spe 98010_new method for production data analysis to identify new opportunities in mature fields

8/12/2019 SPE 98010_New Method for Production Data Analysis to Identify New Opportunities in Mature Fields

1/22

Copyright 2005, Society of Petroleum Engineers

This paper was prepared for presentation at the 2005 SPE Eastern Regional Meeting held inMorgantown, W.V., 1416 September 2005.

This paper was selected for presentation by an SPE Program Committee following review ofinformation contained in a proposal submitted by the author(s). Contents of the paper, aspresented, have not been reviewed by the Society of Petroleum Engineers and are subject tocorrection by the author(s). The material, as presented, does not necessarily reflect anyposition of the Society of Petroleum Engineers, its officers, or members. Papers presented atSPE meetings are subject to publication review by Editorial Committees of the Society ofPetroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paperfor commercial purposes without the written consent of the Society of Petroleum Engineers isprohibited. Permission to reproduce in print is restricted to a proposal of not more than 300words; illustrations may not be copied. The proposal must contain conspicuousacknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

ABSTRACT

Most of the mature fields in the United States have been

producing for many years. Production in these fields started ata time when reservoir characterization was not a priority;

therefore they lack data that can help in reservoir

characterization. On the other hand to re-vitalize these fields

in a time that price of hydrocarbon is high, requires certaindegree of reservoir characterization in order to identify

locations with potentials of economical production. The most

common type of data that may be found in many of the maturefields is production data. This is due to the fact that usually

production data is recorded as a regulatory obligation orsimply because it was needed to perform economic analysis.

Using production data as a source for making decisions havebeen on the petroleum engineers agenda for many years and

several methods have been developed for accomplishing this

task. There are three major shortcomings related to the efforts

that focus on production data analysis. The first one has to dowith the fact that due to the nature of production data its

analysis is quite subjective. Even when certain techniques

show promise in deducing valuable information from

production data, the issue of subjectivity remains intact.Furthermore, as the second shortcoming, existing production

data analysis techniques usually address individual wells and

therefore do not undertake the entire field or the reservoir as acoherent system.

The third short-coming is the lack of a user friendly software

product that can perform production data analysis withminimum subjectivity and reasonable repeatability while

addressing the entire field (reservoir) instead of autonomous,

disjointed wells. It is a well known fact that techniques such as

decline curve analysis and type curve matching address

individual wells (or sometime groups of wells without

geographic resolution) and are highly subjective.

In this paper a new methodology is introduced that attempts toaddress the first and the second, i.e. unify a comprehensive

production data analysis with reduced subjectivity while

addressing the entire reservoir with reasonable geographic

resolution. The geographic mapping of the depletion or

remaining reserves can assists engineers in making informeddecision on where to drill or which well to remediate. The

third shortcoming will be addressed in a separate paper wherea software product is introduced that would perform the

analysis with minimum user interaction.

The techniques introduced here are statistical in nature andfocuses on intelligent systems to analyze production data. This

methodology integrates conventional production data analysis

techniques such as decline curve analysis, type curve matching

and single well radial simulation model, with new techniques

developed based on intelligent systems (one or more oftechniques such as neural networks, genetic algorithms and

fuzzy logic) in order to map fluid flow in the reservoir as a

function of time. A set of two dimensional maps are generatedto identify the relative reservoir quality and three dimensiona

maps that track the sweet spots in the field with time in order

to identify the most appropriate locations that may still have

reserves to be produced. This methodology can play animportant role in identifying new opportunities in mature

fields.

In this paper the methodology is introduced and its applicationto a field in the mid-continent is demonstrated.

INTRODUCTIONTechniques of production data analysis (PDA) have improved

significantly over the past several years. These techniques areused to provide information on reservoir permeability, fracture

length, fracture conductivity, well drainage area, original gasin place (OGIP), estimated ultimate recovery (EUR) and skin

Although there are many available methods identified, there is

no one clear method that always yields the most reliable

answer1. Furthermore, tools that make these techniques

available to the engineers are not readily available. The goal ofthis study is to develop a comprehensive tool for production

data analysis and make it available for use by industry.

Production data analysis techniques started systematically

SPE 98010

New Method for Production Data Analysis to Identify New Opportunities in MatureFields: Methodology and ApplicationMohaghegh, S. D., Gaskari, R. and Jalali, J., West Virginia University


2/22

2 New Method for Production Data Analysis to Identify New Opportunities in Mature Fields SPE 98010

with a method presented by Arps in the 1950s1. Arps decline

analysis is still being used because of its simplicity, and since

its an empirical method, it doesnt require any reservoir orwell parameters. Fetkovich proposed a set of equations

described by an exponent, b, having a range between 0 and 12.

Arps equation is based on empirical relationships of rate vs.

time for oil wells and is shown below3:

( )( )bi

i

tbD

qtq

1

1+

= (1)

In this relationship, b = 0 and b = 1 represent exponential and

harmonic decline, respectively. Any value of bbetween 0 and

1 represents a hyperbolic decline. Although Arps equation is

only for pseudo-steady state conditions, it has been often

misused for oil and gas wells whose flow regimes are in atransient state.

The Fetkovich methodology analyzes oil wells producing at a

constant pressure. He combined early time, analyticaltransient solutions with Arps equations for the later time,

pseudo-steady state solutions. The Fetkovich method likeArps equation, calculates expected ultimate recovery.

Carters gas system type curves were published in 19853.

Carter used a variable identifying the magnitude of the

pressure drawdown in gas wells. A curve with a value of 1corresponds to b=0 in Fetkovich liquid decline curves and

represents a liquid system curve with an exponential decline.

Curves with =0.5 and 0.75 are for gas wells with an

increasing magnitude of pressure drawdown.

Agarwal-Gardner also introduced a method for productiondata analysis in 1998. This technique combines decline curveand type curve concepts for estimating reserves and other

reservoir parameters for oil and gas wells using production

data. 3

Other methods were introduced by Palacio & Blasingame, andAgarwal-Gardner, which provide information on gas in place,

permeability, and skin.

There are also modern analytical methods that do not use typecurves. One of these methods is flowing material balance.

This technique provides the hydrocarbons in place usingproduction rate and flowing pressure data from a reservoir

under volumetric depletion. 3

METHODOLOGY

In this section a step by step process for Intelligent ProductionData Analysis (IPDA) is introduced. Several items must be

mentioned in order to put the degree of reliability of these

techniques in perspective. Please keep in mind that IPDA isdeveloped for situations that only production data is available.

Therefore, in situations that one has access to other data such

as logs, core analysis, pressure tests, geologic models, etc. onemight choose to use other well established techniques.

Nevertheless, it will be demonstrated that as more data such a

those mentioned above are available, one may use them to

increase the accuracy and the reliability of the methodologybeing introduced here.

Furthermore, as it will be explained in this paper, IPDA, at

this time, may come across as a lengthy procedure. This is due

to the fact that IPDA is an iterative and essentially anoptimization technique. The next step in our efforts is the

development of an automatic or semi-automatic procedure thawill make this process much faster and easier to implement. In

the new development the goal is to reduce the user interaction

to a minimum. Figure 1 is a flow chart that describes the

methodology used in this process. As shown in the chart, thereare four distinct and inter-related parts to this process, namely

decline curve analysis, type curve matching, single-wel

reservoir simulation (history matching) and finally relative

reservoir quality indexing and mapping that includes three

dimensional mapping of remaining reserves.

In a nutshell, IPDA is a two step process. In the first step theidea is to simultaneously, interactively and iteratively perform

decline curve analysis, type curve matching and history

matching on the production data of a particular well in the

field until convergence is achieved to a unified set of reservoir

characterizations. Given the fact that each of these techniquesare quite subjective by nature, by letting each one technique to

guide and keep an eye on the other two during the analysis, the

degree of confidence and reliability on the results as well as

repeatability of the analysis will increase.

The second step is to use the well-based results from the firs

step and super impose them on the field as a whole anddevelop two and three dimensional maps of reservoir quality

and remaining reserves. In the following sections each part othis process will be explained.

Decline Curve Analysis

Decline curve analysis is the first step in the process. The

production data is plotted on semi-log scale and decline curve

is fitted. If one uses hyperbolic decline, then the outcome ofthe decline curve analysis would be qi (initial flow rate)

Di (initial decline rate) and b (hyperbolic exponent)

Figure 2 is an example of decline curve analysis performed ona well located in the Golden Trend Fields of Oklahoma.

It is a good idea to plot both production rate and cumulative

production versus time simultaneously and try to match theproduction rate while keeping an eye on the cumulativeproduction. This usually helps in getting a reasonably good

match. Once the match is completed the result would be a set

of decline curve characteristics as mentioned above. Based on

the decline curve characteristics that you have identified inthis step you can easily calculate Estimated Ultimate Recovery

(EUR) for a certain number of years, say 30 years. The 30

year EUR for the well in Figure 2 is calculated to be 1,795MMSCF.

Type Curve Matching

Two things are important in selecting the type curves for this


3/22

SPE 98011 Mohaghegh, Gaskari and Jalali 3

step of the analysis. First, the type curves must be for rate and

not pressure, and second, it would be great if it could provide

some commonality with the decline curve results, such that itwould use some of the results of the decline curve. Something

that would weave the two (the decline curve analysis and the

type curve matching) together. This is essential since the

procedure would call for an iterative process.

In this case a set of type curves for low permeability reservoirs

with hydraulically fractured wells4were used. Figure 3 showsthe type curve match achieved for the same well shown in

Figure 2. The link between this set of type curves and the

decline curve analysis are the hyperbolic exponent b. In

Figure 3 it is shown that the type curves for a particular bvalue are generated and used for matching the production rates

from this well. The b value used for this well (b=0.9) was

the same that was calculated during the decline curve analysis.

This is important since a new set of decline curves can be

developed for any values of b. Figure 4 shows theproduction data from this well plotted against type curves that

have been generated for b values 0.1 (on the left) and 1.5(on the right).

Once the match is accomplished, the results from type curve

matching is a set of parameters such as permeability, fracture

half length and drainage area. Results of these parametersappear on the top left corner of the type curve screen as show

in Figures 3 and 4. On the top right corner of the screen the

EUR (again for 30 years) is calculated and shown. The EUR

from the decline curve analysis was calculated to be 1,795MMSCF while the EUR from the type curve matching is

calculate to be 1,800 MMSCF as shown in Figure 3. The EUR

calculated from the decline curve analysis is used as acontrolling factor for the match of the type curve. This allows

us to develop more confidence on the type curve matching andhave a bit more faith on the values calculated for permeability,

fracture half-length and the drainage area. The EUR values

from the two procedures (type curve and decline curve) areused in order to finalize the match in both cases. As we

mentioned before this is an iterative procedure and sometimes

it may take several iterations before the decline curve analysisand the type curve match would agree with one another.

One important issue that needs to be mentioned at this point isthat the type curve matching process requires knowledge

about a set of parameters. These parameters are used during

the calculation of permeability, fracture half length, drainage

area and EUR. These parameters are listed below:

Initial reservoir pressure; Average reservoir temperature; Gas specific gravity; Isotropicity (kx/kyratio); Drainage shape factor (L/W ratio); Average porosity; Average pay thickness; Average gas saturation;

Average flowing bottom-hole pressure.

Most of the parameters above can be (and usually are) guessed

within a particular range that is acceptable for a particular

field. Usually the initial reservoir pressure for a field or

formation is known with reasonable range or it can beassumed based on formation depth. Formation depth can also

be a good indication of average reservoir temperature. Gas

specific gravity can be easily calculated based on the assumed

average initial pressure and reservoir temperature. In most of

our calculations we assume that the reservoir is isotropicmeaning that the kx/kyratio is equal to 1. The drainage shape

factor is also assumed to be 1 meaning that we are assuming asquare drainage area. Average porosity, thickness and gas

saturation can be calculated for each well from logs, if they

are available. If they are not, then an average value for the

entire field can be assumed. This method allows for bettermatches and results in higher confidence level if wells in the

field have logs. By having access to logs; porosity, thickness

and saturation can be calculated and used individually for each

well during the analysis. If and when such logs are no

available or prove to be too expensive to analyze (whichseems to be the case in many fields in the United States) then

the procedure allows the user to input an average value (as thebest guess) for all wells. In the bottom right corner of Figure 3

you can see two buttons. The button on the top lets the user

input values for the above parameters that will be used as

default for the entire field and the second button will let the

user to input and over-write the default values for any specificwell.

Single-Well Reservoir Simulation

This step of the analysis calls for history matching theproduction data using a single-well, radial reservoir simulator

Reservoir characterization data that is the result of type curve

matching is used as the starting point for the history matchingprocess and the objective is to match the production data of a

particular well. History matching is not a simple and straighforward procedure. Care must be taken in order to preserve the

integrity of the match, in other words, the match that results

from this procedure must make physical sense. One of themain reasons for performing the history matching in

conjunction with the type curve matching and the decline

curve analysis as an iterative process is to preserve theintegrity of the entire process. Many times the results of the

history matching will force us to go back and re-evaluate our

decline curve results and the type curve match. This rigorousiterative process is the key to the successful completion of this

process.

This is performed for all the wells in the field. In most cases amatch should be possible within reasonable ranges of thereservoir characteristics that were used as the starting poin

(result of iterative type curve matching and decline curve

analysis). This may prove to be a long process. The goal is to

automate this process with minimal user interaction.

Once a match is accomplished within a reasonable range of al

the reservoir characteristics, the ground work has beenestablished for another important step in the analysis, namely

Monte Carlo Simulation. Since now we have (most probably)

diverged from the reservoir characteristics that we started

with, it is reasonable to expect that we have converged on a


4/22


range of values for each of the parameters rather than a single

value. In order to get a more realistic look at the capabilities of

a well and its potential future production a Monte Carlosimulation is performed at this point. During the Monte Carlo

simulation each of the parameters that play a role during the

simulation process is represented by a probability distribution

function (instead of a crisp, certain number) with

characteristics that have been identified during the entireintegrated analysis, i.e. decline curve analysis, type curve

matching and history matching using reservoir simulator.

The result of the Monte Carlo simulation is a probability

distribution of the 30 year EUR. If we have done everything

properly, it is expected that the EUR that was calculatedduring the decline curve analysis and the type curve matching

process for a given well would fall within the probability

distribution that has been calculated as the result of the Monte

Carlo simulation. As the values of these EURs get closer to

higher probability values, our confidence on the accuracy ofthe ranges (of the reservoir characteristics) that were used in

the Monte Carlo simulation will increase.

Mapping of the Reservoir Quality

Once the integrated production data matching is completed

and a set of reservoir characteristics are identified, it is time to

produce some maps that would ultimately help operators inidentifying the sweet spots in the field. The two and three

dimensional maps include relative reservoir quality indices as

a function of time as well as some reservoir characteristics and

map of the remaining reserve as a function of time.

Let us look at an example to help demonstrate the process.

First step is to identify the parameter that would be used as thekey for the reservoir partitioning. Figure 5 shows that Gas

has been selected as the fluid of choice and from among theProduction Indicators (PIs) that have been calculated First 3

Months of Production is selected as the output as well as an

attribute. The PI identified as the output (you may only haveone output but multiple attributes) is used as the key to

partitioning the field while attributes are those PIs that their

average values are calculated for each partition and are used asa guide to fine tune the partitioning process. Once these

selections are made, the next step is to partition the field into

different reservoir qualities based on the average values of theoutput for each partition. This is shown in Figure 6.

As displayed in this figure, 85 wells in this field have been

analyzed for this study. Fuzzy pattern recognition5

isperformed on the latitude and longitude as a function of theselected output and then super imposed on one another to

create the partitions as shown in Figure 6. The numbers on the

map indicate the relative reservoir quality with 1 being the

highest quality. In this representation, the relative quality ofthe reservoir rock in terms of hydrocarbon productivity

decreases with higher numbers of RRQI.

The values corresponding to the partitions in Figure 6 are

shown in Table 1. In this table the 85 wells in the field are

divided into five different categories identified as Relative

Reservoir Quality Indices of 1 through 5. The average value of

the PI (First 3 Months of Production) decreases as the index o

the relative reservoir quality increases. This confirms our

partitioning practices. Figure 7 displays a three dimensionaview of the reservoir quality with smooth transitions from one

reservoir quality to the next.

If the output of the problem is changed from a PI representing

early life of the field to one representing a later time in the lifeof the field, then the difference in these figures may represent

depletion in the reservoir or fluid movement. Furthermoreeach of the reservoir characteristics that are generated as the

result of the analysis that was mentioned in the previous

sections can be mapped throughout the field. This will provide

a good visual of the status of the field at different times in itslife and can provide insight on the locations that have high

production potentials for future development. More details on

these mapping procedures and how remaining reserve are

calculated on a field-wide basis is provided in the next section

RESULTS & DISCUSSIONS

The methodology described in this paper was applied toproduction data from 85 wells producing in the Golden Trend

fields of Oklahoma. The only data used to perform the

analysis shown here are the production data that are publicly

available; therefore, all these analyses can be performed on

any field throughout the United States and Canada. This mayprove to be a valuable tool for independent asset valuation

prior to any acquisition.

The first step in the process is performing three different kindsof analysis techniques on each well simultaneously in an

iterative manner in order to converge to one unified set of

matches and EURs. The three analysis techniques are declinecurve analysis, type curve matching and history matching

using a single-well radial reservoir simulation. Figures 8through 12 show results of two analyses, reached

simultaneously (iteratively) on five different wells in the filedThe graph on the left in each figure is the decline curve

analysis and the graph on the right is the type curve match.

The EURs are quite comparable in each of the figures. Table 3shows the parameters that were calculated in this analysis for

the five wells shown in Figures 8 through 12. As mentioned

before, the hyperbolic exponent found in the decline curveanalysis is used to identify the set of type curves that should

be used during the matching process. Then the EUR from the

type curve matching is compared with the EUR calculated

from the decline curve analysis. If the difference is greaterthan a certain threshold, then the characteristics of the declinecurve is modified and the new bvalue is used in the type curve

matching resulting in a new EUR. This process is continued

until the hyperbolic exponent provides the closest EUR

between type curve matching and the decline curve analysis.

Results shown in Table 3 are the final results that have been

achieved upon convergence when all three methods (declinecurve analysis, type curve matching and history matching

using a single-well radial simulation model) are performed

simultaneously and convergence is achieved. Figures 13 and

14 show the history match results achieved for two of the


5/22


wells in the database. Figures 15 an 16 are results of Monte

Carlo simulation performed on the two wells shown in Figures

13 and 14. During the Monte Carlo simulation each of theparameters that were used and calculated during the type curve

matching and history matching process is assigned a

probability distribution function and the history matched

model is run for hundreds of times each time calculating the

30 year EUR. The 30 year EUR calculated during the declinecurve analysis and the type curve matching are also shown in

Figures 15 and 16. As expected (since this was an iterativeprocedure to converge to a common set of characteristics that

would provide a common set of results) the 30 year EUR

values from the other two techniques fall in the high frequency

areas of the probability distribution function of the 30 yearEUR obtained from Monte Carlo simulation.

Once the iterative process of matching the production data of

each well individually using three simultaneous process of

decline curve analysis, type curve matching and historymatching is completed, several production indicators (that are

simply statistical measures of a wells production) arecalculated such as:

Best 3 Months of production Best 6 Months of production Best 9 Months of production Best 12 Months of production First 3 Months of production First 6 Months of production First 9 Months of production First year cumulative production Three year cumulative production Five year cumulative production Ten year cumulative production

Other indicators are calculated as a result of decline curve

analysis and type curve matching such as:

Decline curve related indicators:

o Initial flow rateo Initial decline rateo Hyperbolic exponento Estimated Ultimate Recover

Type curve matching indicators

o Permeabilityo Fracture half length

o Drainage areao Permeability * Thickness (kh)o Estimated Ultimate Recovery

All the above characteristics are now available for all the wells

in the field, in our case for 85 wells. Figures 17 through 20show the change of the relative reservoir quality index with

time. In Figure 17 the Relative Reservoir Quality Index

(RRQI) is shown as a result of First 3 months of production.As mentioned before RRQI 1 indicates the best location in the

field followed by numbers from 2 to 5. Please keep in mind

that all these numbers and quality indicators are relativeas the

name points out.

The RRQIs are identified based on partitioning the fuzzy

pattern recognition performed on latitude and longitude as

shown in the figure. The red and blue lines in the fuzzy patternrecognition part of the figure are the partitioning agents that

can be moved up and down. Each time the partition agents are

moved the partitioning of the field is renewed and the average

Production Indicator (in the case of Figure 17, the First 3

Months cumulative production) for each of the partitions arerecalculated. This process can be used as a guide for

identification of the correct partitions in the field.

Comparing Figures 17 and 18 one can see the migration of

several wells from RRQI 1 to RRQI 2 and from RRQI 2

to RRQI 3. Based on the reservoir characteristics (andsometimes well related issues such as damage) several wells

do not produce as well as others and exhibit differen

behaviors. If it is a well related problem, these would be

isolated wells and in a long term analysis can be identified by

other means such as a process called Intelligent CandidateSelection Analysis6-8 (ICSA). Integration of ICSA with IPDA

is the topic of a future paper. This integration allows theidentification of wells that are in need of remedial processes

and those that cannot be rescued. This integration would unify

the field development strategies as a combination of wel

remedial operations and new drilling opportunities.

Table 3 shows the statistical change of the RRQIs as the

analysis move from first 3 month of production to first 9

months of production. Following items are notable in this

table. The value of average cumulative production in bothcases (3 and 9 months of production) decreases with an

increase in RRQI (please remember that increase in RRQI

indicates a relative decrease in the reservoir quality)Furthermore, the number of wells that are located in RRQI 1

2, and 3 are different in two cases, identifying the migration owells from one RRQI to another which can be an indication o

depletion.

In Figure 19, First 3 years of production is selected as the

output for partitioning. Changes in this figure as compared

with Figure 18 show the movement of RRQIs in the field thatis a clear indication of depletion and fluid movement in the

reservoir. Furthermore, Figure 20 shows the 30 year EUR

forecast for this field if no new wells are drilled. This can helpoperators to identify the locations in the field that would

present good candidate for infill drilling due to the remaining

gas in place.

Similar maps can be generated for other parameters that havebeen calculated as a result of decline curve and type curve

matching analyses. Figure 21 shows the two dimensional map

based on reservoir permeability and Figure 22 is the three

dimensional version of permeability distribution in the filed.

The next step is to track the fluid movement in the reservoir

In case of primary recovery this can simply mean depletion ordetection of remaining hydrocarbon in place. Wells are drilled

at different times and therefore may be in production for

different length of time. This is shown in Figure 23 where

example for several wells is presented. What we are interested


6/22


in is the map of the remaining fluid in the reservoir. This can

help operators (in conjunction with other maps such as those

presented in Figure 17 through 24) to identify the best locationfor drilling new wells or to make decision on the wells that

need to be stimulated or re-stimulated.

As shown in Figure 23 we are going to identify the remaining

gas in place by calculating the 30 Year EUR for each well andthen subtracting the amount of gas that they have produced up

to a certain point in time. This way we can identify theremaining gas in place as of any date of interest. Figure 23

shows the dates at which we would perform the above

calculation. Identification of remaining gas in place based on

the current cumulative production and decline curve forecastis not new. New items that are introduced here are the

identification of fuzzy patterns that evolves in the field and

three dimensional mapping of the remaining hydrocarbon in

place.

Such mapping is shown in Figures 24 and 25. In Figure 24 the

remaining gas in place as of year 2005 is shown and Figure 25is the same map after 15 more years of production. The

difference between these two figures shows the depletion in

the reservoir and identifies the parts of the field that still have

potential for more recovery. Figure 26 shows the reservoir

depletion in 5 year intervals by mapping remaining gas inplace as of January of 2005, 2010, 2015 and 2020.

An important note is that using this technique, new wells can

be virtually drilled in a particular location and their effects(production) on the remaining reserves can be observed. In

other words many What If scenarios can be played out in

order to make the best possible decision on the location of thenext wells.

Application of these techniques to water flooding (and other

enhanced recovery) operation may prove to be very useful in

identification of flood fronts as a function of injection andproduction data when reservoir characteristics of the formation

is complex and not very well understood.

CONCLUSIONS

A new technique for field-wide production data analysis has

been introduced in this paper. This technique takes advantageof an iterative procedure to unify the matching of production

data for each well using three independent techniques, namely

decline curve analysis, type curve matching and history

matching using a single-well radial reservoir simulationmodel. Working with the above three techniquessimultaneously and iteratively a set of reservoir characteristics

emerge that approximately satisfies all three production

matches. In absence of any reservoir characterization studies,

this technique resolves the subjectivity associated with each ofthese techniques by using each one of them as a controlling

factor for the other two in an iterative fashion.

The techniques presented in this paper also provide means for

mapping reservoir quality throughout the field using the

results of the production data matching mentioned above. The

result is a set of two and three dimensional maps that identifies

the potential places for drilling new wells. This technique uses

state of the art in intelligent systems in order to discoverpatterns in the production data using all the wells in the field

simultaneously. This is an important recognizing factor for

this technique as compared to others in the literature tha

allows mapping of the reservoir characterization and

remaining reserves in the field in two and three dimensionagraphs.

Implementation of this technique in its present form for a field

with tens or hundreds of wells is a relatively lengthy process

The research and development team is currently working to

automate this process that would require minimum userinteraction during the iterative process. A new paper

introducing the automated procedure will be presented as soon

as the development process is completed.

REFERENCES1. A Systematic and Comprehensive Methodology for

Advanced Analysis of Production Data, L. Matter,D. M. Anderson, Fekete Associates Inc., SPE 84472,

2003.

2. Useful Concepts for Decline-Curve Forecasting,Reserve Estimation, and Analysis, M.J. Fetkovich,

E.J. Fetkovich, and M.D. Fetkovich, PhillipsPetroleum Co., SPE Reservoir Engineering, February

1996.

3. Analyzing Well Production Data Using CombinedType Curve and Decline Curve Analysis Concepts,Ram G. Agarwal, David C. Gardner, Stanley W.

Kleinsteiber, and Del D. Fussell, Amoco Exploration

and Production Co., SPE 49222, 19984. Advanced Type Curve Analysis for Low

Permeability Gas Reservoirs, Cox, Kuuskraa andHansen. SPE 35595, 1998.

5. Fuzzy Clustering Models and Applications, Studiesin Fuzziness and Soft Computing, Volume 9. M

Sato, Y. Sato, L.C. Jain. Physica-Verlag, Heidelberg

New York.

6. "Benchmarking of Restimulation Candidate SelectionTechniques in Layered, Tight Gas Sand Formations

Using Reservoir Simulation", Reeves, Bastian

Spivey, Flumerfelt, Mohaghegh, and Koperna, SPE63096, 2000.

7. "Development of an Intelligent Systems Approach toRestimulation Candidate Selection", Mohaghegh

Reeves, and Hill, SPE 59767, 2000.8. "Restimulation Technology for Tight Gas Sand

Wells", Reeves, Hill, Hopkins, Conway, Tiner, and

Mohaghegh, SPE 56482, 1999.


7/22


Table 1.Average value of First 3 Months PI for each of the partitions.

RRQI No. Wells % of WellsFirst 3 Months of

Production (MSCF)

1 15 17.65 89,684.5

2 22 25.88 74,832.9

3 9 10.59 51,179.0

4 22 25.88 30,801.4

5 17 20.00 22,141.6

TOTAL 85 100

Figure 1.Flow chart of Intelligent Production Data Analysis IPDA.


8/22


Figure 2.Decline curve analysis of well C-ANY #1-4

Figure 3.Type curve matching of well C-ANY #1-4


9/22


Figure 4.Type curve matching of well C-ANY #1-4 with b values 0.1 and 1.5.

Figure 5.Selecting output and attributes for partitioning the field into zones with different quality.


10/22


Figure 6.Partitioning the field into zones with different quality based on First 3 Months PI.

Table 2.Results of field partitioning for 3 and 9 months production.

Number of Wells Average Cumulative Production

RRQI 3 Months 9 Months 3 Months 9 Months

1 15 9 89,684 270,355

2 22 21 74,832 198,686

3 9 16 51,179 152,250

4 22 22 30,801 100,389

5 17 17 22,141 67,964


11/22


Figure 7.Three dimensional view of the relative reservoir quality partitioning of the field based on First 3 Months PI.


12/22


Figure 8.Simultaneous decline curve analysis and type curve matching of well C-LL#1-28.

Figure 9.Simultaneous decline curve analysis and type curve matching of well C-WBY #1-1.

Figure 10.Simultaneous decline curve analysis and type curve matching of well C-AN #2-27.


13/22


Figure 11.Simultaneous decline curve analysis and type curve matching of well N-CERO 3-3.

Figure 12.Simultaneous decline curve analysis and type curve matching of well T-dle A#1.

Table 3.Results of simultaneous decline curve analysis and type curve matching

shown for five wells in the field.Decline Curve Analysis Type Curve Matching

Well NameQi

(MSCF)Di b

EUR(MMSCF)

K(md)

Xf(ft)

A(ac)

EUR(MMSCF)

C-LL#1-28 40,500 0.051 0.51 1,449.3 3.45 35.8 73.4 1,455.5

C-WBY #1-1 28,130 0.057 1.15 1,709.6 2.09 28.3 48.1 1,671.6

C-AN #2-27 47,830 0.248 1.30 1,314.8 3.78 29.3 17.7 1,328.3

N-CERO 3-3 6,333 0.026 1.80 800.8 0.18 27.2 6.8 809.6

T-DLE A#1 10,894 0.268 1.42 318.9 0.75 8.6 4.2 318.5


14/22


Figure 13.History match results for Well C-ER #1-16.

Figure 14.History match results for Well C-VA #1-34.


15/22


Figure 15.Results of Monte Carlo Simulation study for Well C-ER #1-16.

Figure 16.Results of Monte Carlo Simulation study for Well C-VA #1-34.


16/22


Figure 17.Relative Reservoir Quality Index based on first three months of cumulative production.

Figure 18.Relative Reservoir Quality Index based on first nine months of cumulative production.

Fuzzy Pattern Recognition

FuzzyPatternRecognition


17/22


Figure 19.Relative Reservoir Quality Index based on first three years of cumulative production.

Figure 20.Relative Reservoir Quality Index based on 30 year EUR forecast.


18/22


Figure 21.Distribution of reservoir permeability in the field based on IPDAs integrated technique.


19/22


Figure 22.Three dimensional distribution of reservoir permeability in the field based on IPDAs integrated technique.


20/22


Figure 23.Production schedule for different wells, corresponding production forecast and dates for calculating remaining gas in place

Figure 24.Remaining gas in place as of January 2005.


21/22


Figure 25.Remaining gas in place as of January 2020.


22/22


Figure 26.Remaining gas in place as a function of time.

spe 98010_new method for production data analysis to identify new opportunities in mature fields

Documents