first pass models. Many estimates were required because sonic and density
logs were not acquired, and velocity-pressure tests have not yet been completed.
Such tests would make the modelling more accurate and credible. The results
presented in this paper are based on relationships that have been observed in many
Conclusions and recommendations
94
unconsolidated sediments, and should be close to the real velocity-pressure trends at
the site.
References
95
10 References

Australian Government National Water Commission, n.d., Beenyup reverse Osmosis
groundwater replenishment project, accessed May 19, 2007; http://www.nwc.gov.au/
agwf/wsa/docs/Beenyup.pdf.

Batzle, M., and Z. Wang, 1992, Seismic properties of pore fluids: Geophysics,
57(11), 1396-1408.

Berkman, D.A., 2001, Field geologists manual 4
th
edition, monograph 9: Australian
Institute of Mining and Metallurgy.

Cordsen, A., M. Galbraith, and J. Peirce, 2000, Planning Land 3-D Seismic Surveys,
Geophysical Developments Series No. 9: Society of Exploration Geophysicists.

CSIRO, n.d., Managed Aquifer Recharge: Frequently Asked Questions, accessed
May 19, 2007; http://www.csiro.au/files/files/p2oq.pdf.

Davidson, W.A, 1995, Hydrogeology and groundwater resources of the Perth region
Western Australia, Geological Survey of Western Australia Bulletin 142:
Department of Minerals and Energy.

Dodds, K.J, D.N. Dewhurst, A.F. Sigins, R. Ciz, M. Urosevic, B. Gurevich, and D.H.
Sherlock, 2007, Experimental and theoretical rock physics research with application
to reservoirs, seals and fluid processes: Journal of Petroleum science and
engineering, 57(2007), 16-36.

Evans, B.J., 1997, A handbook for seismic data acquisition in exploration,
Geophysical monograph series, no. 7: Society of Exploration Geophysicists.

Fetter, C.W., 1994, Applied hydrogeology 3
rd
edition: Macmillian College
Publishing Company.

Furre, A., 2002, The effective stress coefficient for wave velocities in saturated grain
packs: 64
th
Annual Conference, EAGE, Expanded Abstracts: cited in Hofmann, R, X.
Xu, M. Prasad, and M. Batzle, 2005, Implications of effective stress laws for seismic:
75
th
Annual International Meeting, SEG, Expanded Abstracts, 1243-1246

Government of Western Australia, 2007, State water plan 2007: Department of the
Premier and Cabinet.

Hacikoylu, P, J. Dvorkin, and G. Mavko, 2006, Elastic and petrophysical bounds for
unconsolidated sediments: 76
th
Annual International Meeting, SEG, Expanded
Abstracts, 1762-1766.

Helgeson, H.C, and D.H. Kirkham, 1974, Theoretical prediction of the
thermodynamic behaviour of aqueous electolytes: American Journal of Science, 274,
1089-1198: cited in Batzle, M, and Z. Wang, 1992, Seismic properties of pore fluids:
Geophysics, 57(11), 1396-1408.
References
96
Huffman, A.R, and J.P. Castagna, 2001, The petrophysical basis for shallow-water
flow prediction using mulicomponent seismic data: The Leading Edge, 20, 1030-
1052.

Kasenow, M, 2001, Applied ground-water hydrogeology and well hydraulics 2
nd

edition: Water Resources Publications, LLC.

Kresic, N, 2007, Hydrogeology and groundwater modelling: second edition: Taylor
& Francis Group LLC,

Li, R., M. Urosevic, and K. Dodds, 2006, Prediction of 4D seismic responses for the
Otway Basin CO
2
sequestration site: 76
th
Annual International Meeting, SEG,
Expanded Abstracts, 2181-2185.

Mavko, G, T. Mukerji, and J. Dvorkin, 1998, Rock physics handbook: Tools for
seismic analysis in porous media: Cambridge University Press: cited in Han, D.H.,
and M.L. Batzle, 2004, Gassmanns equation and fluid-saturation effects on seismic
velocities: Geophysics, 69(2), 398-405.

Playford, P.E., A.E. Cockbain, and G.H. Low, 1976, Geology of the Perth Basin,
Western Australia, Western Australia Geological Survey Bulletin 124:: cited in
Davidson, W.A, 1995, Hydrogeology and groundwater resources of the Perth region
Western Australia: Geological Survey of Western Australia Bulletin 142:
Department of Minerals and Energy.

Reese, R.S., 2002, Inventory and review of aquifer storage and recovery in southern
Florida, U.S. Geological Survey Water-Resources Investigations Report 02-4036:
cited in Kresic, N, 2007, Hydrogeology and groundwater modelling second edition:
Taylor & Francis Group, LLC.

Rosenshein, J.S. and J.J. Hickey, 1977, Storage of treated sewage effluent and storm
water in a saline aquifer, Pinellas Peninsula, Florida: Ground Water, 15(4), 289-293,
cited in Kresic, N, 2007, Hydrogeology and groundwater modelling second edition:
Taylor & Francis Group, LLC,

Sarasty, J.J, and R.R. Stewart, n.d, Petrophysical relationships derived from well logs
in the White Rose oilfield, offshore Newfoundland: accessed September 20, 2007;
www.crewes.org/Conference-abstracts/2003/CSEG/Jaramillo_CSEG_2003.pdf

Sheriff, R.E., 1992, Petrophysical and geophysical background, basic petrophysics
and geophysics; in Sheriff, R.E., ed., Reservoir Geophysics, Society of Exploration
Geophysicists, 37-49

Terzaghi, K., 1943, Theoretical soil mechanics: John Wiley & Sons: cited in
Zimmer, M.A, M. Prasad, G. Mavko, and A. Nur, 2007a, Seismic velocities of
unconsolidated sands: Part 1- Pressure trends from 0.1 to 20MPa: Geophysics, 72(1),
E1-E13.

Turnbull, M (hon), 2007, Australian government injects over $30 million into West
Australian water projects (media release), accessed May 19, 2007; http://www.
environment.gov.au/minister/env/2007/pubs/mr22feb07.pdf.
References
97
Wang, Z. 2000, The Gassmann Equation revisited: Comparing laboratory data with
Gassmanns predictions: in Wang, Z and A. Nur, eds., Seismic and acoustic
velocities in reservoir rocks, volume 3, Recent developments: Geophysics reprint
series 19, Society of Exploration Geophysicists, 8-24.

Water Corporation, 2006a, Sustainable water resource use and management: the
water corporations response to the draft state sustainability discussion paper: Water
Corporation

Water Corporation, 2006b, Water for life report 2006: Water Corporation

Zimmer, M.A, M. Prasad, G. Mavko, and A. Nur, 2007a, Seismic velocities of
unconsolidated sands: Part 1- Pressure trends from 0.1 to 20MPa: Geophysics, 72(1),
E1-E13

Zimmer, M.A, M. Prasad, G. Mavko, and A. Nur, 2007b, Seismic velocities of
unconsolidated sands: Part 2- Influence of sorting- and compaction-induced porosity
variation: Geophysics, 72(1), E15-E25

Zou, Y, L.R. Bentley, and L.R. Lines, n.d, Time lapse modelling for Pikes Peak field,
accessed May 19, 2007; http://www.crewes.org/Conference-abstracts/2003/CSEG/
Zou_CSEG_2003.pdf.

You might also like