Problems For Assessing Oil And Gas Future Production

Assessing Oil and Gas Future Production and the end of Cheap Oil?

by J. H. Laherrere
site: http://www.oilcrisis.com/laherrere
for Canadian Society of Exploration Geophysicists
in Calgary April 6, 1999

Contents:

-Introduction

-Reliability of data and definitions

-Technology and Economics

-Models:

-fractal

-lognormal

-cyclic (bell-shaped)

-diminishing returns = creaming curve

-Forecasts for the next 50 years

-Peak of cheap oil

-Conclusion

How to approach the problem in general:

–”All that is simple is false and all that is complex is useless”         Paul Valery

-Nature is not linear and problems needs multi-solutions: reject any single solution and favour multi- solutions

-Maths does not solve all problems. Trial and error often gives better results. Progress involves making mistakes and recognising them.

do not hide your errors, but try to not repeat them

-Definitions often are vague:

always request definition of each item

-Distributions are harmonious when natural and large in number

select natural and large areas and reject man-made boundaries

            avoid narrow distributions, narrow periods and drastic economic changes

-Data are questionable and accuracy is usually at 10%:

adjust unit to reject any significant digits over 3

How many solutions?

2, 4, 6, ?, ?, ?, ?

rank:      1, 2, 3, …. n-2,  n-1,  n

value:       2, 4, 6, …. N-2, N-1, N

Occam (or Ockham 1285-1349)’s razor = best model = simplest

= minimum complexity = 2,4,6,8,10,   2n

in fact too simple often false:

ever continuous growth is impossible in life

life = principle minimum action (Fermat-Maupertuis) or minimum time (Snell’s law)

            or maximum room (Fibonacci series)

this series 2,4,6, has a beginning and grows, in life it should peak and decline, negative values do not exist

The simplest natural answer is 2,4,6,6,0     N = 3(N-1) – 3(N-2)

-Modeling is just an approximation and a model    should be simple and convenient. Many models which can fit the data, but some give unrealistic extrapolation

try different models and check the extrapolation

select the models which also fit other natural distributions

Nature is characterized by:

-inequality

-self-similarity = fractal

-cyclycity, one or several

-everything is curved by gravitation

-finite without limit

-minimum energy (or minimum time)

-small causes, big effects

-central limit theorem = large number

What data?  How reliable ?

-What  are we talking about?

-supply:           crude oil could be either:

crude oil (conventional)

crude oil +condensate (separator),

crude oil +condensate +NGL (processing),

crude oil +condensate +NGL +unconventional

+synthetic oil +refinery gains

gas could be:

associated

non-associated

wet or dry

gross with inert gas (Natuna gasfield) or only HC or marketable

measured with different standards of temperature (FSU: 5% difference with US)

-demand:         liquids

marketable gas

-oil equivalence for gas:

calorific:                                1 boe = 5.6 or 6 kcf

price:                          1 boe = 10 kcf (IEA forecast)

GTL conversion:        1 boe = 10 kcf (project Exxon Qatar)

Canada burner tip       1 boe = 14 kcf 1989, =20 kcf 1991

we use 1 boe = 10 kcf

+ missing barrels ?

variation from 1 to 1. 2

barrel (of oil) is not an official unit: USDOE is obliged to add barrel of 42 US gallons

few know what is the meaning of bbl (blue barrrel of last century), but use it instead of b

The proportion of liquids derived from gas ranges from 26% (including 21% of NGL) in the United States to 6% in the world as a whole. For Canada, NGL represent 20% in 1995.

Reserves:

-is the total of future (expected) production

-uncertain until the last day of production

-confidential

-conflict between the deterministic approach (one value) and the probabilitic approach (range: three values)

-probability is subjective and poorly understood by most

-only expected values can be added

-only most likely values can be multiplied

-no consensus on definition as publishing reserves is a political act and depends of the desired image

-SPE/WPC definitions: contradictory and not followed, cover only 3 classes out of the 12 from NPD (Norway)

-reserves (expected production) versus resources (potential in the ground)

-different external reserves for the banker, for the shareholder, for the tax agencies, for the OPEC quotas,

-different internal reserves from the geologist-geophysicist, from the petroleum-reservoir engineer, from the economist, from the manager, from the state agency

-initial reserves and remaining reserves at a certain date

-reserves without a date should be initial

Motives for declaring the minimum:

Explorer: for large prospects to avoid being regarded as a dreamer

Engineer: to reduce the risk of being wrong (a mean estimate implies being wrong 60% of the time),

Company: to secure apparent reserve growth over time which presents a more attractive financial image, may reduce tax, and in some cases facilitate its competitive position

Motives for declaring the maximum:

Explorer: to make a small prospect sufficiently economic to pass corporate hurdles,

Company: to augment its share values; sale value;  the stock options of its executives; overcome government constraints to depletion rate (Frigg)

Countries: to provide collateral for debt (Mexico)

OPEC: to increase quota (large increases in the late 1980s)

Soviet Union: to show the maximum theoretical recovery ignoring economic constraints

Motives for declaring the mean (the expected value):

Those seeking a valid national total. The mean value of a large number of fields is the sum of the mean value of each individual field, despite the fact that, statistically, 60% of the cases will prove incorrect.

Comments by the best experts:

“There are currently almost as many definitions for reserves as there are evaluators, oil and gas companies, securities commissions and government departments. Each one uses its own version of the definitions for its own purposes” DeSorcy 1993

“The resource base [of the former Soviet Union] appeared to be strongly exaggerated due to inclusion of reserves and resources that are neither reliable nor technologically nor economically viable” Khalimov 1993

“An industry that prides itself on its use of science, technology and frontier risk assessment finds itself in the 1990s with a reserve definition more reminiscent of the 1890s”      “illegal addition of proved reserves”  Capen 1996

“Why our reserves definition don’t work anymore” Caldwell 1996

“Virtual reserves – and other measures designed to confuse the investing public” Tobin 1996

“The term “reserves” often is treated as if it were synonymous with “proved reserves”. This practice completely ignores the fact that any prudent operator will have, at least internally, estimates of probable and possible reserves”  Ross 1998

Practice on reserves:

-companies listed on US stock market: = SEC rules

= proved reserves = reasonable certainty as FDA for approval of a new product

deterministic: net pay x spacing  x b/acre-foot

-rest of the world, in particular North Sea

= 2P = proved + probable = usually 50% probability

simulation (>10 000 cells)

-correct practice = mean (expected value)

Field growth (or reserves growth or reserves appreciation):

-in US = part of neglected probable becomes proved

= large amount (multiplied from discovery time

by 4 for MMS to 9 for USGS)

= mainly bad practice of reporting

and not technical improvement

-rest of the world = 2P (50%) – mean (40%)

= small amount (0.5%/a)

large field growth = bad practice

Summing proved reserves

Every report on proved reserves:

the proved value of the total = the sum of the proved values of each field or country = wrong.

Adding the Proved reserves of individual fields understates the total.

The proper procedure is to sum the mean values to obtain the mean value of the sum

example: throwing dices

-the”proved” value of throwing 1 dice is 2, as there are 5 chances out of 6 to get at least 2

-the”proved” value of throwing 8 dices is not adding 8 times 2 = 16, but 22

-the mean value of for 1 dice is 3.5 and for 8 dices 28

Multiplying values of several parameters to obtain the reserves of a field

reserves of a field = 6 parameters:

net pay x porosity x area x saturation x1/volume factor x recovery

Multiplying the proved value for each parameter does not give the Proved reserves for the field as a whole: thus multiplying six parameters each having a probability of 90% would give a total of 0.9 power 6, namely 53% = median value

The Most Likely (mode) value for the field is obtained by multiplying the mode value for each parameter.

Fortunately most of the times it is the best value which is taken and the multiplication is almost correct, except that it gives the most likely value and not the proven value.

Norwegian System

The Norwegian Petroleum Directorate operates a particularly thorough and sound procedure recognising as many as twelve categories, of which only three are covered in the SPE/WPC classification, in all cases referring to initial reserves and resources.

0 Reserves where production is ceased

1 Reserves in production

2 Reserves with an approved development plan

3 Resources in a late planning phase (PDO approval within 2 years)

4 Resources in an early planning phase (PDO approval within 10 years)

5 Resources which may be developed in the long term

6  Resources where development is not very likely

7  Resources in new discoveries for which the evaluation is not complete

8 Resources from possible future measures to increase the recovery factor (measures which are not planned, possibly superseding present-day technology)

9  Resources in prospects

10 Resources in leads

11 Unmapped resources

It should be noted that Norway accords Reserve status to no more than 69 percent of its discovered resources.

Recovery Factor:

            -volume-in-place (from seismic and wells) is more uncertain than reserves which are estimated from production decline. When a field is abandoned, reserves are exactly known but OIP is still uncertain.

-usually taken as round number: NPD 50% for oil & 75% for gas

-when estimated from proved reserves, recovery factor grows when probable becomes proved, and not because of technical improvements, but better knowledge

-simulation techniques or production decline give directly reserves and not OIP

-recovery factor is usually a sales promotion item

Grace:             83 Gb undiscovered liquids: it will take 615 years to discover at the present rate of new field discoveries

-reply USGS: with field growth ratio of 10: only 62 years

-comments: ratio is about 4 and take 50 years to reach

In brief:

-US Proved ≈  67% probability = most likely and not 90% as defined by SPE/WPC

-US field growth = difference between mean ( ≈ 40%) and most likely (≈ 65%)

-Rest of the world = 2P ≈  50% probability (median)

-Rest of the world field growth = difference between mean ( ≈ 40%) and 2P (≈ 50%)

–SPE/WPC 1997 reserves definitions: contradictory within the definitions:

– proved = reasonable certainty and high confidence and 90%

– deterministic probable = as likely as not = 50%

– probabilistic proved + probable = 50%

– does not consider resources

– only 3 classes out of the 12 NPD classes

– contradictory with the practice

-proved or proven = most likely

-Accuracy: at best for production and reserves ≈ 10% : giving more than two significant digits implies incompetence, usually the first digit is wrong

-Unit must be adjusted to give only two significant digits

Variations of reported reserves from different sources:

Technology

–From Greek techne  (skill) + logos  (talk) = talk on skills, in the past “techniques”, now “talk on techniques”

-Reserves additions by revisions of old discoveries is usually attributed to progress of technology as 3D, horizontal drilling

-3D is from the 70s (in fact 1966 Shenli with abacus = 200 faults), 3D is a must for appraisal, not for discovery

-horizontal wells are from the 70s

-New discoveries in deepwaters: dynamic positioning ship: first use in Labrador: discoveries of Bjarni, Gudrid 1974, Snorri 1976: BP drillship converted in ODP Resolution by lack of discoveries

-What is called “new technology” was in fact used for the last 20 years everywhere to estimate reserves and to produce faster and cheaper.

-Learning curve is often wrongly called technology progress

-44% of the discoveries of the world outside North America are not yet developed, but they represent only 8% of the oil discovered and 19% of the gas discovered. Most of them are uneconomical discoveries or political discoveries (Tikrit).

-Technology follows the needs as in deepwaters for Brazil

-Some economists believe in “Santa Claus Technology”

What about Y2K bug? >600 G$ for two digits

What about bad use of probability? adding proved 90%

What about poor reserves reporting giving field growth ?

What about poor use of unit?  M (or m NRcan) for k, m or mm or MM for M,

Gm³ ≈ one million the earth volume (should be G.m³)

What about poor reporting on wells counts: EIA 1998 ?

Examples of technologies in UK, US & FSU:

gaslift allows to produce faster, but do not increase reserves:

decrease despite technology available

ultimate stays the same despite new technology as steamflood and horizontal wells

Largest field in FSU: Samotlor (Western Siberia)

new western technology will not change the ultimate

Expectations during the last 30 year for finding a new large oil or gas province

Success:

>50 Gb = 2 years world’s oil consumption:

-North Sea 1969 = Ekofisk

-North Slope 1969 = Prudhoe Bay

other

-Campos deepwater 1984 = Albacora

-West Africa offshore Angola 1971= Takula, 1996 = Girassol

-Yemen 1984 = Alif

-GOM deepwater

-Barentz Russia gas 1988 = Shtokmanov

-Northwest shelf Australia

Disappointments after high expectations

-China offshore

-Tarim

-E.Siberia

-Barents Norway

-Falklands up to now

-Caspian up to now

Economics & ecology

-Economists say that oil reserves will increase with the oil price

it is true for some unconventional reserves as stripper wells, uneconomical fields, tight reservoirs, CBM, extra-heavy oils, oilsands, but not for conventional oil reserves (cheap oil)

it is true for conventional remote gas as transport for gas costs 6 to 10 times as for oil

-Limitations are more physical (time and number of rigs) or ecological than economical:

-The world number of very small fields (<10 Mb) in the ground is about 10 times the number already discovered and represents about 400 Gb: it would take several centuries (even a millenium) to find them with the present number of rigs.

-Canadian oilsands plants needs space and cannot be multiplied by many (wastes = lake of 22 km)

-US oil shales: lack of water, ecology: increased volume of fine

Modeling

Distribution of objects:

Inventory: rank by decreasing size and displayed in a log-log format:

comparison between:

lognormal, stretched exponential, shifted linear and parabolic fractal

All models can be almost identical for the first 500 ranks and quite different beyond:

a model has to be checked on extrapolation for large ranks

Example of US urban agglomerations:

Here the parabolic fractal has the best extrapolation to total population and is the best model as it can easily adjusted to fit the total population

The most important is to select a natural domain:

urban agglomeration defined by morphological boundaries (continuity of building) and not city defined by administrative boundaries

petroleum system and not tectonic basin or country

However when a large number of different domains are added (continent or world) it can be modeled as a natural domain

1500 largest world earthquakes:

Gutenberg-Richter law = power-law= straight line

The data does fit a straight line and the parabola is the best fit

Fractal

–fractal = term invented by Benoit Mandelbrot from

fractus = broken and not from fraction

(fractal dimension of the coast of Brittany = 1.3)

a part describes the whole (cauliflower):

a scale (pencil or man) is needed for a picture of outcrops

= self-similarity

–perfect self-similarity = power law = linear fractal in log-log display size(Sn)-rank(n)            S_n=S₁n^-a                      or        log S_n = log S₁ –a log n

–imperfect similarity = curved display simplest = parabolic fractal

log S_n = log S₁ –a log n –b (log n)²

example: Gulf of Mexico OCS:

Comparison of 3 Petroleum Systems:

Niger delta and GOM are similar as they are dispersed habitat when Saharan Triassic is a concentrated habitat.

Canadian oilpools:

-oilpools are not geological as broken down by units = crooked display

-oilfields display more harmonious

Lognormal

= law of proportionality         dx/x is constant

it is as easy for a millionaire to spend 1000 $ as for a billionaire to spend 1 000 000 $

random = normal (Gauss ore bell-shaped) probability distribution

works well when the mean is large compared to the standard deviation

when the mean is of the same order as the standard deviation, the probability curve is limited by zero and is deformed by this lower limit becoming unsymmetrical

Lognormal plot needs frequency and if more fields are added, frequency change: it is not convenient. If the same plot is kept; it means that as many large fields as small fields have to be added. In mature basins anly small fields are added.

Comparison of 3 Petroleum Systems

Niger Delta and Saharan Triassic are almost straight and similar as few small fields are found, by contrast GOM has a lot of small fields and is linear only for fields larger than 0.1 Mb (88%).

Creaming curve= law of diminishing returns

            In mature basins, most of the large fields are found and only small fields are undiscovered

the cumulative discoveries versus the cumulative number of New Field Wildcats is close to an hyperbola

If for a continent as Africa, frontier (deepwaters) is found, another hyperbola starts:

Cycle: Hubbert ≈ Gauss ≈ one Sine wave

-in probability, adding a large number of asymmetrical distributions gives a symmetrical normal distribution = central limit theorem

-in seismic, a sound is broken down into harmonics with Fourier analysis

Hubbert curve = derivative of the logistic curve (Verhulst 1845) , known as a law of population growth

Hubbert equation        P = 2P_m/(l+COSH(b(t_m-t)))

or        P = 2P_m/(l+COSH(5(t_m-t)/c))             with the ultimate U = 0.8 cP_m

Large countries with a large number of fields and continuous exploration:

only one cycle = US 48 Lower States and FSU

 departure from the model= drastic economic changes (proration, oil shock)

Gauss is lower than Hubbert and data for the beginning of the curve, Hubbert is a better model

Best tool for forecasting

= correlating annual production to shifted annual discovery

several exploration cycles: France, Netherlands, UK, US with Alaska

Two cycles of exploration and two cycles of production shifted by 7 years

World’s liquids production of the 50 next years

reserves = future production : a curve is better than a number

DOE forecast need unrealistic ultimates, based on too optimistic forecasts on demand (and on population increase)

IEA forecast is within our forecast and the peak of cheap oil (conventional oil) is around 2010 (±5)

Notice that the mini and maxi do not add, only the variances (square of standard deviation) are added

How much to discover to delay the peak by one year?

In a Hubbert curve as the ultimate is 0.8. times peak production multiplied by half width, the peak is shifted by one year when the ultimate increases by 0.8 times the annual peak production.

If the peak production is about 45 Gb/a (or 120 Mb/d as in the forecast DOE/EIA IOE 98), 36 Gb new discoveries will delay the peak by one year. A 1 Gb discovery delays the peak by 10 days!

If the peak production is about 32 Gb/a (our forecast), 26 Gb new discoveries will delay the peak by one year. A 1 Gb discovery delays the peak by 14 days!

Population forecast

UN forecasts have consistently reduced since 1990 the estimated size and date of peak. World population growth as a percentage of total population peaked in 1964 at 2.1%/a and is now down to 1.3%/a ; and in absolute terms peaked in 1990 at 90 million and now 78 millions.

Population and HC consumption: global & per capita

-HC has peaked in 1979 and will peak around 2015

-population will peak around 2040

-HC per capita has peaked in 1973, 1979 and will peak around 2005

Conclusion

-reserves data unreliable, as political

-field growth = bad practices of reserves reporting

-technological progress leads to faster and cheaper production, not much impact on conventional reserves revisions as they are already anticipated, but needed for unconventional resources

-oil price increase will raise, little the conventional reserves, more the unconventional resources not yet listed as reserves

-oil price increase will have a drastic effect on diminishing demand

-cheap oil will peak soon in North Sea and for Non-OPEC; for the world around 2005 as the Persian Gulf producers will be in the driver’s seat to increase the price (if they do not fight between themselves)

-population will peak soon in industrialised countries and few decades later in developing countries

-HC demand will be lower than forecasted

-there is a very large resource in energy savings

Mankind will adjust his way of life, as always, to the decline of hydrocarbons. Thanks to Petroconsultants for allowing use of their data.