Sie sind auf Seite 1von 65

Field Testing and Diagnostics of Radial-Jet WellStimulation for Enhanced Oil Recovery from

Marginal Reserves
09123-03
Robert Balch
New Mexico Institute of Mining and Technology
RPSEA Onshore Production Conference: Technological Keys to Enhance Production Operations
June 27, 2013
Wichita, Kansas
1

rpsea.org

Project Information
o Principal Investigator Robert Balch
Research Engineer - Tongjun Ruan
Graduate and Undergraduate students

o Participants:
Harvard Petroleum Co. LLC
Petroleum Recovery Research Center

o Project Cost -

$1,404,804

RPSEA Share: $656,537


Cost Share:

$748,267

o Timeline:
Project Start Date: 3/28/2011
Project End Date: 3/29/2014
2

Objectives

The objectives of this project are to:


o Field-demonstrate a short radius lateral technology
for production enhancement from mature oilfields
o To identify the feasibility and cost efficiency of using
short radius laterals for enhanced oil recovery
o Develop diagnostic methods for tracking the
placement of laterals
o Optimize the design of lateral patterns and fieldvalidate modeling techniques

What is Short Radius Lateral Well Stimulation?


A low cost and low impact drilling system able to make a
right turn in the well-bore and create a short, small radius,
lateral. The most commonly referenced technology is jet
drilling with claimed ranges of ~300ft.

What is Short Radius Lateral Well Stimulation?

o Can replace need


for infill drilling
o Targets bypassed
pay zones
o Targets untapped
compartments

Technologies Jet Drilling

o Casing drill and water jet deployed through a 1


coiled tubing unit
o Weak acid-water pumped through micro jets on a
cutting head
o Jet tip and water supply hose are 0.5 diameter
o Cutting jets on tip eat away at carbonate rock or
carbonate cemented sandstones
o Drive jets push jet tip and pulls hose up to 300 away
from the well bore and helps flush rock residue from
the lateral
o A jagged ~1 jetted lateral increases drainage
6

How Does it Work?


1. Drill Through Casing

7
Source: radialdrilling.com

2. Switch to Jet Assembly

8
Source: radialdrilling.com

3. Jet the Lateral

9
Source: radialdrilling.com

4. Rotate to Next Hole

10
Source: radialdrilling.com

5. Repeat Drilling

11
Source: radialdrilling.com

Major Challenges
I.

Develop modeling techniques to allow for


predictive design of optimal lateral placement
A. Simulate effect of laterals
B. History match response from field tests

II. Develop technologies to determine lateral


placement
A. Sensors
B. Remote sensing

III. Quantify feasibility and cost-effectiveness of using


the technology for EOR
A. Perform three field tests in typical Permian basin wells
B. Measure stimulation response
12

Development of Modeling Techniques

Sensitivity Study
o Four Modeling cases are under consideration:
1.
2.
3.
4.

Simulate with radials along x and y axis of a square grid


Simulate with radials along axes of a radial grid
Simulate with local grid refinements along laterals
Develop a new equation for well-bore radius

o Refinements to these cases include:


1. Optimizing number of laterals
2. Optimizing lateral length for a variety of spacing's

13

Model 1: Square Grid

14

Model 2: Radial Grid

15

Model 3: Grid Refinements

16

Analytical Model

Model basic assumptions:


o

o
o
o

17

The pressures at the inner and outer boundaries


are maintained constant (steady-state) by the
boundary aquifer surround reservoir model.
The fluid is single phase and slightly compressible
Mechanical skin is assumed to be zero
The four optional models used for hydraulic lateral
holes are that of Joshi, Giger, Odeh and MIEX
analytical solution adjusted for anisotropy

Analytical Model

Well definition and constraints:


o A vertical production well placed in the geometric
center of the simulation grid
o First production from Jan 1st, 1990 constraining
minimum bottom hole pressure to 28 psi

Hydro Jetting Lateral Placement:


o Four Laterals placed around the vertical well in a
cross-shaped pattern
o With a 1 inch diameter and 300 ft length
o Laterals emplaced in 1990, 2000, or 2010
o Predictive models run until 2020
18

Results: Model 1

19

Results: Model 2

20

Results: Model 3

21

Comparison of Incremental Oil

Observations:
o
o
o
22

Ultimate recovery is higher with laterals


Ultimate recovery is essentially the same no matter
when laterals are emplaced
Acceleration can occur with earlier placement

Model 2 Refinement: Number of Laterals

23

Comparison of Incremental Oil

Observations:
o
o
o
o

24

~22.7% Incremental by adding a single lateral


~26.6% Incremental by adding a second lateral
~28.9% Incremental by adding a third lateral
~29.9% Incremental by adding a fourth lateral

First Target

o The first target


well is the "Barbie
State NO.1" which
is located in
Millman Field,
Eddy County, New
Mexico
o San AndresGreyburg, mixed
carbonates and
sandstones
25

Building Reservoir Geomodel

o Utilizing Well log data, core data, and depositional


understanding to build a facies based model which
represents our best understanding of the geology
o Will start by history matching production and
pressure to date of the stimulation
o After the stimulation occurs:
Will model the stimulation effect and match 3-6
months of new production data
Determine best type of gridding algorithm to use

26

Digitized Contour Map Top of Structure

27

3D model and Layers


o Layers determined
by log analysis
o 5 layers were used

28

Major Challenges
I.

Develop modeling techniques to allow for


predictive design of optimal lateral placement
A. Simulate effect of laterals
B. History match response from field tests

II. Develop technologies to determine lateral


placement
A. Sensors
B. Remote sensing

III. Quantify feasibility and cost-effectiveness of using


the technology for EOR
A. Perform three field tests in typical Permian basin wells
B. Measure stimulation response
29

Develop Technologies to Determine Lateral


Placement

Some Significant Challenges


o Sensor must be self contained
No direct contact with surface since sensor needs to be
deployed on end of a water hose

o Must operate in a hostile environment


Weak acid water, oil, other chemicals
Heat up to 180oF
Up to 300gs of acceleration as nozzle moves down channel

o Must be very small


Approximately 0.5inch diameter and 1.5 inches long

o Must be inexpensive to fabricate/replace


30

Develop Technologies to Determine Lateral


Placement

Small and durable components exist:


o Piezoelectric accelerometers
Small scale and high g
Can measure very small deviations from a
known point

o Programmable control chip


On-chip memory
Readable by a laptop or portable device at
surface

o Battery

31

Must be able to log continuously for about an


hour while tool is deployed and ran out to end
of jetted lateral

Component Sizes

32

System Architecture
Microcontroller
SCK
SDI
SDO
PIC12F1840
INT
CS1
CS2

3-Axis Accelerometer
SCK
SDO
SDI
INT
CS

LIS331

SCK
SDO SST25VF06
SDI 4C
CS
Serial Flash Memory

Software Design (C and Assembler)


o Downloaded, Installed and Configured MPLAB X
IDE v1.2
o Ordered MPLAB XC8 C compiler
o Collected useful open source, example and sample
codes
o Based on the datasheets of:
PIC12F1840(Micro Controller)
LIS331(3-axies Accelerometer)
LIS331 breakout
SST25VF064C(Serial Flash Memory)
o Application notes and XC8 Manual etc.
34

Why Use C?
o C is an intermediate language. Assembler is a low level
language
o Assembly language is used to program the Microcontroller,
Sensor and SST25VF64C chips.
Hexadecimal, converts to binary on chip
o C can handle pages and math operation
o C code is easy to read and maintain
o C codes efficiency is very good and provides a bridge
between the mind of the programmer and the machine level
of the chips

35

C Program to Assembly Program

CProgram
CCompiler
AssemblyProgram
Assembler
Binarycode
PIC12F1840
36

MPLAB

37

X IDE

Software Flow Diagram

38

Interrupt Service Routine

ReadFromLIS311
WriteintoSST25VF64C
Return

39

Development Tool - REAL ICE

40

Test Circuit on Breadboard

41

Signal wave format

42

SF600 - Serial Flash Memory

o SF600 is a programmer
to Read and Erase the
SPI Flash memory
soldered on board
o Controlled by DediProg
Software on PC
through the USB bus

43

44

45

Build into PCB and Test in LAB

o Cut a piece of PCB


from Prototype
PCB
solder on
PIC12F1840 and
SST25VF064C
o Connect with
LIS331 break out

46

Put into D x 1 L hose

47

Positive Test with Logic Analyzer!

48

Testing the Device


o Testing:
Ran through a
curved PVC pipe a
few times
Took it for a walk
Drove around in car
Dangled and swung

o After several hours


disconnected battery
and read memory

49

Total 2.5+ hours


SF600 read out the
data
Nearly 5.5 MB data.
Battery (3.07v2.9v)/3.07 =0.026
Drop (2.6%)

Raw data read by SF600

50

Data Process algorithm, Software and 3-D


Display/Animation
o Currently Developing Data Processing algorithm
o Software - Using Java to develop
have developed a routine to translate raw data into X-Y-Z
table (Eclipse Java IDE)

o 3-D Display/Animation - Open Source SciDaVis

51

Data Processing using Java

52

Data Translated into an X-Y-Z Table

53

Data Processing Flow Chart

54

Data Transformation and Noise Filter


o Recorded acceleration raw data (integers)
are converted into real numbers (double
precision)
o Then transformed:
o A(t)=A(t)*scale_6g*9.81 for X Y and Z axis
respectively

o A moving average digital noise filter is used


to reduce the noise

55

From Acceleration to Position

Based on the filtered accelerations the position


data can be obtained with a double integration.

56

Tri-Axis Tilt Sensing and Calculation

X, Y and Z axes tilt relative to ground

o and represent the angles that the X and Y


accelerometer maxes make with the fixed reference XY
plane, and is the angle of the Z axis relative to gravity,
these angles can be calculated:
o = arcsin(Ax)
o = arcsin (Ay)
o = arccos (Az)
o Based on the angles we can identify if the sensor was
turning(Tilt) from the vertical well into the jetted lateral
57

Recorded Data (Transformed, Un-filtered)

58

Computed Position

59

SciDaVis Software

60

Current Work

o Finalizing development of the algorithm and software to


process the raw data and display the results
o Draw a PCB design and find a company to make PCB and
solder the devices on PCB
TinyDataLogger
TinyDataLogger with Gyro chip

o Work with drilling contractor on casing


o Test in real world

61

Oilfield tests with Harvard Petroleum

Major Challenges
I.

Develop modeling techniques to allow for


predictive design of optimal lateral placement
A. Simulate effect of laterals
B. History match response from field tests

II. Develop technologies to determine lateral


placement
A. Sensors
B. Remote sensing

III. Quantify feasibility and cost-effectiveness of using


the technology for EOR
A. Perform three field tests in typical Permian basin wells
B. Measure stimulation response
62

Quantify Feasibility and Cost-Effectiveness


o Three sites selected in Permian basin
Site 1: San Andres (Carbonate)
Site 2: Greyburg (Shallow Carbonates/Sandstone)
Site 3: Upper Delaware (Sandstone)

o Each is a work over of an existing well


Allows for immediate measurement of stimulation impact
Provides ample data for simulation and modeling

o Data will be used to calibrate predictive modeling


techniques
Simulation models?
Custom key word for effective radius for jetted laterals?

o Current Status:
63

Finalize contract with drilling company


Waiting field equipment for first test

Questions?

64

Contacts

PI: Robert Balch


balch@prrc.nmt.edu
575.835.5305

65

RPSEA PM: Charlotte Schroeder


cschroeder@rpsea.org
281.690.5506

Das könnte Ihnen auch gefallen