Luck Skill and How We Drill 1 Marc

Luck, Skill, and How We Drill 1 Marc Willerth, H&P Technologies

Playing the Best Poker Player in the World Source: Wizard. Of. Odds. com 2

Better to be Lucky or Good? § My strategy: Look at the first 2 cards – If they are above average, bet everything – If they are below average, fold § No trust in my poker skills § Gives me at least a 1 in 6 chance Source: Wizard. Of. Odds. com 3

Bad Beats: When Luck Overwhelms Skill § My strategy minimized impact of skill, maximized luck – Even perfect decision-making can lose – Winning strategy for the unskilled player § How to make this more likely? – Reduce quantity of information – Reduce quality of information – Minimize number of decision points 4

What Our Advertising Says: Excerpts from the IADD Directory: “Industry leading experience” “Extensive drilling knowledge” “Trusted expert” “Over 100 years of experience” “Knowledgeable, industry experienced professionals” 5 We use skill to succeed!

What Do Our Choices Say? § Cut back on non-magnetic spacing § Use long survey course lengths § Neglect proper survey quality procedures Decisions that rely on luck to succeed! 6

Drillstring Interference for >10, 000 bit runs § Median magnitude: 377 n. T § 95 th percentile – Error model: ± 440 n. T – Real data: ± 1903 n. T § Uncertainty >4 x error model § Spacing has reduced by half 7

Drillstring Interference for >10, 000 bit runs § Median magnitude: 377 n. T § 95 th percentile – Error model: ± 440 n. T – Real data: ± 1903 n. T § Uncertainty >4 x error model § Spacing has reduced by half 8 MWD Error Model distribution

Course Length, Slide–Rotate, and Positional Error Reported Position Survey 2 Error Survey 1 Rotate Slide True Position How the well was drilled 9

Cutting Corners on Survey Quality § Gravity and Magnetic Field data – Often reported with low resolution – unhelpful for QC – Trying to save time and telemetry 10

What have we done to ourselves? 11

Bad Beat #: 1 Lateral Spacing Real-time data: Even Spacing After correction: Large variations Lateral spacing: Corrected well paths Lateral spacing: Reported well paths Wellpath 1 vs 2 Wellpath 2 vs 3 Wellpath 3 vs 4 Wellpath 4 vs 5 Wellpath 5 vs 6 229' 265' 208' 232' 246' No way to succeed in RT with those constraints! 12 Wellpath 1 vs 2 Wellpath 2 vs 3 Wellpath 3 vs 4 Wellpath 4 vs 5 Wellpath 5 vs 6 320' 289' 395' 102' 172'

Bad Beat #2: Vertical Well Placement Two survey sets in the same well show a 40’ TVD difference 13 How can geosteering be expected to succeed if we survey only every 100 ft?

Bad Beat #3: Impossible Troubleshooting Is this a collision risk, failed tool, or just bad decodes? Do you want to trip multiple times to find out? 14

Many Targets are Already Unreasonable For a 10 k lateral with the “Standard MWD” we claim 400 ft 160 ft Driller’s Window 20 ft 50 ft MWD Survey Accuracy Only 1 in 20 actually hit, even if everything goes right 15

If We Properly Modelled Uncertainty… For a 10 k lateral with common practices accounted for 700 ft Driller’s Window 180 ft MWD Survey Accuracy 16 20 ft 50 ft Might as well switch to poker

What Would a Skilled DD Want? § More accurate surveys at greater density § Robust QC data in real-time § Procedures for instrument validation 17

Technical Solutions Exist § IFR 1 & Multi-station Analysis – Improve magnetic survey accuracy § Digital slidesheets and continuous inclination – Enable augmentation of survey density & curvature mapping § These are not cure-alls, they still require good practices – Getting the right data at the right time – Making sure that data is of quality 18

What Do We Really Need to Do? § Have a serious conversation about what directional drillers can deliver § Properly align incentives for true well placement (not reported well placement) § Design workflows where skill can succeed 19

Thank You 20