Professional Documents
Culture Documents
SPE 144383 A Practical Solution For Booking PUD Reserves in A Resource Play Using Probabilistic Methods
SPE 144383 A Practical Solution For Booking PUD Reserves in A Resource Play Using Probabilistic Methods
This paper was prepared for presentation at the SPE North American Unconventional Gas Conference and Exhibition held in The Woodlands, Texas, USA, 14–16 June 2011.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
A practical probabilistic method is presented to administer the evaluation of proved undeveloped (PUD) oil and gas
reserves in resource plays. In 2009, the Security and Exchange Commission (SEC) adopted revisions to the oil
and gas reporting rules which allowed for the opportunity to use probabilistic methods in the evaluation and
disclosure of PUD locations that are more than one spacing unit away from existing production. The revisions to
the “Modernized SEC Rules” created a need for improved methods to evaluate and disclose probabilistic reserves.
The objective of this paper is to present a new and practical probabilistic method for PUD reserve reporting that
addresses the SEC requirements. The paper addresses the issues of proved reserve volumes per PUD location,
proved well spacing, and “sweet-spots”.
The useful method may be applicable to a broad range of subjects, such as professional expertise, individual
ownership positions, and geologic play situations. The figures and principle steps of the method are presented
using the Fayetteville Shale as an example. The method was developed specifically to comply with the Modernized
Rules, to be consistent with statistical principles, and to accommodate ownership situations while maintaining an
expectation of being practical to use.
The first paper (Dobson 2011) in this series of three introduced a new method of describing the proved area in a
resource play. The third paper will present a practical justification for this probabilistic approach. In this paper,
historical production statistics are used to evaluate the method’s ability to comply with reasonable certainty in a
forecast at specific PUD locations.
Introduction
Over time, numerous probabilistic methods for reserve estimation have been presented. The parametric method
(Smith 1993), decision tree analysis, influence diagrams (Schuyler 1998), and modified deterministic methods
(Patricelli 1995) all present probabilistic solutions. Monte Carlo simulation based methods (Hall 2007, Mata 1997,
Moore 1999) might be considered as the traditional industry favorite.
Establishing the basis of the Monte Carlo calculation generally consists of analyzing a best fit probability
distribution function (PDF) for net pay, water saturation, recovery factor and other geologic variables. The terms
are combined through a volumetric calculation and Monte Carlo simulation to estimate both the original in place
and the recoverable volumes at various probability percentages.
Monte Carlo simulation output is displayed in the form of a cumulative frequency plot. Values representing any
probability can be read from the frequency plot. This approach has been shown to be a reliable technology to
estimate recoverable hydrocarbon volumes at a field or geologic feature level and seems to fit the intended
description of probabilistic reserve evaluation.
Each method has inherent strengths and drawbacks and each method will create solutions that constitute a
respectable probabilistic approach to reserve evaluation. The shortfall with the use of methods that calculate a
2 SPE 144383
single volume estimate for the play is that the “answer” provided by those methods is at the field or play level
instead of the individual PUD location level. The field or play level answer is functional if the field is either unitized
or owned 100% by one operator. However, it is insufficient in the calculation of net economically producible
reserves when there is variable ownership, variable marketing, or commercial terms.
While working through specific field cases for resource plays, we observed that although reserve volumes can be
estimated for a field, it is unclear how to divide that volume into appropriate capital projects in order to build an
economic model for each PUD location. Creating an individual economic model for each PUD is critical to the
calculation of economically producible reserves. Likewise, the Monte Carlo simulated reserve volume fails to
establish the proved well spacing density and capital commitment contingent on that spacing. Since both well
spacing and the PUD economic model are required to estimate economically producible reserves for a resource
play, reserve volumes also need to be determined for specific PUDs. Due to these issues, traditional Monte Carlo
methods are beneficial as a check for the reasonableness of another method’s answer but are insufficient for
resource play PUD reserve estimation.
Companies disclosing PUD reserves to the SEC are required to report their net economically producible reserves.
A critical step in developing the required report involves running the economic model. Inputs for the model include
a production schedule, the associated business conditions such as product prices, capital expense, operating
expense, and working and net interest. Individual inputs are needed for each PUD well location in order to enable
the economic model to properly reflect the economically producible reserves. The economic model calculates the
net economically producible volume for each individual project. The aggregation of the individual projects net
reserve volume to the level of field or play results in the reportable volume. Determining a company’s net
economically producible PUD reserves for the assets in a play requires the successful use of detailed economic
models.
Production Statistics
In an established resource play, PDP production statistics, such as EURs, IPs and decline analysis value, are in
abundant supply, form the basis of a statistical dataset and serve as the most direct measurement of economic
producibility. Production statistics from current producers that exist in sufficient quantity and spatial dispersion
form the most reliable probabilistic basis from which to forecast the eventual outcome of future projects. The PUD
projects must be in the same geologic analog (GA) as the PDP wells. These projects are in the same GA if they
share the same geological formation, environment of deposition, geological structure, and drive mechanism.
Additionally they should share a common completion and production method(s) paradigm. Under these conditions,
stochastic analysis of the PDPs can provide a reliable
prediction of PUD performance with reasonable certainty
(SEC 2008).
While the units are from the same geologic analog the net
Map 1—Lateral length opportunity due to unit configuration
EUR reserve volumes available for company Q to report
are significantly different for the three units. In the
example, well performance is calculated to be 885
mcf/FT. The average well lateral length for company Q in Unit A is 3500’ yielding 3.1 BCF per well gross and 2.0
BCF net. Meanwhile, the average well length in Unit B is 2000’ and the gross reserves are 1.8 BCF per well and
only 0.18 BCF net per well. Assignment of reserves based on a volume per well would yield incorrect results
SPE 144383 3
reserve reporting.
If we appropriately measure and analyze the conditions that create performance results from existing PDP wells,
we can use the empirical relationships from that analysis to predict the future results of a similar condition at the
well and unit level. This allows for individual PUD booking estimates at specific locations and ownership situations.
The Proved Area
This paper presents a solution for the booked value at individual PUD locations under the variable business
conditions typically found in a resource play. A brief review of the steps from Paper one, “A Practical Solution to
Describe the Proved Area within a Resource Play Using Probabilistic Methods” (Dobson 2011), are useful in order
to tie the proved area concept into the current subject.
It is necessary to complete the proved area steps prior to continuing with the evaluation to solve for the appropriate
PUD booking value. Stochastic PUD booking is limited to the proved area described in Paper one. PUDs outside of
the proved area are evaluated by traditional “deterministic methods”.
Additional discussion about aggregation, empirical analysis and statistical tools are presented in Paper one and will
not be repeated here. We generally use the term EUR/FT throughout the PDP analysis. Estimated Ultimate
Recovery (EUR) is calculated per PDP well from either decline curve analysis (DCA), simulation, analytical
techniques or a combination of methods. Effective lateral length is calculated as the distance between the most
distal and proximal perforations of the wellbore. This length is represented by FT in the EUR/FT term, while EUR
units are in mcf. For simplicity we standardized on EUR/FT for this presentation; however, other analyses do not
need to be limited to just EUR/FT. Noting that the technique presented in the three paper series will also work in
vertical plays on a per well EUR basis instead of EUR/FT.
Throughout this paper we employ the same use of P^ defined in Paper one. P^ is calculated by averaging the
mean and the P50 of the data. In addition to calculating the P^ it is advised to routinely create and analyze the
cumulative frequency plots for the data to check for lognormality, extreme outliers, and changes in the P10/P90
ratio that would impact the analysis. For sample counts greater than 50, the sum of the P^ values approximates the
aggregated Monte Carlo simulated P90 of the sample and is a reliable and practical measure to compare datasets
(Dobson 2011). The evaluator should develop a clear understanding of aggregation and sample size to effectively
use the techniques.
The groups are plotted by increasing well count as a function of EUR/FT P^ for each group. We observe that the
P^ for one well per section (654 wells) is 616 MCF/FT and two wells per section (508 wells) is slightly better with a
P^ of 667 MCF/FT. Well performance with increasing well density until the peak of six wells per section (1007
MCF/FT @ 132 wells).
4 SPE 144383
% Variance to Analog wells
1000 929
881
780 794 821 805
50% four, three, two or one well per section. All but one
800 and two wells per section perform better than the all
P^ EUR/FT
667 29%
654 24% 24% 30%
616
19% well case. Nine and ten wells per
600 508 13%
405 5% 3% 10%
400
2% 368 335 section perform approximately 19% less than the
‐14% ‐10% peak of six wells per section, yet still perform better
200 132 128
‐21%
49 54 than one, two, or three wells per section and
10
0 ‐30% approximately 4% better than the all well group. With
All Wells 1 2 3 4 5 6 7 8 9 10 a ten wells per section group we can conclude that
Wells Per Section the sample size is too small for stochastic analysis.
P^ EUR/FT Well Count P^ Variance
Since P^ for eight wells per section is 967 MCF/FT
Figure 1—Performance (P^ mcf/FT) of well groups as a function
and the counts are sufficient to satisfy statistical
of well count per section, entire play.
significance we are confident that development of
eight wells per section will achieve the expected
results with reasonable certainty. When compared to the all well P^ of 780 MCF/FT, we would not have statistical
support that well count greater than 9 resulted in reasonable certainty. The All Well data set is the benchmark
because it includes the full range of observed outcomes and all spacing densities.
Well spacing analysis in this play indicates that the sections with more wells generally perform better than the
sections with fewer wells. This has been one of the
more startling observations in the entire field study.
Horizontal Azimuth
Map 3—Horizontal well path, depicting N‐S and 330·
azimuth for PDPs (circle) and PUDS (triangle)
SPE 144383 5
similar to that of the well density plot. If P^ for the groups do not differ by more than 10% they are reasonably
certain to generate comparable results. In this case, the two groups may be treated as a single data grouping. If
they differ by more than 10% then the datasets for N-S and the 330° azimuth would be handled separately
throughout the remainder of the evaluation. As with all of the P^ comparison analysis the authors strongly
encourage evaluators to quality check the entire PDF for shape and size.
The stochastic proved area of a resource play, as it was presented in Paper one, is the area that can be described
empirically and that is reasonably certain (stochastically) of generating economically producible results. The
definition of the area does not suggest that every location is the same, but only that the area meets a minimum
standard where there is a “high degree of confidence” that the area will be economically producible due to geology
and production. The proved area might exhibit patterns or clusters of either greater or lesser production
performance also called “sweetspots”. Identification of sweetspot patterns is often initially recognized through
visual inspection of maps displaying various production data.
Well P^
PROJECT_AREA Count P10/P90 EUR/FT
PA 1 374 3.4 789
PA 2 527 2.9 1089
PA 3 301 3.0 869
PA 4 237 4.6 637
Total Play 1439 4.0
Table 1—Project Area summary for well count and P^ EUR/FT
The next step in calculating the PUD reserve volume multiplies the effective lateral length of the planned PUD with
the P^ of the PDP analysis (Eq 1) from the respective Project Area. The product is the PUD EUR for the location.
This is the gross EUR used in the economics model.
A PUD Gross EUR reserve volume for a location in the PA 1 Project Area and a planned lateral length of 3800’ is
shown in Eq 1.
The gross EUR 2.998 BCF reserve volume is assigned to the PUD economic model for that location and all
business conditions are also applied to the model for that location. The production schedule is adjusted to match
local conditions so that the assigned reserve is developed in the model. The model returns the net economically
producible reserve volume for that PUD.
SPE 144383 7
The procedure is repeated for each prospective PUD in the proved area. Aggregation of the individual PUD cases
returns the economically producible reserves that are reasonably certain of recovery for the probabilistically proven
area of the play or field.
The PUD location(s) must have a geologic and engineering drilling plan where the effective lateral length is
designed for the specific location. The lateral length must be supported by a stochastic analysis showing that the
likelihood of the planned lateral length(s) is reasonable.
Conclusions
A practical approach to book probabilistic PUDs has been presented. Probabilistic reserves in a resource play can
be assigned to individual PUD locations for economic producibility calculations. The method includes steps to
stochastically evaluate the PDP wells as an analog to forecast PUD locations in the same proved Project Area by
1. Establishing the stochastically proved well density (well spacing)
2. Evaluating the subset analog conditions such as horizontal azimuth
3. Evaluating spatial patterns or so called “sweetspots” for division to Project areas
The method describes the steps to calculate and assign stochastically proved reserves to individual PUD locations
for use in the economic model.
We remind the reader that each step of the method is subject to checks against other reliable methods such as
volumetrics, modeling, and recovery factor for reasonable answers. Other results analysis techniques will be
presented in Paper Three.
.
References
Dobson, M.L., Lupardus,P.D, Divine,T.W., and McLane,M.A., A Practical Solution to Describe the Proved Area
within a Resouce Play Using Probabilistic Methods. Paper SPE 142319-PP presented at the Production &
Operations Symposium, Oklahoma City, Oklahoma, U.S.A., March 2011.
Hall, R.K., Evaluating Resource Plays With Statistical Models. Paper SPE 107435 presented at the Hyrocarbon,
Economics and Evaluation Symposium held in Dallas, Texas, U.S.A., April 2007
Mata T., Rojas L., Maraven, S.A. Probabilistic Reserves Estimation of Mara West Field, Maracaibo Basin,
Venezuela: Case Study. Paper SPE 38805 presented at the SPE ATCE held in San Antonio, Texas, USA, October
1997.
Moore, L.R., and Mudford, B.S.; “Probabilistic Play Analysis From Geoscience to Economics: An Example From
the Gulf of Mexico,”. paper SPE 52955 presented at the 1999 SPE Hydrocarbon Economics and Evaluation
Symposium, Dallas, Texas, U.S.A., Mar. 2001.
Particelli, J.A., McMichael, C.L., An Integrated Deterministic/Probabilistic Approach to Reserve Estimations. Paper
SPE 28329 presented at the SPE ATCE, September 1994
Schuyler, J.R., Probabilistic Reserves Lead to More Accurate Assessments. Paper SPE 49302 presented at the
SPE ATCE in San Antonio, Texas, September, 1998.
Smith,P.J., Hendry,D.J, and Crowther,A.R., The Quantification and Management of Uncertainty in Reserves Paper
SPE 26056, presented at the Western Regional Meeting held in Anchorage, Alska, U.S.A., May 1993.
SEC - Modernization of Oil and Gas Reporting, Conforming Version 2008. 17 CFR Parts 210,211,229,and 249
[Release Nos. 33-8995; 34-59192; FR-78; File No. S7-15-08]. US SEC, Washington, DC. (29 December 2008).
http://www.sec.gov.rules/final/2008/33-8995.pdf December 2009