THE ROLE OF DATA IN DRILLING P3 AND WHY QUALITY IS IMPORTANT PEAKE - - PowerPoint PPT Presentation

THE ROLE OF DATA IN DRILLING

P3 AND WHY QUALITY IS IMPORTANT

2

PEAKE DRILLING

CONTINUOUS IMPROVEMENT PLATFORM

Get Data from Source

• f Record Databases

Get Data from Engineer Automate Design Generate Metrics Create Program Send Tasks to Rig/OSC. Rigs/OSC and Update as Completed

Analytics

3

AUTOMATION REQUIRES DATA QUALITY

4

Instrument Digital Sampling Calibrated Data Rig Calculated Channel Rig Corrected Channel

Facts States Activities Events Metrics Expectations Actions Best Practices

Real-Time Data Real Time Analysis Process Analytics Process Improvement

Alarming RTOC Corrected Channel Post- Process Data Score Source of Record

DRILLING DATA PATH

5

GOOD INFORMATION IS A CORNERSTONE

6

Is Our Drilling Data Good?

• Quest

stio ion n 1—Is it a problem we need to address or is magnitude of error lost in the noise

• Quest

stio ion n 2—If 1 is true, how significant are the errors

• Quest

stio ion n 3—If 2 is high, can we even fix it?

QUESTION 1

IS ERROR > NOISE AND IS IT IMPORTANT?

8

WHY IT MATTERS ON THE RIG

Commonly

• nly Found

nd Errors in Drilling illing Data a and Corres esponding

• nding Repercus

ussions ions Variable iable Observed Errors in Field Derivative Variable Errors Small Error Consequence Large Error Consequence Worst Case Scenario Torque ue >100% MSE Rig State 5%-10% Incorrect MSE leading to sub-optimal drilling. Consumption of real- time torque data by applications like Peake Drilling leading to sub-optimal analysis and planning. >20% This would represent ~5000 ft*lb error for most TDS-11 type drives. Bit Failure, Motor Failure, MWD Failure, Tubular Failure, Vibrational Dysfunction, Poor Drilling Performance Loss of Drill String RPM >100% MSE, Rig State 2%-10% Incorrect MSE leading to sub-optimal drilling. Incorrect rotary torque--improper feed- back loop to TD's that require RPM to correctly report torque (Yaskawa, Converteam, etc.) >20% This would represent ~25-40 RPM for most TDS-11 type

• drives. Bit Failure,

Vibrational Dysfunction, MWD Failure, Poor Drilling Performance Loss of Drill String HookLoad Load >100% WOB, MSE, Bit Depth, Rig State 2%-5% Incorrect MSE leading to sub-optimal drilling. Incorrect WOB leading to poor ROP or bit wear/damage. >10% This would represent ~50,000 lb for most sensors. Loss of Drill String

9

WHY IT MATTERS ON THE RIG

Commonly

• nly Found

nd Errors in Drilling illing Data a and Corres esponding

• nding Repercus

ussions ions Variab iable le Observed Errors in Field Derivative Variable Errors Small Error Consequence Large Error Consequence Worst Case Scenario Pump Pressure re >100% ΔP, MSE, Rig State 2%- 5% Incorrect MSE leading to sub-

• ptimal drilling. Poor

managed pressure

• performance. Poor well control
• performance. Wear/damage

to down-hole motors/turbines. >10% 500-750 psi error could lead to potential damage to motors/turbines/MWD. Potential for kicks/fracturing when near pore pressure/frac gradient Blowout Pump Rate >100% MSE, Rig State 2%- 5% Incorrect MSE leading to sub-

• ptimal drilling.

Wear/damage to down-hole motor/turbines. Poor managed pressure

• performance. Poor well

control performance. Poor diagnostic ability with tracers,

• etc. Poor hole cleaning.

>10% 10 strokes/min error could significantly affect well control

• r managed pressure

scenarios. Blowout

10 10

WHY IT MATTERS ON THE RIG

Commonly

• nly Found

nd Errors in Drilling illing Data a and Corres esponding

• nding Repercus

ussions ions Variab iable le Observed Errors in Field Derivative Variable Errors Small Error Consequence Large Error Consequence Worst Case Scenario Block k Po Positi tion

• n

>25% (cumulati ve >50%) Bit Depth, ROP 1%-2% Block Position Error is

• Cumulative. In order to

increase depth, the block must travel at least 2x the distance the drill string travels. Incorrect MD/TVD/Survey Measurements. >5% 10ft error per joint of drilled

• pipe. Could lead to significant

survey errors and TD compromise. Wellbore Intersecti

• n

Flow

• w Out

>100% Rig State, Kick Detection 5%-10% Incorrect MSE leading to sub-optimal drilling. Poor well control

• detection. Poor stuck

pipe/pack off detection >20% Inability to detect kicks, pack

• ff or other problems.

Blowout Pit Volume >5% (100% delta) ΔPit Volume, Kick Size/Density 1%-2% Poor well control detection/performance. >5% 5bbl error could be 100% error in well control calculations Blowout

QUESTION 2

WHAT ARE THE REAL-WORLD ERRORS

12 12

ERRORS FOUND THROUGH CALIBRATION

Rig A Rig B Rig C Rig D Rig E Rig F Rotary Torque 17% 17% 22% 24% 21% 18% Makeup Torque 23% 11% 12% 17% 60% 13% Rotary RPM 1% 1% 1% 1% 2% 1% Pump Rate 1% 32% 1% 1% 40% 1% Block Position 6" <0.5” <0.5” 6ft <0.5” <0.5” Hookload 11% 18% 12% Pit Volumes 15% 12% 18% 16% 15% 22% Pump Pressure 5% 4% 4% 4% 3% 5%

• Every rig has had devices

significantly out of calibration

• Most rigs have rig-ups or

practices that will lead to device error or drift.

• Errors are common to all rigs

and contractors

13 13

TORQUE, HOOKLOAD, MSE

AVERAGE >15% OVERESTIMATION OF APPLIED MSE

14 14

SHARED FIELD OBSERVATIONS

FROM ~10 DIFFERENT RIGS

• Hookload is generally
• ver-reported by hydro-

mechanical gauges.

˃ This means we get less WOB than we think

• Torque is generally over
• ver-reported by flux
• utput of VFD

˃ This means we get less torque than we think

15 15

TORQUE HISTOGRAMS: IR VS IR TESTER

n = = 298 tests

10 10 20 20 30 30 40 40 50 50 60 60 70 70 Number Torque Differ erence ence (ft-lb lbs)

Torque ue Differen erence

+/ +/- 2,000 lbs 20 40 60 80 100

• 50% -45% -40% -35% -30% -25% -20% -15% -10% -5%

0% 5% 10% 15% 20% 25% 30% Number er Torque Differ erence ence (%)

% Differen erence

+/ +/- 10% 10%

>80% of

• bservations are out
• f DS1

recommended tolerance

16 16

TORQUE SCATTERPLOT: IR VS IR TESTER

15,000 17,500 20,000 22,500 25,000 15,000 17,500 20,000 22,500 25,000

Iron Roughne hneck ck (ft-lbs) s) Iron Roughne hnecck cck Tester er (ft-lbs) s)

IR ove verstatin stating torque que (Under der make-up) IR understating erstating torq rque e (over ver make-up)

17 17

MEASURED ERROR IN PIT VOLUMES

PIT VOLUME CHANGES ARE GENERALLY UNDER-REPORTED

• 60%
• 40%
• 20%

0% 20% 40% 60% 10 20 30 40 50 60 70 Error in Volume ume Calculation ulation [bbl/in] l/in] Fluid Level l [in]

Sand Trap Settling 1 Settling 2 Settling 3 Settling 4 Suction Slugging Active Error

18 18

ACTIVE SYSTEM VOLUME

UNDERESTIMATION HAS MANY CONSEQUENCES

100 200 300 400 500 600 700 10 20 30 40 50 60 70 Activ ive e System em Volum ume [bbls ls] Fluid Level l [in] Calculated Measured

19 19

PUMP RATE ERROR

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2000 4000 6000 8000 10000 Volum umetr etric ic Efficie ient nty (%) Pressur ure e (psi)

• Volumetric

Efficiency is not constant

• Effected by

˃ Pressure ˃ Fluid ˃ Swab ˃ Temperature

• Difficult to

measure at

• perating

conditions

• -swab 1
• -swab 2
• -swab 3

20 20

CAN WE RELY ON TODAY’S TOOLS?

• Are they accurate?
• Are they reliable?
• Are they repeatable?

˃ Do they vary from well-to-well and day-to-day

• Can we measure every value of interest?

˃ Makeup torque? ˃ Actual pump rate?

21 21

0% 5% 10% 15% 20% 25% 30% 24-Jul 9-Sep 21-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

Nomac 27

RIG A ERROR TREND

22 22

0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 1-Aug 27-Aug 10-Sep 25-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

RIG B ERROR TREND

23 23

0% 5% 10% 15% 20% 25% 30% 35% 40% 6-Aug 26-Aug 14-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

RIG C ERROR TREND

24 24

Nomac35

0% 5% 10% 15% 20% 25% 30% 35% 1-Sep 19-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

RIG D ERROR TREND

25 25

0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% 27-Jul 19-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

RIG E ERROR TREND

26 26

0% 1% 2% 3% 4% 5% 6% 7% 18-Aug 2-Sep 13-Sep Percentage Error Calibration Date Makeup Torque PumpRate BlockPosition Surface RPM Surface Torque

RIG F ERROR TREND

QUESTION 3

CAN WE FIX IT? YES WE CAN!

28 28

NEW TOOLS ARE NEEDED

• Had to use new tools when old ones did not

satisfy requirements

˃ TTS to Measure Torque and Hookload

• Had to invent new tools when none existed

˃ Iron Roughneck Calibration Tool

• Recommended new types of tools when

calibration is not practical with current tools

˃ Radar Pit Level Sensors ˃ Coriolis meters

29 29

CAN WE CALIBRATE EXISTING TOOLS?

Rig F TOP-DRIVE WAS CALIBRATED TO MATCH TTS SUB

30 30

CAN WE CALIBRATE EXISTING TOOLS?

Vibrations not measured by TD PLC Ammeter Connection Torque Measured by TTS Inertia of Top Drive Slows TD but not DrillString

Current Tools are not sensitive enough and their source measurement(s) mask inertial, electrical and other sources of noise

31 31

WHAT PROCESS IS REQUIRED?

• Comprehensive
• Covers all aspects of quality
• Simple
• Can be performed and communicated by roughnecks
• Economic
• Can be scaled quickly with minimal expense
• Sustainable
• Can be repeated ad infinitum

32 32

P3 COMPREHENSIVE QUALITY GOALS

• Calibration—Trust but Verify

˃ Scheduled and Event Driven calibration

• Collection—Existence and Continuity

˃ Ensure >99.9% availability in Real-Time

• Constraints—Equipment Limits

˃ A system cannot report values in excess of device physical limits

• Consistency—Reality Checks

˃ Example: drilling only occurs on bottom

• Confidence—Bayesian Statistics in Real-Time

˃ Are data relationships maintained across the network of sensors?

• Communication—Check/Act in Real-Time

˃ Set alarms and communicate them objectively ˃ Manage escalation paths to ensure timely resolution

33 33

SIMPLE PHILOSOPHY

• You cannot improve what you cannot measure!
• ALWAYS Calibrate to Actual Units
• ALWAYS Recalibrate IF

˃ The instrument has no calibration records ˃ The instrument is known to be untruthful ˃ The environment in which the instrument operates changes significantly ˃ The recommended calibration interval has been exceeded

• ALWAYS Demand Proof of Calibration

˃ EVERY device must have a certification with a valid date

• ALWAYS Trace Back to NIST
• ALWAYS Calibrate using AT LEAST three measurements
• If It Doesn’t Seem Right, CHECK AGAIN—Follow Your Instincts

34 34

RECOMMENDED PRACTICES

• Developed Recommended

Practices

˃ Meet Following Requirements

‒ Must be roughneck-proof ‒ Must be cost-effective and sustainable ‒ Must add <1 hour total to rig- up

˃ Reviewed by:

‒ DTG Senior Engineers ‒ Drilling Contractors ‒ Equipment Vendors ‒ Company Men/Superintendents

˃ CHK Controlled Document

‒ Version Controlled ‒ Highly Structured

35 35

ORGANIZATION AND TRAINING

• Spent 8 months organizing and training vendors

˃ Some vendor still require a lot of improvement (BOP) ˃ Put ‘teeth’ behind efforts by working with Supply Chain ˃ Certified 10 independent auditors (1/2 day class + field shadowing)

• Provided multiple field-level training

sessions for company men and superintendents

• Trained OSC staff on how to

manage data

36 36

MANAGEMENT OF CHANGE PROCESS

• Device Details

˃ What does it measure

‒ Directly/Indirectly ‒ Units of Measure ‒ Device Thresholds

• Calibration Details

˃ Method ˃ Frequency

• Reporting/Visualization Details
• Alarming Details

37 37

INDUSTRY STANDARDIZATION

• Peer participation is paramount to:

˃ Improve Recommended Practices ˃ Lower Cost ˃ Encourage Vendor Participation, Training, Staffing, etc..

• Several E&P Companies already on board:

˃ Apache, BP, Chesapeake, Chevron, Exxon, Hess, OXY, Statoil

Data Quality is an Industry Problem that Requires and Industry Solution

QUESTIONS?

THANK YOU!