Vous êtes sur la page 1sur 18

Risk, Uncertainty & Economic Analysis for Resource Assessment and Production Forecasting in Shale and Tight Reservoirs

Introduction Probability, Distributions and Correlation Estimating Under Uncertainty Tight Clastics / Carbonate Assessment Shale Assessment Reservoir Flow Valuation Techniques

Have Formal Training In


Mathematics Physical Sciences Computer Science Geologic Principles Scientific Method Engineering

But Who Has Ever Had A Course In


EFFECTIVE ESTIMATING?

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Estimating With Probabilistic Ranges


P99
SMALL CHANCE THE OUTCOME IS MORE THAN P10 VALUE

P01

P90

P10

P50

80 % Confidence Interval

P50

P10
LARGE CHANCE

P90

P01 THE OUTCOME IS MORE


1 THAN P90 VALUE 10 100 1,000

P99
10,000

Estimating P10 - P90 Ranges


Provide your ranges for questions 1- 5 .

from

to
80% confidence

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Estimating With Probabilistic Ranges


Pre-drilling geological estimates of the size of prospect-targets are real-life exercises in estimating under uncertainty. The following exercise is a practical experiment in assessing uncertainty. Even though you probably will not know the exact answer to each of the ten questions below, you should be able to make an estimate as to the correct answer. One way to do this is to try to bracket the correct answer between an upper and a lower limit. Try to set the upper and lower limits such that you are 80% sure that the range you have selected will contain the correct answer. For example, you might be 80% sure that Columbus discovered Latin America sometime between 1450 and 1550 A.D.

P90

P10 How many hotel / apartment rooms did Brazil agree to have for the 2016 Rio Olympics to meet IOC requirements? What is the air distance from Ushuaia, Argentina to Anchorage, Alaska, in KM? When was the epic poem Cantar de Mio Cid written in Spain? What was the population of Bogota in 1912? When was the earliest permanent settlement established (as proved by ceramic dating) in South America?

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

P90

P10
How much shale gas (BCF) in the U.S. was produced in 2009? As of 1998 how many exploration wells were drilled in the 379,357 sq km of the Sao Francisco basin? According to YPF what is the areal extent (sq km) of the Vaca Muerta formation? How much oil (barrels) was produced from the Bakken Formation in North Dakota in 2011? What was the 2009 average daily oil production from Colombia (b / d)?

Estimating With An 80% Confidence Range


P90 Practice Slide A Slide B Slide C P50 P10

EXERCISE 3-2

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

INTUITIVE DECISION MAKING



Involves no formal analysis or process. Impossible to audit for completeness and quality. Based upon past experiences. Everyone processes information differently. Danger of poor decision making is greater when a manager is under stress.

Estimating Under Uncertainty


Biases

Must Know How To Play The Odds Level Of Uncertainty Generally Unknown Biases In Assessing Uncertainty:
Overconfidence Representativeness Availability Anchoring Implicit Conditioning Motivational
Tversky &Kahneman (1981)

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Estimating Under Uncertainty


Biases

Overconfidence
We Think We Are Smarter Than We Actually Are! Setting Predictive Range Too Narrow

Representativeness
Small Sample Size; Examples May Not Be Truly Analogous

Availability
Tendency To More Heavily Weight The Spectacular Examples; Limited Imagination
Tversky &Kahneman (1981)

Biases In Assessing Uncertainty


Overconfidence
Percentage of Misses Type of People Tested Harvard MBAs Employees of a chemical company Managers of a computer co. Physicians Physicists Type of Question Asked Trivial facts Chemical industry and company-specific facts General business facts Company-specific facts Probability that a patient has pneumonia Scientific estimates like the speed of light Ideal Target 2% 10% 50% 5% 5% 0-20% 32% Actually Observed 46% 50% 79% 80% 58% 82% 41%

RussoandSchoemaker (1989)

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Biases in Assessing Uncertainty


Overconfidence

We Think We Are Smarter Than We Actually Are! Setting Predictive Range Too Narrow

We have little comprehension of comparative levels of confidence

Capen (1976)

Biases in Assessing Uncertainty


Representativeness
A certain town is served by two hospitals. In the larger hospital, about 45 babies are born each day, and in the smaller hospital, about 15 babies are born each day. Although the overall proportion of boys is about 50% the actual proportion at either hospital may be greater or less than 50% on any given day.
QUESTION: At the end of one year, which hospital is likely to have the greater number of days on which more than 60% of the babies born were boys? (Check only one answer)

The large hospital The small hospital Neither hospital - The number of days will be about the same (within 5% of each other)

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Biases in Assessing Uncertainty


Availability
For the following pair, choose the one you think causes more deaths in the United States each year:

Lung Cancer

Motor Vehicle Accidents

RussoandSchoemaker (1989)

Estimating Under Uncertainty


Biases

Anchoring
Reluctance to Move Far From Some Initial High or Low Side Value

Implicit Conditioning
We Estimate Based on Our Specific Experience or Expertise - Seek the Input of Others!

Motivational
Overestimate to Sell the Deal; or Underestimate to Err on the Side of Conservatism
Tversky andKahneman (1981)

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Biases in Assessing Uncertainty


Anchoring
Thats my story and Im sticking to it. Dont confuse me with the facts.

Tversky andKahneman (1981)

Biases in Assessing Uncertainty


Implicit Conditioning
Ranking Order
Source of Risk Actuarial Est Deaths League Women Voters 4 6 2 3 18 5 10 22 24 1 College Business Students Club Members 3 7 5 2 19 6 11 17 23 1 4 5 3 1 19 2 9 24 20 8 Moore(1983)

Smoking Alcoholic Bev Motor Vehicles Handguns Electric Power Motorcycles Surgery X-Rays Railroads Nuclear Power

150,000 100,000 50,000 17,000 14,000 3,000 2,800 2,300 1,950 100

Rose & Associates, LLP

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Biases in Assessing Uncertainty


Motivational
Overestimate to sell the deal; or underestimate to err on the side of conservatism; often driven by the rewards system and unwritten rules in place! Rules for Advancement
Written
1 2 3

Unwritten

1 2 3 4 5
Tversky andKahneman (1981)

How Do we Overcome Bias?


Require every estimate to Include a Level of Confidence Provide feedback and training to help people calibrate Ask Disconfirming Questions about your ideas and sources Expose the hidden sources of future problems Limit yourself to the information you can handle Develop Multiple Working Hypotheses

Rose & Associates, LLP

10

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

How Do We Overcome BIAS? Progression of Methods


Ruling Theory Working Hypotheses Multiple Working Hypotheses

Chamberlin(1931)

Progression of Methods Ruling Theory


Single, Dogmatic Opinion Theory Often Formed Prematurely Press New Facts To Fit Theory If Questioned, Owner Takes Personally

Rose & Associates, LLP

11

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Progression of Methods Working Hypothesis


Better than Ruling Theory (a single, dogmatically-held opinion) Used as a vehicle to suggest and guide inquiry, not to find facts to fit the theory: If this is true, then this other fact should follow. Although a working hypothesis may (after appropriate investigation and consideration) legitimately evolve into Ruling Theory, it is all too easy for it to slide into Ruling Theory without suitable support.

Progression of Methods Multiple Working Hypothesis


Pursues simultaneously, none championed Helps pinpoint critical elements common to all possible scenarios Promotes thoroughness Requires discipline and imagination Allows later adjustments

Make more than one map !

What are the alternative scenarios?

What could hurt me here?

Rose & Associates, LLP

12

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Progression of Methods Multiple Working Hypothesis


More than one map is necessary to often characterize the various scenarios you may interpret This is the crux of composite mapping so critical to unconventional play assessment

Where are the sweet spots? What are the alternative scenarios?

Make more than one map !

What could hurt me here?

Pressures Against the Use of Multiple Working Hypothesis



May be seen as wasteful or irresolute Belief that management cant handle uncertainty Difficult to sell a Project if it isnt clear cut Need to justify Project expenditures May need to act decisively before enough data are available

Rose & Associates, LLP

13

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Multiple Working Hypotheses


Numerical Logic:

1, 2, 4 _____ 1, 2, 4 _____ 1, 2, 4 _____ 1, 2, 4 _____ 1, 2, 4 _____

While You Are Doing This, Pay Attention To What Is Going On In Your Head!

A Professional Estimator Should Be

Consistent
Can apply a similar process to different trends

Unbiased
Miss the range on the high side about as often as the low side

Practical
Starting with baseline or analog data, craft the estimate approximate with the amount of uncertainty present

Resourceful
Should make maximum use of the information contained in the data

Calibrated
Track results against forecasts as a basis for improvement

Rose & Associates, LLP

14

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Most Geotechnical Estimates Are Made Under Uncertainty!


Source rock type, volumes and richness Productive area Avg. net pay Effective Porosity HC saturation
More prevalent with unconventional reservoirs

Ro, perhaps density TOC, Gas content (from lab work) % recovery (dont estimate in UCR) Bg, Bo

Geological Chance Factors Productive rates: IP, initial decline, b, terminal decline and rate Pilot duration and completion design Costs: land, drilling, completion, developing, operating Prices, timing

The Basic Equation For Project Evaluation


This part of the Basic Equation is expressed as a cash-flow schedule incorporating net revenue interest (NRI), production decline, time-value of money, and anticipated inflation.

Pc

NRI *

TOTAL EUR

WELLHEAD PRICE

NET FINDING, NET DEVELOPING, & + TAXES OPERATING COSTS

(1- Pc)

NET AFTER TAX FAILURE COST

PROJECT EXPECTED NET PRESENT VALUE @ X%

Rose & Associates, LLP

15

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

Key Points To Remember

Patterns of estimating bias under


uncertainty are endemic, but usually treatable By recognizing this, we can take steps to overcome the biases A good estimator should be:
Consistent, Unbiased, Practical, Resourceful & Calibrated

ESTIMATING IS $ERIOUS BUSINES$ !!


Effective estimating skills arise from

Utilizing the expectation of lognormality; Appropriate Ranges; Plausibility and reality checks; The power of independent multiple estimates

!!! Learning Opportunity !!!


Normally each $1.00 or more each A larger contribution will make a larger prize. Winner takes all!

Rose & Associates, LLP

16

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

METHODS FOR IMPROVING ESTIMATES EXERCISE


1. Working independently, make a median (P50) estimate of the number of beans in the picture (median is 50/50 chance that true number is greater or smaller than your estimate). ........................................................................................................................... BEST GUESS

______________ LOW - P90


2. Range Estimates: Make a "low-side" estimate having only a 10% chance that the true number of beans is less than that; make a "high-side" estimate having only a 10% chance that there are more beans than that. Enter both estimates at right in the dashed box ...... 3. Lognormal fit: plot your P90 and P10 estimates on a log-probability graph and draw a straight line between them. This line is a cumulative lognormal probability distribution. What is its P50 value? .............................................. Check value: P50 = SQRT(P10*P90) 4. Calculate the mean using Swansons Rule as follows: (0.3 x ___________) + (0.4 x ___________) + (0.3 x ___________) = P10 value P90 value derived P50 value Check value: Mean > P50 5. Reality Checks: a. Extend the straight line out to the P99 and P01 points and read off these extreme values: P99 b. Keeping your P99 estimate in mind, inspect an analog "low-side" bean picture, which contains . c. Then return to the first bean picture and make revised estimate of P99 d. Keeping your P01 estimate in mind, inspect an analog "high-side" bean picture, which contains e. Now return again to the first bean picture, and make revised estimate of P01 . 6. Second Lognormal Fit: Plot your revised P99 and P01 estimates on a log probability graph, draw a straight line honoring the data. Determine the P90, P50, and P10 points on that line, then enter your second estimates in box at right .................................................

________________ HIGH - P10


RANGES

DERIVED P50 VALUE

FIRST MEAN

and P01 beans. beans.

P90 _________ P50 _________ P10 _________

7. Calculate the mean just as you did before using your second estimates from item 6 above: (0.3 x ___________) + (0.4 x ___________) + (0.3 x ___________) = P10 value P90 value P50 value 8. Independent Multiple Estimates: a. Compare your P90, P50, P10 and mean estimates with those of your table mates. Enter their second means and average them: 2dM 2dM 2dM Team Average 2dM b. Now, working by yourself, make final P90, P50, P10 estimates by force-fitting them to a straight line on the log probability graph paper c. Once more, calculate a final mean just as you did before using your final estimates: ) + (0.4 x ) + (0.3 x )= (0.3 x P90 value P50 value P10 value Please Print Your Name
Rose & Associates, LLP 17

SECOND ESTIMATES

SECOND MEAN

P90 _________ P50 _________ P10 _________

FINAL ESTIMATE
Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

P99

P01
P02

P98

Rose & Associates, LLP

P95

P05

P90

P10
P20 P30
P40

P80

P70

P60

18

P50

P50
P60 P70 P80

P40

P30

P20

P10

P90
P95 P98

P05

P02

Ch 3 - Uncertainty AAPG Cartagena 2D course, Sept. 2013

P01

P99

10

100

1,000

10,000

Vous aimerez peut-être aussi