Вы находитесь на странице: 1из 64

Reservoir Innovations

V ol ume 1 , No. 2 , 2 0 1 2 A wireline and Perforating Tec h nology Journal

In This Issue

Designing Completions
after Predicting Wellbore
Dynamic-Shock Loads during
Perforating
See pages 4-10

Performance of a New
2.35-in. Wireline or Memory
Quad Combo for Through-Bit
or Small-Hole Logging
See pages 37-46
Executive Steering Committee
Eric Carre

A Message from David Topping Senior Vice President, Drilling & Evaluation

David Topping
Vice President, Wireline and Perforating

As the incoming vice president for Wireline and Perforating, it is my Ahmed Kenawi
pleasure to welcome you to the second edition of Reservoir Innovations. Vice President, Sperry Drilling
My hope is that this edition will help demonstrate the depth and
Greg Powers
breadth of our Wireline and Perforating portfolio.
Vice President, Technology
This issue includes two articles about new additions to our solutions
portfolio; “New Slickline Smart-Release Tool Mitigates Wire-Recovery Editorial Advisory Committee
Issues on Extreme Well Conditions” and “Performance of a New Ian R. Hunt
2.35-in. Wireline or Memory Quad Combo for Through-Bit or Small-Hole Ron Cherry
Logging.” We’re also proud to publicize that the paper on page 53, “Using Computer Modeling Daniel Dorffer
to Generate Accurate Pe Equations,” achieved the SPWLA 2011 Symposium “Best Presentation Andrew Kirkwood
Award.” It covers the impact of computer modeling capabilities on tool response accuracy. Dan Quinn
Sriram Srinivasan
So what is Wireline and Perforating?
Soraya Brombacher
“Wireline” is only describing a means of conveyance. However the range of applications
throughout the well lifecycle can be tremendously varied. In the Open Hole environment this can Managing Editor
include conveying on electric line, drill pipe conveyed or pumped down. While in the Cased Hole Soraya Brombacher
environment it can also include slickline, coiled tubing or tractor conveyed. With collaboration
between the Halliburton team and our customers we can develop unique solutions using the Circulation
most applicable conveyance method to provide successful formation and reservoir evaluation and Nancy Kirkland
efficient well interventions. The paper, “Production Logging in Horizontal Wells: Case Histories
from Saudi Arabia Utilizing Different Deployment and Data Acquisition Methodologies in Open Design
Hole and Cased Hole Completions” is one demonstration of this collaboration. Griffin Creative Company
“Perforating” is not merely punching holes in casing. When it can cost many millions of dollars
This magazine is published biannually
to complete a well, one focuses quickly on the fact that where perforating is concerned, you
by Halliburton Wireline and Perforating.
have only one chance to do it right. Success requires the right equipment, the right perforating
For comments and suggestions,
products, and the right organizational capability to deliver the service. In today’s world, however,
please contact:
even that’s not enough. Success also requires job-design simulation capabilities that deliver
WirelineandPerforating@Halliburton.com
optimized connectivity between the cased formation and the wellbore for production, injection
or stimulation. These simulations also provide the assurance that the dynamic shock loads are
understood and managed to prevent damage to the tubular and completion equipment. Our
first paper, “Designing Completions after Predicting Wellbore Dynamic-Shock Loads during
Perforating” includes three case histories on this subject.
As always, I encourage you to collaborate with our Technical Sales and Formation and Reservoir
Solutions representatives to coauthor papers that raise awareness of your technology challenges,
and to highlight solutions that have created value in your reservoir understanding.
We welcome your interest and feedback in our publications, and we sincerely hope you find them
both valuable and enjoyable.
On the Cover
Halliburton uses a downhole detonation event
predictive model that provides pressure and shock-
Thank you, wave predictions to prevent toolstring damage due to
the firing of perforating guns. More advanced modeling
tools, including 3D finite element analysis (FEA)
techniques, are now being integrated in the product.
This modeling process improves gun system design and
job designs, and also helps ensure that health, safety
and environment needs are met.
David Topping
Vice President, Wireline and Perforating © Copyright 2012 Halliburton
All rights reserved.

2
Contents
4  esigning Completions after Predicting Wellbore
D

4
Dynamic-Shock Loads during Perforating
Mitigating Risks of Deepwater Tubing-Conveyed Perforating

11  dvancements in Carbon-Oxygen Surveillance


A
of the Deepwater Gulf of Mexico Mars Waterflood
Porosity Reduction’s Impact on Pulsed Neutron Saturation

20  ew Slickline Smart-Release Tool Mitigates Wire-Recovery


N
Issues in Extreme Well Conditions

11
Expanding the Range of Applications of Traditional Slickline

26 Production Logging in Horizontal Wells: Case Histories from


Saudi Arabia Utilizing Different Deployment and Data Acquisition
Methodologies in Open Hole and Cased Completions
New-Generation Array Tool Deployment Techniques

37 
Performance of a New 2.35-in. Wireline or Memory Quad Combo
for Through-Bit or Small-Hole Logging

47
Highest-Quality Measurements with Environmental Corrections

 Modern Microwave Formation Evaluation Sensor


A
and Its Applications in Reservoir Evaluation
20
Dielectric Logging Energized by “Cell Phone” Technology

53 Using Computer Modeling to Generate Accurate Pe Equations


Understanding the Pe Measurement

26

37

47

53 3
M i t i g a t i n g R i s k s of De e pwat e r Tubing- Convey ed Perforating

Designing Completions after Predicting


Wellbore Dynamic-Shock Loads During Perforating
Jack Burman, SPE, Exploitation Technologies, LLC; Martin Schoener-Scott, Cam Le, and David Suire, Halliburton
Presented at the SPE 2011 Brasil Offshore Conference on 15-17 June 2010 in Macae, Brazil.
Copyright 2010 SPE. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Abstract
Oilwell casing perforation has been used in the industry since 1932 and was pioneered by the Lane-Wells Company who introduced the bullet-gun perforating
technique. Shortly after the introduction of the bullet gun, explosive jet perforators were introduced, which provided a different and more aggressive way to
perforate casing. With the increased demand for oil and gas over the past decades, operators have been forced to explore deeper, hotter reservoirs to find
the most prolific reservoirs. These deepwater opportunities have required constant changes to equipment and services to increase their technical capabilities
for performing in more critical environments. Perforating with higher-shot densities, propellants, and larger perforating guns has been ongoing to meet these
new challenges.

A major problem with these increases, however, is the difficulty in predicting dynamic wellbore behaviors that cause tubulars to collapse and bend and
packers to unset as perforating guns were detonated. Research to understand the pressure behavior during the perforation event, in addition to the solid
loading that is imparted to the tubulars, packers, and other completion hardware in the perforating assembly, was needed to enable the industry to go
forward with a high level of confidence that wells could be completed safely and cost effectively. This paper discusses a shock-wave computer modeling
program that evaluates the mechanical risk factors of well components to ensure that the health, safety, environment, and service quality needs in a design
are addressed. A time-marching, finite-differences technique is applied as the numerical method for both fluids and solids. The software is installed on a
personal computer and typically executes the models within several minutes to several hours, depending on the complexity of the job design.

The physics-based model has been validated (Schatz et al. 1999 and Schatz et al. 2004) with special high-speed recorders that sense pressure, temperature,
and acceleration at a sampling frequency of 115,000 samples per second. This paper provides data from offshore oil and gas wells in the Gulf of Mexico to
demonstrate the success of the design.

­­Introduction Gauge Specifications


The simulation of dynamic pressure and shock-load responses to wellbore Maximum Acceleration / Vibration ± 50,000 Gs of acceleration/vibration
events, such as perforating, is becoming a standard procedure, especially
Sampling Rate 115,000 data points/second,
in deepwater completions. A shock-wave ShockPro™ computer modeling (High, Intermediate, and Slow Speeds) down to 1 sample every 10 seconds
program is a tool that can be used to better understand the energy created at 12 bits @ 115,000 data points/sec
Data Resolution
gun detonation, enabling more accurate estimates of dynamic shock loading (0.02% accuracy of full scale)

and movement of completion equipment in the wellbore. The software Pressure Range 30,000 psi (peak), 15,000 psi (static)

modeling application determines the dynamic pressure behavior during Power Requirements IES Potted AA alkaline batteries
the perforation event as well as the solid loading imparted to the tubulars, Current Drain 500 uA sleeping / 100 mA sampling
packers, and other completion hardware in the bottomhole assembly (BHA). Temperature Range -40°C to 120°C
A physics-based numerical model accounts for fluid dynamics and dynamic
Sensor Frequency Response 0 to 10,000 Hz
failure of solids by accounting for pressures on surface, drag, internal stress
Memory Capacity 1,048,576 data points
waves and reflections, and gravity.
Computer / Communications 750 MHz or greater Pentium III

The time-marching finite differences technique is applied as the numerical Software is included Windows XP is recommended

method for both fluids and solids. The software is compiled on a personal Shock Mitigator1 - #IES-SM-11116 111⁄16” OD x 17” (8 lb), 17-4 SS

computer and typically requires several minutes to several hours, depending Gauge Dimensions 111⁄16” OD x 50” (22 lb), 17-4 SS
on the job design complexity. The numerical solution accounts for the Gauge Options
following failure modes:
#IES-GAUGE-211 111⁄16” OD Gauge (Pressure and RTD)
• Tubing burst/collapse 111⁄16” OD Gauge (Pressure,
#IES-GAUGE-212
• Packer axial load/differential ±10 G’s Accelerometer, and RTD)

• Tubing axial buckling or bending #IES-GAUGE-213


111⁄16” OD Gauge (Pressure, ±60,000 Gs
and ±10 Gs Accelerometers, and RTD)
• Tubing compressive/tensile yield
• Gun burst/collapse Figure 1. High-speed recorder (http://www.iesrecorders.com).
• Gun compressive/tensile yield
4
BD Sales Team Build Shockpro
Request Shockpro Model Using the
Run & Provide Information
Supporting Provided
Documents

Analyze
Shockpro
Results

Are Provide Changes Halliburton


Changes NO Sales Team NO
Made to Performs
Necessary? with Shockpro Proposal? Perforating Job
Report

YES YES

Record Final
Rebuild Model Rebuild Model Was There a NO Was a NO Model Parameters
with Changes & with Changes & Problem on the Job Fast Gauge
Associated with into Shockpro
Analyze Results Analyze Results Run? Database
Shockpro?

YES YES YES

Are Coordinate with


Changes Rebuild & Run
Investigation Team Model Using Post
Necessary? for a Root Cause Job Information
Analysis

NO

Provide Sales Do Model


Team with Final & Fast Gauge
Shockpro Report Match?

Figure 2. Shrouded firing head and YES NO


high-speed recorder assembly.
Continue Matching
Process Using
Fast Gauge
Guidelines
• Casing burst
• Packer plug effect Record Matched
Model Parameters
• Sump packer/bridge plug axial load into Fast Gauge
Database
• Wireline tensile yield/pull-out
Figure 4. Shock modeling process.
The physics-based model has been
validated (Schatz et al. 1999 and Schatz et al. 2004) many times by using This paper discusses three wells that are located in the East Breaks and
special high-speed recorders (Fig. 1) that sense pressure, temperature, and Garden Banks areas of the Gulf of Mexico (Fig. 3). East Breaks is located
acceleration at a sampling frequency of 115,000 samples per second. off the coast of Texas in the western Gulf of Mexico in water depths that
range from 600 to 4,500 ft. The Garden Banks area is located off the
One of the many lessons learned when modeling dynamic shock loads is Texas/Louisiana coast in water depths that are similar to those of the
understanding how it affects various components in the completion string. East Breaks area.
A critical component in a shoot-and-pull operation is the firing head that is
coupled directly to the perforating gun assembly. Many times this connection Case Histories
is the weak point in the BHA, and something needs to be done to make this Case History 1 The first case history describes the process of using the
connection robust enough to withstand the forces created at gun detonation. new shock-wave modeling software program (Fig. 4) on a typical shoot and
The technique adopted is to place the firing head in a shroud (Fig. 2) that pull in the Gulf of Mexico. This gas well is located in Garden Banks in a
would make the connection between landing string and the gun assembly, water depth of 2,715 ft. It is a two-stage completion, with perforations at
thus eliminating the tensile and compression loads from the firing head to 19,750- to 19,793-ft measured depth (MD) (Stage 1) and 19,474- and
the shroud. 19,504-ft MD (Stage 2) and anticipated BHPs of 10,028 and 9,994 psi,
respectively. A job proposal (including the planned toolstring) was provided,
as well as the well trajectory, well schematic, and input sheet with critical
parameters for dynamic shock modeling. In the upper zone of a two-zone
completion, it is imperative that no damage occurs to the packer plug set in
the lower frac-pack packer.

The shock-wave computer modeling software simulations are performed for


the planned tubing-conveyed perforation (TCP) jobs in accordance with the
following BHAs:
Stage 1 – Lower Zone
• Workstring to surface
• 95⁄8-in. packer
• 1–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing
• Below packer safety joint
Figure 3. East Breaks (EB) and Garden Banks (GB).
5
M i t i g a t i n g R i s k s of De e pwat e r Tubing- Convey ed Perforating

• 1–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing and the points tagged are for illustration only; for each time step, the model
• 3½-in. fill disc assembly checks for dynamic failure at each depth in the completion.
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing
• Shrouded firing head assembly Fig. 6 shows a plot of pressure at various nodes vs. time; the maroon curve
• 43-ft loaded 6½-in., 14 spf RDX Super Hole shows the differential pressure on the packer in the annulus as a function
• Sump packer 10 ft below bottom shot of time. The peak differential pressure at the packer is ±1,700 psi at 26 ms
• PBTD 246 ft below bottom shot into the run. This peak differential is a result of the pressure in the annulus
Stage 2 – Upper Zone below the packer that goes on a vacuum for a short period of time as fluid in
• Workstring to surface the annulus and reservoir flows into the guns, and the hydrostatic pressure
• 95⁄8-in. packer on the backside remains constant. Although pressure is normally applied
• 1–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing to the backside to protect the packer or keep it in place from a hydraulics
• Below packer safety joint standpoint, the addition of annulus pressure generally serves to increase the
• 1–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing differential pressure on the packer elements. The peak pressure at the top
• 3½-in. fill disc assembly perforation (green curve) and the bottom perforation (olive) is ±15,400 psi
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing as a result of gun detonation; the pressure decreases exponentially as the
• Shrouded firing head assembly distance from the event increases.
• 30-ft loaded 6½-in., 14-spf RDX Super Hole
• Frac-pack packer 10 ft below bottom shot with packer plug installed
Note: the BHA above the packer is not modeled because the packer
is the anchor.

The well is perforated in 10.5-ppg calcium chloride completion fluid. The


shock-wave modeling simulation is used as a screening tool to identify
components of the BHA that are at higher risk of failure in a dynamic
mode and to confirm upward and downward movement of the guns
after detonation.

The graphical output from the simulation, shown in Fig. 5, shows a plot of
motion (left y-axis) and packer load and tension/compression (right y-axis)
as a function of time (0 to 0.5 sec). A point was tagged in the tubing above
the fill disk at 19,720-ft MD; the tubing joint moves up ± 0.16 ft and down Figure 6. Stage 1 pressure at selected nodes.
± 0.08 ft and then oscillates like a spring during gun detonation (red curve).
From motion at the tubing and packer (black line), the corresponding tension Fig. 7 shows a plot of pressure vs. time across the perforated interval; pres-
or compression can be calculated. In this case, the tubing (light blue curve) sures are averaged across the entire perforated interval. The peak pressure
experiences a maximum compressive loading of ± 345 klbf and a tensile is 15,268 psi, and the minimum pressure is 9,983 psi; a dynamic surge is
loading of ± 77 klbf. The packer (navy blue curve) at 19,662-ft MD initially predicted to be greater than the planned perforating state. This dynamic surge
shows a maximum upward force of ± 319 klbf and then a downward force is attributable to the completion and reservoir fluids that rush into the 6½-in.
of ± 176 klbf. The simulation is a finite difference, time-marching model, gun carriers. The gun carriers are at atmospheric pressure internally before

Figure 5. Stage 1 motion, packer load, tubing compression, and tension. Figure 7. Stage 1 average pressure in perforated interval.
6
gun detonation and cause a local vacuum after holes are punched in the gun
scallops. These three plots are typically included in the shock-wave modeling
software report. No issues were predicted in this case. Results were logged
into the modeling software database; the job was implemented as planned
with no damage to downhole equipment.

For the Stage 2 upper completion, the main concerns were damage to the
packer plug and excessive tubing movement because the frac-pack packer
is 10 ft below the bottom shot. With the packer plug installed, it will yield
a clearance of approximately 4 ft from the packer plug running tool. The
graphical output from the simulation, shown in Fig. 8, provides a plot of
motion (left y-axis) and tension/compression (right y-axis) as a function of
time (0 to 0.5 sec). A point was tagged in the tubing above the fill disk at
19,454 ft MD (red curve); the tubing joint moves up ± 0.11 and down ± 0.09 ft.
This peak oscillation is within the 4-ft clearance, which ensures that the Figure 9. Stage 3 Iteration 1 motion, packer load, tubing compression,
packer plug will not be affected by the running tool upon gun detonation. and tension.

No issues were predicted to occur as in the previous stage. The upper zone
was completed with no damage to downhole equipment.

Figure 10. Stage 3 Iteration 2 motion, packer load, tubing compression,


and tension.

Note: For Iteration 2, the 30-ft joints were replaced with 10-ft pups; total
Figure 8. Stage 2 motion, packer load, tubing compression, and tension. tubing length was left unchanged.

Case History 2 This case history is an oilwell located in Garden Banks with The plots in Fig. 9 and 10 are essentially identical; the difference is that in
a water depth of 2,095 ft. It was originally planned for three stages with changing the rigidity of the tubing increases its threshold to resist bending.
perforations at 21,509–21,589 ft for Stage 1, 21,138–21,218 ft for Stage 2, The model will flag bend if this possibility should occur, but calculations will
and 20,721–20,781 ft for Stage 3. Preliminary models were run on these continue as if the tool string is still intact.
three stages with results showing no issues with the first two stages.
Stage 3 had possible bending in the tubing from 20,590–20595 ft, which can When the actual well was drilled and cased to TD, the lower zone had
be seen in Fig. 9. Fig. 10 shows the results from the second iteration that no indication of hydrocarbons and was, therefore, dropped form the
replaces the 30-ft joints to 10-ft pup, which resolves bending. The BHA for completion plan. Models were then updated with the new intervals
Stage 3 is as follows: (21,132–21,152 ft and 21,044–21,080 ft) and reservoir parameters. Findings
• Workstring to surface from previous models (use of 10-ft pups) were implemented in the updated
• 95⁄8-in. packer models. Final pre-job models showed no indications of failure. Both stages
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing were completed with no damage to downhole equipment.
• Below packer safety joint
• 2–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing Case History 3 This case history is also a gas well, and it is located in East
• 3½-in. fill disc assembly Breaks. It has a water depth of 3,382 ft with planned perforations at 7,630
• Shrouded firing head assembly to 7,660 ft MD and anticipated BHP of 4,166 psi. As the first case history
• 60-ft loaded 6½-in., 14-spf RDX Super Hole described, the required information was provided and a preliminary model
• Frac-pack packer 10 ft below bottom shot with packer plug installed was run with no flags. The dynamic load results are displayed in Fig. 11.

7
M i t i g a t i n g R i s k s of De e pwat e r Tubing- Convey ed Perforating

Figure 11. Iteration 1 initial proposed BHA without BPSJ. Figure 12. Iteration 2 with addition of BPSJ and tubing.

This case history was modeled in accordance with the following BHAs:
• Workstring to surface
• 95⁄8-in. packer
• 1–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing
• 3½-in. fill disc assembly
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing
• Shrouded firing head assembly
• 30-ft loaded 6½-in., 14-spf RDX Super Hole
• PBTD 10 ft below bottom shot

It was recommended to add a Bellow Packer Safety Joint (BPSJ) to the


BHA for a controlled disconnect in the event that the gun was to become
stuck. The well was then drilled to total depth (TD). An updated well
survey, reservoir parameters, and the BHA with a BPSJ were changed in
Iteration 2. Reservoir parameters that changed include the BHP (from 4,166
Figure 13. Iteration 3 with addition of 30 ft of tubing between packer and
to 5,878 psi) and the BHT (145 to 160°F). The perforations were changed BPSJ. All tubing was changed to 10-ft pup joints.
to 9,102 to 9,122 ft MD with the PBTD at 9,206 ft MD. The updated BHA
included the following: • Workstring to surface
• Workstring to surface • 95⁄8-in. packer
• 95⁄8-in. packer • 6–10 ft pup joint 3½-in., 12.95-lb/ft P-110 tubing
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing • Below packer safety joint
• Below packer safety joint • 3–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing • 3½-in. fill disc assembly
• 3½-in. fill disc assembly • 3–10-ft pup joint 3½-in., 12.95-lb/ft P-110 tubing
• 1–30-ft joint 3½-in., 12.95-lb/ft P-110 tubing • Shrouded firing head assembly
• Shrouded firing head assembly • 20-ft loaded 6½-in., 14-spf RDX Super Hole
• 20-ft loaded 6½-in., 14-spf RDX Super Hole • Sump packer 10 ft below bottom shot
• Sump Packer 10 ft below bottom shot • PBTD 84 ft below bottom shot
• PBTD 84 ft below bottom shot
The Iteration 3 dynamic load results, shown in Fig. 13, predict no issues with
The Iteration 2 dynamic load result, shown in Fig. 12, indicates possible the final BHA. The 10-ft pup joints increase the rigidity of the tubing, which
tubing bending (green flags) at depths between 8,980 and 8,982 ft and lessens the likelihood of bending. The additional 30 ft of tubing reduces the
tubing compressive yielding (yellow flags) at 8,990 to 8,991 ft. To eliminate peak compressive loads in the tubing. The TCP job was implemented, and
these possible failures, the BHA was adjusted by adding an additional 30 ft BHA was retrieved with no damage. The shock-wave software modeling
of tubing between the packer and BPSJ and by changing the 30-ft joints to results were then logged into software database.
10-ft pup joints. The BHA for Iteration 3 included the following:

8
For Case 2, which has fewer guns than Case 1 Stage 1, it is counter- questions requires examining the pressure results and determining how
intuitive that the case with fewer guns had the higher loads. Why would they correspond to the loads. Figs. 14 and 15 show the pressure plots from
it have higher loads? What is causing this to occur? Which parameters Case 2 with elapsed time of 0.25 sec to provide more visual detail of the
are critical to dynamic loading? These are some of the questions (and pressure response. Fig. 14 plots the pressures at selected nodes, and Fig. 15
opportunities) made apparent by this numerical analysis. Addressing these plots the average pressure in the perforated interval.

When comparing the peak pressure of the top and bottom perforations,
Case 3 had a peak of 21,660 psi, as compared to 15,400 psi in Case 1
Stage 1. Regarding the differential pressure across the packer, Case 3 had
a maximum of 3,200 psi resulting from the bottom of the packer pushing
it up. Case 1 Stage 1 had a maximum of 1,700 psi resulting from a surging
effect of a partially loaded gun on top of the gun assembly, which lessens
the effect of the upward force. In both cases, the average pressure in the
perforated interval was similar, but the reservoir pressures were very
different. Case 3 was 5,878 psi, as compared to Case 1 Stage 1 with
9,994 psi. This difference in pressure would affect the containment of
gun gas pressures and the differential across packer, tools, and guns. In
addition, the volume available below the packer must be evaluated; Case 1
Stage 1 had 162 ft more rat hole than Case 3. This difference would enable
more pressure to dissipate and prolong the returning pressure wave from
Figure 14. Case 3 Iteration 2 pressure at selected nodes. PBTD. All of these factors contributed to the higher packer loads and tubing
compressive loads in Case 3.

Conclusions
• As shown in this paper, an experienced user of this software can model
the dynamics of tubing-conveyed perforating operations and eliminate
potential problems before execution.
• The oil and gas industry now has a numerical model to predict dynamic
behavior during perforating events. With this tool, operators can design
well completions ahead of time to mitigate operational challenges and to
minimize the risk of downhole tubular failure during execution.
• When the probability of an operational issue is identified and determined
to be largely unavoidable because of well conditions, contingency
operations to rectify the issue can be developed beforehand to minimize
the economic effect should the worst case arise.

Figure 15. Case 3 Iteration 2 average pressure in perforated interval.

9
M i t i g a t i n g R i s k s of De e pwat e r Tubing- Convey ed Perforating

Authors
Jack Burman is President Martin Schoener-Scott is the Cam was later transferred to Houston Business
of Exploitation Technologies, Business Development Technical Development in 2007 as Technical Advisor for GoM
a Houston, Texas-based Manager for Tubing Conveyed operations. His role as a Product Champion includes
consulting firm that specializes in Perforating. He started his career managing dynamic modeling services. Cam is a member
worldwide deepwater and shelf with Halliburton Energy Services, of YPSPE and SPE.
well completions, production in Victoria, Texas, after graduating
engineering and field implementation. He graduated from Saint Mary’s University San Antonio with a David Suire is the Gulf of Mexico
with a BS in Mining Engineering from Virginia Tech business degree in 1981. He later obtained his MBA in Service Quality Manager for
in 1978 and an MS in Petroleum Engineering from 2004 from Regis University in Denver, Colorado, and Halliburton Testing and Subsea,
the University of Wyoming in 1982, and is a licensed progressed through various operational and managerial responsible for the implementing
professional petroleum engineer. Jack’s career includes positions in well testing, perforating, cement and and continued development of
positions at Chevron, Conoco, Newfield Exploration, stimulation in a diverse range of operating conditions the Quality Management System
and Snyder Oil Corp before forming Exploitation both onshore and offshore. Martin is also a member for GoM TSS Operations. In 2001 he received a BS in
Technologies in 1999. He has experience on three of the SPE and coauthor of several technical papers Mechanical Engineering from the University of Louisiana
world-record water depth deepwater developments: presented at SPE conferences around the globe. at Lafayette. David started his career as a field engineer
Vastar/BP “Horn Mountain” Spar, Anadarko “Marco with Halliburton in 2001 and was later transferred to
Polo” TLP, and Anadarko “Independence Project” Cam Le is the TCP Perforating Houston Business Development in 2005 as a Technical
Subsea, as well as numerous other deepwater and Product Champion for Halliburton Advisor for GoM Operations. In 2006 David became an
shelf projects. Jack has authored 15 SPE papers and Wireline and Perforating, Account Representative for GoM Operations in Tools,
has made numerous industry presentations on various responsible for implementing and Testing, and Tubing Conveyed Perforating. He is a
completion and production engineering topics. Jack is commercializing new technologies, member of SPE and API.
a member of the SPE, and was awarded the SPE Gulf and providing technical support for
Coast North America Drilling and Completion Engineer global operations. He holds a BS (2002) in Chemical
of the Year award in 2007. Engineering from Lamar University, and started his
career as a field engineer with Halliburton in 2004.

Acknowledgements References
The authors would like to thank Halliburton for supporting this effort Canal, A., Miletto, P., Schoener-Scott, M., Medeiros, J., and Barlow, D.
and for permitting the publication of this paper. They would also like to 2010. Predicting Pressure Behavior and Dynamic Shock Loads on
Completion Hardware During Perforating. Paper OTC-21059 presented at
express special gratitude to Halliburton customers and associates who the Offshore Technology Conference, Houston, Texas, 3-6 May.
supported the development of this technique to improve the probability
Schatz, J.F., Folse, K.C., Fripp, M., and Dupont, R. 2004. High-Speed
that the job was done right. Pressure and Accelerometer Measurements Characterize Dynamic
Behavior During Perforating Events in Deepwater Gulf of Mexico. Paper
Nomenclature SPE 90042 presented at the SPE Annual Technical Conference and
Exhibition, Houston, Texas, 26-29 September.
BHA Bottomhole assembly
Schatz, J.F., Haney, B.L., Ager, S.A. 1999. High-Speed Downhole Memory
BHP Bottomhole pressure Recorder and Software Used to Design and Confirm Perforating/Propellant
BH Big hole guns Behavior and Formation Fracturing. Paper SPE 56434 presented at the SPE
Annual Technical Conference and Exhibition, Houston, Texas, 3-6 October.
ID Pipe inner diameter
in. Inches
Lp Perforation length
mD Millidarcy unit
MD Measured depth
OD Pipe outer diameter
OMS Odorless mineral spirits
Perm Permeability
psi Pound square inches (pressure)
SPF, spf Shot per foot
PBTD Plug back total depth

10
Porosity Red uction’s Impa c t on Pulsed Neutron Sa tur a tion

Advancements in Carbon-Oxygen
on-board (POB) constraints. Pulsed neutron tools
in the Mars field are often run continuously in
excess of 40 hours to achieve the objectives of the

Surveillance of the Deepwater Gulf of logging program due to the number and thickness
of reservoirs being surveyed. Since other logs,

Mexico Mars Waterflood like formation compaction and production logs,


are often combined with pulse neutron logging, a
typical multiwell campaign can last 10-14 days.
Matt Cuttitta and Jeff Weiland, Shell; Suparman, Phil Fox, Ismar Setiadi, Halliburton
Presented at the SPWLA 52nd Annual Logging Symposium held in Colorado Springs, CO, May 14-18, 2011
Copyright 2010, held jointly by the Society of Petrophysicists and Well Log Analysts (SPWLA) and the submitting authors
Over the years, methods of spectral data
assurance have been developed to ensure the
highest level of data quality. Log analysts have
Abstract refined these practices through years of pulsed
The primary purpose of the Mars (Mississippi Canyon 807) waterflood in the deepwater of the Gulf of neutron experiences, particularity C/O. These
Mexico (GOM) is to increase recovery efficiency in three main reservoirs. A robust surveillance logging practices include quality assurance during
program has been conducted in the field since 1996. Measurements have included strain, flow profile, acquisition and post-acquisition. Real-time quality
reservoir layer pressure, casing inspection, and multicomponent fluid saturation evaluations. The assurance allows immediate repeat logging if
fluid saturation surveillance results derived from log measurements have historically been based on needed to ensure sufficient data are acquired.
original porosity estimates. More recent analysis techniques have included reductions in porosity due to
compaction to better understand fluid distributions and eliminate anomalous saturation indications. The Surveillance of the Mars field is critical. The team
newly developed modeling can observe and more accurately identify injected seawater at the observation has taken, or plans to take, the following actions
wells. This fluid identification is accomplished through recently developed interpretation methods using as a result of the surveillance data:
pulsed neutron sensors including inelastic carbon-oxygen (C/O mode) and capture sigma (Σ mode). The
method is demonstrated with several Mars examples including discussion of interpretations. 1. Reduce injection pressure in one of the injectors
due to non-uniform sweep.
Surveillance of the Mars field is critical to optimizing recovery. The cased hole logging data are integrated 2. Re-drill an injector due to preferential injection
with other subsurface data, including 4D seismic, additional openhole logs from new wells, production into one layer.
data, and injection data. This dataset allows the team to properly operate existing wells, identify 3. Attempt different completion strategies,
additional opportunities, and plan future activities. Carbon-oxygen and sigma surveillance has provided especially on injectors, to better understand
an understanding of the fluid changes within the reservoirs under the influence of compaction, water which is most successful.
injection and aquifer movement. Through experiences in several reservoirs, anomalous tri-fluid (oil, native
formation water and injected waterflood seawater) saturation estimates were identified. The proper Continued surveillance of the Mars field is
evaluation of multiple fluids is dependent on the porosity of the rock matrix system. After application of required to enhance recovery and profitability.
measured strain and porosity reduction modeling, estimates obtained were more representative than what
conventional analysis would derive. While log measurements investigate the near wellbore, reservoir Model Description
modeling can predict a compaction profile spanning from most dramatic in the near wellbore region to Fluid saturation modeling derived from pulsed
lesser impact further in the reservoir. Techniques are demonstrated that help identify fluid and porosity neutron measurements (MarsSat) was initially
changes in the volume observed by the pulsed neutron tool. developed for the Mars waterflood in 2004. The
basic model logic consists of two steps. The first
Selected Mars well log examples are described in detail to highlight the results of compaction and fluid step is calculating the oil volume from the inelastic
saturation with time. This discussion is focused on best practices learned during the 14-year surveillance carbon-oxygen (C/O) responses. Then the volumes
program. Included are highlights on how recent log responses have encouraged the development of more of formation water and injected seawater are
advanced methods of interpretation. In some cases the interpretation became problematic since log determined from formation capture sigma (Σ)
responses did not indicate a unique solution. The new techniques presented have provided more accurate measurements. A schematic of the basic model
multi-fluid volumetric evaluation and assisted in quantification of compaction estimates in wells lacking logic is displayed in Fig. 1. Historically saturation
strain surveillance. Potential added benefits from this recent method include fluid saturation evaluation computed with the first-generation MarsSat was
in complex lithology reservoirs with enhanced recovery. The new method better defines fluid types based on the original porosity evaluation derived
and allows a more independent determination of porosity reduction. The use of improved methods will from openhole log data.
hopefully increase interest in reservoir surveillance and allow waterfloods to be operated more effectively.
During more recent evaluations some reservoirs
­­Introduction are identified. While a regular logging schedule showed the unexpected presence of injected
Surveillance using pulsed neutron carbon-oxygen would be ideal, it is not always possible. seawater. The interpretation of injected seawater
(C/O) logging can be very useful in the GOM. This Opportunities must be prioritized based on their was disputed by geologic models, waterflood
is a large part of the Mars surveillance program. value, access to the well, which can be restricted reservoir models, and production data. These
Each year a number of surveillance opportunities due to rig and construction activities, and persons- opinions were based on reservoir simulations and

11
Po r o s i t y R e d u c t ion’s Impact on P ul se d Ne utr on Saturation

OPEN HOLE C/O ∑ OH

SAND SHALE Voil (C/O) Vfw (∑) Visw (∑) BVI

OPEN HOLE VOLUMETRICS


Sand Volume
Shale Volume
Total Porosity
Effective Porosity
Original Water Saturation

C/O OIL EVALUATION


Bulk Volume Oil
Bulk Volume Total Water
Oil Saturation
∑-BASED WATER CLASSIFICATION
Bulk Formation Water
Bulk Seawater
Formation Water Saturation
Seawater Saturation Figure 3. Formation water has displaced oil in the
M2 sand in Mars A-15 ST01. A sigma comparison
between the 2006 and 2010 surveys is displayed in
Figure 1. First-generation MarsSat processing logic to determine oil, formation water and injected the far right track.
seawater volumes.
and oil (Σoil=19.6 cu). With such a large contrast,
the position of the surveyed interval relative to the
sigma (Σ) logging is an established GOM method
injector and other nearby wells, some of which
for evaluation of reservoir saturations when
serve as observation wells through reservoir layers
dealing with these two fluids. If another fluid,
of interest.
such as seawater (Σisw=34 cu), becomes a part of
the environment, it is no longer a simple matter
These discussions are common among the multi-
to evaluate fluid saturations with sigma alone.
disciplinary subsurface team. Review is conducted
The equations below demonstrate the MarsSat
at several stages validating provisional and
water classification logic. It determines volumes
revised modeling of the log responses. A series
of formation and injected seawater after an oil
of meetings are held during the interpretation
saturation has been derived.
to better understand and communicate the data
and assumptions. Sharing of other data and
Σtotal =
interpretations, such as reservoir simulation
ΣsandVsand + ΣshaleVshale + ΣoilVoil + ΣfwVfw +ΣiswVisw
results, helps drive a deeper understanding of
the log responses. Figure 2. Cross-plot of sigma log responses and 1 = Vsand + Vshale + Voil + Vfw + Visw
volume of oil derived from C/O under constant
porosity and a single fluid replacing oil.
A cross-plot of the first-generation MarsSat The oil volume (Voil) is determined from the C/O
fluid saturation modeling is displayed in Fig. 2. formation water (blue) or injected seawater (black), measurement, sand (Vsand) and shale (Vshale)
The basic saturation model boundary lines are replacing oil with no change in porosity over time. components are obtained from the openhole
for a fixed-formation porosity and a single fluid The actual log data are expected to fall within the evaluation. Exact parameter choices are made
replacing oil. envelope of these two lines for a given porosity consistently within geologic horizons as correlated
(diagram displayed in Fig. 2 is using a formation between well penetrations.
There are two boundary conditions indicated by porosity of 32 porosity units).
the blue and black lines. These represent the upper Fig. 4 is a cross-plot of C/O oil saturation and
and lower projected limits of sigma responses Typical Examples of MarsSat Analysis sigma in the M2 sand of Mars A-15 ST01.
cross-plotted with the independent oil saturation Formation Water Replacing Oil: The porosity used was from the openhole log
determined from C/O. Mars A-15 ST01 M2 Sand 2010 evaluation. In the figure, the data trend along the
Formation water replacing oil is a classic fluid upper boundary indicating the presence of oil and
The upper boundary condition is indicated by the substitution encountered. This occurred in the M2 formation water with no injected seawater.
blue line, which defines oil and formation water sand in Mars A-15 ST01 as displayed from the
as the only fluids. The lower boundary condition is 2006 and 2010 surveys in Fig. 3. Seawater Replacing Oil:
indicated by the black line, which defines oil and Mars A-10 ST01 G4 Sand 2010
injected seawater as the only fluids. This is the There is a significant contrast between capture Seawater was detected in the lower portion of
expected sigma response for a single fluid, either cross section values of formation water (Σfw=80 cu) the G4 sand. The G4 is not currently, nor has been,
12
A-10 ST01 well, another pass-through well, or an is located down dip of most pass-through wells.
injector well. However, we feel this is unlikely Typically this sand is logged in conjunction with
due to cement isolation and the location of A-10 deeper sands penetrated in the same wellbore.
ST01. Further investigation including additional
surveillance logging is ongoing to help explain
this response.

A crossplot of sigma and C/O oil saturation of


the above example with 10 pu removed from the
original porosity is displayed in Fig. 6. The section
16,270’-16,252’ is displayed to highlight the
interval having the most detected seawater. The
data trend along the injected seawater boundary
suggesting the water component is injected
seawater rather than formation water.
Figure 4. Crossplot of C/O oil saturation and
sigma in Mars A-15 ST01 M2 with the original
openhole porosity.

a waterflood reservoir. This apparent seawater


breakthrough was unexpected. The team did
sensitivity analysis to see if the interpretation Figure 7. Mars E2 modeled 2010 water saturation
map with well penetrations.
could be a result of porosity reduction rather than
seawater influx from injection. The composite Baseline formation compaction logs were run in
profiles in the right portion of Fig. 5 show 1996 in the E sands. Some wells have as many as
computed saturations using original porosity eight runs monitoring the compaction based on
and reductions of 5 porosity units (pu) and 10 pu strain observed in the casing. Pulsed neutron C/O
moving to the right. saturation logging began in 2004 and subsequent
surveys were conducted in the years since in
various wells. Several wellbores were logged in
order to establish reference saturation logs in the
E sands allowing more accurate differentiation of
future fluid changes. The wells are highlighted in
Fig. 7 and comprise a broad span of observation
Figure 6. Crossplot of C/O oil saturation and sigma points for monitoring waterfront advancement.
in Mars A-10 ST01 G4 with a porosity reduction of
10 pu from the original porosity. No Saturation Change:
Mars A-14 ST03 E2 Sand 2006
Waterflood Surveillance in Mars A-14 ST03 was logged in 2006 to determine
Mars E2 Reservoir saturations of oil, formation water and seawater.
Introduction to Mars E2 Reservoir This well is one of the closest to the A-18 ST02
Figure 5. Mars A-10 ST01 G4 saturation profile run
in 2010 showing unexpected fluids. The analysis is The Mars E2 reservoir is a well-sorted, fine- injector. With limited injection time and volume,
based on original porosity in track 4, 5 pu reduc- grained, high net-to-gross Pliocene sandstone seawater advancement was not expected to reach
tion in track 5, and 10 pu reduction in track 6. with an average porosity of 33 pu. The log this well.
character is two massive sand lobes separated by
A reduction of 10 pu from the original porosity a shale break. It has been penetrated by more than MarsSat results from the 2006 E2 sand pulsed
is extreme and beyond what can be expected in 20 wells in the area and is currently the shallowest neutron survey are displayed in Fig. 8. The final
these sands. However, with a reduction of 10 pu, injection horizon in the Mars waterflood. Fig. 7 step in the waterflood saturation interpretation
there is still an indication of seawater present. is a saturation map of the E2 including well method is facilitated with the MarsSat processing
penetrations. Production began in 1997 and two model. All saturation log displays are from
The presence of seawater may be an artifact of producing wells, Mars A-1 ST01 and Mars A-18 this method. The displays include an openhole
the acquisition due to 1) insufficient environmental ST02, produced 52.3 MMBO and 25.3 BCFG to lithology-porosity evaluation and gamma ray in
corrections, 2) incorrectly assumed lithology and/ date. Current production is 7,900 BOPD. In 2005, track 1. Moving to the right is an overlay envelope
or mineralogy, or 3) an unknown shortcoming of the A-18 ST02 well was converted to an injection of C/O ratio with water-saturated formation and
the analysis and interpretation. Seawater may be well. Current injection is 12,000 BWPD with a oil-saturated formation boundaries. The next track
present due to a flow path behind casing in the total injection of 25.5 MMBW to date. The injector is C/O-derived oil saturation shaded green. Next is

13
Po r o s i t y R e d u c t ion’s Impact on P ul se d Ne utr on Saturation

saturation evaluations for the E2 sand with 2004


C/O baseline, 2006 and 2010 surveys displayed
from left to right.

Figure 9. Mars A-10 ST01 E2 sand monitor


saturation profiles run in 2005 and 2010 confirming
expected fluids.

the nature of pulsed neutron measurement


statistics, variations in measurement resolutions
Figure 10. Mars A-4 E2 saturations in 1996,
between openhole resistivity logs, openhole 2002, 2004, 2006 and 2010 showing no change
Figure 8. Mars M-14 ST03 E2 in 2006 showing no
change as expected. porosity logs and cased hole pulsed neutron logs. as expected.
Additionally there are bed boundary effects,
an overlay envelope of sigma with formation water Seawater Replacing Oil: Mars A-3 E2 Sand 2010
depth-matching effects, accuracy of lithology
and injected seawater boundaries, followed by the Evaluation with 32 pu Porosity
and porosity estimates, and the ability to exactly
formation water saturation derived from sigma. Mars A-3 is located further updip than the first
environmentally correct the pulsed neutron
The panels on the right are the fluid saturation three examples. The logging objective was to
measurements. The environmental correction
volumes for oil, formation water, and injected provide a reference baseline for future logging.
depends on borehole geometry including well
seawater, if present. The time progression is Looking at the well position in Fig. 7, the first
trajectory, formation washout, relative positioning
earliest on the left to most recent on the right for three well examples (A-14 ST3, A-10 ST1, A-4)
of casing to borehole center, completion tubing
monitor and surveillance displays. are between A-3 and the A-18 ST02 injection well.
relative position to casing and borehole/formation.
There was unchanged saturation in these wells,
Fluid consistency within the completion and within
The evaluation of Mars A-14 ST03 in the E sand so the expectation for Mars A-3 was the same.
the casing-completion annulus is also of concern.
showed very little saturation change compared to Fig. 11 shows that this was not the case.
This consistency can be affected by deposits
when the well was drilled. No injected seawater such as scale and drilling fluid solids. These
was detected, which matched the expectation. factors are accounted for as much as possible
based on the pulsed neutron spectral responses
No Saturation Change: and previous experiences.
Mars A-10 ST01 E2 Sand 2005-2010 Monitor
Monitor pulsed neutron logging was conducted No Saturation Change:
in the Mars A-10 ST01 in 2010 to provide a Mars A-4 E2 Sand 2004-2010
time-lapse snapshot to compare with the 2005 One of the most frequently surveyed wells with
survey and serve as an additional baseline for pulsed neutron measurements is Mars A-4. This
future logging. This well is situated farther from is due to its proximity to the A-12 injector in the
the injector and more updip than A-14 ST03. The M sand and the ability to log all three waterflood
expectation was that oil would be present without reservoirs in A-4.
seawater. Saturation evaluation from the 2010 run
Figure 11. Mars A-3 E2 saturation showing
is compared with the 2005 baseline pulsed neutron In the E2, reservoir models predicted oil without injected water replacing oil in the bottom of
survey in Fig. 9. seawater because A-4 is far from the A-18 ST02 the upper lobe when analyzed using a porosity
of 32 pu.
injector. The results were as expected with
Results from MarsSat were as predicted similar the slight seawater indication being within the
to the previous well example. Oil saturation The MarsSat calculations indicated a significant
measurement uncertainty.
was practically identical to the openhole profile, amount of seawater in the lower portion of the
and no injected seawater was detected. The upper lobe and nearly original saturation in the
Track 3 includes monitor sigma saturations
minor indications of seawater as illustrated by lower lobe. This result prompted an assurance
spanning from the 1996 baseline to the current
the light blue in the two right tracks are within review of the modeling method. Data quality
survey. The right most 3 tracks in Fig. 10 show
the measurement uncertainty. This is due to was reviewed, and environmental conditions
14
were investigated looking for potential space contains oil, formation water, and no displayed in the color coded key on the right border
contributions to erroneous data. Processing injected seawater. of the display. Well locations are labeled with
parameters were reviewed and compared to the value extracted from the model. The porosity
other wells in the E sands. No error was found Mars E2 Sand Porosity Reduction (Compaction) displayed was used to analyze logs run in 2010.
in the modeling method. Map and Well Locations This porosity reduction is being estimated for all
As a result of production, the average reservoir reservoir layers in order to utilize a more accurate
Discussion then focused on porosity reduction pressure in the E2 has reduced approximately 3000 porosity for pulsed neutron log analysis. There is
because the reservoir is highly compressible. psi. This reduction in pore pressure means more one formation compaction survey in A-4 in the E2
Processing was conducted over intervals of of the overburden must be supported by the rock sand that helps calibrate the estimate. It measured
interest by removing fixed values of porosity as an matrix frame since less is supported by the fluid in a maximum strain of 3.5%.
initial quick-look method to verify basic modeling the system. This phenomenon is described as an
logic. With a reduction in porosity, there is an increase in vertical effective stress. Since the rock
associated increase in matrix contribution matrix frame is now subject to greater stress, it
(Σsand = 9 cu) to the bulk measured formation compacts, reducing its porosity and thickness.
sigma (Σ) capture cross section. With an
increasing matrix component, sigma will decrease The Mars subsurface team has taken an integrated
in a constant fluid system with a reduction of approach to understand the severity of this
porosity. If not properly accounted for, the model compaction. The compaction model, displayed in
determines a false increase in seawater to Fig. 13, is based on whole core measurements,
compensate for the lower porosity. Seawater has sidewall cores, formation compaction surveys,
a relatively low sigma capture cross section value ultrasonic imaging and casing inspection logging,
(Σisw = 34 cu) and will be substituted for formation acoustic properties of the formation measured by
water to balance the porosity reduction effects. LWD and wireline tools, GOM trends, and reservoir
This occurs when the porosity used in the model simulations. This is accomplished by a common
is too large. understanding by reservoir engineers, geologists,
petrophysical engineers, and research physicists.
Oil saturation derived independently from the The effects of this compaction are significant. Figure 14. Mars E2 modeled 2010 porosity
C/O responses is cross-plotted with sigma in Compaction can damage pass-through wellbores reduction map with well penetrations.
Fig. 12 as a composite presentation for the two and producing wellbores. It can make drilling
sand lobes displayed in Fig. 11. depleted sands difficult and also can cause No Saturation Change: Mars A-3 E2 Sand 2010
a larger uncertainty in production rates. The Evaluation with 27 pu Porosity
seafloor below the Mars TLP has subsided The Mars A-3 E2 original interpretation showed
approximately seven feet, likely as a result of seawater breakthrough in Fig. 11. A quick
formation compaction. As this paper describes, pass removed 5 pu from the original porosity
it also complicates openhole and cased hole to understand the sensitivity in porosity. The
log interpretation. However, compaction is not recomputed saturations from this exercise are
always a negative effect. The compaction provides displayed in Fig. 15. The evaluation revealed a
reservoir energy resulting in higher production significant reduction in the amount of seawater
rates and better recovery efficiency. with the 5 pu porosity reduction compared to
the original evaluation. The small volume of
seawater was within the measurement uncertainty
and could be disregarded as an indication of
advancement of the seawater flood front in
this wellbore.

The team considered developing a more complex


Figure 12. Cross-plot of C/O oil saturation and evaluation method to help identify differences
sigma in Mars A-3 upper (black circles) and
between injected seawater and porosity reduction
lower (blue circles) lobes of E2 with the original
openhole porosity. effects. To this point in time the first six years of
Figure 13. Compaction model for the E2 sand
representing porosity reduction as a function of multi-mode tri-fluid pulsed neutron interpretation
The data points from the upper lobe in black reservoir pressure depletion. had been based on using the original openhole
along the lower boundary indicate the pore space log based porosity. This new evaluation method
contains oil, injected seawater, and no formation An estimate of porosity reduction from initial is now being regularly used by the team and is
water. The data points from the lower lobe in production to 2010 for the E2 sand is displayed described in more detail later in this paper.
blue along the upper boundary indicate the pore in Fig. 14. Scaled porosity reduction in pu is

15
Po r o s i t y R e d u c t ion’s Impact on P ul se d Ne utr on Saturation

Mars M1/M2 Sand Porosity Reduction


(Compaction) Map and Well Locations
Similar to the E2 sand, a compaction model was
also developed for the M1/M2 reservoir and is
displayed in Fig. 17.

Figure 15. Mars A-3 E2 saturation showing Figure 19. Mars A-15 ST01 MarsSat in the M1
seawater replacing oil when analyzed using a sand showing seawater break-through as
porosity of 27 pu. observed in 2010.
Figure 17. Compaction model for the M1/M2 sands
Waterflood Surveillance in Mars M1/M2 representing porosity reduction as a function of A crossplot of C/O oil saturation and Σ in the
Reservoirs reservoir pressure depletion. Mars A-15 ST01 is displayed in Fig. 20. The
Introduction to Mars M1/M2 Reservoirs data for the selected interval, 17,544’-17,515’,
The Mars M1/M2 reservoirs are late Miocene An estimate of porosity reduction from initial trend along the lower boundary indicating the
sandstone deposits with an average porosity production to 2010 for the M1/M2 sand is presence of oil and injected seawater with no
of 28 pu, lower net-to-gross than the E2, and displayed in Fig. 18 below. Scaled porosity formation water.
often consists of many thin layers with shale reduction in pu is displayed in the color-coded
breaks in between. A number of wells in the key on the right border of the display. Well
area have penetrated the M1/M2, and it is the locations are labeled with the value extracted
most frequently surveyed waterflood reservoir from the model. The maximum strain measured in
in the Mars field. Production began in 1997, and wells with formation compaction logging is also
four producing wells, Mars A-12, Mars A-14, displayed. There are two formation compaction
Mars A-15 ST01 and Mars A-14 ST03, produced surveys in A-4 and A-10 ST01 in the M sand that
91.0 MMBO and 87.6 BCFG to date. Current help calibrate the estimate. The surveys measured
production is 7,000 BOPD. In 2004, the A-12 a maximum strain of 5.5% and 6.3%, respectively.
well was converted to an injection well. Current
injection is 12,000 BWPD with a total injection
of 15.4 MMBW to date.

Three wells in the M1 layer are displayed in


the following section as examples of injected
seawater sweep. Fig. 16 is a water saturation
map from a reservoir simulation. There is a key of
saturation values related to the color shading in
Figure 20. Crossplot of C/O oil saturation and
the map on the right-hand border.
formation capture cross section in Mars A-15
ST01 M1 sand.

Unfortunately this is the only C/O log run in Mars


Figure 18. Mars M2 modeled 2010 porosity
reduction map with well penetrations. A-15 ST01 well to date. A change in log response
over time yields more confident interpretation. This
Seawater Replacing Oil: particular log was run to obtain a reference log for
Mars A-15 ST01 M1 Sand 2010 comparison to future log runs. The following two
MarsSat saturation results for Mars A-15 ST01 examples illustrate this, and they highlight the
are displayed in Fig. 19. The well is the second- C/O oil saturation decline over time.
closest well to the M1/M2 injector.
Seawater Replacing Oil:
Figure 16. Mars M1 modeled 2010 water saturation A large portion of the sand has been swept of oil Mars A-4 M1 Sand 2004-2010 Monitor
map with well penetrations.
and filled with seawater at the time of the 2010 This is the most frequently surveyed well in the
log. The majority of oil in the lower section has Mars M sand, primarily due to its proximity to the
been flushed to a nearly residual saturation. injector, A-12. The baseline saturation log was run
16
in 2004 coinciding with the startup of the Mars Seawater Replacing Oil: Mars A-10 ST01 M2.1
waterflood. This survey provides an excellent and M2.3 Sands 2005-2010 Monitor
reference for all future surveillance logging. An example of seawater replacing oil is displayed
Fig. 21 displays the time-lapse runs for in Fig. 23. The surveillance shows decreasing
Mars A-4 to date. oil volume between 2005 and 2010. The earlier
2005 baseline survey shows minor injection influx
in the lower portion of the M2.3 sand. The latest
2010 evaluation indicates the entire M2.3 and
the majority of the M2.1 sand have been swept
with seawater.

Figure 24. Cross-plot of C/O oil saturation


and sigma in Mars A-10ST01 M2.1 shows a
combination of porosity reduction and injected
Figure 21. Mars A-4 M1 saturations in 2004, water responses.
2006, and 2010 showing an increasing volume
of seawater.
The upper half of the chart represents a formation
Figure 23. Mars A-10 ST01 M2.1 and M2.3 undergoing formation water sweep and the lower
The second run in 2006 was the first injection half represents a formation undergoing porosity
saturations in 2005 and 2010 showing an
observed by pulsed neutron logging in the Mars increasing volume of seawater. reduction. Different cases of oil volume and
field. A later run in 2010 is displayed on the far formation water saturation are displayed in the
right of the display, and data from this run are The previous logs demonstrate that the relation- middle of the chart. The column on the right of the
displayed in Fig. 22. ship between C/O and sigma studied so far can chart is the summation of individual components
reasonably determine the conditions when a of the modeled cases shown as ΣTOTAL.
The MarsSat processing and associated water significant amount of injected seawater is
classification boundaries in the cross-plot are present in the log responses.
constructed using the porosity reduction drawn Cases for Original Porosity
from the simulation map in Fig. 18. The cross-plot When the data trend along the boundaries, the TPOR Voil SFW ΣTOTAL

supports the analysis, which indicates seawater is interpretation is well defined. The data from 32 PU & 90% Oil 0.32 0.2520 0.1000 17.95
present in 2010. A-10 ST01 M2.1 sand show a different trend. 32 PU & 25% Sweep 0.32 0.2100 0.2500 20.47
The points fall in the area between the formation
32 PU & 50% Sweep 0.32 0.1400 0.5000 24.67
water and injected seawater boundaries, as
Cases for Porosity Reduction (Compaction)
displayed in Fig. 24.
TPOR Voil SFW ΣTOTAL

Here the logs are responding to a combination of 32 PU & 90% Oil 0.32 0.2520 0.1000 17.95
porosity reduction and a mixture of water types. 30 PU & 90% Oil 0.30 0.2340 0.1000 17.61
Likely the seawater injection front is transitioning 28 PU & 90% Oil 0.28 0.2160 0.1000 17.27
through the original formation water. In this
instance, there is not a clearly defined set of Figure 25. Modeling for formation water sweep
(upper) and porosity reduction (lower) cases. Right
conditions the pulsed neutron responses can column displays the bulk sigma for cases listed in
deliver to separate the porosity reduction from left column. Center columns indicate oil volumes
saturation change. Fig. 25 compares the modeled and formation water saturations for each case.
responses due to water sweep and due to
porosity reduction. There are different sigma (ΣTOTAL) values for both
cases, a reservoir undergoing formation water
The left column indicates fluid-content of a series sweep and a reservoir undergoing compaction.
of cases including original oil-filled reservoir, Fig. 26 demonstrates this by showing the change
water sweep, and porosity reduction conditions. in ΣTOTAL values for each case.
Figure 22. Cross-plot of C/O oil saturation and
sigma in Mars A-4 M1 for the latest surveillance
run in 2010.
17
Po r o s i t y R e d u c t ion’s Impact on P ul se d Ne utr on Saturation

35 section contrast between low-salinity seawater


∆Θ=3pu-ΔΣ=2cu
Oil Sweep Compared to Porosity Reduction
30
and hydrocarbons, there is a significant contrast
SIGMA MODEL W/O WATER INJECTION
25 between water and hydrocarbons in the C/O
∑ TOTAL
20
15
ORIGINAL SIGMA POINT
measurement. However, this analysis is sensitive
10
to the porosity as demonstrated by this paper.

Sigma
5
0
Further complicating the situation, the Mars sands
SIGMA WITH FORMATION
17.95 20.47 24.67
WATER SWEEP
SIGMA WITH FORMATION
are typical of many deepwater GOM sands and are
17.95 17.78 17.61
COMPACTION
highly compressible resulting in porosity reduction
0.0
1.0 Silver So 0.0 with depletion. For a robust interpretation, the best
Figure 26. Modeled sigma (ΣTOTAL) values
Silver
0.0 Sw 1.0

increase for a reservoir undergoing formation estimate of current porosity must be used.
water sweep (blue line) and slightly decrease for Figure 28. Red line represents modeled data
a reservoir undergoing compaction (red line). trend resulting from formation water replacing Measurement uncertainties and limitations of
oil (no injected water) and porosity reduction analysis methods are very important. Invalid
For sigma interpretation, it is difficult to confuse from pressure decline.
interpretations due to uncertainty mask the
porosity reduction and formation water sweep. The
significance of valid interpretations. Additional
measurement is robust even without correcting for From sigma and C/O modeling, a series of
work is ongoing to 1) better understand the
porosity reduction. However, the same is not true curve families can be developed as displayed in
uncertainties and 2) improve the analysis methods
of C/O interpretation as displayed in Fig. 27. Fig. 28. These curve families are consistent with
to deal with these uncertainties.
the observed data trend displayed in Fig. 24.

Oil Sweep Compared to Porosity Reduction The Mars field is a world-class oil field with a
These log responses are being studied further to
0.30 world-class infrastructure. The waterflood is one
0.25 develop a deeper understanding of the data trends.
C/O 0.20 method to improve hydrocarbon recovery and
Presently it is unclear why the data sometimes has
VOLUME 0.15
help maintain reservoir integrity with pressure
OF OIL
0.10
the shape of these curve families, but additional
0.05 maintenance. Surveillance of the waterflood
0.00 work is in progress to understand these responses.
Voil WITH
is critical to determine where the seawater is
WATER SWEEP 32 PU & 90% OIL 32 PU & 25% SWEEP 32 PU & 50% SWEEP

Voil WITH FORMATION


and to predict where it is going. Pulsed neutron
32 PU & 90% OIL 30 PU & 90% OIL 28 PU & 90% OIL Conclusions
COMPACTION
sigma and C/O logging is one tool to monitor
Surveillance using pulsed neutron sigma logging
that movement. As illustrated in the paper, C/O
has proved extremely valuable in the GOM. This
Figure 27. The volume of oil decreases for a logging is complicated with inherent uncertainty.
reservoir undergoing water sweep (blue line) is due to the large capture cross section contrast
The uncertainty varies depending on sand quality,
and also decreases for a reservoir undergoing between high-salinity formation water and
compaction (red line).
porosity estimates, and the wellbore environment.
hydrocarbons. By assuming original rock properties
Reservoir models, baseline C/O and sigma logs,
from openhole logs, saturation changes can be
and compaction data reduce the uncertainty and
For C/O interpretation, it is easy to confuse easily calculated and the influx of formation water
should be obtained early in field surveillance.
porosity reduction and water sweep. This is is easily discernable. The measurement is very
because C/O measures the volume of oil and robust, and examples in the paper illustrated the
Through enhanced multidisciplinary teamwork,
in both cases the volume of oil decreases. The fact that reduced porosity has little effect on
reservoir surveillance can lead to an increased
measurement must be corrected for porosity the analysis.
understanding of aquifer water movement,
reduction in order to be robust. In situations
waterflood water movement, injection perform-
encountered like these, bringing together Surveillance using pulsed neutron carbon-oxygen
ance, and compaction. This understanding allows
multiple disciplines yields a better interpretation. (C/O) in conjunction with sigma logging is also very
operators to better develop existing fields and
Understanding porosity reduction from other useful in the GOM. The Mars waterflood began in
use these learnings to understand what should be
methods is critical to developing a multi-fluid 2004 and included three reservoirs: E2, M1/M2,
expected of new fields.
interpretation with confidence. Currently work is in and N/O. This introduced a third fluid, seawater,
progress to refine porosity reduction maps for all to the existing hydrocarbons and formation
reservoirs. All existing pulsed neutron evaluations water. Although there is not a large capture cross
will be revised with these adjusted porosity
estimates. The team will review new modeling
results and apply them appropriately.

18
Authors
Matt Cuttitta is a Petrophysical Suparman is a Lead Log Analyst Ismar Setiadi is a Log Analyst
Engineer with Shell Exploration for Halliburton in the Formation and for Halliburton in the Formation
& Production Company in New Reservoir Solutions group. He holds and Reservoir Solutions group. He
Orleans, LA. He is a member of a BS in physics from the University has been in the oil industry since
SPWLA and SPE, and he holds BSE of Indonesia. Suparman has more 1973, when he joined ELNUSA in
and MSE degrees in mechanical than 27 years of experience in the Petro-Service group in Jakarta
engineering from the University of Michigan. His current the wireline logging industry, starting with Gearhart Indonesia. He transitioned to log analyst in 1984 with
responsibilities include petrophysical evaluation, as a field engineer. He joined Halliburton in 1987 and Gearhart and later Halliburton. He held assignments in
operations support and reservoir modeling support for entered the evaluation group in 1994. His experience various areas of Indonesia and Thailand before joining
the Mars field. is in various areas, including Indonesia, Australia, and the Gulf of Mexico team in 2007.
most recently in the Gulf of Mexico.
Jeff Weiland is a Senior Staff
Petrophysical Engineer with Phil Fox is a Technical Advisor
Shell Exploration & Production with Halliburton supporting
Company in New Orleans, LA, formation evaluation job design
currently working the Mars field. and interpretation for the Gulf
He joined Shell after earning a BS of Mexico. He holds a BS in
in geological engineering from the South Dakota mechanical engineering from the
School of Mines and Technology in 1985. He later University of Colorado, joining Halliburton in 1981. His
acquired an MBA from Loyola University. Jeff has previous responsibilities include field engineering, log
worked extensively in primary development, mature analysis and technical support in Southeast Asia, South
development, and exploration in the GOM and off the America and the United States.
north coast of Alaska.

Acknowledgements
The authors thank the management of Shell, BP and Halliburton for their Eyvazzadeh, R., Nguyen, C., Fox, P., Poernomo, B. 1999. Nuclear Magnetic
support and approval to publish this paper. Resonance and Carbon/Oxygen Logging Improve Determination of Residual
Oil Saturation: A Case Study from the Minas Field. Paper SPE 54338-MS
presented at SPE Asia Pacific Oil and Gas Conference and Exhibition,
References Jakarta, Indonesia, 20-22 April.
Schutjens, P.M.T.M., Hanssen, T.H., Hettema, M.H.H., Merour,J., de Bree, Fox, P.E., Adnyana, G., Setiadi, I. 1999. Applications of Carbon/Oxygen
P., Coremans, J.W.A., Helliesen, G. 2004. Compaction-Induced Porosity / Logging in Indonesian Reservoirs. Paper SPE 54353-MS presented at SPE
Permeability Reduction in Sandstone Reservoirs: Data and Model for Asia Pacific Oil and Gas Conference and Exhibition in Jakarta, Indonesia,
Elasticity-Dominated Deformation. SPE Res Eval & Eng 7 (3): 202-216. 20-22 April.
SPE 88441-PA.
de Kock, A. J., Johnson, T. J., Hagiwara, T., Zea, H. A., Santa, F. 1997.
van der Horst, J., Rambow, F., Frisch, G., Quirein, J., Fox, P. 2005. Formation GOM Offshore Subsidence Monitoring Project with a New Formation
Compaction Surveillance in Deepwater Gulf of Mexico: Applications, Compaction Monitoring Tool. Paper OTC-8529 presented at Offshore
Experiences and Best Practices. SPWLA 46th Annual Logging Symposium, Technology Conference, Houston, Texas, 5-8 May.
New Orleans, Louisiana.
Simpson, G., Fox, P., Chafai, N., Truax, J. 2001. Introduction Experiences
of a New High Accuracy Through-Tubing Pulsed Neutron Reservoir
Management Solution in Asia-Pacific. Paper SPE 68713-MS presented
at SPE Asia Pacific Oil and Gas Conference and Exhibition, Jakarta,
Indonesia, 17-19 April.

19
E xpand i n g t h e r ange of appl icat ions of t rad itional s lick line

New Slickline Smart-Release Tool


were improved, the technologies were adapted
to electromechanical slickline tools (Gazda et al.,
1995; McDaniel et al., 2008; Goiffon, J., 1996;

Mitigates Wire-Recovery Issues in Gazda et al., 1995)

Extreme Well Conditions Another limitation for slickline service for


many years has been the concept that slickline
intervention was applicable to vertical or near
Jack Clemens, Darrell Moore, John Schram, and Jerry Foster, Halliburton;
vertical wells only. Therefore, wells with depths
Johnnie Garrett, Shell
greater than 15,000 feet and deviations greater
Presented at the 2009 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana,
USA, 4–7 October 2009
than 70 degrees were believed to exceed the
Copyright 2010 SPE. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. capability of slickline well intervention.

One of the primary concerns when considering


Abstract
slickline well intervention in these deep or
Traditionally, when wells in extreme environments required service, slickline, which had always been
deviated wells was that the wire might break, and
considered as the most cost-effective service option, was unable to provide the services because the
this would leave thousands of feet of wire in the
available slickline tool options could not maintain integrity in the extreme well environments now
well that would require fishing operations that
being explored and developed. To address the requirements of the new scope of conditions in extreme
could result in considerable added expense. The
environments, new methods that would meet a greater scope of needs were developed. The innovations
fishing job could very well require coiled tubing or
include pre-job wire-tension modeling software and a state-of-the-art timed-release tool that provides
tubing fishing, which, in addition to service costs,
successful intervention in difficult scenarios that normally could not have been considered as slickline
could result in days or weeks of lost rig time and
candidates.
lost production.
This paper will discuss these developments and how they have enabled slickline to be a more cost-
This paper discusses pre-job preparation and new
efficient alternative for the operations that had to be performed by other service options when intervention
tools that have been designed to mitigate the
was required in extreme depths or in highly deviated wells.
above mentioned risks in challenging wells. Even
when all pre-job tasks are performed properly,
A tool that can be timed to release the wire from the tool string at the rope socket without surface
in some cases, there are still well conditions
intervention has been developed to reduce the risks associated with stuck strings and dropping of
that create obstacles that are beyond control
conventional wire cutters. In addition, pre-job wire-tension modeling software is now available. How the
of normal operating windows and preparations.
newly developed slickline techniques can be used to mitigate the risks associated with the challenges
When the unforeseen occurs, the slickline tool
created by the extremely deep and highly deviated wells will be discussed along with the enhanced cost
strings can become immobilized and will require
efficiency and safety provided by these innovations. Implementation of the smart-release tool into the tool
fishing operations. While pre-job preparations help
string will also be discussed.
mitigate risks, they cannot foresee all situations,
and, therefore, cannot eliminate all risks.
­­Introduction cutting tools from successfully cutting the wire
Ultradeep and highly deviated wells have created when necessary?
Commercially Available Wire
many challenges when wells require slickline
Tension Software
services. Because of the depths and deviations This paper discusses the new methods and tools
Software programs that model wire, wire-tension,
now being attempted, service questions have that were designed to address the challenges
tool string, well parameters, and fluids are
surfaced as to the capabilities of slickline mentioned above, since the primary needs for any
commercially available. These programs are run
to perform adequately in the more extreme operation are to reduce non-productive time and
in advance of running the slickline service and are
environments. There are many issues that have operating costs as much as possible.
capable of predicting the wire tension throughout
been raised along with questions, such as:
the wireline trip into and out of the well. Pre-job
1. Can the tool string reach target and be History of Slickline Capabilities in
wireline-tension models greatly reduce the risk
pulled out? the Oil Field
of ‘over pull’ on the wire. In conjunction with
2. Can work be performed at the target depth? Slickline well intervention has been the industry’s
pre-job wire-tension modeling, pre-job wire
3. What happens if a tool string becomes stuck in basic low-cost well intervention service for many
management also reduces wire breakage risk.
a deep or deviated wellbore? years (Larimore et al., 1997a; Larimore et al.,
Eddy current inspection of the wire prior to running
4. How can the tool string be released at the rope 1996; Larimore et al., 1997b; Foster et al., 2001).
it into the well detects damaged and improperly
socket to prevent costly fishing operations? Until recently, well intervention conducted on
manufactured wire and eliminates this wire from
5. Can the entire length of wire be successfully slickline was limited to mechanical surface-wire
being used for well services. Wire-management
recovered if the tool string becomes stuck? manipulation of downhole tools. As battery
software programs that record and locate highly
6. Might deviations prevent conventional wire- technology and packaging of electronic equipment
fatigued and over-stressed wire locations predict if

20
the wire will be capable of successfully performing
additional jobs.

Fig. 1 is an example of a data chart prepared with


wireline-tension software. The chart has four
lines. The red line is the “compression” line that
portrays the lowest wireline tension without travel
stopping. The green line is the “run in the hole
(RIH)” line. The values on this line can be validated
by comparison to the RIH hanging weight. The
blue line is the “pull out of the hole (POOH)” line
that portrays the pick-up weight. Its values can
be validated by frequently comparing the weight
indicator pick-up weight. The vertical line on the
right side of the chart is the maximum pull of the
slickline based on the pre-job safety factor.

History derived from wireline tension programs


has shown that data produced are very close to the
actual recorded tensions.

Wire Breakage Figure 1. An example of a data chart prepared with wireline-tension software.
Slickline operations without proper pre-job
of the tubing bore, and thus, gravity-fed wire Smart-Release Tool. The smart-release tool was
planning in challenging wells can increase wire
cutters are prevented from traveling down the designed and developed after several discussions
breakage risks, since wells with high deviations
entire length of wire to the target location. Other with Shell in the Gulf of Mexico. Shell was
often will increase the drag/tension of the wire.
pressure-and-tension-releasable rope sockets are instrumental in helping to assess the needs for a
Operators without pre-job wire-tension modeling
also available, but they require a force or pressure tool that could mitigate the risks associated with
capabilities have only the surface load cell to
applied at the rope socket. The traditions options running slickline into deeper, more deviated wells.
monitor. High deviations and extreme depths may
are discussed below, along with a new option that The goal of the project was to develop a tool that
cause readings at the surface load cell that may
was developed recently to address the difficulties would decrease the risk associated with dropping
confuse inexperienced operators. When the tool
that occur when a tool string becomes stuck. tools to cut the wire.
string becomes stuck, jarring may be applying
abnormal loads and stress at the tool string. If the
Slickline Wire-Cutting Tools. When the slickline The smart-release tool was developed as an
tool string is inadvertently stuck, the best scenario
tools could not be retrieved to the surface, the automatic-release mechanism to release the
is to release the wire at the rope socket on top
traditional option was to cut the wire. One of the slickline at the top of the tool string without
of the tool string and fish the tool string using
cutting tools used for slickline wire cutting is a manual manipulation from the surface. Deepwater
conventional fishing methods. Causes of wire
slotted stem tool with a fishing neck. This tool and deviated wells are considered risky appli-
breakage can usually be grouped into the
can be attached to the slickline via a small strip of cations for conventional gravity-fed/surface-
following classes:
metal pinned in the slot to keep the wireline from dropped wireline cutting tools, and therefore, more
• Excessive Tension – line tension exceeds coming out. When it is dropped from the surface, reliable contingency alternatives were needed.
wire yield point it will slide down the wire until it hits a restriction
• Wire fatigue – wire had bent too many or the top of the rope socket. The tool will cut The smart-release tool is electromechanical and
times in same location the slickline at that point, allowing the slickline is operated by a timer-based release mechanism.
• Chemical-induced failure – well fluid to be retrieved. This tool is battery-powered and is pre-job
attacks wire programmed to release the slickline at the rope
• Weak-point failure – wire has been If the tool cannot reach the rope socket at the top socket after a predetermined time has lapsed.
mechanically damaged at a high-stress point of the tool string, the wire may be cut at another The time allowed depends on the slickline
Any of the above conditions may cause the point, but this will often leave a long length of operation being performed. When the time has
slickline to break during operations. Traditional wire remaining above the rope socket. With the lapsed, the timer powers an electric DC motor that
wire-cutting methods have limitations. Wire challenges facing the operators in the deeper well retracts a ramp that supports the release dogs.
cutters that transverse the wire using gravity scenarios, it became obvious that a new concept
have shown only limited success in wells with that would lessen the current operational risks Various surface timers are available, and the
high deviations. In highly deviated wells, the wire was needed. selection of these timers is based on job-specific
is commonly stretched tightly against the inside parameters. The timers activate after normal

21
E xpand i n g t h e r ange of appl icat ions of t rad itional s lick line

slickline procedures have failed to retrieve the


stuck tool string. Currently, each tool is shipped
with three timers: one timer for 6 hours, one for
24 hours, and one for 48 hours, and as discussed
above, the appropriate timer is pre-job selected
to allow adequate operational time for the
well intervention service that is selected. The
activation of the smart-release tool results in the
retrieval of the complete length of wire from the Short Reverse Over
Circuit Voltage Voltage Over Heat Drop Rupture
well, and the top of the tool string/fish that is left
in the hole has an internal fish neck to facilitate
future fishing operations. Also, the tool is “fail Figure 2. Abuse testing on the rechargeable batteries.
safe” in that if there is any failure in the system;
were also subjected to temperature and vibration and assembly features that would help eliminate
i.e., battery, motor, timer, or any other part, the
with “Highly Accelerated Life Testing (HALT).” potential binding of the release piston were
tool will not release.
The nano-structured rechargeable batteries are incorporated into its design. Each size tool has
inherently safer than the lithium thionyl-chloride application in a variety of services, with both
One of the components that has been
batteries that are commonly used in the oil field. tools having been designed to mechanically
instrumental in the development of the new
These batteries exhibit no thermal runaway and release the tool string below the rope socket
release tool has been the improvements in
have an intrinsic overcharge protection. The cells without the need to mechanically manipulate
electromechanical power supplies. Over the
can be short circuited without damage to either from surface.
past several years, battery technology has
the battery or to the environment. Depleted cells
evolved, providing greater capacity and more
pose no disposal concerns; cells can be tossed The 1.88-OD tool is also an electromechanical
power for downhole electromechanical tools.
into general trash or can be recycled like any battery-driven tool with printed circuit board (PCB)
The smart-release tool provided the opportunity
other lithium battery. The improved safety of activation logic. As with the earlier smart-release
to design a tool that would take advantage of
the nano-structured rechargeable battery aids in tool, once the desired time lapses, the dogs
the advancements in battery capability. Several
their transportation as they can be transported as release from the fish neck adapter, which remains
types of batteries are available, and each has
carry-on luggage aboard aircraft or shipped by air. on the tool string in the well, and the rest of
distinct advantages. Alkaline batteries have high
the smart-release tool is removed from the well
energy storage rates, are commercially available,
The primary advantage of the type of release without operator action required to initiate the
and have no transportation restrictions, but they
mechanism that the smart-release tool provides is wire disconnect.
are limited by ambient temperature. Primary
that regardless of well depth, deviation, or other
(non-rechargeable) lithium batteries have a high-
well parameters, the separation of the tool string When the slickline intervention is accurately
energy density but require controlled disposal
occurs at the top of the slickline tool string, and pre-planned and a smart-release tool is used
after use, and there are concerns about
an internal fish neck remains in the well for future but an unforeseen catastrophic event does
their safety.
fishing. No external force or pressure is required not occur, the new tool will not function, and
to activate the release tool. In the event the tool therefore, the tool string can be run in and out of
Advancements in phosphate-based lithium
activates in the well, it can be reused, but it first the well without the tool string being released.
chemistries provided the opportunity to
must be disassembled and reassembled. Fig. 3 is a graphic of the smart-release tool,
implement rechargeable battery capabilities
which is placed below the rope socket and above
into downhole electromechanical tools. The
The first tool developed had a 1.69-in. OD. the stem. Fig. 4 shows the smart-release tool in
nano-structured rechargeable battery used in the
Recently, a new 1.88-in.-OD tool has been a wireline configuration.
smart-release tool is a high-rate cell that is good
released that was developed for heavier tool
for downhole operations where temperature is
strings. Because the tool would require an Fig. 4a shows the configuration of the tool
less than 130°C, and duration of its use will be
increased diameter and capabilities to support and how the components operate, and Fig. 4b
less than 2 weeks. The key features of the battery
heavier tool strings, some changes such as a shows how the tool looks during the actual
are high current, low internal resistance, long life
larger internal fish neck, a heavier-duty gearbox, running procedure.
cycle, and low cost.

Before implementation, the performance of the


nano-structured rechargeable battery had to
be established through extensive preliminary
testing. Tests that measured capacity, voltage
profile, cycle life, soak time, self discharge, and
hydrostatic crush were performed. The cells Figure 3. Smart-release tool.

22
Top External Fishneck Smart-Release Tool Lower Internal Fishneck
The complete electronic module, motor, and
battery were tested repeatedly while cycling
between room temperature and maximum
operating temperature to predict battery life.
Figure 4. Illustration of the smart-release tool in a wireline configuration.

Battery Motor Dogs Released

Dogs

Figure 4a. Tool operation.

Actual Testing Steps Performed for


Each Procedure
A. Pressure test to 10,000 psi
The fully assembled tool (without motor
and battery) was placed in a pressure test
chamber and pressurized with water to
10,000 psi. Pressure was held at 10,000 psi for Figure 5. Tension test.

Battery
5 minutes and then slowly released. The tool
was removed and inspected for leaks as well D. Logic validation test
as to see if the dog mechanism had moved. The basic logic for this tool is that it will turn
No leakage and no release of dog mechanism the motor on after a pre-selected time has
indicated a successful test. lapsed. The prototype PCB has only one time
Timer Module
setting. Changing one resistor on the PCB
B. Tensile pull test to 5,000 pounds creates another time setting. The logic was
Subsea Release (SSR) was mounted in the tested with different resistors to create all
Motor Gearbox
Instron 1332 tension tester (See Fig. 5) that 10 timer settings. Proper function of logic at
is located in the metallurgy lab. Tension was all times was considered successful.
Drive Motor
applied to the SSR mechanism in 1000-pound
increments until 5000 pounds was reached. E. Jar testing
Dogs Tension test is considered as successful if no Creating a jar test that would meet all field
damage occurred in the SSR after being pulled parameters was not possible. This jar test was
5000 pounds. performed to validate that the SSR tool would
operate after being jarred.
C. Temperature test to 300°F [150°C]
Electronic and battery assembly requires During jar testing, the SSR was located above
Figure 4b. Additional drawings depicting the extensive temperature testing to determine the jars. Therefore, the lower section of jars
smart-release tool during a release operation. the battery life and effects of temperature was attached with a vise. A 5-ft weight bar
on the components. Lithium batteries create was placed above the jars, and the SSR was
Testing
a passivation layer when operated at high placed above the weight bar.
The general testing to validate the design of the
temperatures, which is then difficult to
prototype tool is outlined below:
overcome when operated at room temperature. In the test for activating the logic; the stem
A. Pressure tested to 10,000 psi
Even though this assembly will normally is attached, and the SSR is attached above
B. Tensile pull tested to 5,000 lbf
operate at fairly high temperatures, it may be the stem. Then, the 1.5-in.-OD x 5-ft stem
C. Temperature tested to 300°F [150°C]
required to operate at room temperature after is attached. The 1.5-in.-OD x 20-in. jars are
D. Logic validation tested
exposure to high temperature. Therefore, this attached and manually jarred up and down
E. Jar tested
tool will be temperature-cycle tested. 25 times; the function of the tool was checked.
F. Release tested

23
E xpand i n g t h e r ange of appl icat ions of t rad itional s lick line

nonproductive time that could occur if a tool • Rigged up the tool string with a smart-
string becomes stuck. Case History A illustrates release tool. The smart-release tool was
the difficulties that can be experienced when the set for 24 hours.
traditional cutting and fishing methods are used • The lubricator was tested and run into the hole
compared to a job when the preliminary planning with a wire finder to the top of the surface-
includes a smart-release tool in the tool string. controlled subsurface safety valve (SCSSSV) at
5903 feet. Then, pulled out of the hole, bled off
Case History A scale inhibitor (SI).
This is a case history of a high-risk operation • Run in hole to 20,236 feet at 72-degree
where the tools became stuck in spite of the fact deviation. Pulled out of hole.
that great care had been taken in setting up all • Stopped at 3000 feet. Pick-up weight was
pre-job preparations. Below is an outline of well 425 lb. Picked up again and wire stuck.
parameters and the process that was required to 1500-lbf pull was applied, but there was
free the tool string. no movement.
• Determined that tool string was stuck in
This job was performed on a floater rig in the the hydrate.
Gulf of Mexico in 3,500 feet of water. The well • Pumped ethylene glycol (MEG) with cement
had 5,000 shut-in tubing pressure (SITP) and units to free hydrate. Pumped 91 bbls, shut in
bottomhole pressure (BHP) of 11,500 psi. The the well, and waited.
Figure 6. Smart-release tool tension test.
service to be performed was a clean-out of the • Repeated pumping and working wire for
well with slickline. This well was considered a several days.
No damage to SSR after Test E was considered high-risk well due to its history of asphaltenes, • Rigged up the coiled tubing unit.
as a successful test. This test was repeated and suspected gravel-pack failures. • Freed hydrate after 19 days. Rope socket
10 times. and smart-release tool recovered. Wire had
Coiled tubing (CT) washed to the bottom and become wrapped around the coiled tubing
F. Release testing then was pulled out of the hole. The slickline several times.
The purpose of this test was to determine if tool string was prepared. A 1.88-in.-OD conven- • The remaining tool string was fished with a
the SSR would function properly when placed tional slickline tool string that consisted of a pulling tool.
under a tension load. The logic was set for the rope socket, a 5-ft stem, a knuckle joint, a
shortest period of time. The SSR was hung 5-ft stem, spang jars and a 2-in.-OD drive bailer The complete length of wire was recovered
vertically from an overhead crane. A 600-lbm was used. The tool string was run into the hole without breakage, and the tool string below
weight was attached below the SSR. The test to 11,400 feet, and the bailer stuck. the smart-release tool was successfully
was to determine if the logic would activate recovered. Inspection of the tool string showed
and if the SSR would unleash when under The process of fishing the tool string, which that no damage was sustained from using the
tension load. required obtaining a cutter from the shore to pulling tool.
the platform, required 25 hours because of the
If the SSR was released properly, the test was weather conditions. Three drop-bar-type cutters Conclusions
considered as successful. Fig. 6 is a photo of the were used before the wire was finally cut at Shell has run the new tool in its Gulf of Mexico
tension test setup. 10,800 feet. 400 feet of wire remained above operations numerous times as a contingency
the rope socket. A coiled tubing unit (CTU) was method in the event that a tool string should
The prototype of the 1.69-in.-OD smart-release deployed to fish the cutters, wire and tools. The become stuck. As of the writing of this paper,
tool successfully passed all the tests CTU fish job required 23 days. The cost of lost rig only one situation had occurred in which Shell
outlined above. time was in the hundreds of thousands of dollars. wanted the smart-release tool to initiate a
disconnect process. This occurred when the tool
After lab testing, the prototype tool was tested in Although the smart-release tool would not have was used to do a soft release of memory BHP
a test well in New Iberia, LA. The tool string was prevented the tool string from getting stuck, had gauges, but the tool failed to release. It was
assembled, and the timer was set on 6 hours. it been used, it would have removed all the wire determined that the cause of the failure was a
The tool string was run to 500 feet and released above the stem, and this would have reduced the dead battery, and the tools are now supplied to
at 6 hours on two separate runs. fishing time by approximately 20 days. the field offices with battery testers. Since the
tool is “fail safe,” any failure in the system will
Case Histories Case History B prevent its releasing. As a worst-case scenario,
While the smart-release tool cannot prevent the This case history concerns a deepwater slickline if this situation occurs, you would be in the same
tool string from becoming stuck, the following job that was run to a depth of 20,236 feet. situation as if you had not run the tool, but since
two case histories show a comparison of Following are the steps followed in performing it is fail safe, there is no concern that the tools
the impact on the job from the economic and this job: can be released early in a failure situation.

24
The auto-disconnect benefits of the tool are The tool has applications in most slickline opera-
particularly advantageous when applied: tions. In 2008, Shell used the tool approximately
1. In deepwater wells where wire fishing 25 times in the GOM. To date, another major
operations could result in expensive rig costs operator also has used the tool in the Gulf of
2. In deviated wells where a dropped cutter might Mexico, but as of this writing, the tool has not
have difficulty reaching the rope socket been required to release a stuck tool string.
3. During services when using .125-in. wire However, both operators feel that the insurance of
diameter and larger, making fishing operations having the tool should a catastrophic occurrence
more difficult take place is well worth the cost of its use.

Authors
Jack Clemens earned a BSME technical papers, Darrell is published in many industry Jerry Foster is the Slickline
from the University of Arkansas periodicals. He is a New Orleans native and earned a Technical Applications Manager
and is a registered Professional degree in Electrical Engineering from the University of for Halliburton Wireline and
Engineer in the state of Texas. Texas at Arlington. Perforating, responsible for the
He is a Technical Advisor in the development of new products and
Halliburton Wireline and Perforating John Schram is a Technical services for the Integrated Cased
Group. Jack has over 25 years with Halliburton and Professional in the Packers product Hole business entity. A 1970 graduate of the University
other oilfield-related jobs prior to joining Halliburton. He line of Halliburton. In this role, John of Texas at Arlington, Jerry has held various well
has 11 patents and has co-authored numerous technical provides customized permanent intervention engineering and management positions
papers. Jack’s areas of interest within Halliburton are packer designs tailored to the since joining Halliburton in 1978.
downhole electromechanical tools that are used in the specific conditions of customer
Wireline and Perforating Group. wells. Since joining Halliburton in 2006, John has also Johnnie Garrett is a Well
worked in the Advanced Slickline Technology and Coiled Interventions Engineering Technical
Darrell Moore has been a Tubing Surface Equipment product service lines. Lead with Shell, having started in
design engineer in the oil and 1984. He later worked in production
gas industry for over 15 years. operations for 13 years and as a
He holds a number of patents and well interventions planner for two
industry awards for his engineering years. Johnnie has authored several articles, and is a
innovations. The author of many member of SPE.

Acknowledgements
The authors wish to thank the management of Shell for their input during Goiffon, J.: 1996. Setting Tool Increases Safety and Decreases Setting Time
development of this tool as well as their usage and faith in its capabilities for Packers and Bridge Plugs. WORLD EXPRO 96 (1).
for all their GOM jobs. The authors also want to thank BP for using the tool. Larimore, D.R., Fehrmann, G.Z. 1997. Field Cases of Cost Efficient Slickline
Finally, the authors wish to thank the management of Halliburton for their Well Interventions. Paper OTC 8530 presented at the Offshore Technology
Conference, Houston, Texas, 5-8 May.
encouragement in both tool development and the writing of this paper.
Larimore, D.R., Goiffon, J.J., Bayh, III, R.I. 1996. Low Cost Solutions for
Well Interventions through Advanced Slickline Services. Paper SPE 35236
References presented at the SPE Permian Basin Oil & Gas Recovery Conference,
Bargawi, R.A., Dean, D., Clemens, J., Whitmire, C. 2008. New Electro- 27-29 March.
Mechanical Perforating Technology Reduces Cost and Increases Safety in Larimore, D.R., and Fehrmann, G. Z. 1997. Field Cases of Cost Efficient Well
Workover Operations. Paper SPE 113805 presented at the SPE/ICoTA Coiled Interventions Performed with Advanced Slickline Technology. Paper SPE
Tubing and Well Intervention Conference and Exhibition, The Woodlands, 38097 presented at the SPE Rocky Mountain Regional Meeting, Casper,
Texas, 1-2 April. Wyoming, 18-21 May.
Foster, J., Clemens, J., Moore, D. 2001. Slickline-Deployed Electro- McDaniel, D., Cromb, J., Walton, J., Clemens, J., Moore, D., Huggins, J.
Mechanical Intervention System, a Cost-Effective Alternative to Traditional 2008. Case History: Extended-Stroke Downhole Power Unit Successfully
Cased-Hole Services. Paper SPE 67201 presented at the SPE Production and Pulls Subsea Wellhead Plugs. Paper SPE 113806 presented at the SPE/ICoTA
Operations Symposium, Oklahoma City, Oklahoma, 24-27 March. Coiled Tubing and Well Intervention Conference and Exhibition,
Fripp, M., Kyle, D. Hamid, S., Moore, D., Caja, J. 2008. Development of The Woodlands, Texas, 1-2 April.
a High-Temperature Rechargeable Battery for Downhole Use in the
Petroleum Industry. Paper OTC 19621 presented at the Offshore Technology SI Metric Conversion Factors
Conference, Houston, Texas, 5-8 May. ft x 3.048* E - 01 =m
Gazda, I, and Goiffon, J.J. 1995. A Battery-Operated Electro-Mechanical in. x 2.54* E + 00 = cm lbm x 4.535.924 E - 01 = kg
Setting Tool for Use with Bridge Plugs and Similar Wellbore Tools. Paper psi x 1.589 873 E – 01 = kPa lbf x 4.448 222 E + 00 = N
SPE 29459 presented at the SPE Production Operations Symposium, bbl x 1.589 873 E – 01 = m3 *Conversion factor is exact.
Oklahoma City, Oklahoma, 2-4 April.

25
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

Production Logging in Horizontal Wells:


Case Histories from Saudi Arabia
Utilizing Different Deployment and Data
Acquisition Methodologies in Open Hole
and Cased Completions Figure 1. Typical well completions utilized in
Ali R. Al-Belowi, Mohamed K. Hashem, Mohammed A. Al-Mudhhi, Saudi Aramco, SPE; Saudi Arabia.
Francisco Arevalo, Marvin Rourke, Tarek El Gamal, Juan Torne, Halliburton, SPE
Presented at the Abu Dhabi International Petroleum Exhibition & Conference held in Abu Dhabi, UAE,
November 1-4, 2010
Copyright 2010 SPE. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Abstract
The horizontal wells drilled during the last few years are coming to a mature state in different fields
and locations. This has created the requirement to evaluate the performance of the wells and determine
water sources for proper reservoir management.
In horizontal wells tool deployment, multiple sensor data acquisition and the interpretation of these
results is a challenge, particularly when there are long intervals of open hole completions and where Figure 2. Ideal and actual flow profile example.
fractures are often encountered. The horizontal cased completions use multiple downhole valves and
external packers to provide isolation between intervals. for each section or ICD or inter-packer zone is
This paper reviews the two primary deployment techniques of coiled tubing and tractor conveyance and constructed based on shut-in (zero surface pro-
highlights their advantages and limitations for each one. In the paper we propose a general guideline duction) and flow rates at different choke sizes.
for their utilization.
A particular flow profile can explain lateral hetero-
The case histories in the paper evaluate the different sensors utilized and provide recommendations geneities of the formation, different flow units,
particularly for low flow rate conditions. Two interpretation methodologies are evaluated and results open fractures, high-permeability intervals arising
are compared along with the interpretation from other sensors. from dolomitization in carbonates, among others.
The case studies examples in the paper demonstrate the value of integrating all available information The permeability model derived from open hole log
and the value of different data acquisition and evaluation techniques. The conclusions highlight analysis and core data can be confirmed as well.
the operational considerations, the data interpretation and added value in the evaluation of well Flow profiles from multiple wells in a field can be
performance to maintain production and improve reservoir management. used to define reservoir patterns that may provide
useful information for future horizontal in-fill well
placement to avoid intervals of low permeability,
Introduction This is the case of a perfect homogeneous non-
to promote intersecting open fractures and to avoid
Production logging in horizontal wells is primarily damaged reservoir rock with the same vertical
water production zones. In addition, known water
used to determine a flow profile and any water permeability along the entire wellbore.
zones could be isolated between packers, flow
sources. These wells are normally completed
from high-permeability zones can be restricted by
in Saudi Arabia with barefoot open hole in the This theoretical model also assumes there are
ICDs, and many other strategic decisions can be
horizontal section or with a liner, packers and no shale sections, impermeable zones, faults or
made to improve field productivity.
Inflow Control Devices (ICD) (Fig.1). The ICD fractures crossing the wellbore and the horizontal
completion functions to control fluid entry rate wellbore does not cross different formations. In
If the horizontal well is producing some water, it
along the horizontal section or to isolate water addition, this model assumes there is minimum
is important to know the water source. Oil and
producing zones. pressure drop between the heel and toe.
water flow segregation by gravity when flowing
in horizontal sections of the wellbore has been
Horizontal wells are designed to maximize the To achieve the objective of determining a flow
extensively documented. Even if water does not
contact between the borehole and the reservoir profile and to quantify the productivity index
flow to the surface it can remain static within the
permeable zones. Thus, ideally the flow rate of each section of the well, multiple PLT logs
wellbore on the lower side. Stratified flow regime
increases linearly from the toe to the heel (Fig. 2). are acquired at different flow rates. An IPR plot
with the lighter oil flowing on top and heavier
26
water flowing on bottom is the most often Capacitance Array Tools
encountered flow regime at downhole conditions. The current fluid ID array tool type can be
Slugging and bubbling flow regimes are typical capacitance or resistivity. For Saudi Arabia
when the oil flow rate is low and are considered oil reservoirs, the capacitance hold-up array
unstable flow regimes (Fig. 3). tool has been used. In open hole logging a
dielectric tool typically operates at about 1 GHz
A bubble flow regime is characterized by high to properly distinguish between oil and water.
Figure 3. Oil-water flow regime in horizontal section.
fraction of water with oil bubbles flowing along A center sampling capacitance tool operates at
the upper side of the hole. The water fraction frequencies lower than 100 KHz, and they are
may be flowing or can be static (z4 in Fig. 3). If well known to be mostly a qualitative hold-up
the water fraction is not flowing, a mini-spinner measurement for multiphase production logging.
located in the low side of the hole will measure They clearly indicate oil entry in water continuous
zero rps. A mini spinner in the high side of the medium, but the response is not proportional
borehole would behave similar to a vertical to the fraction of oil entering the flow stream.
bubble flow regime. The oil bubbles impinging The response to water entry when oil is the
the small impeller vanes immersed predominantly continuous medium is reliable and when properly
in water will create an impeller response that calibrated can quantify the water-and-oil
would not represent accurately the velocity of the fractions of a water in oil emulsion.
oil bubbles. A distributed fluid ID sensor tool will
respond to only water in the lower sensors and to Despite the center-sample capacitance tool
alternate water and oil in the sensors located in Figure 4. Measurement coverage of circular limitation at high water hold-ups in vertical
the high side of the borehole. A center-sampling distributed array tools. wells, when these sensors are calibrated in
spinner will measure the water velocity and a “segregated” water, oil, and gas within the
center located fluid ID (Capacitance or Density) is not considered reliable. These regimes usually wellbore, they show very good repeatability.
tool will be immersed in water. occur at low to very low oil flow rates and In horizontal wells, where the “segregated
can make these cases the most challenging stratified flow regime” prevails, a circular array
As more oil is added into the wellbore (z3) a to evaluate. of capacitance sensors performs very well and
slugging flow regime would be present. Water with the added advantage of proper response to
in the lower side of the borehole may oscillate Circular Array tools are one type of multi-spinner water-in-oil emulsions as shown in case study #1.
in direction as the oil slugs flow intermittently and multi-fluid ID devices where the sensors Here, there is a water-in-oil emulsion flowing
along the high side of the hole. A mini-spinner are arranged in a circular distribution around the upwards along the high side of the hole with
located along the low side of the borehole will wellbore as depicted in Fig. 4. From left to right, static water at the low side of the hole.
measure an alternating positive and negative a 12-sensor array tool is first shown measuring
angular velocity pattern or simply will not turn the fluid at 7 vertical levels making repeat The mini-sensors response of a capacitance array
at all, depending on the impeller sensitivity and (redundant) measurements in the 5 intermediate device in a stratified flow regime is shown in
threshold velocity. vertical levels. The mini-sensor No. 1 (shown Fig. 5. The upper four mini-sensors are immersed
in red) is located at the high side of the hole, in oil, and the lower eight mini-sensors are
Additional oil influx will result in a stratified and the relative bearing (angular displacement immersed in water. A colour 2D map is used
flow regime (zone z2 and z1 in Fig. 3). If the measured in degrees referenced to the high to present the hold-up measurements, which
water is not flowing or flowing at a low side of the hole) is zero degrees. In the second
velocity, then a water re-circulation pattern sketch, 12 vertical levels are measured, and
will be established in the low side of the hole. this is the most frequent coverage that can be
A mini-spinner in the low side may respond to observed. The last diagram shows another case
this recirculation flow pattern depending on its where the tool position measures at 6 vertical
specific height in the borehole. A mini-spinner levels with full redundancy. A coverage plot
located in the upper portion of the hole will indicating the vertical levels sampled at each
measure the oil phase velocity. tool angular position is also presented. Note that
more often the multi-array tool is sampling 12
Multi-Sensor Array Devices vertical levels except at some particular angles
Multi-spinner devices and fluid identification when redundant measurements are made. Slow
array tools measure the phase velocities and tool rotation provides redundant measurements,
water-oil fraction at all depths within the logging which improves reliability of the measurements,
interval. They work well in stratified flow that particularly when the sensors alternate from one
dominates the cases in horizontal wells; however, fluid to another. The use of swivel joints in the
Figure 5. Capacitance array – borehole tool
in bubble or slug flow regimes their performance tool string slows down the tool rotation. position and hold-up map.
27
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

allows better visualization of the layers within required for proper measurements: each mini-
the stratified flow regime. From the 2D map at a spinner needs to be parallel to the tool axis,
particular depth a localised average water height the tool must be well centralized in the wellbore,
is calculated and an associated uncertainty. This, and tool movement should be smooth with a
in turn, is used to compute the water hold-up constant speed.
and uncertainty along the horizontal interval.
While the response of all the mini-sensors is very Deployment Techniques –
similar, a normalization process is used to make General Considerations
all sensors read the same in gas, oil and water. Currently E-Line Coiled Tubing (CT) and tractor
tools are used to deploy the PLT tools in
Since the mini-sensors are located 0.35 inches horizontal wells. Both methods have advantages
from the borehole wall, there is minimum water and disadvantages and several applications Figure 6. Dual mini-spinner array aligned to the
hold-up that can be resolved. It is important to have been documented2. The tractoring force is hold-up tool.
display the average water hold-up and associated applied through a special high-technology design
uncertainty which has some dependency on tool of the wheels; the operation of the tractor has 3. Run and discuss a pre-job deployment modeling
angular orientation (i.e., relative bearing). When been extensively documented2. Regardless of the software that can predict the involved forces
the 2D map shows that all sensors are immersed selected deployment methodology, it is essential and estimate if the CT or tractor can reach
in water, it is possible that there is some oil that every horizontal production-logging job TD. In the case of the well tractor, the cable
flowing along the high side of the borehole, have extensive pre-job planning, which typically weak point needed to come out of the hole
which bypasses the upper sensors. Conversely, requires several pre-job preparation meetings safely should be designed. In the case of CT,
when all sensors show oil, there may be some between the different parties involved. The the weak point must be designed based on the
water bypassing the tool along the low side of service company specialist for horizontal logging tool breaking strength. Plan for maximum tool
the borehole. involved in operations, quality control and data rigid length based on the dogleg and borehole
interpretation should attend these meetings. diameter at the specific depth.
It is well documented6 that very small changes We also recommend the use of the latest 4. Pre-job planning meeting(s) to review
in wellbore inclination, as small as 1° deviation developments in real-time operations and data deployment method, tool to be deployed,
from perfectly horizontal, dramatically change transmission to ensure integrity of the acquired and operation sequence. A risk assessment
the oil and water velocities and hold-ups. data in real time. is important considering the following: open
Therefore, in addition, to have accurate water or cased completion, directional survey with
hold-up measurements, it is essential to have Both deployment techniques have been used and clear definition of doglegs and wellbore
accurate well inclination and well-profile the choice is determined based on economic or diameter, mechanical completion of the well
measurements (ID) to obtain reliable logistical reasons; the following technical items including restrictions and special devices
flow profiles. should be considered: like Y-tool for downhole pumps, the latest
production information, presence of corrosive
Spinner Array Tools 1. If the horizontal section is extended reach and fluids including very saline water, well head
A multi-spinner array tool consists of 6 small CT may lock up prematurely even with the help pressure, and bottom hole temperature.
spinners arranged in a circular pattern similar of other mechanical or chemical means, then 5. Once the operational sequence is completed,
to the capacitance fluid ID array tool. Two the wireline tractor can be used to reach the perform safety and risk analysis defining focal
spinner array tools are stacked and aligned total depth (TD) points personnel.
to the hold-up array tool to provide 12-phase 2. If there is a requirement to nitrogen-lift the
velocity measurements (Fig. 6). In the case well fluids, then only CT can be used
study examples here, every mini-spinner has a 3. On an offshore platform where space and
corresponding mini-capacitance sensor aligned maximum weight are limited, a tractor is Integration of All Logs,
Mechanical, Production
Pre-Job Deployment
Planning Meeting where all
so that the fluid in which every mini-spinner is normally the best alternative and Reservoir Information
to define and achieve
information is reviewed
and a drag and torque
immersed is known at all depths. 4. If corrosive fluids are present, then probably objectives modeling software is used
(1) (2)
CT is the only choice, especially if there are
Since spinners respond differently in different large amounts of H2S
fluids, it is required to perform multiple passes
at different tool speeds in static oil and water to In general, both deployment techniques follow
calibrate the spinners. This is usually done with the same recommended planning process (Fig. 7): Complete and review data
acquisition program with all
the well in shut-in conditions. Once the spinners parties involved and define
Focal Point for proper and
are calibrated, each mini-spinner response can 1. Define the objective of the job effective communication
(3)
be converted to a phase velocity for oil or water, 2. Collect existing well data such as logs,
depending on which fluid the mini-spinner production history, drilling reports, well Figure 7. Horizontal well logging and tool
is immersed in. There are several conditions schematic and survey deployment planning process.

28
The risk of damaging the tool is much smaller
than CT. It is important to avoid running the
tractor above 50% maximum load to minimize
the risk of electric failure.
3. The equipment and personnel required for
using wireline tractor are much less than with
CT. The wireline tractor rig up operation is
as simple as adding a large sinker bar on the
string while rigging up 20 feet more lubricator
in average. The CT operation implies the
mobilization and rig up of special lubricators or
special deployment system (deployment bars),
special large BOP, a special large-capacity
crane, additional pumping truck and fluid
storage tanks without counting the personnel
that can exceed 25 people over maximum
6-8 people for tractor operations. In offshore
operations, the CT equipment load and size
could exceed the limit of the platform, while
tractor operation does not differ too much from
a standard wireline requirement.
4. While CT can run as fast as 60 ft/min, it is
recommended not to exceed 40 ft/min. The
wireline tractor will normally run passes
at line speeds between 20 and 40 ft/min,
Figure 8. Tractor deployment modeling and planning software; well mechanical description and depending on the wellbore conditions. CT
directional survey. allows log down and log up operations, but
it could induce a swabbing effect on the well
The most important input in the planning process the probabilities to retrieve the tool if proper
while going upward, it restricts production
is the CT or tractor deployment simulation procedures are followed are greater when CT
while logging downward. This depends on the
using specialized software. There are different is used because of the larger pulling and
tubing size compared with the CT size. Large
commercial packages available. One of the most maneuver capacity:
flow disturbances have been observed with
important simulation parameters, is the friction
a 2-in. OD CT running inside a 31⁄2-in. tubing
factor to be used, which varies from open hole 1. Both systems apply tension and compression
size. The same CT size created only small
to cased hole and other wellbore conditions. to the tool, but the forces over the tool are
flow disturbances inside 41⁄2-in. production
In the case of CT, the software will indicate the less and easier to control using the wireline
tubing. Tractors have the advantage of
maximum reaching depth before buckling and tractor. While tractor compression is limited to
creating smaller flow disturbance excepting
the typical pulling force considering drag. In the applied current and the grip of the wheels
for the tractor itself. Inside the tubing, the
the case of wireline tractor, the software will to the wall (normally 500 lb compression), the
tractor could act as a downhole choke, but in
indicate the maximum depth that can be reached compression applied with the CT can exceed
the larger-diameter horizontal section, there
with the specific cable used and the maximum several thousand pounds in a few seconds
should be much lower flow disturbance. New
overpull at TD to verify that it is within the safe when running in hole. Most of the compression
technology tractors allow “logging while
range for the cable. An actual planning case forces go into the coil as buckling and friction.
tractoring” or downward logging. This is a
(Fig. 8) shows how the tractor reaches the The larger forces observed at surface could
very important feature, because the tool is
minimum tension towards the TD of the well and mask the smaller forces applied to the more
running against the flow giving larger effective
the pull-out tension approaches the maximum fragile PL tool.
velocity on the spinners. Typically, in upward
permissible tension for the specific cable (50%). 2. CT requires continuous attention from the
passes, there would be some zones where the
operator and engineer to prevent large forces
line speed is similar to the fluid velocity. In
Each technique has advantages and on the tool, in 2-3 seconds up to 10,000 lb
these particular zones, the spinners will not
disadvantages, but most of them need to be compression or tension can be applied on the
respond proportionally to the fluid velocity. The
considered case by case since most of the CT at surface. In the case of wireline tractor,
acquired information then would not be useful
time the well conditions could force one of the it is important to keep constant tension on the
to quantify phase velocities, rendering an
choices with their own limitations and the risk line when running in hole to prevent damage of
incomplete flow profile.
needed to be balanced. Apparently, the tractor the wireline; the operating window is relatively
5. The wireline tractor has some hole size
presents large advantages over the CT, but small and also requires continuous attention.
limitations, depending on the production tubing

29
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

restrictions and open hole size. In some cases, internal pressure of the CT slightly higher than interval. The flow profile is generated from the
a 3.125-in. tractor was used to pass through the wellbore pressure. Standard wireline cable downward pass. The upward pass is used for well
3.625-in. restrictions, to log in 6.125-in. and 0.321-in. size has been used in Saudi Arabia stability assessment by observing the pressure
8.5-in. diameter horizontal open hole. tractor operations successfully, but in high- between the up and down passes, Fluid ID maps
Tractoring tools more than 5000 ft of horizontal percentage H2S environment it will be required and temperature repeatability. In addition, a
open hole section have been achieved. to use special steel material H2S 0.222-in. quality check is made for similarity of the spinner
Normally, the tool is gravity deployed until the cable; this cable has lower pulling capacity response in zones when a particular mini-spinner
65° deviation section of the well. The CT size and larger line resistance that limit the is immersed in the same fluid type.
has to be selected according to the tubing size capabilities of the wireline tractor operations.
and the completion restrictions. Larger CT is Larger wireline cables for H2S are not available Log Interpretation
preferred to reach farther, but it could create because they are normally very expensive and The objective of the interpretation is to obtain
large flow disturbances. It is important to plan require special manufacturing. a flow profile in open hole completions and in
the flow rates for a particular CT size, by using cased hole completions to verify ICD performance.
the proper simulators that take into account Data Acquisition Techniques The required input data to determine a flow
tubing size, open hole size, and the created Data acquisition techniques for horizontal wells profile includes water and oil hold ups (Yw, Yo),
flow disturbance. are an adaptation of standard production logging measured with the fluid ID array tool and phase
6. It is always recommended to have swivel and techniques. The typical horizontal PL tool string is velocities (Vw, Vo) measured with the mini-
a knuckle or flex joints in both operations, centralized for the array section and isolated from spinner array tool. The following formulas are
but while the wireline tractor applies limited the deployment tool (CT or wireline tractor) by used to compute the flow rates
compression to the tool, the CT can break the use of a set of flex joints, knuckle joints, and (Qw, Qo) (Eq. 1 & 2):
it in the weakest point, which is normally swivels subs.
the knuckle joints and centralizers. It is not Qo = 1.4ID 2VoYo (1)
recommended to use knuckle joints with CT, Doing multiple passes, which is routine in
excepting for a high-strength flex joint. The standard PL in long horizontal sections, is not as Qw = 1.4ID 2VwYw (2)
use of swivels is very important to isolate the effective as in vertical wells because the mini-
torque and buckling effects with the CT and spinners would not be in the same fluid at the Where:
minimize tool rotation with wireline tractor. same depth in the separate passes due to tool ID = Open hole borehole internal diameter or
The CT cable head has incorporated a swivel rotation. Since the mini-spinners are calibrated pipe ID in inches
joint. Tool tension and especially compression, with multiple up and down passes in segregated Vo = Oil phase velocity in ft/min
is monitored using a downhole tension- oil and water during shut-in conditions, only two Vw = Water phase velocities in ft/min
compression tool. This tool is set above all PL passes, an up and down at low logging speeds, Qo = Oil downhole flow rate in (B/d)
tools and should be monitored continuously for are done for each flow rate test. Typically, the Qw = Water downhole flow rates in (B/d)
excess compression while running in hole. well is flowing with an intermediate choke Yo = Oil hold-up
7. Cable slack management is very important in setting while the tool is conveyed down the Yw = Water hold-up
CT operations; normally around 2% excess production tubing to the heel of the horizontal
is pumped into the CT to minimize the section. Once the well is stabilized, two passes The basic model is shown in Fig. 9.
internal tension of the cable due to curvature are made down and up in a single descent at low
bending when running in hole. If proper slack line speed, typically 30 ft/min. The line speed Methodology
management is not performed, the cable will has to be slow due to the restrictions imposed There are two available methods to perform
be disconnected from the head while running by logging in an open hole completion and CT or flow rate computations. They are depicted in a
in hole, losing communication with the PLT. Tractor limitations. The stationary measurements simplified flowchart in Fig. 10.
This is not a problem with Tractor; instead the are performed during the upward pass to save
wireline tension needs to be controlled from operational time. The tool is then repositioned Both methods start with data preparation, which
the logging truck by slacking cable to allow the at the heel and a full choke test is performed includes file conversion and plotting the raw data
tractor running in hole free. following similar procedures described above.
8. It is important to consider during the design of
the job the H2S gas content and other corrosion The final step is the shut-in log. The well is
fluids that could be present in the wellbore. shut in for a short period to allow the fluid to
Depending on the H2S percentage, it will be segregate and log-down and log-up passes to
required to use either a special H2S cable be made. Then a horizontal section is selected
or corrosion-resistant CT. In the case of CT based on the shut-in log, where oil and water
logging, the cable is protected and corrosion are stagnant, to calibrate the mini-spinners and
affects the CT only. In H2S environments a the regular center spinners in both fluids. This is
corrosion inhibitor is recommended to be achieved by making multiple passes at different
pumped while running in hole, keeping the line speeds (typically 10 to 12) through this short Figure 9. Basic interpretation model.
30
Where:
I = Depth
J = Capacitance sensor number
CO(J) = Frequency of Jth sensor when
immersed in oil
CW(J) = Frequency of Jth sensor when
immersed in water
RACP(I,J) = Frequency of Jth sensor at
depth “I”
NCAP(I,J) = Normalized value of Jth sensor at
depth “I”

After the normalization process is complete,


when any sensor is immersed in water, it
will read NCAP=1.0, and when any sensor is
immersed in oil, it will read NCAP=0.2. Any
intermediate readings correspond to a slow
response of the sensor or emulsified oil and
water as has been sometimes observed. CO and
Figure 10. Processing flowchart. CW frequencies for each capacitance sensor
can be obtained by computing maximum and
for quality control (QC) purposes. For a typical for packer/ICD completions. Typically the minimum values throughout the log or by using
log, the down and up passes are plotted for each X-Y caliper log shows borehole enlargement surface calibration values.
flow condition (surface choke), plus the shut-in in sections drilled uphill (>90°) and close to
condition (“Shut-in plot”). A “Composite plot” bit size in the downhill sections (<90°) in An alternate normalization method is shown in
including all conditions is useful at this point open hole completions. The caliper log also the right-hand side of Fig 10. This is a procedure
in the analysis. would show borehole enlargement close to where a specific frequency for each fluid is used.
casing shoe. For example, 50 Hz for water and 150 Hz for oil.
The QC and data preparation process includes: 7. The flow units or formation tops could be Using this procedure for all probes will result in
1. Depth matching for each individual flowing included in the plot as annotations, if they a normalized reading with all of them responding
or shut-in plot; depth matching between are available. with the same frequency value when immersed in
the three plots within the composite plot. A 8. Review open log image log for the existence a particular fluid.
graphical well sketch is useful when attached of open fractures and verify original drilling
to each plot. reports to determine zones of mud losses. The normalized hold-up map is then built using
2. Well stability assessment verifying there is 9. Review the open hole log interpretation, a convenient color scale and industry-standard
minimal pressure difference between upward including porosity and saturations to identify colors to represent water, oil and gas. The map
and downward passes for a particular survey. permeable formations. represents the unfolding of the circular wellbore
Also the total pressure drop between toe and 10. Verify stations depths, duration and to a flat strip where the sensors located at
heel is used as reference. correlation with passes at each well condition the low side of the borehole, as indicated by
3. Constant and consistent surface tension to (flowing and shut-in). relative bearing, are in the center of the map.
ensure the cable speed is equal to the tool Typically, these sensors will be immersed in
speed. Difference between up and down Hold-up Calculation water indicated by a normalized value close to
tension is because of drag effects on the After completing the QC procedure, the 1.0 and seen as a blue color in the map. The
cable. processing to determine the flow profile for the sensors located at the high side of the borehole
4. Inclination measured during the PLT compared first flowing condition can start. The process should have a normalized value close to 0.2 and
to the original inclination included with the starts with the hold-up computation. Each are shown with a green color in the map. Hence,
well survey. capacitance array sensor response for oil and the color map should help to understand the flow
5. Generating a TVD plot using the inclination water is similar but not identical for all of them, regime and other phenomena occurring in the
measured to build an accurate downhill/uphill hence normalization is required. wellbore. The color scale as shown in Fig. 11
well profile in the horizontal section. helps to visually diagnose the fluid type and
6. Compare measured PLT X-Y caliper with The procedure is based on the following distribution in the borehole. Most maps in an
the bit size and the open hole caliper log if relationship (Eq. 3): oil-water environment will show both dark blue
available in the case of barefoot completion and dark green because of the stratified nature
CO(J ) − RCAP(I, J )
or check the packer/ICD internal diameters NCAP(I, J) = 0.2 + 0.8 (3) of the flow regime. In high flow-rate wells, the
CO(J ) − CW (J)
and its correspondence with the well sketch interface has a significant width and would be

31
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

Vthphase+ = Threshold velocity for oil/water when the well is shut in and there are static
and for positive angular velocities segregated fluids representing clean water and
(ft/min) clean oil.
Vthphase- = Threshold velocity for oil/water
and for negative angular velocities
(ft/min)

Line speed (LSPD) is considered negative when


Figure 11. Color scale for the hold-up map.
logging down and positive when logging upward.
Vth is positive for positive angular velocities
represented by an intermediate band of light
and negative for negative angular velocities.
green between dark blue and green colours. In a
The slope of the linear regression denoted as
stable emulsion, all sensors will be light coloured,
Slope represents the impeller sensitivity and
either green or blue, depending on the emulsion
the threshold velocity (Vth), its inertia. At lower
type. A clear distinction between oil and water is
angular velocities (ω) the spinner response is
needed to define in which fluid every mini-spinner
known to be nonlinear, and therefore those
is immersed. A threshold value of 0.6 is normally
equations (Eq. 4 & 5) are no longer valid. Thus, a
used to differentiate between oil and water.
new parameter ωL should be introduced to take Figure 12. Spinners’ responses in static fluid when
into account the lack of linearity (Eq. 6 & 7). running in hole.
Once the processing has generated the color map
and differentiated fluid types, then a water height Vphase =
ω
+ LSPD +Vthphase+ ⇒ ω > ω L
Slope phase+
(6) Once the spinners are properly calibrated every
and hold-up can be computed. The second method logging pass can generate a flow profile and
uses all the normalized values to generate a Vphase =
ω
+ LSPD +Vthphase− ⇒ ω < −ω L
Slope phase− (7) even a stop check will provide a flow rate at that
frequency hold-up map. From the position of each particular depth, time and well status.
particular color in the hold-up map, the water and
Where:
oil distribution is computed. Both methods have
ωL = Minimum linear spinner angular Flow Rate Calculations
been shown to give similar results. The flow rates at downhole conditions are
velocity (rps)
computed once hold-ups and phase velocities
Velocity Calculations are obtained. The first method uses the simple
A typical response for the different spinners in
The fluid velocity is calculated from each center
static oil is shown in Fig. 12 where the x-axis relationship between hold-up and phase velocities
spinner tool and mini-spinner in the array tool. to calculate flow rates (Eq. 1). The second method
is logging speed (LSPD) and the y-axis is the
Velocity calculations include in-situ spinner calculates the flow rates downhole by multiplying
angular velocities (ω) of the spinner impeller.
calibration to convert from the impeller angular the hold-up profile by the velocity profile; this
The CFS and ILS are both center spinner devices,
velocity to fluid phase velocity. It is assumed process is done by subdividing the borehole cross-
the first being a continuous and second being an
there is a linear relationship between the flow section into multiple horizontal slices where each
inline spinner tool. SPIN1 through SPIN6 are the
meter response (ω) in revolutions per second (rps) slice has a velocity and fluid hold-up.
six mini-spinner responses.
and the tool logging speed (LSPD) in feet/minute
(ft/min) according to the following equations It has been verified that the results from both
We consider two methods to resolve fluid phase
(Eq. 4 & 5): methods are similar. The second method is
velocity from the mini-spinner tool. In the first
ω method, these formulas (Eq. 4 & 5) are applied more suitable if fluid viscosity and density are
Vphase = + LSPD +Vthphase+ ⇒ ω > 0 (4)
Slope phase+ individually as each mini-spinner has a separate similar between the oil and water phases since
ω slope and threshold in both oil and water. the spinner calibration parameters are similar.
Vphase = + LSPD +Vthphase− ⇒ ω < 0 (5)
Slope phase− Whereas in the second method, an average slope It is also useful when the well is not completely
and threshold is calculated for the 6 mini-spinners stable; under these well conditions the first
Where: method is not applicable.
in oil and water during the in-situ calibration.
Vphase = Oil or water phase velocity (ft/min) Then these averaged calibration values are
ω = Spinner angular velocity (rps) applied to each mini-spinner response. In both Center Sample Measurement Integration
LSPD = Line speed (ft/min) cases the measured hold-up from the array device Center sample spinners (continuous and inline)
Slopephase+ = Slope for oil/water and for is used to determine the appropriate values for have an impeller diameter similar to the center
positive angular velocities slope and threshold so that valid phase velocities hold-up sensors, which give a consistent
(rps/ft/min) are calculated. measurement and respond to the velocity of
Slopephase- = Slope for oil/water and for the fluid in the axis of the borehole when they
negative angular velocities The slope and threshold values for oil and water are properly centralized. The center sample
(rps/ft/min) can be obtained by an in-situ calibration of capacitance and fluid density tools respond to
multiple up and down passes at different speeds the fluid type at the axis of the borehole.
32
These tools are run in addition to the array tools
that follow a circular pattern giving an additional
measurement in the center of the borehole. If
we determined that the oil-water interface is
well below the center impeller blades, then that
particular center flowmeter response is reliable
and representative of the oil velocity at that
depth, and therefore can be used to enhance
the results.

The center sample capacitance is used as a


criterion to select in which fluid the center sample
spinners are immersed. If they are immersed in
oil, the standard spinners will be turning in oil
measuring the oil phase velocity. If the center
sample capacitance indicates that they are
immersed in water, the standard spinners will be
measuring the water phase velocity. Figure 13. Change in flow patterns and Figure 14. Temperature and pressure profile
emulsified flow. between passes.
There is a required distance from the interface
to ensure the center spinner response is not the single-phase emulsion with oil. Above all wellbore, indicating a large ICD pressure drop. A
affected by this fluid interface. This distance is ICDs the stable emulsion contents are about 20% similar effect is observed at ICD7; above this, the
measured from the low side of the impeller’s downhole water cut, which results in 25% water temperature changes reflect the pressure drop in
blade to the interface in case of oil phase velocity cut at surface conditions. Notice that ICD1 and each ICD, but in this case, convection effects mask
measurements and from the upper side of the ICD2 should be producing the same flow rate those. The increase in temperature at the packer
impeller’s blade to the interface in case of water according to the completion design. The same at x780 ft is an indication of packer leaking.
phase velocity measurements. This distance is can be said for the pairs 3&4 and 5&6. The two The temperature response in the stratified flow
typically set as 0.2 in. for low flow rates and packers between ICD7 and ICD8 are isolating a regime region (x650 ft – x100 ft) does not repeat
0.5 in. for high flow rates, where the interface possible water production zone. very well. This could be related mainly to the
width tends to increase. different production at these ICDs (7 and 8) during
Observe also that mini-spinners 1 and 6, which downward and upward passes, created by the
Case Histories are at the upper side of the pipe per the relative CT swabbing/restricting effect and/or to the water
The following four case study examples from bearing, confirm the emulsion is flowing. This is a recirculation at the lower side of the pipe.
Saudi Arabia demonstrate the new sensors and unique feature for the hold-up capacitance array
interpretation techniques described above. The tool. Excursions to the right in all spinner curves Well Study H/2 – Multirate Analysis and
most important observations are presented for in are due to smaller ID at the packers. Static Flow Pattern Comparison
each case. water at the lower side of the hole, probably re- This is an open hole horizontal well drilled to TD
circulating, prevented the mini-spinners 2, 3, 4 and @ x2460 ft MD, using a 6.125-in. bit with a plug
Well Study H/1 – Change in Flow Patterns 5 from turning in this stratified emulsion-water back at bottom. The well was logged to x1800 ft
and Emulsified Flow flow regime. MD. The 7-in. casing shoe is located at x490 ft,
This is a cased hole horizontal well completed and the 4.5-in. tubing is located at X050 ft. The
with packers and inflow control devices (ICDs) in A pressure differential between the up and well production was evaluated at three different
a 4½-in. liner. (Fig. 13). The deployment technique down passes was observed. This is related to choke sizes (172/64 in., 92/64 in., and 50/64 in.).
was CT. The well production was reported as oil the swabbing effect of the 2-in.-OD coiled tubing
and water 25% average water-cut. moving in a 3.9-in.-ID monobore completion The oil production was increasing proportionally
(Fig. 14). The pressure drop from toe to heel is to the choke size, but this did not happen with
This well exhibited stratified flow regime near about 18 psi; the pressure difference between up the water cut, which ranged from 9 to 13%.
the toe and single-phase emulsion towards the and down passes is about 40 psi. The deployment methodology was wireline
heel. A stable emulsion entered the wellbore at tractor, 31⁄8-in. OD using the latest logging while
ICD8. This emulsion is flowing in the upper side The temperature logs in this well show an tractoring technology (LWT) to acquire the down-
of the hole with static water at the lower side important effect related to the performance of the log and an up-log.
of the hole. The addition of oil at ICD7 changes ICD completion. The temperature changes (Fig. 14)
the flow regime to single-phase emulsion (all 12 in front of each one of the producing ICDs. The This well exhibited stratified flows through
mini-capacitances reading the same). The addition deepest large change at ICD8 shows heating the entire logging interval as shown in Fig. 15.
of oil through ICDs 5&6, 3&4 and 1&2 enriches Joule-Thomson effect as the emulsion enters the The flow profiles for three different flow rates

33
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

Well Study H/3 – Response in Low


and High Flow Rate
This is a long-reach open hole horizontal well
that was logged using a single spinner array
tool, tractor deployed, and only up passes were
acquired due to tractor technology limitations
imposed at the time of data acquisition. The well
was drilled using a 6.125-in. bit size, and it was
completed with 7-in. casing down to x730 ft. The
production tubing of 4½-in. was landed at x260 ft.,
and a nearly 5000-ft section was logged (Fig 17).

The well production was unstable. Two passes


at 50- and 30-ft/min were performed. The total
computed rate from the 50-ft/min pass was
close to actual well production. The total flow
rate while logging at 30-ft/min cable speed was
almost zero; therefore, the 50-ft/min pass was
used for the flow profile computation.
Figure 15. Flowing profile for the three rates.
The pressure response between the two passes
Table 1. Productivity Index Table
as shown in Fig. 17 confirmed that the well was
unstable since they do not follow each other
Inflow Performance Relationship
particularly towards the heel (starting at x0500 ft).
Pressure (psi) 2.292.00 2.209.00 2.072.00 1.932.00
The pressure difference is normal below x990 ft,
Condition Shut-In Q50 Q92 Q172 PI P R^2 and it is confirmed by the similar fluid map that
Units rb/d rb/d rb/d rb/d b/d/psi psi the fluid velocity is still low. The difference in
p1 0 1.000 2.050 3,500 9.50 2,299 0.995 pressure profile is very large above x500 ft, and it
p2 0 300 800 1,250 3.49 2,295 0.999
is shown as the different flow profile above this
depth, making the pass at 30-ft/min unreliable.
p3 0 200 600 1,000 2.81 2,287 0.999
This confirms the need for real-time operations
p4 0 500 1,250 1,950 5.41 2,297 0.999 where actual plots are delivered and reviewed
p5 0 200 500 800 2.21 2,296 1.000 before pulling the tool out of the hole.
Total 0 2,200 5,200 8,500 23.37 2,296 0.999
Surface 0 1,746 4,127 6,746 18.55 2,296 0.999 This well exhibits different flow patterns,
including stratified and slugging flow while
showed that the oil hold up increased as the rate IPR plots for oil and water were constructed for the fluids were almost static near the toe. The
increased. The flowing profiles include mini- this well. The oil and water IPR plots are shown directional profile is related to the different
spinner angular velocities and phase holdups. in Fig. 16, for each oil and water producing zone flow distribution and patterns observed; in the
From x0800 ft to PBTD, hold up and phase velocity and for the total well production. Table 1 shows uphill production sections, the water fraction is
profiles are essentially the same. Evidently, this is the Productivity Index (PI) for the five zones high, and we could observe some oil slugging
an indication of quasi-static fluid at the toe of the producing oil. on the upper side of the borehole (x750 ft). In
well. In the lower flow-rate log, the oil and water the downhill sections the fraction of oil is much
flow rate were small enough to keep stratified A similar analysis was performed for the water higher. Above x920 ft, however, a stratified
flow regime inside the tubing. Notice that was zones, indicating a lower productivity index, flow regime with close to 50% water fraction
considered safe to enter inside the 41⁄2-in. tubing which explains the direct relationship between oil is observed, indicating a large oil entry at this
with the 31⁄8-in. tractor considering the low flow fraction and total production. point. Only emulsified flow (water in oil emulsion)
rate was not enough to blow up the tool string. is observed inside the production tubing due to
higher fluid velocities.

34
Figure 16. Oil and water IPR plots.

This case history confirms the close relationship 3. Logging down is important to acquire proper
between flow profiles and the directional survey, flowing profiles as confirmed in the different
hold up, and fluid velocity previously documented cases studies. Acquiring production profiles
in multiple studies and papers. at different rates is important to have a
better understanding of the well performance
Conclusions through IPR plots and determine the different
1. Horizontal production logging requires inflow zones. It is important to have proper
an integrated approach from planning to selection of the flow rates. It is recommended Figure 17. Composite of logging up at 30 and
final delivery to meet the objectives. It is to acquire a minimum of two flow rates and a 50 ft/min using one choke and pressure
comparison showing the well unstable.
very important to consider the deployment shut-in pass.
methodology as part of the planning process 4. Temperature and pressure response analysis
and use simulators to estimate torque and in horizontal wells should be analyzed profiles in horizontal production loggings. The
forces when deploying the tools. Tool lifting independently of the production profile. ability of selecting sensors based on expected
forces should also be considered, given the Once there is a clear understanding of their downhole conditions using flow regime
large flow rates of several horizontal wells. response, they should integrate with the simulators is a challenge, and it presents
It is also important to use a flow regime flowing profile to have a more complete opportunities for development in the future.
multiphase simulator to assess the reliability understanding of the well behavior. 7. The study of emulsions should be considered
of the interpretation results. 5. The introduction of new sensors, deployment in the future for horizontal wells in addition
2. The introduction of array sensors and the methods, and interpretation techniques has to further studies of water in near-static
complete understanding of their response been positive in the evaluation of horizontal oscillating or recirculating conditions.
have improved the flow profile evaluation of well performance in Saudi Arabia. Different Measuring the net water flow when there is
horizontal wells but still new sensors need to flow patterns ,including stratified flow, recirculation is still a great challenge.
be developed to complement the ones currently confirm the unique performance nature of
used. The sensor distribution in the wellbore each horizontal well.
is important to maximize the investigation 6. The development and studies of flow regimes
of the different phases and velocities along profiles and data acquisition and interpretation
the wellbore, especially at the interface in techniques in horizontal production logging
stratified flow regime, or when fluids are still present challenges. The research and
not fully stratified in slugging and bubbling development of new sensors should continue
flow regimes. along with a better understanding of the flow

35
N ew - G e n e r a t i o n Array Tool De pl oyme nt Te ch niques

Authors
Ali Al-Belowi is the Supervisor Mohammed A. Al-Mudhhi is a Tarek El-Gammal is responsible
of the South Area Petrophysics Petroleum Engineering Specialist in for Halliburton Wireline and
unit in the Reservoir Description the Southern Area Petrophysics Unit Perforating Business Development
and Simulation Department for in the Reservoir Description and (Integrated Cased Hole Operations)
Saudi Aramco. He is responsible Simulation Department in Dhahran, in Saudi Arabia. He graduated from
for ensuring that open hole and Saudi Arabia. He joined Saudi the Higher Technological Institute-
cased hole logging programs are optimized, quality Aramco in 1978, and has 34 years of experience in Egypt in 1995 with a BSC in Mechanical Engineering
log data are acquired, and logs are analyzed on time various disciplines of the petroleum industry, including (Power Section), and has 17 years of experience in
utilizing fit-for-purpose petrophysics methods. He has Production Engineering, Reservoir Management and wireline logging operations and sales. He was awarded
done extensive work on petrophysical analysis in both Reservoir Description. Mohammed holds a BS degree the Halliburton WP Global Top Sales Award in 2010,
exploration and development fields. Ali has 22 years of in Petroleum Engineering from Tulsa University in and also received the UAE, SPE Award for Technical
experience in the industry, and he holds a BS degree in Oklahoma, and is a member of SPE. Support (Rigless Solutions - Adding Extra Barrels)
Petroleum Engineering from the King Saud University in 2011.
in Riyadh. Francisco Arevalo is the
Halliburton Wireline and Perforating
Mohamed Hashem is a Senior Formation Reservoir Solutions
Petroleum Engineering Consultant Manager in Iraq. Soon after
with Saudi Aramco, Dhahran, Saudi graduating from the Universidad
Arabia. He began his career in Nacional de Ingenieria (Lima-Peru)
1975 as a Production Engineer for with a degree in Electrical Engineering, he started his
GUPCO-AMOCO in Egypt and left career with Gearhart, which was later acquired by
the company as the head of the Wireline Department Halliburton. With 30 years of experience, Francisco has
to join Saudi Aramco as a Production Engineer in 1981. also worked in Colombia, Ecuador, Peru, Venezuela and
Mohammed has a BS degree in Petroleum Engineering Iraq. He has authored several papers, and is a member
from Cairo University, and has authored several of SPE and SPWLA.
publications in well stimulation, scale mitigation,
LWD logging, wireline logging, production logging and
logging tools conveyance.

Acknowledgements Yo Oil hold-up


The authors would like to thank Saudi Aramco and Halliburton for the Yw Water hold-up
support to introduce new technologies for horizontal logging and the ω Spinner angular velocity (rps)
permission to submit this paper. ωL Minimum linear spinner angular velocity (rps)

Nomenclature References
Salam, S.P., Al-Mubarak, H.K., Al-Malki, S., Al-Ajmi, A.M. 2003. Impact of
CO(J) Frequency of Jth sensor when immersed in oil Comprehensive Horizontal Well Production Logging and Testing Program in
CW(J) Frequency of Jth sensor when immersed in water Shaybah Field, Saudi Arabia. Saudi Aramco Journal of Technology (4)
I Depth Al-Buali, M.H., Dashash, A., El-Gammal, T., Arevalo, F., Torne, J.P. 2010.
ID Open hole borehole internal diameter or pipe ID in inches Intelligent Sensors for Evaluating Reservoir and Well Profiles in Horizontal
J Capacitance sensor number Wells: Saudi Arabia Case Histories. Paper SPE 137202 presented in
LSPD Line speed (ft/min) CURPEC, 21-23 October.
NCAP(I,J) Normalized value of Jth sensor at depth “I” Wang, J., Chace, D., Georgi, D., Maxit, J. 2000. Interpreting Spinner Response
Qo Oil downhole flow rate in (B/d) in Multiphase Bubble Flow. Paper SPE 63188 presented at the SPE Annual
Qw Water downhole flow rates in (B/d) Technical Conference and Exhibition, Dallas, Texas, 1-4 October.
RACP(I,J) Frequency of Jth sensor at depth “I” Pilman, D., Skinner, T., Denton, R., Hansen, M., Torres, D. 2009. Cost
Slopephase- Slope for oil/water and for negative angular velocities Effective Reservoir Characterization, Utilizing Dielectric Logging
(rps/ft/min) Measurements, San Joaquin Basin, California. Paper SPE 124206
presented at the SPE Annual Technical Conference and Exhibition, New
Slopephase+ Slope for oil/water and for positive angular velocities
Orleans, Louisiana, 4-7 October.
(rps/ft/min)
Vo Oil phase velocity in ft/min Wade, R.T., Cantrell, R.C., Poupon, A., Moulin, J. 1965. Production Logging-
The Key to Optimum Well Performance. Paper SPE 944-PA. Journal of
Vphase Oil or water phase velocity (ft/min)
Petroleum Technology 17 (2): 137-144.
Vthphase- Threshold velocity for oil/water and for negative angular
velocities (ft/min) Theron, B.E. and Unwin, T. 1996. Stratified Flow Model and Interpretation in
Horizontal Wells. Paper SPE 36560 presented at the SPE Annual Technical
Vthphase+ Threshold velocity for oil/water and for positive angular
Conference and Exhibition, Denver, Colorado, 6-9 October.
velocities (ft/min)
Vw Water phase velocities in ft/min

36
h igh est-quA lity meA suremeNts with eNViroNmeNtA l c or r ec tioNs

Performance of a New 2.35-in. Wireline


suite of environmental corrections is available at
the wellsite in real time.

or Memory Quad Combo for Through-Bit or The Natural Gamma Ray Measurement
A sodium iodide scintillation counter in the dual-

Small-Hole Logging purpose telemetry gamma ray sub provides a


measurement of naturally occurring formation
radioactivity. The gamma ray pit at the industry-
Jerome Truax, Jim Galford, Gordon Moake, David Torres, Ron Cherry,
standard API Calibration Facility (Belknap et al.
Batakrishna Mandal, Arabinda Mishra, Luis San Martin, and Alberto Quintero,
1959) in Houston, Texas, is the primary calibration
Halliburton Energy Services
for the detector sensitivity. Thorium sleeves
Presented at the SPE Annual Technical Conference and Exhibition held in Denver, Colorado, USA,
30 October–2 November 2011
constructed from sheets of silicone impregnated
Copyright 2010 SPE. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with monazite sand that have been calibrated
to the primary standard are used as secondary
calibrators and field verifiers.
Abstract
A new UltraSlim™ quad combo has been developed and deployed with log quality very comparable to
Standard logging conditions for the gamma ray
that of larger tools. The suite of logging tools is backed up by a full complement of swivels, knuckles, flex
measurement consist of an eccentered tool in a
joints, springs, weights, cable head, and latches adaptable to the situation. The string can be conveyed by
freshwater-filled borehole. When the borehole
wireline using conventional power and telemetry, either through-and-below drill pipe or in a conventional
size and/or fluid do not match the standard
open hole log run. The combo can also be run in the hole by slickline inside drill pipe, then pumped out and
condition, the attenuation of gamma rays in the
latched beneath the bit, using its downhole battery and memory.
borehole will be different and a correction is
The tool string has been logged successfully in more than 20 commercial operations in operator wells and required to obtain accurate values. The influence
two of Halliburton’s test wells. The combo and its peripheral equipment have performed without incident. of different attenuation conditions can be
Some of these log runs were made back-to-back with standard-sized logging equipment for the purposes characterized in terms of a borehole absorption
of certifying that the log quality is comparable. The string features an array induction with a number parameter, which is a function of the borehole
of coils and spacings identical to its larger siblings, array sonic with dual transmitters, deployable-pad fluid density and the borehole diameter
density with Pe and caliper, dual-spaced neutron, gamma ray, and borehole environmental measurements. (Wahl 1983, Galford 2005). To apply the
correction, Fig. 3 is entered at the bottom with
This paper describes the UltraSlim logging system and provides details of its reliability, repeatability,
a calculated borehole absorption parameter to
accuracy, and petrophysical performance in comparison to standard-size logging tools, by means of visual
locate a point on the appropriate tool position
and statistical log analysis.
relationship. The observed gamma ray reading
is then multiplied by the correction factor found
Introduction the toolstring with cable head is 68 ft [20.7 m] on the axis at the left. Eq. 1 can be used to
As U.S. drilling activity has increased over the long. Use of the in-line bow-spring eccentralizer calculate the borehole absorption parameter,
two-year period beginning in the third quarter for the neutron tool adds 5 ft [1.6 m] to the x, for a given borehole diameter, dbh, and mud
of 2009, shown in Fig. 1, most of the increase length. Swivels, knuckles, flex joints, springs, and electron density, ρe_mud.
has been in horizontal drilling. This trend is weights can be added as the situation dictates.
expected to continue throughout the remainder The telemetry-GR-accelerometer section also Drilling Activity by Rig Type
of 2011 (Spears 2010). The situation affirms contains memory for slickline operations, which 1400
the need for a slim logging system that can requires the addition of a battery sub. Landing 1200 Horizontal

be deployed through drill pipe, below the bit. and shock absorber subs are used for latching 1000
Directional
Vertical
Other applications include new or re-entry slim in a special drillstring sub; these assemblies are 800
holes as small as 31⁄8-in. [7.94 cm] and barefoot currently undergoing assembly and testing. 600
completions, common throughout the Mideast 400
and other areas, where the logs can be used to In the past half-dozen years, much experience 200
monitor the reservoir or assess flood progression. has been gained in construction of a new full-size 0
quad combo string of nominal 35⁄8-in. [9.2-cm] Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

Six original series tool strings have been deployed, diameter and a hostile environment quad combo 2009 2010 2011 2012

and six production series tool strings are under string at 31⁄8-in. [7.94-cm] diameter. For the new Sources: BHI, Spears

construction. Tools are rated at 14,000 psi 2.35-in. [5.97-cm] string, all this expertise has Figure 1. Spears and Associates, Inc. show the
[103 MPa] and 300°F [149°C]. The four sensor been brought to bear to construct the equipment breakdown of drilling activity by rig type, with
sections are depicted in Fig. 2, along with their and fully characterize the measurements to a most of the increase deriving from rigs outfitted
for horizontal drilling. This trend is expected to
dimensions and measure points. When run as a level equal to that of the other strings. The full
continue in the second half of 2011.
quad combo in a standard wireline configuration,
37
H i g h e s t - Q u a l i t y Me asure me nt s wit h Environmental Corr ec tions

Figure 3. The UltraSlim gamma ray borehole


correction factor derived from this chart
accounts for the attenuation of formation gamma
rays for various combinations of borehole size and
mud density.

Figure 4. Mud electron density for the gamma ray


borehole absorption parameter calculation can
be estimated by entering the chart mud density in
pounds per gallon for muds weighted with barite,
hematite, or formate.
Figure 2. Four sensor sections of the UltraSlim™ quad combo are depicted. The overall length of the stack
is less than 68 ft [20.7 m] with wireline cable head. Auxiliary components, such as knuckles, flex joints,
swivels, battery, landing, or shock absorber subs are not shown in this figure.
The charts in Fig. 5 provide FKmud relationships
for several combinations of borehole size and
x = 1.27 ( dbh − 2.35) ρe _ mud (1)
occurs because the radioactive isotope 40K
mud weights for tools eccentered or centered
(which accounts for 0.0117% of naturally
in the borehole. To determine the borehole
Mud electron density can be found by entering occurring potassium atoms) produces 1460-keV
potassium contribution to the uncorrected gamma
the chart in Fig. 4 at the bottom with mud density gamma rays through one of its decay modes.
ray reading, enter the charts at the bottom with
in pounds per gallon to locate a point on the Eq. 2 is used to obtain a borehole absorption
mud weight and locate the borehole diameter by
appropriate relationship corresponding to the mud and mud potassium corrected gamma ray value,
interpolating between trend lines as necessary,
weighting material. GRc, by removing the borehole bias from the
then read FKmud on the vertical axis of the chart. It
uncorrected gamma ray reading, GRu, before
should be pointed out the gamma ray corrections
Various mud additives are used to achieve applying the borehole absorption correction
outlined above are applied by the logging
mud properties that lead to improved drilling factor, Cf, from the chart in Fig. 3.
software and both the uncorrected and corrected
efficiency, reduced formation damage and
clay swelling, and so forth. In many instances, GRc = (GRu − FKmudWK )C f (2)
gamma ray outputs are available in real time.

these additives only affect the gamma ray The Neutron Porosity Measurement
response through their contribution to borehole The borehole potassium bias in Equation 2 is
represented by the product of the API units per The thermal neutron porosity measurement is
attenuation of formation gamma rays. However, made with a standard 15Ci Americium-Berylium
if the additives contain potassium, the formation weight percentage of potassium in the mud,
FKmud, and the weight percentage of potassium neutron source and 3He proportional counters
gamma ray signal will be biased by an additional deployed in a dual-detector array. The detectors
signal from the borehole. This borehole signal in the mud, WK.

38
response for water- and oil-based mud systems
is dependent on various parameters. Individual
corrections have been derived for water- and oil-
based mud systems weighted with either barite
or hematite; for oil-based mixtures, corrections
have been derived for diesel, mineral, olefin,
paraffin, and ester-based oil.

Fig. 6 shows the derived ratio-to-porosity


transforms for the three primary rock matrix
types. Open circle symbols show Monte
Carlo modeling results, and triangle symbols
indicate measurements carried out in TRAC
Lab formations, including a water tank large
enough to represent an infinite water medium.
The ratio-to-porosity transforms are continuous
functions with continuous derivatives that
Figure 5. These charts show borehole potassium gamma ray contributions in API units per weight span the full range of porosity from zero to 100
percentage of potassium in the mud for eccentered tools (left-hand panel) and centralized tools (right- p.u. Although there are no formations with
hand) panel run in various combinations of borehole size and mud weight. less than zero porosity, there are formations
such as salt and anhydrite that produce ratios
provide a near-to-far ratio porosity response corrections were defined in terms of apparent smaller than the ratios of zero porosity quartz
comparable with standard-sized dual-spaced limestone porosity; additional corrections for sandstone, limestone and dolomite formations.
devices. Monte Carlo modeling with the MCNP formation fluid salinity or formation thermal To accommodate those situations, the porosity
program (Briesmeister 2000) was used to select neutron capture cross-section/hydrogen index, transforms extend below the zero porosity point
detector sizes, gas pressures, and source-detector which are dependent on the rock matrix were for interpretation purposes. Fig. 7 is a porosity
spacings for optimal performance and acceptable defined for the three primary matrix types. equivalence plot that shows the relationship
sensitivity to borehole environmental influences. between the porosity transforms for the three
Most of the environmental effects are dependent principal rock types. Accuracy and statistical
The basic response is defined as the ratio of on a single variable, however, there are a few precision of the neutron measurement are listed
near-to-far detector count rates and porosity exceptions. The influences of tool standoff and in Table 1 for three formations, three logging
at standard conditions consisting of a clean, formate mud density on the UltraSlim neutron speeds, and two resolution options.
freshwater-filled formation with the tool porosity readings are strongly dependent on the
eccentered in a smooth 8-in. [20-cm] diameter borehole size. The effect of mud weight on the Crossplots of bulk density vs. neutron porosity are
borehole filled with fresh water wherein the entire useful interpretation aids to estimate formation
system is at ambient temperature of 68°F [20°C]
and pressure of 14.7 psia [101 kPa]. Individual
ratio-to-porosity transforms were determined for
each of the primary rock matrix types (limestone,
quartz sandstone, and dolomite) from Monte
Carlo calculations that were normalized to
measurements performed in Halliburton’s Tool
Response and Characterization Laboratory (TRAC
Lab) (Gadeken et al. 2000) facility.

Environmental corrections are required for


non-standard logging conditions. Data from
approximately 2500 Monte Carlo simulations
were used to derive corrections for the UltraSlim
neutron tool for borehole diameters ranging from
4 to 16 in. [10 to 41 cm], mudcake thickness up
to 1 in. [2.54 cm], borehole fluid salinity, mud Figure 7. The true porosity for a quartz sandstone
Figure 6. The algorithms for converting near-to- or dolomite formation can be determined by
weight, formation temperatures up to 300°F far detector ratio to porosity for each of the three entering this chart with environmentally corrected
[150°C], formation pressure up to 20 kpsi primary types of rock matrix are derived from a apparent limestone porosity. The reverse may also
[138 mPa], and tool standoff. These traditional combination of lab measurements and Monte be accomplished with the chart.
Carlo modeling of neutron transport and detection.
39
H i g h e s t - Q u a l i t y Me asure me nt s wit h Environmental Corr ec tions

porosity and identify mineralogy. A crossplot for


freshwater-filled formations is given in the left-
hand panel of Fig. 8. The right-hand panel of
Fig. 8 provides a mapping of bulk density and
neutron porosity for saltwater-filled formations
that takes into account the differences in the
density of the pore fluid and the formation salinity
effects on the neutron log for each of the three
rock types.

Traditional environmental effects, (e.g., bore-


hole size, mudcake thickness, borehole salinity,
mud weight, formation temperature, formation
pressure, and tool standoff) can be treated
independently in limestone porosity units as
previously noted (Galford et al. 1988). Figure 8. Bulk density vs. UltraSlim neutron porosity crossplots for freshwater-filled formations on the left
Figs. 9 and 10 show examples of correction and saltwater-filled formations on the right.
nomographs for these environmental effects.
It is not necessary in most instances to use in the domain of interpretation because the salinity. Moreover, formation salinity corrections
these charts to make routine corrections to logs volumetric fraction of water in the pore space are different from the traditional environmental
because all of the environmental corrections is required to properly apply the correction. corrections because the amount of correction
can be applied in real-time while logging, or The formation salinity nomographs presented for a given formation porosity and water salinity
alternatively as a post-acquisition adjustment. in Fig. 10 apply to situations where the rock depends on the rock matrix. To determine the
Formation salinity corrections actually belong pores are completely filled with water of a given amount of correction for each condition, the

TABLE 1
Accuracy and Statistical Precision of Neutron, Density, and Photoelectric Measurements
Listed for Three Formations, Three Logging Speeds, and Two Processing Options
Measurement Bulk Density Pe Neutron
Principle Gamma Attenuation Neutron – Thermal Neutron
Range 1.0 to 3.1 gm/cc 0 to 20 -2 to 100 p.u.
Vertical Resolution (90%) Standard: 18 in. ( 46 cm) Standard: 19 in. (48 cm) 24 in. (standard), 12 in. (enhanced)
Enhanced: 9 in. (34 cm) Enhanced 9 in. (34 cm) 61 cm (standard), 30.5 cm (enhanced)
High Resolution: N/A High Resolution: 6 in. (15 cm)
Depth of Investigation (50%) 1.5 in. (3.81 cm) 0.5 in. (1.27 cm) 6 in. (15.2 cm)
Precision (1 SD) (g/cc) (p.u.)

ρ =2.60 gm/cc, Pe=5.0, φ =3 p.u. Standard* Enhanced # Standard* Enhanced # Standard Enhanced #
At 10 ft/min ± 0.007 ± 0.010 ± 0.08 ± 0.11 ± < 0.1 ± < 0.1
At 30 ft/min ± 0.010 ± 0.016 ± 0.12 ± 0.16 ± < 0.1 ± <0.1
At 60 ft/min ± 0.013 ± 0.021 ± 0.15 ± 0.21 ± < 0.1 ±< 0.1

ρ =2.20 gm/cc, Pe=3.5, φ =20 p.u. ± 0.006 ± 0.008 ± 0.05 ± 0.07 ± 0.3 ± 0.3
At 10 ft/min ± 0.008 ± 0.012 ± 0.07 ± 0.10 ± 0.6 ± 0.6
At 30 ft/min ± 0.010 ± 0.016 ± 0.09 ± 0.12 ± 0.9 ± 0.9
At 60 ft/min
± 0.005 ± 0.006 ± 0.03 ± 0.04 ± 3.0 ± 3.0
ρ =1.69 gm/cc, Pe=1.8, φ =60 p.u. ± 0.006 ± 0.009 ± 0.04 ± 0.05 ± 5.3 ± 5.3
At 10 ft/min ± 0.008 ± 0.011 ± 0.05 ± 0.06 ± 7.5 ± 7.5
At 30 ft/min
At 60 ft/min

Accuracy (with no standoff) ± 0.015 gm/cc +0.010* [Mud density ± 0.1 in light muds with Pe<5.5 ± 5% or ± 1 p.u., whichever is greater
(g/cc) – 1]
Primary Curves RHOB, DRHO, Pe, CORM, CORP, Caliper Neutron Porosity on Limestone (NPHL),
Sandstone (NPHS), and Dolomite (NPHD)
Matrix
Secondary Curves DPe , QLS, QSS, GR, Caliper, Standoff, Pad Extension (PCAL), Arm Extension Near-to-Far Detector Count Rate Ratio,
(ACAL), and Density Porosity (DPHI) Near-and-Far Detector Count Rates
* Measured using the standard 1.75 ft filter, # High axial resolution mode
40
pad housing to let gamma rays pass into the
detectors unimpeded. The collimators are plugged
at the pad’s surface with low-density windows
to prevent borehole fluids from entering the tool.
The window over the near detector is composed
of low-atomic-number materials, which facilitates
the passage of low-energy gamma rays that are
required to make a photoelectric Pe measurement.

Each detector is comprised of a sodium-iodide


crystal attached to a photomultiplier tube.
When gamma rays interact in the crystals,
light is emitted that is converted to electronic
signals by the photomultipliers. Electronics
amplify the signals and store them according
to their amplitude, which is proportional to the
Figure 10. Tool standoff and formation salinity energy deposited in the crystals. The voltage
correction charts for UltraSlim neutron logs.
signal generated by a gamma ray varies with
Figure 9. Environmental correction charts for
UltraSlim neutron logs are also available in temperature due to variations in the physical
When designing density tools, it is important
interactive format. properties of the crystal and photomultiplier tube.
to recognize that gamma rays coming from
To compensate for that variation, small 137Cs
the back and sides of the tool are sensitive
charts are entered with the indicated uncorrected sources are located inside the crystal housings.
to the borehole diameter and mud properties.
neutron porosity index and the corresponding They provide 662 keV gamma rays, some of which
Corrections can be applied to account for these
environmental parameter; uncorrected values for down-scatter to lower energies. Electronics
sensitivities, but uncertainties in these variables
all three matrix types are included in the standard adjust the system gain to keep the 662 keV
make it impossible to compute exact corrections.
set of output curves for the UltraSlim neutron gamma rays at the same voltage level. As a
Consequently, it is desirable to shield the
tool. Each correction is then found by following result of this stabilization process, the voltage
detectors from these gamma rays. Because this
the trend curves to the standard condition, which signal generated by a gamma ray has very little
tool has such a small diameter, the pad housing
is marked with an asterisk on the right of each sensitivity to temperature.
was fabricated from a tungsten alloy, which
chart. The total correction is found by summing
provided the maximum amount of shielding.
the individual values. Voltage signals from each detector are sorted
into eight different groups, which correspond
The tool measurement is made with a 137Cs
The Density and Photoelectric to eight energy ranges of the gamma rays. In
source that has a nominal strength of 1.78 Ci
Measurements each measurement period, eight count rates that
[65.9 GBq] when new. Two detectors placed at
Instrumentation for the density measurement is correspond to these energy ranges are generated
different distances from the source are used to
housed in a pad that is approximately two feet for each detector. Fig. 11 illustrates these
detect gamma rays that scatter back into the tool.
[61 cm] in length. The pad is retracted into windows with spectra from a marble formation.
The tool is designed to focus the measurement
a mandrel when not logging. With the pad Windows W5-W8 are used in the stabilization
on the front of the pad, which is touching the
retracted, the mandrel can pass through a process just discussed, and windows W1-W4
formation. Collimators are machined into the
2.35 in. [5.97 cm] circle. When logging, springs
force the pad out of the mandrel and up against
the borehole wall while simultaneously extending
a backup arm out the other side to eccenter the
entire tool string. Under normal operation, the
pad extension is limited to about one-half inch
[1.3 cm] so as not to induce standoff on the
neutron detectors. However, when washouts
are encountered, the pad will extend out up to
two inches [5.1 cm] to maintain contact with
the borehole wall. The pad and backup-arm
extensions are measured independently and
added to the mandrel diameter to provide a
borehole-caliper measurement. The pad extension
Figure 11. Spectra were recorded by two density detectors in a marble formation. The energy windows
is an optional output, which is useful as a
W5-W8 are used for gain control of the detector system and windows W1-W4 are used to measure
diagnostic tool. density and Pe.

41
H i g h e s t - Q u a l i t y Me asure me nt s wit h Environmental Corr ec tions

are used to compute density and Pe. Several [1.2, 1.5, and 1.9 g/cm3], and an 18.3-lbm/gal The Borehole Sonic Array Measurement
corrections are made to the count rates in [2.2-g/cm3] hematite mud. Accuracy of the density The piezoelectric sensor section of the UltraSlimTM
W1-W4 before they are used in computations. measurement for various standoffs is shown in borehole sonic array tool, or SBSAT, consists of
These account for dead time, pulse pile-up, Fig. 12 for marble and fused-silica formations. five receivers between two transmitters. The
background from the stabilization sources, (The fused silica formation is man-made from receivers are 0.5 ft [0.15 m] apart (see Fig. 2) and
streaming from the 1.78 Ci source, and small SiO2 and has a matrix density of 2.2 g/cm3. It are positioned to achieve transmitter-to-receiver
temperature drifts that arise from changes in has some porosity, so its density is similar to spacing from 3 to 5 ft [0.91 to 1.52 m]. The
detector resolutions. a 42-p.u. formation. The borehole is sealed to transmitters fire alternately 40 ms apart, with one
prevent borehole fluids from invading.) The errors pair of firings occurring in every 200-ms frame.
The tool is capable of measuring formation are larger for the marble formation, because the Waveforms are recorded in 4-μs sample intervals
density because the number of gamma rays contrast in density between the formation and with 512 samples per detector. At the standard
detected in windows 3 and 4 depend primarily on borehole fluid is larger than for the fused-silica wireline rate of 4 samples/ft, a logging speed of
the electron density of the formation. However, formation. The errors are minimal out to a quarter 75 ft/min is the fastest logging speed allowable
measured count rates in these windows also of an inch [0.6 cm], but become significant for the to achieve one pair of firings per depth sample.
depend on the photoelectric cross section of the marble formation at half an inch [1.3 cm]. Fortun- For metric depth units, the standard sample rate is
formation, as well as the binding energy of the ately, mudcakes are rarely thicker than 0.25 in. 25% larger at 10 samples/m, so a logging speed
electrons to their atoms and the cross section for
coherent scattering (Moake 2011). In addition,
in the presence of mudcake (or standoff), the
count rates are dependent on the density and Pe
of the mudcake, as well as the thickness of the
mudcake. Thus, the count rates have a strong
dependence on five different variables. Originally,
two-detector density tools were processed with
the spine-and-ribs technique, which is a two-
dimensional solution. That clearly is inadequate,
especially with the wide variety of muds in
use today. Consequently, the four-dimensional
solution developed by Moake (1991) is employed. Figure 12. Errors are shown at various standoffs for the density measurement in marble and fused silica
This works well for small standoffs, where the (similar to 42 p.u. sand) formations.
five variables can be combined into only four
independent variables, one of which is the The borehole correction was studied using of 30 m/min achieves exactly one pair of firings
formation density. Using this technique, windows marble, aluminum, quartzite, and magnesium per depth sample.
W2-W4 from each detector are combined to formations with five different diameters ranging
compute the formation density. An additional from 5.8 to 14 in. [14.7 to 35.6 cm]. The correction For this slim tool, there was a particular
correction to account for the borehole diameter increases with the density contrast between challenge in designing a transmitter that could be
and mud properties is still required, and it is the formation and mud, and it increases as accommodated in the 2.35-in. [5.97 cm] maximum
applied to the density computed with the four- the borehole diameter deviates from the 8-in. tool diameter. After extensive laboratory and
dimensional solution. diameter in which the tool was characterized. modeling work, it boiled down to a choice
The correction was found to correlate with the between two excellent transmitter designs, on
The Pe measurement is made using windows parameter vbh, which is defined as par with transmitters of the larger tools, with
W1 and W4, along with the measured density. center frequencies at 22 kHz and 28 kHz. In a
⎛ 1 1 ⎞
The technique, which is described elsewhere
vbh ≡ ( ρe − ρe _ mud )ρe ⎜ − ⎟ (3) large water tank, transmitter output of each
(Moake 2011), makes extensive use of computer ⎝ dbh 8 ⎠
modeling to derive a response equation that
works for a wide range of materials. In light muds where ρe is the electron density of the formation,
with good borehole contact, the measurement is ρe_mud is the electron density of the mud, and dbh
accurate from 0.1 to 15. is the diameter of the borehole. The correction
is plotted in Fig. 13 for all the muds and
The tool was characterized using six formations formations studied. The correction is less than
that ranged in density from 1.68 to 3.02 g/cm3, 0.02 g/cm3, which is very good, especially
as well as a water tank, which had a density of considering the limited space available for
1.00 g/cm3. The boreholes were filled with five shielding. The accuracy and statistical precision
different fluids: water, barite muds with of the density and Pe measurements are given Figure 13. Borehole correction for the density
weights of 10.0, 12.9, and 16.0 lbm/gal in Table 1. measurement.

42
Figure 14. Frequency distributions for two specimens each of two candidate detectors were determined
using hydrophones at a close spacing in a large water tank.
Figure 15. In the Odessa test well with fast
carbonate formations (left), the 28 kHz transmitter
model was recorded using hydrophones spaced slowness logs in the test wells, shown in Fig. 15,
(red) yields higher peak amplitudes. However, in
1 ft [0.3 m] from the source. This produced the but larger signal amplitudes in the slightly slower the somewhat slower sand-shale sequences in the
frequency distributions shown in Fig. 14. formations and the steady peak frequency seen in Ft. Worth test well (right), the 22 kHz transmitter
the receivers favor the 22 kHz transmitter. The (blue) yields higher peak amplitudes.
To choose the correct transmitter, it was decided 22 kHz transmitter was chosen for use in all tools.
uencies, the slope of the line is unity, which
to construct tool #1 with the 22 kHz transmitter
demonstrates a perfect correspondence between
and tool #2 with the 28 kHz transmitter and The Array Induction Measurement
the two measurements.
record some logs in a pair of Halliburton’s test The Slim Array Induction Tool (SACRt) is the
wells. The Odessa test well has fast carbonate latest member of the array induction family.
Log Examples
formations and the Ft. Worth test well has Its design follows the same methodology as
Fort Worth Test Well. Initial logging tests were
somewhat slower sand-shale sequences. Both previous arrays (standard-size ACRt and hostile
carried out in a test well located at Halliburton’s
transmitters enable the correct compressional environment HACRt) except that it is a thinner
Training Center in Ft. Worth, Texas. The well
tool to be compatible with the UltraSlim family of
was drilled with an 8.75-in. [22.2-cm] bit and
tools. It has the same array coil configuration and
Packets
9.2-lbm/gal [1.1-g/cm3] low-salinity mud. This
spacings, and works at the same frequencies as
4 samples per foot
Time to Depth WLI: T

Logging Time Pre-Processing


ACRT
well was selected because it provided an
all the other induction arrays of the family (Xiao
Calibration Values Calibration Application opportunity to compare the UltraSlim quad-
et al., 2006). Processing of the raw data, depicted
Temp Coefficients Temperature Correction combo toolstring responses with logs previously
in Fig. 16, is identical to the processing of the
Depth Alignment WLI: I recorded using larger, standard-size wireline
ACRT
other tools, except that a new borehole correction
Data Library
Skin Effect Correction
tools. In addition, repeat passes were made using
WLI: R
ACRT database has been calculated to accurately
Borehole Correction
Debug two separate tool string configurations in these
Parameters
RT90, RT60, RT30, etc. remove the borehole effect considering the new
Software Focusing
WLI: R tests to evaluate repeatability performance.
Radial Inversion
RT, RXO, DI ACRT tool diameter.
Fig. 19 presents the results from these tests in
Figure 16. The processing methodology for the The new working envelope of the tool is
an overlay format for each sensor. Track I shows
new induction tool is the same as that used for the presented in Fig. 17. In practice, the slim tool
induction tools – standard size and hostile. four gamma ray passes spanning a sand-shale
has identical response to the standard-size
formation sequence. Presented in Tracks II – VI
tool within its working envelope. This can be
are results from the dual-spaced thermal
evidenced on Track V of Fig. 19, where responses
neutron, bulk density, photoelectric absorption,
of the deep-reading resistivity array (Rt90) of
90-in. [2.3-m] array resistivity at 2 ft [0.6-m]
the two tools are compared. It can be seen that
resolution, and sonic compressional travel time
the response of the new slim tool overlays that
measurements, respectively. All of the UltraSlim
of the standard-size tool. This figure also shows
logging passes were recorded at an elevated
the repeatability of the new tool to be excellent.
logging speed of 45 ft/min [13.7 m/min]. The
Cross-plots of raw data from the three longest
overlays in each track include corresponding logs
arrays of the two tools are displayed in Fig. 18
Figure 17. Within the working envelope of the new
obtained with standard, 3.625-in. [9.2-cm]
for all the frequencies and for the longer arrays.
tool, the response is identical to that of the other LOGIQ tools.
induction tools in the family. It can be seen that for all the arrays and freq-

43
H i g h e s t - Q u a l i t y Me asure me nt s wit h Environmental Corr ec tions

The interval includes a sandstone reservoir and a similarly affected by these minerals whose bulk
shaly carbonate together with neighboring shales. densities are greater than that of quartz.
A non-productive low-porosity shaly carbonate
occurs near the bottom of the interval from Summary and Conclusions
XX236 – XX190. The reservoir between XX146 A new quad combo suite of 2.35-in. [5.97-cm]
and XX080 is a Pennsylvanian Age complex- diameter logging tools and associated peripheral
mineralogy sandstone deposited in a deltaic equipment has been designed, constructed, and
environment that initially produces gas and deployed. The suite is designed to be conveyed
transitions to oil over time. The neutron-density in normal wireline operations or through drill
gas crossover that would normally be observed pipe. The full complement of sensor data can be
in a productive gas zone is suppressed in this telemetered to the surface via wireline or saved
example by the presence of auxiliary minerals in memory downhole. Battery and latching subs to
such as siderite, biotite, glauconite, and varying allow deployment at the BHA without a wireline
amounts of illite clay. The gas crossover is are being assembled and tested.
diminished because all of these minerals have
substantial neutron responses that increase the Tool design, algorithms, and software capabilities
apparent neutron porosity and the density log is are on par with standard-size equipment and a
Figure 18. The three deepest-reading slim tool
responses are comparable to the standard-size
tool responses for all three frequencies. Numbers
on the left refer to the array length.

As seen in this example, excellent repeatability is


achieved for the resistivity and acoustic logs. The
nuclear logs are additionally subject to counting
statistical variations. Even so, the repeatability
of the nuclear measurements is quite good.
Moreover, the agreement between the UltraSlim
log responses and those of the larger LOGIQ tools
is exceptional.

Commercial Log Example. The new toolstring has


been run in more than 20 commercial wells thus
far. In at least one instance, the UltraSlim tools
were successfully run through drill pipe following
a bridge-out with standard-size tools. Shown in
Fig. 20 is a portion of triple combo log that was
recorded below drill pipe during one of the first
commercial through drill pipe operations. The
well was drilled with an 8.75-in. [22.2 cm] bit and
9.8-lbm/gal [1.2 g/cm3] water-based mud whose
salinity was equivalent to a 7500-ppm NaCl brine.
Bottom hole temperature in the well was 193°F
[89°C]. The log was recorded at a nominal logging
speed of 30 ft/min [9.1 m/min].

Track I of Fig. 20 shows gamma ray and caliper


logs, Track II presents the array resistivity
curves at 2 ft [0.6-m] resolution, Track III shows
neutron porosity and bulk density presented
on a compatible limestone porosity scale, a
photoelectric absorption (Pe) curve is shown in
the left-hand half of Track III, and the bulk density
Figure 19. Overlay of repeat logs obtained with two separate UltraSlim toolstrings from a test well located
correction (DRHO) curve is provided in the right- at Halliburton’s Training Center in Ft. Worth, Texas show good repeatability. The responses compare
hand half of Track III. favorably with those of the standard-size LOGIQ tools.

44
new hostile suite. Each of the sensor sections has
accuracy and precision similar to its larger sibling.
In the case of the compressional slowness
measurement, the transmitter section provides an
improved signature over the full-size equivalent
tool. The full suite of environmental corrections is
available in real time.

The toolstrings have performed very well on


about 20 logging runs. Comparison runs have
been made with standard-size equipment in
test wells and they demonstrate excellent
repeatability. Essentially the only differences
between test-well logs acquired with standard
equipment and the new slim tools are the
statistical variations associated with counting
statistics of the nuclear logs.

An example commercial log is indicative of


the good log quality seen on more than 20
commercial logging operations.

With six toolstrings busily deployed and six


more under construction, much experience will Figure 20. UltraSlim triple combo log from a
through drill pipe operation carried out in a
be gained in the coming months in deviated commercial well. The zone includes a complex-
and horizontal holes, slim holes, and barefoot mineralogy sandstone reservoir near the top of the
completions. Some will be below drill pipe and interval and a shaly carbonate formation near the
others in normal wireline operations. bottom of the displayed interval.

Authors
Jerome Truax is the Halliburton patents related to conventional nuclear logging methods in the oil industry. He began his career in 1980 with
Petrophysics Technology Fellow and NMR logging applications. Jim is a member of the Gearhart Industries in Brazil, and has worked as a
and Manager of the Petrophysics Society of Petrophysicists and Well Log Analysts and field engineer and a technical manager. David moved
Technology Development group. the Society of Petroleum Engineers. to Texas and has held several positions within the
Jerry has a BS degree in geological organization, including director of training, technical
engineering (1978, University Gordon L. Moake is a Chief assistant to the vice president of operations, and
of Minnesota), an MA degree in oceanography Scientific Advisor for Formation research. He holds several patents and has authored
and limnology (1991, Western Connecticut State Evaluation in the Halliburton various papers in signal and image processing for the
University), and a PhD in petrophysics (1995, Delft Drilling and Evaluation division. logging industry. David holds an EE degree and an MSEE
Technical University), and has been with Halliburton His primary focus is the degree from the University of Madison, Wisconsin.
since 1998. His experience includes field operations, development of wireline and
engineering, research, and log interpretation in both LWD density and neutron-porosity tools, although Ron Cherry is the Halliburton Open
wireline and logging while drilling, mostly specializing he sometimes works on non-nuclear projects. Before Hole Strategic Business Manager.
in pulsed neutron logging technology and other nuclear joining Halliburton in 1984, he worked four years After receiving a BS degree in
applications. Jerry has written or contributed to a dozen at Baker Tubular, developing electromagnetic flaw electrical engineering from the
patents and many technical papers, and is a member of detectors. Gordon obtained BS degrees in math and in University of Tulsa in 1975, Ron
SPWLA and SPE. physics from the University of Wisconsin, and MS and joined Schlumberger Well Services
PhD (1979) degrees in physics from Purdue University. where he held various positions managing operations,
Jim Galford is a Chief Scientific A member of SPWLA and SPE, he holds 18 U.S. patents marketing, and technology. In 1994, he joined NUMAR
Advisor in the LWD and WL Sensor related to the oil and gas industry. and served as the general manager until the company
Physics group at Halliburton. was acquired by Halliburton in 1997. Ron has been
Previously, he worked on nuclear David Torres is the Resistivity responsible for new product developments associated
magnetic resonance (NMR) Global Product Champion for with NMR (MRIL products) for the last several years,
petrophysical applications and Halliburton. In this role, David is and is now responsible for all open hole logging
interpretation. He holds a BS degree in physics from responsible for overseeing the products and services.
West Virginia University. Jim has written several development, testing, and marketing
technical papers on various logging applications for of wireline resistivity logging. For
conventional nuclear and magnetic resonance imaging more than two decades, David has worked in various
logs. In addition, he has contributed to a number of management and research and development positions

45
H i g h e s t - Q u a l i t y Me asure me nt s wit h Environmental Corr ec tions

Authors, cont’d
Batakrishna Mandal is a Chief Arabinda Mishra is a Project Alberto Quintero is the Project
Scientific Advisor in the Sensor Manager in the Wireline and Development Manager at the
Physics group of Halliburton Perforating Technology group, Singapore Technology Center,
Wireline & Perforating. He holds a managing the development projects responsible for the development
BS degree in Physics from Calcutta for UltraSlim tools and the UltraSlim of new technology. He joined
University, India, an MS degree Conveyance and memory mode Halliburton as a Maintenance
in Exploration Geophysics from the Indian Institute of operation. Before starting his current role in April 2010, Engineer in 1984 and worked in several field operations
Technology and a PhD degree in Geophysics from Saint Arabinda held several positions on land and offshore, including Colombia, Ecuador, Sudan, Dubai, Turkey,
Louis University. He was a Postdoctoral Associate starting as a Field Engineer in 1991. He holds a Master’s Oman and Egypt. Alberto moved to Houston to work in
(1987-1989) and Research Scientist (1990-1993) at Degree (1983) in Engineering from the Indian Institute of the Technology Center in 1997 and then to the Electrical
Massachusetts Institute of Technology. Since 1994, Technology at Kharagpur, India and has over 25 years of Engineering Development Group in 1998. During this
Bata has been involved with various developments of experience in the oil and gas industry. time, he obtained a patent for the development of the
acoustic and ultrasonic tools. He holds 26 patents, and ADSL Wireline Telemetry, used in the INSITE logging
has published more than 15 papers in referred journals, Luis San Martin is a Senior tools. He was also involved with the development of
and presented over 30 times at international meetings. Technical Professional Leader at several other tools, including, the ACRt, CAST-M and
Halliburton. He joined Halliburton in UltraSlim tools.
1998, after he received his PhD in
physics from the University of Illinois
at Urbana-Champaign. Luis works in
the Sensor Physics Group, where his main interest is in
electromagnetic instruments, both in modeling
and hardware.

Acknowledgements References
Deployment of the UltraSlim services and construction of this paper Belknap, W.B. et al. 1959. API Calibration Facility for Nuclear Logs. Amer.
owe much to the cadre of technicians working in the Tool Response and Petr. Inst. Drilling and Production Practice. 289-316.
Characterization (TRAC) lab at Halliburton’s facility in Houston. Tim Plasek Briesmeister, J. F., Editor. 2000. MCNP – A General Monte Carlo N-particle
is the manager of the lab, and full-time technicians Aldo Garcia and Shilo Transport Code, Version 4C. LA-13709M.
Black contributed to the work. They also supervised former Texas A&M Gadeken, L. et al. 2000. New Facilities for Characterizing and Validating
engineering students Dustin Durbin and Donald Durbin, who performed Logging Instrument Performance. Paper presented at the 41st Annual
SPWLA Logging Symposium, Dallas, Texas, 4-7 June.
much of the natural gamma ray measurements. The twin brothers were
so enthralled with the oil field that, after their graduation in May 2011, Galford, J.E. et al. 1988. Improved Environmental Corrections for
Compensated Neutron Logs. SPE Formation Evaluation 3 (2): 371-376.
they signed on with Halliburton in Duncan, Oklahoma. Finally, Sperry field
SPE 15540-PA
engineers and true professionals Tiko Davis and Terry Bossier helped us
Galford, J., Truax, J., and Moake, G.L. 2005. Borehole and Formation
out during a slack time in the Gulf of Mexico in late 2010 and early 2011.
Invasion Effects of Formate-Based Mud Systems on Nuclear Logs. Paper
presented at the 46th Annual SPWLA Logging Symposium, New Orleans,
Nomenclature Louisiana, 26-29 June.
GRc Gamma ray borehole absorption correction factor Moake, G.L. 1991. A New Approach to Determining Compensated Density
dbh Borehole diameter, inches and Pe Values with a Spectral-Density Tool. Paper 1991-Z presented at the
32nd Annual SPWLA Logging Symposium, Midland, Texas, 16-19 June.
FKmud Mud potassium gamma ray signal per K concentration unit,
API units/Wt. % K Moake, G.L. 2011. Using Computer Modeling to Generate Accurate Pe
Equations. Paper 2011-MMM presented at the 52nd Annual SPWLA
GRc Corrected gamma ray, API units
Logging Symposium, Colorado Springs, Colorado, 14-18 May.
GRu Uncorrected gamma ray, API units
W K Potassium concentration in the mud, Wt. % Spears and Associates, Inc. 2010. Drilling and Production Outlook 12
x Gamma ray borehole absorption parameter for gamma ray Xiao, J. et al. 2006. A New Asymmetrical Array Induction Logging Tool.
log corrections, g/cm2 Paper SPE 101930-MS presented at the SPE Annual Technical Conference
and Exhibition, San Antonio, Texas, 24-27 September.
vbh Borehole parameter for density log correction, g2/(cm6-in.)
Pe Formation photoelectric value, unitless Wahl, J.S. 1983. Gamma Ray Logging. Geophysics 48 (11): 1536–1550.
ρe Formation electron density, g/cm3
ρe_mud Mud electron density, g/cm3

46
Dielec tr ic Logging Energized by “Cell Ph one” Tec h n ology

A Modern Microwave Formation


Hence, the measurements have a shallow depth
of investigation and probe only the invaded zone.
When combined with Rxo devices, dielectric

Evaluation Sensor and Its Applications measurements can be used to compute the
cementation exponent, m, through the use of the

in Reservoir Evaluation Archie equation. Dielectric logging tools provide


high vertical resolution, which is beneficial in
thinly laminated zones.
Michael Bittar, Jing Li, Gary Kainer, Ron Cherry, and David Torres, Halliburton;
David McCoy, Vintage Production California
The use of dielectric measurements waned
Presented at the SPWLA 51st Annual Logging Symposium held Perth, Australia, June 19-23, 2010
Copyright 2010, held jointly by the SWPLA and the submitting authors
through the years, and by the beginning of the
21st century, their use had become quite limited,
largely because of their sensitivity to borehole
Abstract rugosity, mudcake, and standoff effects. More
In the development of a reservoir, it is critical to know the hydrocarbon saturation within the pore space. recently, with the advent of modern miniaturized
Generally, this can be obtained by measuring the resistivity of the formation and applying Archie’s electronics developed for mobile communications,
equation to estimate the water saturation. In many instances, such as when the reservoir connate the components for these logging tools can easily
water is fresh or when the water salinity has been changed by injection water, estimating the water fit within an articulated pad. This development
saturation from resistivity logs becomes very difficult. However, water has a unique physical property provides better contact with the formation, and
that differentiates it from any other reservoir fluid. Its dielectric constant is an order of magnitude higher when applied with symmetric transmitter to
than any other material in the reservoir. This property is important in the propagation of electromagnetic receiver configurations, the detrimental effects
signals in the microwave range. When a very high frequency signal in the microwave range is used to of borehole rugosity are substantially reduced.
stimulate the formation, the propagation properties of this signal is highly affected by the amount of
water in the formation. Consequently, the measurement of the propagation properties of microwaves can Why Logging at Microwave Frequencies
be used to measure the amount of water in a formation. An electromagnetic (EM) wave propagating
through a homogeneous material changes its
This paper discusses the use of a new formation evaluation sensor that operates in the microwave amplitude and phase as (Eq. 1).
range (1 GHz) to differentiate the fluids in the reservoir. The sensor consists of three receiving antennas
positioned between two transmitting antennas. All five antennas and the related electronics are mounted E ( z , t )  E 0 e i ( kz t )
in an articulated pad that is extended toward the formation. The electronics and signal processing (1)
provides logs of formation dielectric constant and formation conductivity, which can be used to calculate k    0  *
the bulk volume of water in the flushed zone. The configuration of the antennas offers compensation for Where ω = 2πƒ and ƒ is the frequency of the
tool standoff, mud cake effects, and borehole rugosity. In addition, the state-of-the-art electronics has EM wave, μ is the magnetic susceptibility of the
taken advantage of the latest developments in modern mobile communication technology to improve the material, ε0 = 8.85x10-12 is the dielectric constant
operating range and accuracy of the measurement. of free space. ε* is a property of the material,
referred to as its complex dielectric constant; it is
This new modern microwave formation evaluation tool can be used to perform the following: defined as the following complex number:
• Determine flushed zone water saturation (Sxo) and oil mobility
• Determine water saturation (Sw) in fresh water, water injected, and heavy oil reservoirs   
• Analyze thin beds and low resistive pay *  '  i     dl  (2)
• Quantify V-shale on sand shale sequences   0 
• Perform carbonate analysis, calculation of cementation exponent “m” The real part of this equation ε’ is the dielectric
• Provide an alternative to micro-resistivity logs constant of the material in the traditional sense.
The imaginary part is a function of the frequency
Introduction resistivity contrast between water-bearing zones of the EM wave ƒ and two properties of the
Measuring the dielectric properties of and hydrocarbon-bearing zones is very small. material in question: its conductivity σ and its
formations began in the late 1970s and showed Consequently, the uncertainty of the computed dielectric relaxation losses εdl.
promise in the ability to distinguish water from water saturation is very high. When the formation
hydrocarbons. Because the dielectric properties water salinity is unknown or has been altered Fig. 1 provides a graphical representation of
of water are significantly different from the by injection into nearby wells, alternative the behavior with frequency of the real and
dielectric properties of hydrocarbons and from interpretation methods that do not rely on imaginary parts of ε* for water of different
those of the formation matrix, dielectric logging resistivity measurements become necessary. resistivities. ε’ (red dashed line) is constant,
tools can provide a measurement of the quantity however, the imaginary part (solid lines) changes
of water in the region investigated by the tool. The transmitter-to-receiver spacing of a dielectric with frequency and conductivity and leads to the
When the water in the formation is fresh, the tool is small, and it works at high frequencies. following observations:

47
D i e l e c t r i c L o g g i ng Ene rgiz e d by “ Ce l l P hone” Tech nology

~
• For frequencies up to approximately 100 KHz, equivalent resistivity R defined by: the desired properties (R, ε’) can be performed by
the imaginary term is 100 times greater than means of look-up tables. The complex dielectric
18 ⎛ 18 ⎞ R
the real part, meaning that the propagation of
= ⎜ + ε dl ⎟, ⇒ R = (4) constant of the formation is a function of these
the EM wave is dominated by the imaginary R ⎝ R
 ⎠ R two quantities; consequently, measuring the
1+ ε dl
part and the dielectric constants of the material 18 attenuation and phase shift experienced by a
do not play a significant role on its propagation. 1GHz EM wave enables us to determine the
Consequently, the conductivity of the formation Fig. 2 shows the relationship between the complex dielectric constant of a formation ε*for.
is most important at these frequencies. Modern true and equivalent resistivities, described by
and traditional induction and laterolog tools Equation 4, for water with different resistivities. 700

~ R=0.1 εr =90
operate at these frequencies. For low resistivity values R ≈ R and for large 600
R=0.2
R=0.15 εr =60
~ ~ R=0.3 εr =30

resistivity values, R tends asymptotically to Rmax


500 R=1000 R=0.5 εr =5
εr =3

Phase (Deg)
400
100000 given by:
300

 
10000
 R  18 200

~
Rmax  lim   100

(5)
1000
R   R   dl 0

 1   dl
0 10 20 30 40 50 60 70 80
 Attenuation (dB)
100  18 
Figure 3. Relationship between attenuation
10 For example, in the case of fresh water, εdl = 3.7, (x axis) and phase shift (y axis) with respect to the
~
then Rmax= 4.86. material dielectric constant and its resistivity; pink
1 lines indicate lines of constant resistivity, and blue
103 106 109
In summary, Equations 1 and 3 show that an lines indicate the dielectric constant.
Frequency (hz)
EM wave propagating through a material will
Figure 1. Max ε’ (red dashed) for water. The be attenuated and experience a phase shift. Table 1 shows typical values ε’, ε’’ for common
change of the imaginary term of ε* with frequency At 1GHz, the amount of attenuation and phase materials encountered in formation evaluation.
and water of different resistivities. shift are functions of intrinsic properties of The dielectric constant for water is at least 10
the material, including its dielectric constant, times greater than the dielectric constant of any
• In the MHz range, depending upon the conductivity, and dielectric relaxation losses. other material. This means that at 1GHz, water
water resistivity and frequency, both the By measuring the attenuation and phase shift will be the dominant material in the behavior
real and imaginary parts can be important; experienced, it is possible to obtain Resistivity of the EM wave; consequently, the response of
consequently, both the dielectric and the R and the dielectric constant ε’ of the material. a tool working at this frequency will be mostly
conductivity of the material contribute to the At high resistivities, the relaxation dielectric influenced by the water content of the formation.
propagation of an EM wave. losses must be considered to obtain an accurate
• At 1 GHz, the relative importance of the R reading. TABLE 1
Dielectric constant for several materials
real and imaginary parts reverses. Now, the encountered in formation evaluation
dielectric constant of the material ε’ has a The relationship between (attenuation and
Material ε’ ε’’ (1GHz)
stronger influence in the complex dielectric phase shift) and (R and ε’) can be modeled and is
mapped in Fig. 3. These curves can be stored as Water (T, KPPM) fig 4 56 to 80 0.2 to 20
constant. Equation 2 becomes:
tables in a computer, and the mapping from the Air, Gas 1 0.0
*  18 
' measured quantities (attenuation and phase) to
    i   dl  (3) Oil 2.0 to 2.2 0.0
R  Sandstone 4.65 0.1

100 Limestone 7.5 to 8 0.01


where R denotes formation resistivity. In
Dolostone 6.8 0.01
addition, at this frequency and low resistivities,
the dielectric relaxation losses do not affect the Anhydrite 6.3 0.001
10
R (ohm-m)

imaginary part of the complex dielectric constant Pyrite 80 200


(both curves with and without εdl lie on top of one Berea Sandstone, 12 7.0
another); however, at higher resistivities (such 1 ϕ=0.2, 30kppm, 77°F
as 10 ohm-m), the dielectric losses become an
important part of the imaginary term (both The dielectric constant for water changes as a
curves diverge). 0.1 result of salinity and temperature. Fig. 4 shows
0.1 1 10 100 this relationship. As the salinity increases,
R (ohm-m) the dielectric constant decreases; when the
To further understand the effect of the
dielectric relaxation losses, the imaginary Figure 2. The effect of the relaxation dielectric temperature increases, the dielectric constant of
term of Equation 3 can be replaced by an losses at 1 GHz for different water resistivities. water decreases.

48
ε*matrix = parameters whose values are local investigation and the minimum signal that can
εwater(Kppm, ˚F)
90
knowledge (see Table 1). be accurately measured, especially in highly
80
50˚F
100˚F
ε*HC = parameters whose values are local attenuating formations.
knowledge (see Table 1).
Dielectric Constant

70 150˚F
200˚F
60
250˚F
ε*mf = complex dielectric constant of mud
50
40
300˚F filtrate in the zone investigated by the
30 tool; as previously indicated, its value is
20 a function of temperature and salinity as
10 shown in Fig. 4.
0
0 50 100 150 200 250 300
ϕ = total formation porosity that can be
Water Salinity (Kppm) obtained by density/neutron tools.
Sx0 = water saturation of the flushed zone and Figure 5. Dielectric pad showing the location of
Figure 4. The dielectric constant of water as a can be obtained by solving equation 7: transmitters, receivers and temperature sensor.
function of its salinity and temperature.

ε *for − (1− φ ) ε matrix


* *
− φ ε HC Fig. 6 shows the extended dynamic range of
Relationship of the complex dielectric constant Sx 0 = (8) using a three-receiver design. The region in
of a formation ε*for and water saturation Sxo.
We have established that at 1 GHz, the complex
φ ( *
ε mf *
− ε HC ) the upper right of the chart represents the area
where two-receiver tools yield acceptable
dielectric constant of the material is the measurements of conductivity and dielectric
Dielectric Porosity. The porosity of a matrix that
property that defines the behavior of the EM constant. The intermediate region is the
is filled with only water ϕwater can be obtained by
wave propagating through it. What happens in additional area in which the three-receiver
making Sx0 = 1. This porosity can be obtained by
a formation when more than just one material design can provide accurate measurements, but
solving for ϕ in equation 8 when Sx0 = 1:
(matrix, water, hydrocarbons) are present? where the two-receiver tools cannot. With this
Several theories model how the individual ε* configuration, the tool provides unique solutions
of each material are combined to provide the
ε *for − ε matrix
*
in a much wider range of formation conditions.
φ water = (9)
complex dielectric value of the mix of materials. * *
ε mf − ε matrix
These models are usually referred as mixing
models. At 1 GHz, the most widely used model
By definition, the dielectric porosity is the
is the complex refractive index method (CRIM)
porosity filled with water, and can be obtained
model. The CRIM model states that the square
by the measurement of the attenuation and
root of the complex dielectric constant of the mix
phase shift at 1GHz. This is the purpose of the
is the weighted average of the square root of ε*
microwave formation evaluation tool (MFET) that
of the various elements that make up the mixture,
is described in the following paragraphs.
and the weights are simply the relative amounts
of materials in the mix, i.e.:
The Microwave Formation Evaluation Tool.
Figure 6. The extended dynamic range of three
* V The high frequency dielectric tool (MFET) operates receivers design.
ε mix = ∑ i εi* at 1 GHz; the sensor is mounted on an articulated
i VTotal pad that is forced against the formation (Fig. 5).
(6) Likewise, the inter-receiver spacing is optimized
where VTotal = ∑Vi The pad contains the transmitter and receiver
to obtain good vertical resolution and to avoid
antennas and its associated electronics for firing phase wrapping. As the distance between
i
the transducers and measuring the received receivers is increased, the likelihood that the
Vi are the volumes of each of the materials in
signals. The pivot point of the pad is almost at measured phase is greater than 360 increases.
the mix, and ε*i are their corresponding complex
the center and behind the receivers to minimize This will produce non-unique solutions that
dielectric constants.
pad tilt problems. It has two transmitters, complicate the inversion scheme and limit
three receivers, and a temperature sensor. The the dynamic range of the tool. For this reason,
For a formation filled with a mixture of mud
temperature sensor is thermally isolated from a three-receiver spacing was chosen, which
filtrate and hydrocarbons, this equation becomes:
the pad body to measure the temperature of the enables an increase in the dynamic range and
formation, which is critical for steam injection
ε *for = (1− φ ) ε matrix
*
+ wells in which the thermal gradients have been
maintains a good vertical resolution.
(7) altered. The transmitters are sequentially fired,
φ (1− Sx 0 ) ε HC
*
+ φ Sx 0 ε mf
*
and the signal at each receiver is digitized and
The new tool produces six phase and attenuation
measurements, three for each transmitter/
sent up-hole. receiver. They are later combined to produce
Where:
robust measurements with a different depth of
ε*for = measured log value obtained by measuring
The spacing between the transmitter and nearest investigation. Fig. 7 shows how the measured
the phase and attenuation of the EM wave.
receiver is optimized between the depth of signals are combined to get the final results.
49
D i e l e c t r i c L o g g i ng Ene rgiz e d by “ Ce l l P hone” Tech nology

Vertical Resolution @ .2” per Sample


25
ε from Sn
ε from Sf
20 ε from S

Dielectric Constant
15

10

0
0 2 4 6 8 10
Depth (in)

Vertical Resolution @ .2” per Sample


5

Resistivity (Ohm-m)
3 R from Sn
R from Sf
2 R from S
R Formation

0
0 2 4 6 8 10
Depth (in)

Figure 8. Tool response (dielectric and resistivity)


to a bed 2-in. thick.
Figure 7. How signals are combined in the dielectric tool to produce a set of compensated measurements.

Data Acquisition and Processing. The 2 tool • EM travel velocity (m/nanosecond) 0.9
consists of 2 transmitters and 3 receivers. The • Dielectric constant
data acquisition and processing is as follows: • Resistivity
ε =10, R=10
• Each time a transmitter is on, the electronics • Water porosity med 10, 10
far 10, 10
measure the in-phase and quadrature • Quality curve ε =25, R=1
med 25, 1
signals at each receiver with respect to 0
far 25, 1

the transmitter excitation. From these Vertical Resolution. Fig. 8 shows the response
0 2 4 6 8 10
measurements, the received amplitude (A) of the tool for a 2-in. thick bed. The dielectric and Radial Depth (in)

and phase (α) are obtained: resistivity obtained by the Sf and Sn signals read
its peak value, whereas the S signal does not. Figure 9. Integrated radial sensitivity response of
⎡ LR MR UR ⎤ ⎡ LR MR UR ⎤
⎢ ALT , ALT , ALT ⎢ α LT , α LT , α LT
Consequently, the vertical resolution of the tool the basic signals (transmitter to receivers).
⎥, ⎥
is more than 2 in.
⎢ AUT
⎣
LR
, AUT
MR
, AUT
UR ⎥
⎦
LR
⎢ αUT
⎣ , αUT
MR
, αUT
UR ⎥
⎦
Where the subscript refers to the transmitter, Fig. 9 shows the percentage of contribution
and the superscript indicates the receiver. to the total signal coming from a radial
distance from the face of the tool. The depth
• For each transmitter, a new set of signals is
of investigation changes drastically with the
obtained by taking the ratio of amplitudes of
formation characteristics; in essence, it changes
two receivers and their difference in phase. In
with the attenuation power of the formation.
this way, the new signals are independent of
For example, for a formation with ε’=10 and
transmitter output power or transmitter phase
R=10, 90% of the signal comes from a distance
(shown as dotted blue, pink, and yellow signals
of 7.5 in. for the Sn signal and from 8.5 in. for Figure 10. Radial depth of investigation of the
in Fig. 7).
the Sf signal. However, for a more attenuating differential signals.
• The signals from the two transmitters are formation (ε’= 25, R= 1) these values change
averaged to obtain the desired final outputs borehole rugosity and mudcake effects. In the
from 2.5 in. to 3 in.
(shown in solid blue, pink, and yellow in worst conditions, the 90% point is approximately
Fig. 7). These signals are compensated for 4 in.; consequently, the tool has a depth of
For the same conditions plotted in Fig. 9,
tool tilt, and will be referred to as Sf, Sn, investigation of more than 4 in.
the differential measurements have a radial
and S. The following outputs are available for behavior, as shown in Fig. 10. In the differential
these signals: Mudcake Effects. Fig. 11 shows that the tool is
measurements, the formation on front of the pad
• Attenuation (db/m) not very sensitive to mudcake up to thicknesses
does not contribute much to the total signal; this
• Phase shift of 0.3 in. Both the measured dielectric constant
creates a measurement that is less affected by
50
Mud Cake Effect (εt=10, Rt=10; εm=25, Rm=1 Ohm-m)
40
ε from Sn
35 ε from Sf
Dielectric Constant

30
ε from S

25
20
15
10
5
0
0 0.2 0.4 0.6 0.8 1
Mud Cake Effect (in)

Mud Cake Effect (εt=10, Rt=10; εm=25, Rm=1 Ohm-m)


100
R from Sn
R from Sf
R from S
Dielectric Constant

10

1 Figure 12a. Data from Test Well 9 in Ft. Worth, TX.

0.1
0 0.2 0.4 0.6 0.8 1
Mud Cake Effect (in)

Figure 11. Effect of mudcake thickness.

and resistivity are very near the true value of the


formation properties.

Tool Calibration. Air has very steady electrical


Figure 13. Data from field test in California
properties. Its dielectric constant and resistivity (Bakersfield area).
are almost independent of temperature and
humidity, which makes it an ideal calibration and that they track completely and respond equally
verification point. to the thinly layered sequence of the formation.
Finally in the last track the porosity seen by the
Suppose an initial phase shift in each channel dielectric tool matches the cross-plot porosity of
is generated by the circuitry and other the density neutron, indicating that these zones
environmental factors, i.e., αU0 at the upper are filled with water.
receiver, αM0 at the middle receiver, and αL0 at
the lower receiver, respectively. In this instance, Fig. 13 shows the results of a field test in
Figure 12b. Data from Test Well 9 in Ft. Worth, TX.
the resulting phases of three receivers become: California (Bakersfield area). Again the resistivity
⎧ TXU TXU
removed by using a set of air-hang data. The and dielectric curves exhibit high resolution and
⎪⎪ α RU = α RU + αU 0 air-hang-data can be saved in the tool memory matching between the different compensated
TXU TXU for future use. The air-hang also serves as a
⎨ α RM = α RM + αM 0 (10) measurements. In this well the difference
⎪ TXU TXU
procedure to perform a pre-log verification. between the dielectric porosity and the total
⎪⎩ α RL = α RL + α L 0 porosity is shaded in green, and indicates the
Field Example presence of residual hydrocarbons. Cores were
when the lower transmitter (TX_U) fires, and Fig. 12 shows both the high vertical resolution taken in many of these zones, and a fluorescence
of the tool and the match of the 3 resistivity analysis was performed in these cores. The
⎧ TXL TXL '
⎪⎪ α RU = α RU + αU 0 curves and the dielectric measurements. The results show that in those zones where the
TXL TXL
+ α M' 0 well was drilled with fresh mud, and under these core fluorescence indicates the presence of
⎨ α RM = α RM (11)
⎪ TXL TXL conditions the resistivity measured by dielectric hydrocarbons, the difference between the
'
⎪⎩ α RL = α RL + α L 0 tool will be lower than the true formation dielectric porosity and the total porosity also
resistivity since the dielectric resistivity does not indicated hydrocarbons. Likewise, in those zones
when the lower transmitter (TX_L) fires. include corrections for the effect of the relaxation where the cores do not show hydrocarbons, the
dielectric losses. Track 3 shows the measured dielectric porosity indicates that the zone was
The initial phase errors generated by the circuitry dielectric constant seen by the different water wet.
are independent of the outside formation compensated measurements. It can be seen
properties. They can be easily determined and

51
D i e l e c t r i c L o g g i ng Ene rgiz e d by “ Ce l l P hone” Tech nology

Conclusions in the reservoir and compute water saturation depths of investigation with extended dynamic
The microwave formation evaluation tool when the formation water salinity is unknown or range and offers compensation for tool standoff,
measures the formation dielectric constant and has been altered. mudcake effects, and borehole rugosity. The tool
formation conductivity at 1GHz. The dielectric also measures the temperature of the formation,
constant measurement is used to calculate The tool has plurality of transmitters and which is critical for steam injection wells in which
water-filled porosity to differentiate the fluids receivers that enable measurements at varying the thermal gradients have been altered.

Authors
Michael Bittar is the Formation Gary Kainer works as a Program David Torres is the Resistivity
Evaluation Director for Halliburton’s Manager for Halliburton Wireline Global Product Champion for
Drilling and Evaluation Division. and Perforating Services. He joined Halliburton. He is responsible
Michael joined Halliburton through Halliburton in 1988 and worked as a for overseeing the development,
Sperry-Sun Drilling Services in 1990. field engineer before becoming the testing, and marketing of wireline
Throughout his career, Michael has technical operations leader in the resistivity logging. For more than
contributed to the advancement of logging-while-drilling Gulf of Mexico. Since joining the technology group in two decades, David has worked in various management
and wireline sensor technology and interpretation. 2002, he has managed several projects in the resistivity and research and development positions in the oil
Michael received his BS, MS, and PhD degrees in family, including the XRMI (Extended Range Micro industry. He began his career in 1980 with Gearhart
electrical engineering from the University of Houston Imager) and the OMRI (Oil Mud Reservoir Imager). He Industries in Brazil, and has worked as a field engineer
and has more than 70 patents and publications relating received an associate’s degree in electronics from Texas and a technical manager. David moved to Texas and has
to formation evaluation and downhole measurement. State Technical College in 1987. held several positions within the organization, including
Michael is a long-term member of the SPWLA and director of training, technical assistant to the vice
SPE and the recipient of the 2006 SPWLA Technical Ron Cherry is the Halliburton Open president of operations, and research. He holds several
Achievement Award. Hole Strategic Business Manager. patents and has authored various papers in signal and
After receiving a BS degree in image processing for the logging industry.
Jing Li is a Principal Scientist for electrical engineering from the
Halliburton’s Drilling and Evaluation University of Tulsa in 1975, Ron David McCoy is a Senior
Division. Jing joined Halliburton in joined Schlumberger Well Services Petrophysical Advisor for Occidental
2008. Before joining Halliburton, where he held various positions managing operations, Petroleum’s Vintage Production
he worked as a research scientist marketing, and technology. In 1994, he joined NUMAR California where he is responsible
in the well logging lab at the and served as the general manager until the company for formation evaluation in the
University of Houston. He received his BS, MS, and was acquired by Halliburton in 1997. Ron has been San Joaquin Basin. David joined
PhD in Electrical Engineering from Xidian University, responsible for new product developments associated Occidental Petroleum in 2004 as a petrophysicist to
Southwest Jiaotong University, and Xian Jiaotong with NMR (MRIL products) for the last several years, work in the Rocky Mountain Asset Team. In 2007, he
University, China, respectively. and is now responsible for all open hole logging transferred to California, where he worked for OXY Elk
products and services. Hills before coming to Vintage in 2009. Earlier in his
career, David worked for Phillips Petroleum and Amoco
in Houston as a petrophysicist and geologist.

Acknowledgements
The authors would like to acknowledge the contribution of the Rau, R., Davies, R., Finke, M., Manning, M. 1991. Advances in High
development team and of all who were involved in the field testing. They Frequency Dielectric Logging. Presented at the SPWLA 32nd Annual
Logging Symposium, 16-19 June.
especially acknowledge the support of David Sandoval, who performed
the field testing of the tool. The authors would also like to thank Vintage Bittar, M.S., Rodney, P.F., Mack, S.G., Bartel, R.P. 1993. A Multiple-Depth-
of-Investigation Electromagnetic Wave Resistivity Sensor: Theory,
Production for permission to publish the data. Special thanks to the Experiment, and Field Test Results. SPE Formation Evaluation 8 (3): 171-176.
management of Halliburton Energy Services, Inc. for their full support of
Pilman, D., Skinner, T., Denton, R., Hansen, M., Torres, D. 2009. Cost
this project and for encouragement to publish this work.
Effective Reservoir Characterization, Utilizing Dielectric Logging
Measurements, San Joaquin Basin, California. Paper SPE 124206-MS
References presented at the SPE Annual Technical Conference and Exhibition, New
Archie, G.E. 1952. The Electrical Resistivity Log as an Aid in Determining Orleans, Louisiana 4-7 October.
Some Reservoir Characteristics. In Petroleum Transactions of AIME 146: Hizem, M., Budan, H., Devillé, B., Faivre, O., Mossé, L., Simon, M. 2008.
54-62. Dielectric Dispersion: A New Wireline Petrophysical Measurement. Paper
Meador, R. A. and Cox, P. T. 1975. Dielectric Constant Logging, a Salinity SPE 116130-MS presented at the SPE Annual Technical Conference and
Independent Estimation of Formation Water Volume. Paper SPE 5504 Exhibition, Denver, Colorado, 21-24 September.
presented at the Society of Petroleum Engineers Conference, Dallas,
Texas, 28 September - 1 October.

52
Un d er s tand ing th e Pe Meas ur em en t

Using Computer Modeling to Generate


confused the industry about the true definition
of Pe and its limitations. The “Pe for Molecules
and Mixtures” section of this paper clarifies the

Accurate Pe Equations definition of Pe. This section demonstrates that


the equations used by the industry to compute Pe
Gordon L. Moake, Halliburton are based on an approximation, and explores the
Presented at the SPWLA 52nd Annual Logging Symposium held in Colorado Springs, CO, May 14-18, 2011
accuracy of that approximation.
Copyright 2011, held jointly by the SWPLA and the submitting authors
Some methods for computing Pe that are
contained in the literature are discussed, along
Abstract
with their limitations. In addition to inadequate
Over the years, several equations have been used to compute the photoelectric log of a formation (Pe).
equations used to compute Pe, there are inherent
Some of these equations lack a density term, resulting in errors; this is obvious when the tool is placed in
limitations to how well photoelectric absorption
a tank of water. In addition, the lack of quality test standards or marker beds makes Pe accuracy difficult
can be isolated from other effects.
to verify in all but a few conditions. This leads to the question: what is the proper response equation for
computing Pe?
Modeling is used to determine a mapping of a
ratio of high-energy to low-energy count rates
To answer that question for a wireline density tool, a Monte Carlo computer-modeling study was
to Pe for a wireline logging tool. A wide range
undertaken. Two features of computer modeling make it ideal for studying this problem. First, any
of formations, both realistic and imaginary,
formation can be easily simulated, so a wide variety of data points can be used to determine the response.
were used to develop the mapping and analyze
Second, non-physical formations can be modeled, which enables the density and Pe responses of the tool
its accuracy. Log examples are presented to
to be studied independently. Because the physics behind the Pe measurement is essentially the same for
illustrate the accuracy of the new algorithm
all density tools, the form of the final equation should be applicable to many tools.
and how it differs from the previous algorithm.
Finally, the new algorithm is shown to also
This study used the ratio of high-energy to low-energy count rates as the primary variable for the
apply to other density tools that have completely
calculation. For a given density, it is possible to obtain an accurate Pe response from 0.4 to 15 using a
different designs.
simple function of the ratio. As expected, the generalized equation that is appropriate for all rock densities
requires a density term. Accuracy of the technique breaks down for Pe > 15.
Many cross sections are plotted throughout this
paper to help explain the measurement and to
The new Pe equation is illustrated in several log examples that compare results obtained with the new
analyze its accuracy. All cross sections presented
algorithm to those obtained with a previous algorithm.
in this paper were computed with version 3.1 of
the program XCOM, which is distributed by the
A review of the physics and concepts behind the Pe measurement explains why the measurement breaks
National Institute of Standards and Technology of
down when Pe > 15, and clarifies why the Pe measurement should be treated as a unitless quantity.
the United States.

Introduction to use to convert tool measurements to Pe. In Physics of Photoelectric Absorption


In 1981, a spectral-density tool was introduced the past, the characterization of tool responses Photoelectric absorption refers to the absorption
to the industry (Bertozzi et al. 1981). Like the relied primarily on experimental measurements. of a photon by an electron. Photons cannot be
previous density tools that were prevalent at Unfortunately, it is difficult to obtain quality absorbed by an isolated electron because the
the time, it used a gamma source to bombard standards that span a wide range of densities interaction could not conserve both momentum
the formation with 662 keV gamma rays. Two and Pe values. However, tremendous advances and energy, which is required of any interaction.
detectors measured gamma rays that returned to in computer processing make it possible to use Thus, photons can only be absorbed by electrons
the tool. Unlike previous tools, the energy of the computer modeling to generate the required that are bound to atoms; the electromagnetic
detected gamma rays was measured. In addition data. The work described in this paper used connection with the atomic nucleus enables
to measuring density, the tool was used to those advances to determine a mapping from tool momentum and energy to be transferred to the
measure a new petrophysical parameter called Pe. measurements to Pe that is valid for a wide range nucleus in a manner that conserves both.
Pe is a measure of the importance of photoelectric of materials.
absorption relative to Compton scattering in a The stronger the connection between the electron
formation, and it helps to differentiate between To understand what equations are reasonable and nucleus, the easier it is for the transfer to be
lithologies. The measurement caught on, and and what limitations they may have, it helps to made and the more likely it is that the electron
most major service companies now offer the understand the theory of the Pe measurement. will absorb a photon. Binding energy, which is the
measurement with their density tools. Therefore, this paper begins by discussing the energy required to free an electron from an atom,
physics of photoelectric absorption, and then is a measure of this strength. For a given atom,
One of the challenges associated with making the discusses the concept of Pe as introduced by the binding energies of its electrons are pre-set
measurement is determining the proper equation Bertozzi et al. (1981). To some degree, their paper by nature, with the electrons nearest the nucleus

53
Un d e r s t and i n g t he Pe Me asure me nt

having the largest binding energies. Therefore, sudden decrease in cross sections for 40 keV and per electron. Furthermore, they reasoned that
the nearest electrons have the highest probability 50 keV photons around an atomic number of 60, because Pe is proportional to σp /Z, it, too, should
of absorbing a photon, although the photon as shown in Fig. 2. have those units. Unfortunately, they neglected
energy must be greater than the binding energy to consider that the proportionality constant also
of the electron. The electron is ejected from the had units. In fact, it is quite clear that Eq. 1 is
atom after absorbing the photon. unitless; it depends only on the atomic number of
the atom, which is unitless.
The photon energy also affects the probability
of absorption. The closer the photon energy is Pe for Molecules and Mixtures
to the binding energy, the less residual energy Eq. 1 applies only to elements. To derive the
there is after the absorption. Thus, less energy appropriate equation for molecules and mixtures,
must be transferred to the nucleus, making Bertozzi et al. argued that the ratio of count rates
the transfer easier. Consequently, the closer in a low-energy window to those in a high-energy
the photon energy is to the binding energy, the window should be proportional to the ratio of
Figure 2. Photoelectric absorption cross sections
greater the probability that the photon will be the macroscopic cross section for photoelectric
as a function of energy for 40 keV and
absorbed — as long as the photon energy is 50 keV photons. absorption to that for Compton scattering.
slightly greater than the binding energy. This is They then derived equations for molecules and
why the cross section for photoelectric absorption The Pe Concept mixtures, assuming that Pe is proportional to that
depends strongly on the energy of the photon and The Pe concept described by Bertozzi et al. (1981) ratio. This assumption ignores their definition
increases as the energy decreases, as indicated was intended to provide a measured quantity of Pe given in Eq. 1, and it assumes that Pe is
by Fig. 1. that would differentiate between different measured with a ratio of count rates. It also shifts
lithologies. They realized that different rocks are the definition from being based on atomic number
composed of different mixes of elements and that to being based on cross sections. However, they
photoelectric absorption varies significantly from then used an approximation for photoelectric
element to element. Thus, a measurement based cross sections that is based on atomic number.
on photoelectric absorption can differentiate The following paragraphs provide a more rigorous
between various rocks, such as calcite, quartz, derivation of their equations that clarifies the
and dolomite. required assumptions and does not rely on the
method with which Pe is measured.
Because photoelectric absorption of gamma rays
increases dramatically as their energy decreases, Definition of Pe. To derive the equations provided
the measurement must be made with low-energy by Bertozzi et al. for computing Pe of molecules
Figure 1. Photoelectric absorption cross sections
for silicon (Z=14) and calcium (Z=20). gamma rays. However, even the low-energy and mixtures, Pe must be defined differently
gamma rays are strongly influenced by Compton than in Eq. 1. According to Bertozzi et al., the
Larger atoms have more electrons, which scattering. Bertozzi et al. realized that by dividing photoelectric cross section for an atom in
contributes to an increase in the probability of count rates obtained for low-energy gamma sedimentary rocks is well approximated by
photoelectric absorption with atomic number. rays by those obtained for high-energy ones,
barns
In addition, the binding energies of the inner sensitivity to Compton scattering could be greatly σ p ( E ) ≈ 12.1 E −3.15Z 4.6 −3.15 (2)
electrons increase as the number of protons in reduced, so that the ratio depends primarily on (keV )
the nucleus increases, which also contributes to photoelectric absorption.
an increase in the probability of absorption with where σp (E) is the photoelectric cross section for
atomic number. Consequently, the cross section For simplicity, Bertozzi et al. defined Pe as a the atom and E is the energy of the photon. To
for photoelectric absorption increases rapidly function of the atomic number of the element (Z), honor both Eq. 2 and the desired Pe response of
with atomic number, as shown in Fig. 2. 3.6 Eq. 1, Pe is defined as
(
Pe ≡ Z 10 ) (1) −3.15
σ C (E) (keV ) ( )
Σp E
Electrons with binding energies that are greater Pe ≡ (3)
The value of 10 in the denominator gives the
than the photon energy are unlikely to absorb the
measurement a convenient range. Bertozzi et
12.1×10 3.6
E −3.15
barns ( )
ΣC E
photon because they cannot leave the atom, and
al. observed that the photoelectric absorption
the atom is unlikely to have a vacant orbit with a where ΣP(E) and ΣC(E) are the macroscopic
cross section (σp) for most atoms in formations
binding energy that exactly matches the energy cross sections for photoelectric absorption
is approximately proportional to Z4.6, so Pe as
that the electron would have if it absorbed the and Compton scattering, respectively. σC(E)
defined in Eq. 1 is proportional to σp /Z. Since
photon. That is why the photoelectric absorption is the Compton cross section for an individual
cross sections are often quoted in units of
dependence on atomic number sharply decreases electron. Note that these cross sections depend
barns and there are Z electrons in the atom,
when binding energies begin to exceed the on the energy, E, of the gamma ray. This
they reasoned that σp /Z has units of barns
photon energy. The effect is illustrated by the equation merely states that for a given energy,

54
Pe is proportional to the ratio of the macroscopic NA is Avogadro’s number, vi and ρi are the
photoelectric cross section to the macroscopic volume fraction and bulk density of the ith
Compton cross section. With this definition, Pe is compound, and σi and Ai are the cross section
clearly unitless. and atomic mass of the molecule composing the
ith compound. Using Eq. 9 in Eq. 3 gives
Pe for an Element. For materials composed of a
vi ρiσp, i ( E )
single element, the ratio of macroscopic cross σ C (E)
−3.15 ∑
Pe =
(keV ) i Ai
(10)
sections is the same as the ratio of atomic cross
sections. Using that fact and ZσC for the cross
12.1×103.6 E −3.15 barns

( vi ρi Ziσ C ( E ) )
i Ai
section of the atom, Eq. 3 becomes
−3.15 Using the approximation of Eq. 6, Eq. 10 becomes Figure 3. Percent error in the approximation for
σ C (E) (keV ) σ p (E) the photoelectric absorption cross section given in
Pe = (4) vi ρi Zi Pe,i Eq. 2 as a function of atomic number.
12.1×103.6 E −3.15 barns Zσ C E ( ) ∑ Ai
Pe ≈ i
(11) atomic numbers less than 24 could have errors on
Using Eq. 2 to approximate σp, one obtains vρZ
∑ i Ai i the order of 5%. Materials that contain elements
3.6 i i with atomic numbers greater than 30 could
(
Pe ≈ Z 10 ) (5)
experience very large errors.
where Pe,i represents Pe of the molecules in the ith
which is the same as Eq. 1, except that it is now compound of the mixture.
Measuring Pe
clear that the equation is an approximation for Pe
Ratio Processing. Bertozzi et al. proposed
and not its definition. It is common to define electron density (ρe) as
measuring Pe by measuring energy spectra in a
2Z detector and dividing the spectra into discrete
For convenience in the following derivations, ρe ≡ ρ (12)
A energy windows. One window covered the
Eq. 5 can be inserted into Eq. 2 to yield the
high-energy portion of spectra (called hard), and
approximation Using Eq. 12 in Eq. 11 yields
another covered the low-energy portion (called
barns soft). They proposed computing Pe from the count
σ p ( E ) ≈ 12.1×103.6 E −3.15ZPe −3.15
(6) ∑v ρ P i e,i e,i
(keV ) P ≈ i rates in those windows using
e (13)
∑v ρ i e,i CS A
i = (14)
Pe for a Molecule. For materials composed of a C H 1+ α Pe + β Pe2
single molecule, the ratio of macroscopic cross which is a generalization of the equation Bertozzi
where CS and CH are count rates from the
sections is the same as the ratio of molecular et al. obtained for two compounds, although
soft and hard windows, respectively, and A,
cross sections. For a molecule that has ni atoms it is now clear that this equation is only an
α, and β are constants. In practice, the first
for element i, Eq. 3 becomes approximation.
implementation of the measurement omitted the
σ C (E) (keV )
−3.15
∑n σ ( E) i p,i
The equations obtained for Pe are all independent
second-order term in the denominator (Ellis et al.
Pe = i
(7) 1983). One flaw in this equation is that it lacks a
12.1×103.6 E −3.15 barns ∑n Z σ ( E)
i i C of the energy of the gamma ray. Thus, the
i density term.
concept is valid if the measurement is made using
where σp,i is photoelectric cross section for a range of gamma-ray energies, as long as the
Computing Pe from a ratio has two advantages. It
element i, and Zi is the atomic number of element approximation in Eq. 2 is valid.
reduces sensitivity to Compton scattering, and it
i. Using Eq. 6, this becomes
reduces sensitivity to standoff effects. However,
Limitations of the Pe Concept. Eq. 5, Eq. 8, and
∑n Z P i i e,i
Eq. 13 are the equations used by the industry to
neither dependence is completely canceled.
P ≈ i
(8) Standoff effects can be well compensated in light
e
∑n Z compute Pe. They all hinge upon the validity of
i i
mud, but small amounts of standoff in barite mud
i the approximation given in Eq. 2. Fig. 3 shows the
can overwhelm the measurement. Nor will the
where Pe,i is the Pe of element i. This is the exact accuracy of that approximation; the percent error
ratio completely cancel Compton effects, because
equation given by Bertozzi et al., except that it is in the approximation is plotted vs. the atomic
Compton cross sections for high-energy gamma
now clear that it is only an approximation. number for the photon energies that typically
rays are different from those for low-energy
span a Pe measurement.
gamma rays. Consequently, Pe measurements that
Pe for a Mixture. The macroscopic cross section for
do not account for density effects tend to have
a mixture of compounds, where each compound is Although errors for atomic numbers less than six
significant errors in formations with densities
a collection of identical molecules, is given by are large, they generally have a negligible effect
significantly different from those used to calibrate
on the Pe of most formations or muds. That is
vi ρiσ i because their contributions to Pe are themselves
the measurement.
Σ = N A∑ (9)
i Ai negligible. Materials composed of atoms with

55
Un d e r s t and i n g t he Pe Me asure me nt

Logarithmic Processing. An alternative to ratio Pe, the measurement is limited in its ability to
processing relates the count rate of a given isolate photoelectric effects. That is because
window to the density of the formation and there are other interactions at low energies
a lithology term (L) that is computed from that are comparable in strength to photoelectric
photoelectric and Compton cross sections (Moake absorption and depend on atomic number.
1991). The relationship for any window can be Coherent scattering, in which photons scatter off
approximated by of more than one electron at a time, has larger
cross sections than photoelectric absorption for
lnC = a0 + a1ρ + a2 ρ 2 + a3ρ 3 + a4 L + b1x1 + b2 x2 (15)
light nuclei. Also, incoherent scattering is not
where C is the count rate in the window, really Compton scattering, which assumes that
x1 is related to standoff and the density contrast the electron is free. Because the electrons are
between the formation and mud, and x2 is related actually bound to an atom, the binding energy Figure 4. Ratio of the coherent scattering cross
to standoff and the L contrast between the affects incoherent scattering cross sections. section to the photoelectric cross section and
ratio of binding-energy corrections to incoherent
formation and mud. This approach clearly handles Fig. 4 shows the relative magnitude of these
scattering to the photoelectric cross section
the density dependence of the measurement. effects. Unfortunately, there is no way to plotted vs. atomic number for 60 keV photons.
L can be computed from the equation for the differentiate these effects from photoelectric
low-energy window and then converted to Pe. x1 absorption, so any Pe value computed from tool it is necessary to count over a time span in which
and x2 are computed from the density correction measurements will not agree perfectly with the a trillion or more gamma rays are generated
and known mud parameters. A variation of this Pe value predicted from the equations typically by the source. Unfortunately, even modern
approach combines the logarithms of the hard used in the industry. Some of the error will be computers are not fast enough to track that many
and soft windows to reduce the x1 dependence. calibrated out, but there will always be residual particles in a reasonable amount of time, so
The simple dependence on L in Eq. 15 was error associated with the measurement. mathematical methods have been developed to
determined experimentally with a handful speed up the process. These methods are called
of density standards. Although elegant, the Which detector is best for measuring Pe? In variance reduction, and many different methods
modeling in this study indicates that it is principle, either the near or far detector could of variance reduction are available. Part of the art
not exact. be used to measure Pe. Photoelectric absorption of modeling is choosing the best methods for the
mainly affects low-energy photons. Most of problem at hand.
What is the best way to compute Pe? It seems the gamma rays that eventually reach the
best to combine the count rates from the low- detectors do not reach low enough energies to Two different versions of MCNP were used in
energy and high-energy windows to reduce be significantly affected until they are near the the study. Modeling for the first tool considered
standoff effects as much as possible. Combining detectors. Thus, the depth of investigation of the was performed with MCNP4, which could not
logarithms provides more control in canceling Pe measurement is shallow and approximately generate energy spectra while using variance
standoff effects than a ratio because the two the same for both detectors. The tool reported by reduction. Consequently, the energy and angle
windows can be weighted differently. However, Ellis et al. measured Pe using the detector farthest of incidence of the gamma rays that entered
in practice, the weighting tends to be close to from the source. However, the wide collimation the detector were stored. A separate program,
one-to-one, which is equivalent to using the that is typically used with a far detector makes GAMRES, was then used to estimate an energy
logarithm of the ratio. In principle, an equation the measurement overly sensitive to the borehole, spectrum for the detector from the stored
that accurately relates the ratio to Pe will so the near detector can provide a better information. Modeling for the other tools used
have comparable sensitivity to standoff as an measurement (Moake and Schultz 1987). Also, MCNP5 version 1.51, which could generate
equation that accurately relates the logarithm the near detector usually has a higher counting spectra while using variance reduction.
of the ratio to Pe. Thus, it makes little practical rate, which translates to better statistical
difference which technique is used. Because Pe precision. Thus, the near detector is usually a Calibrating the model. Although the model
is most commonly computed from a ratio, that better choice for the measurement. successfully emulated the tool response, there
combination is used in this study. are unknowns and approximations made in the
Deriving a Response Equation tool geometry that cause small deviations from
Bertozzi et al. used the ratio of low-energy to with Modeling the response of a physical tool. There are also
high-energy count rates to compute Pe, and About the Model. Modeling was performed using variations in physical tools from their nominal
others have used the inverse of that ratio the program MCNP, which was developed at Los design. Thus, it was necessary to calibrate the
(Schultz 1985). The difference is academic. The Alamos specifically for nuclear modeling. The model so that it would give the same results
critical part of the problem is to determine an model emulates nature by tracking each particle as a physical tool. It was calibrated to match
equation that accurately converts the ratio to Pe. that is generated by a source. Each interaction experimental data that was used to develop the
That is accomplished in the next section. is selected randomly in such a manner as to original density and Pe equations for the tool.
reproduce, on average, the cross sections of Thus, equations developed with the modeled data
Limitations of the measurement. In addition to the the elements involved. To obtain statistically can be directly inserted into logging software.
approximations used in the equations to compute significant measurements with a physical tool,

56
The initial study was performed with the LOGIQ® standards. Except for the magnesium point,
version of the Halliburton Spectral Density Log a linear mapping of the modeling does an
Tool (SDLT-IQTM). Fig. 5 shows a typical spectrum excellent job of reproducing the measured
from the near detector. A 662 keV peak at the values. The blue line is a linear fit to the other
far right side of the spectrum comes from a data points. The deviation of the magnesium
small source near the detector that is used to point is probably attributable to the uncertainty
maintain a constant gain in the system. The four in the zinc content of the magnesium alloy that
colored regions of the spectrum signify the energy was used in the model, which has a large effect
windows that are used in computing density and on the photoelectric absorption. The magnesium
Pe. Pe is computed from the ratio of W4 to W1. point is more consistent in the high-energy
window, W4, as shown in Fig. 7. Because Figure 7. Calibration data for the high-energy
As part of the normal tool processing, a variety of photoelectric absorption is much less important window of the near detector of the SDLT-IQ.
modifications are applied to the measured data. for high-energy gamma rays, uncertainty in the
First, they are corrected for dead time and pulse zinc content is less important. The second part of the study determined the
pile-up. Then background from the stabilization relationship between the density-corrected ratios
source is subtracted. The corrected count rates To calibrate the high-energy count rates, a linear and Pe. The formations chosen for this study were
should be equivalent to the modeled data. All of transform was applied to logarithms of the W4 those of most importance to reservoirs: calcite,
the measured data plotted in this paper have had count rates using the solid line shown in Fig. 7. dolomite, and silica. Porosities of 0, 20, and
those corrections applied. Good elemental analysis existed for the marble 40 p.u. were used for each rock. Water was used
and quartzite standards, their impurities were to extend the low-Pe region. Other materials
The tool used NaI detectors, which convert small, the two standards span a large portion were also chosen to study the accuracy of the
gamma rays to light pulses in a non-linear fashion. of the normal Pe range, and they are the most Pe measurement. Table 1 provides the full
Consequently, the energy thresholds of the relevant standards to logging. Consequently, list of materials. The densities used for these
windows shown in Fig. 5 are not exactly where these were the only standards used to calibrate formations were the natural densities.
they appear to be. It was necessary to account the low-energy window. To calibrate W1, a linear
for this non-linearity before comparing to data transform was applied to the logarithm of its Extracting Equations from the Data. Fig. 8
computed with the model. count rates such that the ratio of calibrated W4 to plots Pe against the ratio of high-energy to low-
calibrated W1 was identical to the measured ratio energy count rates for the materials in Table 1.
Fig. 6 illustrates the relationship between the for those standards. All ratios in this paper were This clearly shows that a good Pe value cannot
logarithm of the measured count rates and computed from calibrated count rates. be obtained from the ratio alone, even at low
those of the modeled count rates for various Pe values.
Choosing the Formations to Model. The beauty
of computer modeling is that a wide variety of
formations can be studied, including those that
are not physically possible. To take advantage
of this, the modeling study was divided into two
parts. First, eight materials with a wide range of
Pe were selected, and formations were created
for each material. The materials were calcite,
dolomite, quartz, water, siderite, 40 p.u. calcite
filled with water, 40 p.u. dolomite filled with
Figure 5. Near-detector spectrum from SDLT-IQ. water, and 0.9(CaCO3)•0.1(FeCO3). Computations
for six formations were made for each material Figure 8. Plot of Pe vs. the ratio of high-energy to
using densities of 1.0, 1.5, 2.0, 2.5, 3.0, and low-energy count rates for the SDLT-IQ.
5.0 g/cc for the formations. The analysis was
performed in electron density. Because the Fig. 9 shows plots of ratios computed for
difference between electron density and bulk 2.5 g/cc formations vs. the ratios computed for
density is very small for all of the materials except the same materials with a different density.
water, the bulk densities were set at those listed The relationships are all linear. A convenient
above with one exception. The bulk density of way to remove density effects is to choose a
water was selected so that the electron density specific density to be the standard density, and
was equal to the listed values. The resultant then determine the ratio-to-Pe transform for that
modeling data was then used to determine the density. Ratios measured for other densities
Figure 6. Calibration data for the low-energy density dependence of the measurement. must be converted to the ratio that would be
window of the near detector of the SDLT-IQ. obtained at the standard density before Pe can be

57
Un d e r s t and i n g t he Pe Me asure me nt

Table 1. using the natural densities for the formations. The


Materials used in this Pe study computed ratios were then mapped to the ratios
Pe Electron of 2.5 g/cc formations using Eq. 16 and Eq. 17.
Material Ratio Density R(2.5) Pe The results are plotted in Fig. 11. This correlation
Coal 1.48 1.44 1.19 0.16 between Pe and the density-corrected ratio is
Water 1.74 1.11 1.30 0.36 much better for low-Pe values than that for the
uncorrected ratio shown in Fig. 8. However, the
Magnesite 1.44 3.03 1.50 0.83
correlation is still poor for Pe values greater
40 p.u. silica 1.89 2.03 1.82 1.49
than 15.
20 p.u. silica 1.91 2.34 1.89 1.67
0 p.u. silica 1.93 2.65 1.95 1.81 Because the most important Pe region in
Magnesium 2.10 1.72 1.98 1.93 formation evaluation is for values less than
Aluminum 2.28 2.60 2.29 2.57
5.1, a response curve for only that region was
determined. Formations of calcite, dolomite, and
40 p.u. dolomite 2.31 2.16 2.28 2.57
silica with porosities of 0, 20, and 40 p.u. were
20 p.u. dolomite 2.37 2.51 2.38 2.90 used in the fit, along with water and anhydrite.
0 p.u. dolomite 2.43 2.86 2.46 3.14 Fig. 12 shows the data; the solid line represents
Silicon 2.53 3.75 2.61 3.36 the fit to the data. The equation is a simple
Figure 9. Plots of the ratio of high-energy to low-
Gypsum 2.82 2.37 2.80 3.99 energy count rates for the SDLT-IQ. The x-axis is quadratic given by
the ratio for the density stated, and the y-axis is 2
40 p.u. calcite 2.85 2.07 2.81 4.07 Pe = c0 + c1R2.5 + c2 R2.5 (18)
the corresponding ratio in a 2.5 g/cc formation.
20 p.u. calcite 3.01 2.39 3.00 4.64
where R2.5 is the ratio of a 2.5-g/cc formation.
Halite 3.01 2.07 2.98 4.65 Although it is not shown in the plot, Pe computed
Anhydrite 3.10 2.96 3.13 5.06 with this equation for the experimentally
0 p.u. calcite 3.11 2.71 3.12 5.08 measured water point was within 0.04 of
Sulfur 3.24 2.32 3.23 5.43 the true value.
30% hematite 3.34 2.45 3.34 6.03
and 70% water Eq. 18 underestimates Pe above 5.1, so a different
90% calcite and 3.38 2.80 3.40 6.15
response equation was used for that region.
Figure 10. Plot of the offset values (a0) and slopes
10% siderite (a1), obtained for the transforms plotted in Fig. 9. Because the correlation between the density-
Sylvite 3.96 1.92 3.95 8.51 corrected ratio and Pe breaks down for Pe values
density, so the transform derived from the plot greater than 15, those points were ignored when
Rutile 4.15 4.05 4.25 10.08
will extrapolate well to large densities. For deriving the response. A quadratic equation was
Siderite 4.76 3.75 4.83 14.69
convenience in determining a0 and a1, 0.4 was used to represent the transform in this region.
Ilmenite 4.84 4.46 4.97 16.63 It was chosen so that the response curve and
subtracted from the inverse density in the plots.
Pyrite 4.76 4.83 4.92 16.97
Thus, the abscissa for the reference density of its derivative are continuous at Pe = 5.1, and the
Ti 5.03 4.50 5.17 17.09 2.5 g/cc is zero, so the corresponding a0 should curve passes through the siderite point. The
Hematite 4.95 4.99 5.12 21.48 be zero and a1 should be one. Both variables combined response curve for the low and high Pe
can be well represented by a cubic equation, regions is represented by the solid line in Fig. 11.
Magnetite 5.20 4.88 5.36 22.08
10% barite and 4.95 1.08 5.23 22.18
which is represented by the solid lines in Fig. 10.
90% water Combining these equations yields a transform Sensitivity of the Pe measurement degrades
that converts the Pe ratio measured at a given significantly above a value of 6, which is why
25 p.u. silica 5.23 1.32 5.48 23.48
invaded with density to the density that would be obtained for extrapolation of the fit determined with low-Pe
15.3 ppg Cs
the same material at a density of 2.5 g/cc, data underestimates large-Pe formations. The
formate mud
cause originates with the low-energy count rate,
⎛ 1 ⎞ which does not continue to decrease as rapidly
x ≡ ⎜ − 0.4 ⎟ (16)
computed. A value of 2.5 g/cc was chosen as the ⎝ ρ ⎠ as one might expect. That is because there
reference density. is a source of low-energy gamma rays that is
R ( 2.5) = b1x + b2 x 2 + b3 x 3 not affected by photoelectric absorption in the
The mapping from one density to another is (17) formation. A close examination of Fig. 5
+⎡⎣1+ b4 x + b5 x 2 + b6 x 3 ⎤⎦ R ( ρ )
linear, but coefficients of the line vary with the shows a narrow peak in the spectrum that occurs
densities involved. Let a0 and a1 be the offset where the bi are constants defining the lines in the low-energy window (W1). That peak
and gain of a line. Their values for the lines in Fig. 10. To determine the mapping for ratios corresponds to 70 keV x-rays that are generated
plotted in Fig. 9 are plotted in Fig. 10. They are of 2.5 g/cc formations to Pe, modeling was in the tungsten shielding around the detector.
plotted in terms of inverse density rather than performed for the formations listed in Table 1 These x-rays occur when gamma rays knock an

58
Accuracy of the Measurement experiment or modeling. For example, the Pe
As explained in the “Derivation of Mixing measurement could be used to help quantify
Equations” section, the equations used to the amount of invasion that exists with cesium
compute Pe have errors associated with them. formate mud (Berg et al. 2007), because cesium
In addition, as explained in the “Measuring Pe” has a very large photoelectric cross section.
section, the measurement itself cannot isolate
photoelectric absorption effects from coherent Log Examples
scattering and binding-energy effects. In practice, The following log examples compare the new Pe
the calibration of the tool and the choice of algorithm to the existing one in several different
response equation minimizes these errors for lithologies. The existing algorithm was based on
the most important formations. To understand experimental measurements in density standards.
Figure 11. Plot of Pe vs. the ratio corresponding
to a 2.5 g/cc formation for the materials listed in the magnitude of the residual errors, a wide The standards included relatively pure quartz
Table 1. range of formations were modeled. Fig. 14 and relatively pure calcite, both with very low
shows the results for the normal logging range porosity. A dolomite standard was available,
of 0 to 5.1. Unusual formations are listed in but it was far from pure. Elemental analysis of
order of increasing Pe. Silicon exhibits the the standards was available and used to assign
largest error, 0.17. Pe values to the standards. Nevertheless, the
dolomite standard was only used as a check on
As expected, Pe errors become larger at the algorithm because the accuracy of the Pe
higher Pe values, as shown in Fig. 11. Clearly, value assigned to it was dubious.
Pe measurements will not be very useful for
distinguishing between unknown materials that Fig. 15 is a log from a test well in Fort Worth,
have Pe values greater than 15. Nevertheless, Texas. Density porosity and neutron porosity
it should be possible to use it to determine the are plotted in track 3 on a sandstone matrix.
amount of mixing of two known materials that They do not quite overlay in the sand bed
Figure 12. Plot of Pe vs. the ratio corresponding to have very different Pe values. However, the between 1,460 and 1,480 ft, indicating that the
a 2.5 g/cc formation for select materials.
expected Pe values could not be computed from bed contains some shale. The new Pe curve
atomic numbers; they must be determined from described in this paper is plotted with the
electron out of an inner orbit of the tungsten and previous Pe curve (labeled old) in track 2 on an
another electron falls into the vacancy. They can expanded scale. They agree very well in the sand,
be generated by any gamma ray greater than reading about 2.2, which is reasonable for a shaly
70 keV. Most of the x-rays generated in this sand. The bottom portion of the interval is shale.
fashion will be blocked by the tungsten, but In that interval, the new Pe reads approximately
many of those generated near the collimator can 0.2 less than the previous algorithm.
enter the detector without passing through much
tungsten. The magnitude of this effect will vary Fig. 16 shows a log example in carbonates,
with the collimator design. obtained in a Midland, Texas, test well. In this
case, density and neutron porosity curves are
Muting the Response. The response equation plotted on a limestone matrix in track 3. They
becomes more aggressive at larger ratios to overlay very well at depths below 5,230 ft,
account for the tungsten x-rays just described. Figure 13. Plot showing the effect of muting the Pe indicating a clean limestone. The two Pe curves,
That aggressiveness was not present when the response curve.
which are plotted on an unusual expanded scale,
algorithm was determined solely from Pe values agree very well when they are above 4.5. They
less than 5.1. A side effect is that the new read between 4.8 and 5.0 below 5,230 ft, where
algorithm is more sensitive to barite mudcake the porosity reads between 5 and 8 p.u. The
or standoff in barite muds. (The barium in barite theoretical Pe of clean limestone in that range
has an extremely large photoelectric absorption is 4.9 to 5.0.
cross section.) To minimize this sensitivity, the
Pe measurement is muted for values above 5.1, Fig. 17 shows another log example from the
using the equation Midland test well. The porosity curves are plotted
(
Pe,muted = 5.1+ 0.5 Pe − 5.1 ) (19) on a dolomite matrix in this case. The porosity
curves are very close in the intervals 4,710 to
Fig. 13 shows the effect of the muting. Logging 4,725 ft and 4,765 to 4,795 ft, indicating that
software includes an option to provide the these intervals are primarily dolomite. The
Figure 14. Accuracy of the Pe response curve for
unmuted value when needed. both normal reservoir rock and unusual materials. Pe curves, which are plotted on an expanded

59
Un d e r s t and i n g t he Pe Me asure me nt

3.11 at 3 p.u. and 3.04 at 9 p.u. Although these


formations are undoubtedly not pure dolomite or
free of shale, the new values are more believable
than the old ones.

Application to Other Tool Designs


A study similar to the one previously described was
performed for two other Halliburton density tools
that had completely different designs. The Hostile
Density Neutron Tool (HDNTTM) was designed for
very high pressures and temperatures, and the Slim
Density Neutron Tool (SDNTTM) was designed to
pass through drill pipe. In addition to mechanical
Figure 15. Log example illustrating a sand bed and differences, their energy-window thresholds differ
shale. Porosity is plotted on a sandstone matrix.
from one another and from the SDLT-IQ. The
models were calibrated to the experimental data in
the same manner as for the SDLT-IQ.

Modeling the HDNT. The high-energy to low- Figure 18. Plots of the ratio of high-energy to low-
energy ratio computed for 2.5 g/cc formations is energy count rates for the HDNT. The x-axis is the
plotted against the corresponding ratios computed ratio for the density stated, and the y-axis is the
corresponding ratio in a 2.5 g/cc formation.
for formations with the same material but
different densities in Fig. 18. Formation materials
of water, quartz, dolomite, calcite, siderite, and
0.9(CaCO3)•0.1(FeCO3) were modeled. Unlike the
case for the SDLT-IQ, the mapping is not quite
linear across the entire range of ratios. However, a
good linear mapping can be obtained if the siderite
point is ignored, which is what was done. The solid
lines in Fig. 18 represent the resultant mappings. Figure 19. Plot of the offset values (a0) and slopes
The small errors in the density correction that result (a1) obtained for the transforms plotted in Fig. 18.
Figure 16. Log example illustrating a limestone from this approximations will affect only large Pe
bed. Porosity is plotted on a limestone matrix. values, and most of those errors will be removed
when the ratio-to-Pe transform is determined.

Details of the density dependence of the offsets


and slopes of the lines in Fig. 18 are slightly
different than for the SDLT-IQ, but they have the
same general trend, as shown in Fig. 19. As for the
SDLT-IQ, they can both be fit by a cubic polynomial.
In fact, the slope (a1) is well represented by a
quadratic equation.
Figure 20. Plot of Pe vs. the ratio corresponding to
Fig. 20 shows the mapping from the ratio for a a 2.5 g/cc formation for the HDNT.
2.5 g/cc formation to Pe for this tool. Data used in
the fit consisted of quartz, dolomite, and calcite of for formations with the same material but different
various porosities, as well as water, anhydrite, and densities in Fig. 21. The formation materials are
Figure 17. Log example illustrating a dolomite bed. the experimentally measured values for quartzite the same as those used for the HDNT. As with the
Porosity is plotted on a dolomite matrix. and marble formations. Although it is not plotted, SDLT-IQ, the transforms mapping the ratio at one
the modeled water point was essentially identical density to that at 2.5 g/cc are linear across the
scale, differ in this case, with the new Pe to the measured one. entire range of materials.
reading approximately 0.2 lower than the
previous algorithm. The new value ranges from Modeling the SDNT. The high-energy to low- The offsets and slopes of the lines plotted in
approximately 3.05 to 3.4 in these intervals. The energy ratio computed for 2.5 g/cc formations is Fig. 21 have the same general trend as for the
theoretical value of a pure, clean dolomite is plotted against the corresponding ratios computed other two tools, as shown in Fig. 22. As with

60
and marble. In this case, the experimental ratio was systematically studied. As expected,
water point was used rather than the modeled there is a significant density dependence. The
water point because there was a significant ratio can be corrected for density by mapping
difference between the two. The difference it to the ratio that would be expected if the
may be attributable to imperfect dead-time formation had a density of 2.5 g/cc. Although
and pulse-pile-up corrections, which are large the mapping equation is simple, the parameters
for the water point. The choice of water point of the equation vary with the true density of the
does not significantly affect the mapping of the formation. Consequently, six parameters are
other formations. required to describe the density correction.

Summary A wide variety of formations were modeled to


The true definition of Pe, one that enables it determine the mapping of a density-corrected
to be computed for molecules and mixtures, is ratio to Pe. For convenience, two quadratic
not Eq. 1 based on atomic number, but Eq. 3, equations are used: one for the normal Pe
which is based on cross sections. The standard range and one for Pe values greater than 5.1.
equations used by the industry to compute Pe are The equations are constrained such that both
actually only approximations that hinge on the the mapping and its derivative are continuous.
validity of Eq. 2, which relates photoelectric cross Consequently, there are four free parameters
Figure 21. Plots of the ratio of high-energy to low-
sections of atoms to their atomic number. These in the mapping that must be determined. When
energy count rates for the SDNT. The x-axis is the approximations are at best valid to approximately added to the six free parameters used in the
ratio for the density stated and the y-axis is the 5%, and break down for atomic numbers less density correction, there are 10 parameters that
corresponding ratio in a 2.5 g/cc formation. than 6 or greater than 30. Furthermore, all of the must be determined to define the Pe algorithm.
expressions used to compute Pe are unitless, It would be virtually impossible to accurately
so the practice of assigning units of barns per determine these parameters experimentally.
electron to the measurement is incorrect. Thus, modeling is required to calibrate
the algorithm.
There are also limitations on how accurately
measurements can be combined to compute The Pe response of three different density tools
Pe. That is because low-energy photons are with very different designs can be represented
Figure 22. Plot of the offset values (a0) and slopes sensitive to interactions other than photoelectric by the equations presented in this paper. Pe
(a1) obtained for the transforms plotted in Fig. 21. absorption. Cross sections for coherent scattering values computed with this algorithm match the
are greater than 10% of those for photoelectric theoretical values computed with the standard
absorption for most elements found in reservoir equations extremely well for typical reservoir
rock. Although a considerably smaller effect, rock. Errors observed for other materials with
the binding energy of electrons to atoms causes Pe values less than 5 were less than 0.2. As
scattering from electrons to deviate from simple expected from the theory, Pe values obtained from
Compton scattering. That effect on the incoherent the tool do not agree well with values computed
cross section exceeds 10% of the photoelectric from atomic numbers when Pe exceeds 15.
absorption cross section for atomic numbers of
13 or less. Mapping of the ratio of high-energy to low-
energy count rates to Pe is affected by tungsten
The nuclear modeling program MCNP was used x-rays. The x-rays are created when gamma
Figure 23. Plot of Pe vs. the ratio corresponding to to compute the tool responses for a wide range rays of sufficient energy pass through tungsten
a 2.5 g/cc formation for the SDNT. of formations for several wireline logging tools. shielding in the tool, and they are collected in
The results of the computations were calibrated the low-energy window. Even if photoelectric
the HDNT, the offset (a0) is well represented to reproduce the measurements originally used absorption in the formation obliterated all low-
by a cubic equation in the inverse density, to characterize the tool. Consequently, equations energy photons that could enter the tool, many
and the slope (a1) is well represented by a derived to fit the modeling data could be used tungsten x-rays would still be detected. Thus,
quadratic equation. directly in logging software. the ratio becomes less sensitive to Pe as Pe
increases. Because previous Pe algorithms were
Fig. 23 shows the mapping from the ratio for a This paper studied equations that compute Pe not generally determined with high-Pe formations,
2.5 g/cc formation to Pe for this tool. The plotted from the ratio of high-energy to low-energy count they are not likely to account for this loss in
formations include calcite, dolomite, and quartz rates from the near detector. By modeling a wide sensitivity. Consequently, this algorithm tends
of various porosities; anhydrite; water; and variety of formations, most of which could not to read higher values than the previous ones at
experimentally measured values of quartzite physically exist, the density dependence of the large Pe values. Unfortunately, this feature also

61
Un d e r s t and i n g t he Pe Me asure me nt

causes the algorithm to read much larger values it agrees very well with the previous algorithm
when barite mud is present. To present a more in sand and limestone, but reads approximately
normal log, the standard Pe curve is muted above 0.2 lower in dolomite. This is not surprising.
5.1. However, an option exists to provide the The previous algorithm was characterized in
unmuted value. good quartz and calcite formations, but a good
dolomite formation was not available. This result
Log examples showed that the new algorithm underscores the need for modeling in determining
provides reasonable results in sandstone, a Pe algorithm.
dolomite, and limestone. They also showed that

Author
Gordon L. Moake is a Chief
Scientific Advisor for Formation
Evaluation in the Halliburton
Drilling and Evaluation division.
His primary focus is the
development of wireline and
LWD density and neutron-porosity tools, although
he sometimes works on non-nuclear projects. Before
joining Halliburton in 1984, he worked four years
at Baker Tubular, developing electromagnetic flaw
detectors. Gordon obtained BS degrees in math and in
physics from the University of Wisconsin, and MS and
PhD (1979) degrees in physics from Purdue University.
A member of SPWLA and SPE, he holds 18 U.S. patents
related to the oil and gas industry.

References
Berg, P.C., Pedersen, E.S., Lauritsen, A., Behjat, N., Hagerup-Jenssen, S., Moake, G.L. and Schultz, W.E. 1987. Improved Density-Log Lithology
Howard, S., Olsvik, G., Downs, J.D., Harris, M., Turner, J. 2007. Drilling, Identification Using a Borehole-Compensated Photoelectric Factor.
Completion, and Openhole Formation Evaluation of High-Angle Wells in Paper 1987-FF presented at the SPWLA 28th Annual Logging Symposium.
High-Density Cesium Formate Brine: the Kvitebjorn Experience, 2004-2006. Moake, G.L. 1991. A New Approach to Determining Compensated Density
Paper SPE/IADC 105733 presented at the SPE/IADC Drilling Conference, and Pe Values with a Spectral-Density Tool. Paper 1991-Z presented at the
Amsterdam, The Netherlands, 20-22 February. 32nd Annual SPWLA Logging Symposium, Midland, Texas, 16-19 June.
Bertozzi, W., Ellis, D.V., Wahl, J.S. 1981. The Physical Foundation of Schultz, W.E. et al. 1985. Dual-Detector Lithology Measurements with a
Formation Lithology Logging with Gamma Rays. Geophysics 46 (10): New Spectral Density Log. Paper 1985-DDD presented at the SPWLA 26th
1439-1455. Annual Logging Symposium.
Ellis, D., Flaum, C., Roulet, C., Marienbach, E., Seeman, B. 1983.
The Litho-Density Tool Calibration. Paper SPE 12048 presented at the
58th Annual Technical Conference and Exhibit of the Society of Petroleum
Engineers, San Francisco, California.

62
63
www.halliburton.com/ReservoirInnovations

H09338 07/2012
© 2012 Halliburton. All Rights Reserved.

Вам также может понравиться