2015 TD Securities Calgary Energy Conference Modelling Risk and Value in Unconventional Plays Tyler Schlosser, P.Eng. Director, Commodities Research July 2015
Understanding risk is more important than ever The good old days June 9, 2015
Was this a warning sign
Low-cost producers continue to increase production Saudi and Iraq production increased 1 million bopd in March This is approximately one North Dakota Bakken
while everyone else strives to become a low-cost producer EOG Eagle Ford Productivity Improvements Source: RBN Energy, EOG
Is 65 the new 80 Efficiency Technology Cost deflation 2015-2016 Some plays are demonstrating better ROR in 2015 at 60 oil than in 2012 at 90 oil
How can we answer these questions 1. What is the chance of realizing a 10%-discounted NPV greater than zero for an 8-well horizontal drilling program spread across the Kerrobert area Viking What if all 8 wells are drilled in the same…
With probabilistic analysis Identify Generate Run Monte analogous revenues Carlo simulation wells 100,000 iterations Build Apply costs, Confirm Monte production royalties & other Carlo results forecasts adjustments…
Which factors are known Components of a Half-Cycle Economic Analysis Market prices are hard to predict Normally predictable to 70% with 95% confidence one year out Prices Hedging can improve certainty Production Well…
Sample Area Kerrobert Viking Choosing a sample area Geologically similar 30 wells is best Representative of your area of interest
Most relationships in resource analysis are nonlinear Spearmans Rank Correlation vs. Pearsons Correlation Pearsons Correlation Assumes constant variance Tests fit to straight line Is the R in the familiar R2 Pearsons R = -0.48…
Decline parameters are not independent It is nearly always incorrect to move a type curve up or down proportionally to IP the EUR to IP relationship is nonlinear
Sampling with dependence - Kerrobert Viking 100,000 times Initial Rate Initial Decline Rate Transient Exponent (1-Di)
Sampling with dependence - Kerrobert Viking EUR Could also analyze on a per-frac or per-length basis Mstb
Confirming production forecast results Real Data MC Sim Results Looks good P10/P90 ratios are within 4%
Calculating revenues with stochastic pricing Each of the 100k production forecasts run with a stochastic price forecast anchored by TD forecasts (75 WTI and 3.50 Henry Hub) PF 1 Revenue 1 PF 2 Revenue 2 PF 3 Revenue 3
Stochastic price forecast: the Ornstein-Uhlenbeck process A modified random walk with a mean reversion tendency Has four parameters: Xo: initial price : equilibrium price : volatility : rate of shock dissipation Source:…
Choosing suitable parameters for the O-U process Rate of shock dissipation: 5 years (based on historical data, Equilibrium detailed in SPE-162629 paper) Price: 75 USD/bbl Volatility Initial price: 60 Date
WTI 68% confidence that price will be between 40/bbl and 84/bbl through 2016 Implied volatility can be calculated from the prices of options sold by banks 68% confidence
Henry Hub 68% confidence that price will be between 2.05/MMBtu and 5.25/MMBtu through 2016 68% confidence
Possible Future WTI Price Realizations Possible future price O-U equilibrium realizations prices: USD/bbl WTI: 75/bbl HH: 3.50/MMBtu Date
From revenues to cash flows Capital 750k-1000k Operating Costs Costs GLJ Database + Probabilistic Sampling Price Byproduct Surface Adjustments Yields Loss
Single-well time to payout Mean: 27 mo P90: 7 mo 36% never pay out P50: 17 mo P10: 60 mo
Expectations tighten with more wells Wells P50 1 137k Here, every 2 280k well drilled is geostatistically 5 409k independent 10 472k from every 20 507k other well 50 525k (shotgun blast scenario)
What if all wells are drilled in one small area Medicine Hat/Milk River Shallow Gas Example Wells drilled nearer to each other are likely to perform more similarly than wells drilled further apart
Expectations dont tighten as much if wells are all drilled in one small area R=0.0 P50: 466k If 8 wells are all R=0.1 P50: 454k drilled in the R=0.3 P50: 435k same section, then some of the Kerrobert R=0.5 P50: 415k law of…
Answers 1. What is the chance of realizing a 10%-discounted NPV greater than zero for an 8-well horizontal Viking drilling program spread across the Kerrobert area 90% What if all 8 wells are drilled in the same section 84% 2. How…
Applications Asset valuation Comparing Hedge evaluation with intelligent drilling programs sensitivities Backstopping Portfolio and reserves work on Applications efficient frontier major fields analysis Generating…
Loading...
