+ All Categories
Home > Documents > SPE-8280-PA

SPE-8280-PA

Date post: 01-Jun-2018
Category:
Upload: tomk2220
View: 219 times
Download: 0 times
Share this document with a friend

of 5

Transcript
  • 8/9/2019 SPE-8280-PA

    1/9

    Combined nalysis

    o

    Postfracturing

    Performance

    and

    Pressure Buildup Data

    for Evaluating

    an

    MHF as Well

    James N. Bostic, SPE, Amoco Production Co.

    Ram O. Agarwal,

    SPE, Amoco Production Co.

    Robert D. Carter,

    SPE, Amoco Production Co.

    Introduction

    Low-permeability less than 0.1 md) gas wells must

    be

    stimulated with massive hydraulic fracture MHF)

    treatments to be commercial. Once a well has been

    stimulated, it

    is

    important to determine the length

    and conductivity

    of

    the resulting fracture as ac

    curately as possible to predict rates and reserves and

    evaluate the effectiveness

    of

    the stimulation.

    Two types of data normally are available to

    perform the required analysis. The first type

    is

    routinely collected data such as gas flow rates and

    flowing wellhead pressures. The second type

    of

    data

    is

    that from specially designed tests such as pressure

    buildup tests. There are various methods available

    analyzing both kinds

    of

    data,

    but

    all

    of

    the

    methods have certain limitations.

    Current analysis techniques are satisfactory for

    finite-flow-capacity fractures when formation flow

    capacity kh

    is

    known and sufficient production data

    without shut-in periods

    is

    available to perform a

    type-curve match as shown by Cinco

    et al

    1 and

    Agarwal

    et al

    2

    However, one disadvantage of

    MH

    treatments

    is

    that they often cause the stimulated

    wells to produce large volumes

    of

    load water

    returning fracture fluids) for several weeks or

    months. Production data obtained during this

    0149·2136/80/0010·8280 00.25

    opyright 198 SO iety o Petroleum Engineers

    cleanup period usually are not applicable for type

    curve matching purposes with the currently available

    type curves. Since type curves normally are plotted

    on a log-log scale

    and

    exhibit the most character at

    small dimensionless times, the loss of the early-time

    production data disproportionately reduces the

    length

    of

    the available data band and

    is

    critical to the

    problem

    of

    obtaining a satisfactory match.

    The purpose of this paper is to present a method of

    combining the production

    and

    buildup test data so

    that their concurrent analysis overcomes some

    of

    the

    limitations

    of

    either individual

    data

    set. The

    superposition principle is used in this combination,

    and the analysis

    of

    the resulting

    data

    set should

    provide a more accurate result than previously

    possible. This combination has the potential to

    generate field data curves

    of

    sufficient length that the

    shape

    of

    the data curve may be sufficiently definitive

    to determine formation flow capacity by type-curve

    analysis. Additionally, this method makes it possible

    to use production data that contain occasional shut

    in periods

    of

    long duration.

    The superposition principle, on which this paper

    is

    based, was described by van Everdingen and Hurst

    3

    and was used by Hutchinson and Sikora,

    4

    Mueller,

    5

    and Coats

    et

    al

    6

    to calculate pressure functions for

    the analysis of aquifers associated with water-drive

    oil reservoirs. Those pressure functions were called

    This paper presents a method

    o

    combining

    post

    racturing performance data with

    post racturing buildup data

    or

    analysis on the same constant-rate type curve. This

    combination

    o

    buildup and production data is accomplished through the use

    o

    the

    superposition principle. The method allows the use

    o

    production data that contain

    occasional shut-in periods

    o

    long duration.

    OCTOBER 198

    1711

  • 8/9/2019 SPE-8280-PA

    2/9

      resistance,,,4 response,

    5

    and influence

    6

    functions and are similar to the pressure function

    P

    F N

    presented in this paper.

    n

    1965 Jargon and van Poollen

    7

    applied the

    superposition principle to single-well drawdown test

    data to generate a pressure function they called the

    unit response function. They then used this unit

    response function in the analysis of variable-rate

    variable-pressure drawdown tests of oil wells.

    n

    1975

    Ridl

    ey

    8

    discussed a method of unified

    analysis

    of

    well tests which used pressure data from

    both drawdown and buildup periods. Ridley used a

    statistical regression analysis method of parameter

    estimation, but he assumed the fundamental form of

    the pressure solution.

    The method described herein extends these con

    cepts to the combined analysis of buildup and

    production data f ;'om MHF-stimulated gas wells.

    t

    also appears to offer certain advantages in both

    accuracy and ease of use, in comparison with the

    currently accepted analysis techniques.

    Methods and Limitations

    o

    Postfracturing nalysis

    A number

    of

    techniques have been presented in the

    petroleum engineering literature for the analysis of

    post fracturing pressure and/or rate data. Raghavan

    9

    presented a very comprehensive summary of pressure

    behavior of fractured wells in 1977. Cinco and

    Samaniego

    10

    and Lee and Holditch

    11

    also have

    discussed the advantages and limitations

    of the

    various analysis methods. While some discussion of

    the various available analysis techniques is included,

    this paper is not intended to duplicate other com

    petent and recent work by examining in detail all the

    various analysis methods.

    Basically, the various analysis techniques can be

    divided into two major categories: conventional

    analysis techniques and type-curve matching

    techniques. The majority of the conventional

    methods are applicable only to infinite-flow-capacity

    fractures. (The modified Millheim-Cichowicz meth

    od proposed by Lee et at 11

    is

    an exception to this

    restriction, as

    is

    the analysis of the bilinear-flow

    period data using the one-fourth root of time plot

    suggested by Cinco et at 10 Both will be discussed in

    more detail.) Of the various available type-curve

    methods, some are applicable only to infinite-flow

    capacity fractures, while others can be used with both

    infinite- and finite-flow-capacity fractures.

    1,2,10

    Defining dimensionless fracture flow capacity

    after Agarwal

    et

    at 2 as

    kJ :

    D

    = , . . . . . . . . . . . . . . . . . . . . . . . . . .

    1)

    k f

    a fracture performs as an infinite-flow-capacity

    fracture only for high FCD values. Since MHF

    treatments usually are associated with long finite

    flow-capacity fractures, any technique used to

    analyze MHF-stimulated wells should be applicable

    to finite-flow-capacity fractures.

    As mentioned previously, the only two con

    ventional analysis techniques which address the

    7 2

    finite-flow-capacity concept are the modified

    Millheim-Cichowicz method and Cinco's one-

    fourth root analysis of the bilinear-flow-period

    data. However, as Lee

    et

    at 11 and Cinco

    et

    at 10

    point out, both of these techniques require that

    formation permeability be known a priori. However,

    if this condition

    is

    met, the type-curve matching

    techniques also are simplified greatly and both the

    reliability and uniqueness of their answers are im

    proved, as discussed in Ref.

    2.

    Since type curves are fairly reliable where for

    mation permeability

    is

    known and still can be used, if

    somewhat less effectively, in the absence of a known

    permeability,

    we

    prefer the use of type-curve

    methods over other methods for the routine analysis

    of MHF-stimulated wells. This especially applies

    when time considerations will not permit the ap

    plication of all available methods. Agarwal

    et at

    2

    have shown that their finite-flow-capacity vertical

    fracture type curves produce the same solutions as

    the Cinco et at 1 type curves. For simplicity

    throughout the remainder

    of

    this paper, all examples

    with and references to finite-flow-capacity vertical

    fracture type curves will use the Agarwal

    et

    at

    type

    curves.

    Two sets of finite-flow-capacity vertical-fracture

    type curves for

    an

    infinite reservoir were presented in

    Ref. 2. One set of type curves was for a vertically

    fractured well producing at a constant weI/bore rate

    The other set of type curves was for a vertically

    fractured well producing at a constant weI/bore

    pressure

    While the latter set of curves

    is

    much more

    likely to approximate a realistic producing scheme

    for an MHF-stimulated well, the former set also is

    valuable as it can be used to analyze pressure buildup

    data.

    There are some limitations on the use of these type

    curves. One

    of

    these limitations

    is

    that separate type

    curves must be used for the analysis of buildup and

    production data. Trying to match two sets of field

    data on two separate type curves can be a difficult

    and time-consuming procedure, especially in the

    absence of a dependable estimate of formation flow

    capacity kh

    A second limitation applies to the Agarwal et at

    2

    technique of matching

    1/

    q (rate - 1) vs. time with the

    constant wellbore pressure type curve. This

    limitation

    is

    that the production must be continuous

    from the time the well

    is

    placed on production. While

    this limitation was not stated specifically in Ref. 2, it

    is

    obvious since the rate data

    is

    to be matched with a

    type curve generated under the assumption

    of

    a

    constant flowing bottomhole pressure. However,

    experience has indicated that many wells producing

    in low-permeability gas plays are subject to oc

    casional shut-ins of extended duration due to

    demand restrictions or other problems.

    Additionally, there are some limitations that are

    inherent in the use of either buildup test data or long

    term production data. One limitation of pressure

    buildup data

    is

    that it

    is

    normally impractical or

    impossible to obtain long buildups. Since the buildup

    tests require the well to be shut in, economic con-

    JOURNAL OF PETROLEUM

    TECHNOLOGY

  • 8/9/2019 SPE-8280-PA

    3/9

    siderations may dictate that buildup time should not

    be excessive. n addition, since the buildup data

    is

    to

    be used for type-curve matching, the production time

    immediately before the buildup must be much longer

    than the buildup time.

    12

    Raghavan

    13

    has discussed

    this limitation in detail and has provided guidelines

    for estimating the extent

    of

    acceptable buildup data

    based

    on

    prior production time.

    While the buildup data suffers from

    an

    absence of

    late-time data, the production

    data

    has the opposite

    limitation. Early-time production data on MHF

    stimulated wells

    is

    normally not analyzable due to the

    production of large volumes of returning fracture

    fluids during the first several weeks following

    stimulation. Since type-curve matching normally is

    done

    on

    a log-log basis and it is the early-time

    data

    which have the most characteristic curve shape, the

    loss

    of

    this early-time data has an especially severe

    effect

    on

    the quality

    of

    the analysis.

    Superposition

    The principle of superposition has been applied to

    petroleum engineering problems for many years. Its

    mathematical basis was described in the petroleum

    literature by van Everdingen and Hurst in 1949.

    3

    Superposition can be used to apply known pressure

    solutions for single-well constant-rate systems to

    multiwell and/or multirate systems.

    The general form of the superposition equation

    describing the pressure history of

    an

    undamaged well

    in an infinite system producing a slightly com

    pressible fluid

    is

    n

    E [ q jB j -q j - IB j - I ) [PD tn- t j - I )DJ] ,

    j=1

    (2)

    where qo =0, to

    =0,

    and

    p to)

    =Pi . n SI metric

    units, the numerical constant in the numerator

    is

    1.84.

    For a fractured gas well, Eq. 2 becomes

    1,424T

    m[p to)] -m[p tn ) ] ~ m [ p t n ) ] =

    n

    E [ q j -q j - I ) [PD tn -

    t

    j-I)DXj1),

    ......

    3)

    j=1

    where m p) is the real gas pseudopressure as

    defined by AI-Hussainy

    et

    al.

    14

    and

    tn

    - t j - I )Dx

    is

    a dimensionless time difference based on the fracthre

    half-length. n SI metric units, the numerical con

    stant in the numerator

    is

    1.28 x

    10

    -

    3 .

    evelopment

    o

    FCN

    Since

    PD tn - t

    j

    _

    1

    DXf

    = [m[p to)] -m[pc tn - t j - I )J }

    kh)

    1,424Tqc

    where q

    c

    is a constant rate and Pc is the bottomhole

    OCTOBER 1980

    flowing pressure that would exist

    if

    the rate qc were

    imposed on the well, Eq. 3 can be written as

    1,424T

    n ( kh )

    ~ m [ p t n ) ] = ~ j ~ ~ q j - q j - I ) ~ 1,424T

    [m [p tn) )

    m [ c

    tn

    - t j - I ) I ]

    /qc] .. 4)

    n

    SI metric units, the constant 1,424

    is

    replaced by

    1.28 X 10-

    3

    .

    Simplifying Eq. 4 yields

    n

    ~ m [p tn)]

    =

    E

    qj -q j_ I ) [m [p to) 1

    j=1

    - m [Pc tn - t

    j

    _

    l

    ) l}/qc .   5)

    Define a new term as

    PpCN t) = m [p to) 1

    m

    [Pc

    t)

    l}/qc-   (6)

    Thus, PpCN t) can be thought of as the

    m

    [p to) 1 m [pc t) that would be generated

    by a constant flow riite of 1

    MscflD

    (or 1 std m

    3

    /d in

    SI metric units to avoid carrying a numerical con

    stant in the following equations). Substituting Eq. 6

    into Eq. 5 yields

    n

    ~ m [p tn)] = E qj

    -q j - I

    )PPCN tn - t

    j

    _

    l

    )

    j=1

    n

    =

    q

    1 - q 0 ) PPCN tn -

    to)

    + E

    j=2

    qj -q j - I)PPCN tn

    -

    t

    j

    _

    I

    (7a)

    Since qo =0 and to =0,

    n

    ~ m

    [p tn)

    1

    =ql

    PpCN

    t ~ )

    +

    E

    qj

    - q j - I )

    j=2

    PPCN tn t

    j

    _

    I

    )· (7b)

    Solving Eq. 7b recursively for

    PpCN t

    n

    ), we.

    obtain

    f o r n ~

    P

    PCN t

    n

    ) = [

    ~ m [p tn

    ) ] -

    q

    -

    q

    - 1 )

    PPCN tn

    t

    j

    _

    I

    ) ] /ql .   (8a)

    For

    the special case

    of

    n

    =

    1 Eq. 8a becomes

    and PpCN is then the transformation to the

    equivalent variable pressure history with a constant

    rate. PPCN

    is

    similar to the influence function of

    Coats et al.

    6

    or the unit response function

    of

    Jargon

    et

    al.

    7

    Example

    The following example shows how PpCN is

    calculated for a fractured gas well which has

    produced continuously

    at

    a constant flowing bot-

    1713

  • 8/9/2019 SPE-8280-PA

    4/9

    -- L

    CTU L PRODUCTION

    HISTORY

    ---1 APPROX

    IM TE

    PRODUCTION

    HISTORY

    ~ ________________________ ______

    o

    TIME t

    Fig -

    Production history for

    P

    FCN

    example.

    tom hole pressure

    Pwj.

    For this case, the pressure

    drawdown in terms

    of

    real gas pseudopressure is

    Am [p

    t

    n ]

    =

    m [p t ) ] - m (p

    wj

    =

    constant

    =Am

    (Pwj)

    The rate history for this example is given in Fig. 1

    where the solid line

    is

    the actual producing history

    and the dashed lines represent a stair-step ap

    proximation

    of

    this history where the size

    of

    the time

    length of this approximation can be chosen as small

    as necessary. I f this well had been produced at an

    e ~ u i v l e n t constant rate

    of qc

    = 1

    MscflD

    or 1 std

    m /d to avoid carrying a numerical constant), then

    its pressure history would have been

    n

    PFCN(tn) =

    [Am

    (Pwj) - E (q j -q j - l )

    j=2

    PFCN(tn - t

    j

    -

    1

    )] Iq l ' ......... 9a)

    for

    n ~ a n d

    PFCN

    ( t l )

    =

    Am

    (Pwj) Iq l '

    ..............

    9b)

    for

    n=

    1, where

    PFCN(tn)

    is defined by Eq. 6. Note

    that Eqs. 9a and 9b differ from Eqs. 8a and 8b only

    in the substitution

    of

    the constant

    Am (p

    wj) for the

    function Am [p( t

    n

      ] .

    Use of P

    FCN

    The P

    F N

    vs. time data now generated using

    drawdown data can be matched with the constant

    rate type curves of Agarwal et al

    2

    or

    o t h e r ~ .

    Comprehensive instructions

    on

    the use

    of

    type curves

    are given by Earlougher

    15

    and will not be repeated

    here. Note that dimensionless pressure can be ex

    pressed in terms

    of

    P FCN ( t

    n)

    as

    PD =kh PFCN(t

    n

    )

    /l,424T, ............. 10)

    where

    PFCN(tn)

    is defined by Eq. 6. In SI metric

    units, the numerical constant in the denominator is

    1.28 x 10-

    3

    In the special case of constant drawdown

    production, no advantage is obtained by using P

    FCN

    vs. time as opposed to 11q vs. time with the constant

    wellbore-pressure type curve. The use

    of

    P

    F N

    vs.

    time will be preferred when 1) a well is produced

    with varying flowing bottomhole pressures, 2)

    1714

    available buildup data are combined with the

    production data, and

    lor

    3) there are shut-in periods

    in the production data.

    Combination o Production

    and uildup Data

    The superposition transformation which results in

    the calculation

    of

    P

    F N

    t) allows constant wellbore

    pressure production data to be used with constant

    well-rate type curves. This is valuable because

    pressure buildup data also are analyzed with con

    stant-well-rate type curves; thus, pressure buildup

    data and production data can be evaluated on the

    same type curve.

    Since buildup tests in the absence of wellbore

    effects) provide early-time data but may be too costly

    in terms of lost production to run for long periods,

    and production

    data is

    generally available for long

    times but is unanalyzable at early times due to

    fracture fluid cleanup

    and/or

    other production

    problems, these two types of data tend to com

    plement each other. Together, they can provide a

    much longer data band for type-curve matching.

    Another way to use the superposition principle is

    to convert pressure buildup

    data

    to the equivalent

    rate-time data. Equivalent rate data can be calculated

    from Eq. 8 for any desired drawdown Am

    [p(tn)]

    if

    P

    F N

    t) is known. I f the preceding producing

    period is much longer than the buildup period,13

    P

    FCN

    t) can be estimated from the buildup data

    using

    PFCN(At) =[m[p(At) ] -m[p(At=O)lJlqj . . 11)

    where At is the buildup time and q j is the final

    stabilized flow rate before buildup. Once obtained,

    these calculated early-time rates are treated like any

    other rate data and, thus, can be used in the

    P

    F N

    calculations.

    When this early-time data is combined with

    production data for about a year, the resulting data

    band will be significantly longer than the data bands

    available from either buildup or production data

    alone. This may help to minimize the problem of

    non uniqueness of results which may exist with type

    curve analysis techniques.

    Use o the Real Gas Pseudotime

    It is important to note that to use the buildup data

    accurately with draw down-generated type curves, it

    was necessary to apply the real gas pseudotime

    ta

    (p) concept

    of

    Agarwal

    16

    to the buildup data.

    This

    is

    similar to the

    real

    gas pseudo pressure

    m(p) of

    AI-Hussainy

    et

    al

    4

    ta(p) compensates

    for the variations

    of

    the effective Jl t product with

    pressure, which appears in the dimensionless time

    term. This quantity is given by

    t , ( p )= r

    p

    ~ )

    Jpo Jl(p)c

    t

    (p)

    dp ............ 12)

    and replaces

    Atl Jlc )

    i in the dimensionless time

    term.

    JOURNAL

    OF

    PETROLEUM TECHNOLOGY

  • 8/9/2019 SPE-8280-PA

    5/9

    40 ------------------------------

    38

    0..

    34

    z

    U

    :

    c.. g32

    30

    10

    20 30 40 50 70

    100

    200 300 400

    t DAYS)

    Fig 2 - Analysis

    of

    simulated radial

    flow

    data

    using P

    FCN

    technique.

    Extension Over Shut In Periods

    With Unavailable Bottomhole

    Pressure Measurements

    Eqs. 7 and 8 can be used to obtain the

    PpCN

    function

    for wells that have shut-in periods extending over

    several days or weeks. When the bottonhole pressure

    is available during these periods, the procedure

    described previously can be used in a straightforward

    way for this purpose, treating the data as in a

    standard variable-rate situation. However, when

    bottomhole pressure information

    is not available

    during such shut-in periods, an extrapolation

    technique is employed to synthesize the missing data .

    An inspection of

    any

    of

    the dimensionless pressure

    vs

    dimensionless time solutions for fractured wells

    shows that beyond very early time the 10gPwD vs log

    tDx curve is approximately linear over a short in

    terval. This interval

    is

    usually

    at

    least twice the

    t

    Dx

    value

    of

    the last available pressure point. This ob

    servation, therefore, allows estimation

    of PpCN

    as a

    function of time during shut-in periods, provided

    that the shut-in time does not exceed the previous

    producing time. Unacceptable extrapolations are

    manifested in a computed P PCN curve that is

    oscillatory.

    This extension

    of

    the

    PpCN

    technique over long

    shut-in periods is very significant. It serves to extend

    the technique to a much larger group of candidate

    wells and enlarges the amount of usable data on

    many more wells.

    Some Practical Considerations

    Although the equations used to calculate P PCN t )

    appear simple in concept, they can become extremely

    tedious in application.

    In

    practice, a computer is

    needed to perform the calculations. Preparation of a

    program to perform these calculations is relatively

    simple, and for convenience the program can include

    routines to calculate m(p)14 and t

    a

    (p).16 f these

    routines are included, the required data need only

    include times, pressures, rates, and gas property

    data

    to generate PpCN t ) . As with any computer program

    or analysis technique, the final results depend on the

    OCTOBER 1980

    TABLE NON·MH APPLICATION

    - SIMULATED EXAMPLE

    Input

    Parameters

    Formation permeability k md

    Formation

    thickness

    h

    ft

    (m)

    Reservoir temperature T 0 R

    (K)

    Analysis Results

    Semilog s lope m, [ psi

    2

    /cp) McflD)]/cycle

    [ MPa

    2

    /Pa·

    s)/(m

    3

    /d)]/cycle

    I

    Formation

    flow

    capacity

    kh,

    md·ft

    md·m)

    0.004

    50 15)

    740 411)

    6.05

    X

    10

    6

    10.1)

    0.2 0.06)

    quality

    of

    the input data. Inaccurately measured

    rates or pressures will yield inaccurate values of

    PpCN(t ) .

    The calculation>

    of

    P PCN

    t )

    using Eq. 8 requires

    repeated interpolation

    of

    some type to obtain the

    terms

    P p c N ~ t n

    j

    _

    l

    ) for j= to n We agree with

    Jargon et

    al that

    a logarithmic interpolation is pref

    erable here, due to the nearly linear nature

    of

    the log

    PwD

    vs. log t

    Dx

      relationship over the portion

    of

    the

    curve normally subject to interpolation. Although no

    support

    is

    presented in this paper, we also agree with

    the observation

    of

    Jargon et al

    7

    that it is possible to

    obtain erratic and oscillatory results

    n

    the

    calculation of

    PpCN

    when drastic rate and/or

    pressure chages are encountered.

    N on MHF Applications

    This technique should be applicable to any situation

    in which the superposition principle can be used. In

    particular, it should be possible to use this approach

    to analyze gas and liquid flow data from wells that

    have been given at most a small stimulation treat

    ment.

    Since PpCN corresponds to a constant rate of 1

    MscflD or

    1 std m /d this function plotted for

    radial flow analysis on a semilog graph yields a

    straight line

    of

    slope m from which formation flow

    capacity can be calculated using

    kh=

    1 637T/m

    .......................

    13)

    In SI metric units, the numerical constant in the

    numerator is 1.47 x

    10

    - 3 .

    Fig. 2 is a plot

    of

    P PCN vs. t on semilog paper

    which illustrates this analysis method. The rate vs.

    time data used in the calculation

    of PpCN t)

    for this

    figure was obtained from a one-dimensional radial

    flow reservoir simulator. 17 The important input

    parameters and results of

    the analysis are given in

    Table 1.) This method can be a useful way to analyze

    wells which, for various reasons, could not be shut in

    for buildup tests or held at constant flow rates for

    drawdown tests.

    1715

  • 8/9/2019 SPE-8280-PA

    6/9

    T BLE

    2 -

    RESERVOIR PROPERTIES

    -

    SIMUL TED EX MPLE

    Initial

    reservoir pressure

    Pi

    psi a MPa)

    5,500(38)

    0.004

    50

    (15)

    0.035

    740(411)

    Formation permeability k md

    Formation

    thickness

    h ft m)

    Hydrocarbon

    porosity

    ¢HC

    Reservoir temperature T oR K)

    Initial

    gas

    viscosity

    Ui,

    cp

    Pa·s)

    Initial

    gas

    compressibility cgi

    0.0284 (2.84 x 10-

    5

    )

    psi-

    1

    Pa-

    1

    )

    Half

    fracture length

    x

    ft

    m)

    Fracture flow

    capacity

    k w, md-ft

    md·m)

    0.0001286 (1.86 x 10-

    8

    )

    1,000 (305)

    40 (12)

    T BLE

    3 -

    TYPE·CURVE M TCHES

    -

    SIMUL TED EX MPLE

    Match

    k

    X, k,w

    No. • md)

    ft)

    m)

    md-ft) md· m)

    l

    2

    3

    - -

    0.004 995 303 39.8

    0.0032 1325 403 42.4

    0.0072 740 225 26.6

    • Preferred match .

    12.1

    12.9

    8.1

    5 0 c 0 ~ - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

    q

    (MCFD)

    2000

    0

    500

    >-

    =>

    :r

    Vl

    200

    0

    40

    240

    t

    DAYS)

    Fig. 3 - Simulated example - production history.

    10

    ~ l

    w

    oo

    =>

    V>

    V>

    1

    V>

    V>

    Z

    V>

    Z

    w-

    e

    P

    FCN

    ( J l S i ~ C P

    \

    I MCFD 7

    oMENS IONl£SS

    FRACTURE

    CA PAC

    ITY FC D

    I O ~

    : I I

    I

    I

    I

    -----,

    MATCH

    POINT

    :

    I I

    I I I

    I : I

    _ _____ _____ L ____

    I. 0 10 100 1000

    , (DAYS) .k w

    FCD kif

    DIM

    ENS IONl£SS TIME, DX,

    Fig. 4 - Simulated example - type·curve match.

    1716

    320

    Note

    that

    due

    to

    the length

    of

    the time interval

    covered by the production

    data

    to be analyzed, in

    terference or boundary effects frequently will be

    observed in the analysis

    of

    normal-permeability

    wells.

    Type Curve Matching

    Although type-curve matching

    is

    now an accepted

    pressure transient analysis technique, the problem

    of

    sometimes locating more than one possible match

    remains. Agarwal et

    l

    2

    suggested that if

    kh

    is

    known from a pre fracturing buildup test, then a

    y-

    axis value

    of

    the type curve can be calculated for a

    given y-axis value

    of

    the tracing paper overlay (field

    data). With this relationship fixed, the tracing paper

    overlay is moved in the x direction only until the best

    match is obtained. This simplifies the type-curve

    matching

    and

    usually results in only one acceptable

    match.

    When no acceptable match is obtained, it may be

    because

    of

    inaccuracies in the prefracturing testing

    or

    because the prefracturing test was representative

    of

    a

    much smaller portion of the reservoir than that

    examined by the post fracturing buildup and

    production data. When this occurs, a

    two

    dimensional type-curve match must be attempted.

    One method

    of

    limiting the range

    of

    possible type

    curve matches is to calculate a value of kh from a

    plot of P

    F N

    vs. log t as discussed in the section on

    non-MHF

    applications. For most MHF-stimulated

    wells, the calculated kh from this type

    of

    analysis will

    be somewhat optimistic since the well probably will

    have not reached pseudoradial flow yet. This

    calculated value, therefore, should be the maximum

    possible value

    of kh for the well, and a PwD value

    calculated from it could then serve as a maximum

    PwD value. This should minimize at least partially the

    freedom of movement during the matching process

    and perhaps provide better answers.

    Once one or more potential solutions have been

    obtained from type-curve analysis, the acceptablity

    of

    the solutions can be checked using a two

    dimensional reservoir simulator such as the one

    described by Agarwal et

    al.

    2

    This is done by making

    a comparison

    of

    actual vs. simulated buildup and

    production

    data

    to determine which

    of

    several

    possible solutions is preferable. Lee et

    al.

    have

    suggested

    that

    a reservoir simulator could be used in

    place

    of

    other techniques to determine

    k xf

    and

    kfw, but experience indicates that such trial-and

    error history matching techniques can be more time

    consuming and expensive than those suggested here.

    Of

    course, systems with multiple layers, areal

    heterogeneities, variable fracture flow capacities

    or

    other nonuniform conditions may be analyzable,

    at

    present, only through trial-and-error simulation.

    Examples

    Simulated Example

    To

    test the

    P

    F N

    analysis technique, the two

    dimensional reservoir simulator mentioned pre

    viously was run with the reservoir properties given in

    Table 2. A well was produced with a constant flowing

    JOURNAL OF PETROLEUM

    TECHNOLOGY

  • 8/9/2019 SPE-8280-PA

    7/9

    bottomhole pressure of 1,238 psia (8.538 MPa) for

    150 days, shut in for 30 days, and then produced at

    1,238 psi a (8.538 MPa) for another 130 days. The

    rate vs. time results

    of

    this simulation are shown in

    Fig. 3.

    Simulated rate-time

    data

    and buildup pressure

    time

    data

    obtained at the end of the final production

    period along with the appropriate gas properties then

    were processed through a computer program

    to

    compute PpCN

    I).

    Next, a data plot of PpCN vs.

    t

    was made on the same scale as the log-log type curve.

    The resulting type-curve match is shown in Fig. 4.

    Match point data are

    PpCN) M = 10

    6

    (psi2/cp)/(McflD)

    [1.66

    x 10

    3

    (MPa

    2

    /Pa·s)/(m

    3

    /d)],

    I)

    M =

    100 days

    =

    2,400 hours,

    tDXf) M = 2

    x

    10 -

    2

    ,

    PwD) M

    = 0.19, and

    FCD = 10.

    Fracture flow capacity, permeability, and fracture

    length then were calculated from Eqs. 1, 10, and 14,

    where Eq.

    14 is

    where

    t

    is in hours. In SI metric units, the numerical

    constant in the numerator is 3.6 x

    10 9 .

    The results

    are given in Table 3 as Match 1.

    Although these results were in excellent agreement

    with the simulator input data, it should be noted that

    two other type-curve matches with the FCD =

    10 and

    F

    CD

    =

    5 curves were possible. This is the

    non-

    uniqueness

    of

    match problem

    that

    frequently has

    hampered type-curve matching methods. Analysis

    of

    these other matches would have yielded the results

    indicated as Matches 2 and 3 in Table 3.

    These three sets of results illustrate the typical

    range

    of

    answers that can be obtained from type

    curve matching when estimates

    of

    kh from

    prefracturing buildups are not available. f a

    pre fracturing buildup estimate

    of

    kh had been

    available, only the first (correct) type-curve match

    would have been possible.

    Field Example

    An actual field example

    of

    the use

    of

    the

    PpCN

    analysis technique also is given. Fig. 5 shows the

    reported 3-year producing history

    of

    a low

    permeability gas well. Both prefracturing

    and

    post fracturing buildup tests were available on this

    well. The reservoir properties estimated for this well

    are given in Table 4, and the post fracturing buildup

    data are contained in Table 5.

    The available rate-time and pressure-time data

    were used

    to

    calculate

    PPCN t).

    A type-curve match,

    which honored the

    kh

    calculated from the

    prefracturing buildup test, then was made,

    and

    the

    results are shown in Fig. 6. (Note that the buildup

    portion of the data is indicated

    on

    Fig. 6.) The match

    point obtained was

    OCTOBER 1980

    ~ r ~

    \.

    q

    (MCFDI

    \ ----

    000

    500f-

    1975 1976

    YEARS

    1977

    Fig. 5 - Field example -

    production

    history.

    TABLE

    4 - RESERVOIR PROPERTIES - FIELD EXAMPLE

    I

    nitial

    reservoir pressure Pi

    psia MPa)

    Flowing

    bottom

    hole pressure Pwf,

    psia MPa) -

    5,100(35)

    1,273 (8.777)

    0.0027

    56 (17)

    0.035

    725 (403)

    Formation permeability

    k

    md

    Formation

    thickness h, ft m)

    Hydrocarbon

    porosity

    cPH

    Reservoir temperature T, 0 R K)

    Initial

    gas

    viscosity

    Ui,

    cp Pa

    ·s)

    Initial gas

    compressibility

    e

    g

     

    0.0264 (2.64 x 10 -

    5)

    psi - 1 Pa - 1 )

    I

    0.0001465 (2.12 x 10 - 8)

    TABLE 5 - POSTFRACTURING BUILDUP DATA

    - FIELD EXAMPLE

    Time Pressure

    hours) psia) MPa)

    0.0 1,273 8.78

    0.1 1,415 9.76

    0.4 1,567

    10.80

    1.0 1,692 11.67

    2.0 1,803 12.43

    4.0 1,947 13.42

    8.0 2,040 14.07

    12.0 2,193 15.12

    16.0 2,292 15.80

    20.0 2,358 16.26

    24.0 2,415 16.65

    32.0 2,507 17.29

    40.0 2,583 17.81

    48.0 2,647 18.25

    56.0 2,703 18.64

    63.0 2,746 18.93

    100.0 2,929 20.19

    118.0 2,996 20.66

    130.0 3,037 20.94

    134.3 3,051 21.04

    Final rate before shut·in: 1,600 McflD 45 760 m

    3

    /d .

    1717

  • 8/9/2019 SPE-8280-PA

    8/9

    10

    DIMENSIONLESS

    FRACTURE CAPAC ITY FCD

    7

    1

    T

    -- ----

    I

    1

    I

    l[ t

    N ~ ~ I VI

    I

    6

    VI

    l O I - . ~ -  

    - . J :

    Q

    10-

    1

    I

    I 3

    ,

    ,

    I

    I

    ...

    I

    l h 0 _

    10-

    I I I

    _.1 . ______

    1. ______

    J

    10 100 1000

    t

    IDAYS)

    F •

    kfW

    CD

    kx

    f

    10 -3 ------::_--- --;-_--- ---::-_---- ---::-_--- --::-_--- ------

    10 '

    10-4

    10-

    3

    10-

    2

    10-

    1

    DIMENSIONLESS

    TIME, to

    x

    f

    Fig. 6 - Field example - tYPEl·curve match.

    P

    FCN

      M =

    10

    6

    (psi

    2

    /cp)/(McflD)

    [1.66x 10

    3

    (MPa

    2

    /Pa·s)/(m

    3

    /d)],

    t)

    M

    =

    10 days

    =

    240 hours,

    tDX/)M

    = 2.2x

    10-

    4

    ,

    PwD)

    M

    =

    0.146,

    and

    FCD = 50.

    The results, calculated as before using Eqs. 1, 10,

    and

    14, were

    k

    =

    0.0027 md,

    x f = 2,400 ft (730 m), and

    kfw = 324 md-ft (99 md ·m).

    The two-dimensional reservoir simulator then was

    run using these parameters and the production

    schedule of this well. The results of this simulation

    are shown in Fig. 7. To predict reserves

    and

    future

    rates for this well, the reservoir simulator

    then

    was

    run

    in a predictive

    mode

    for

    the

    expected life

    of

    the

    well with an appropriate drainage area assigned.

    onclusions

    Based

    on

    the

    work

    presented in this

    paper, the

    following statements appear valid.

    1.

    Superposition can be used to generate

    P

    FCN

    t ) , which

    is

    the transformation to the

    equivalent variable pressure history with a constant

    rate

    of

    a well/reservoir system

    that

    has

    produced

    a

    variable rate with a known pressure history.

    2. Production

    and buildup

    data can be combined

    into a single P

    FCN

    vs. time curve which can be used

    for type-curve matching.

    3.

    The combination

    of

    production and buildup

    data offers a significant advantage because it

    produces a much longer field data curve, which

    greatly enhances type-curve matching by curve

    shape.

    4. The P

    FCN

    technique can be used to extend type

    curve matching

    to

    wells with intermittent

    production

    periods.

    5. The PFCN technique also should be applicable

    to

    data

    from non-MHF

    wells.

    6. The r e ~ u l t s of a type-curve match can be

    checked and long-term rates

    and

    reserves can be

    1718

    2000

    q

    MCFD)

    1000

    500

    CASE CUM. PROD.

    \

    SIMULATED

    -

    8 8

    MMCF

    'lr ACTUAL

    - 809

    MMCF

    100 L..L..J....L..L-:-:19=75

    .........

    ..L..L.L...L..

    .........

    ':-'::'--'-J....L..I...L.J....L...L..'-:1L, 97='7 ...L...L.1...LLJ....l...J...J

    Fig. 7 - Field example - simulated/actual production

    comparison.

    predicted with a two-dimensional reservoir

    simulator.

    Nomenclature

    =

    formation

    volume factor, RB/STB

    (m

    3

    /m

    3

    )

    gas compressibility at initial reservoir

    pressure, psi - 1 (Pa - 1)

    total compressibility at initial reservoir

    conditions, psi -1 (Pa -

    1 )

    F CD

    dimensionless fracture flow capacity

    h formation thickness, ft (m)

    k = formation permeability, md

    k

    f

    =

    fracture permeability, md

    m = slope on a semi log plot, [(psi2/cp)/

    (McflD)]/cycle {[(MPa

    2

    /Pa·s)/

    m p)

    t::.m p)

    (m

    3

    /d»)/cycle

    J

    real gas

    ~ s e u d o p r e s s u r e

    psi2/cp

    (MPa /Pa·s)

    difference in real gas pseudopressure,

    psi2/cp

    (MPa

    2

    /Pa

    ·s)

    P =

    pressure, psia

    (MPa)

    Pi

    = initial reservoir pressure, psia

    (MPa)

    PwD =

    dimensionless wellbore pressure

    Pwf

    =

    flowing bottomhole pressure, psia (MPa)

    P

    FCN

    = pressure function, (psi2/cp)/(McflD)

    [(MPa

    2

    /Pa·s)/(m

    3

    /d»)

    flow rate, MscflD (std m

    3

    /d)

    producing time, days

    real gas pseudotime, hr cp . psi - 1

    (hr/Pa.s. Pa -1

    dimensionless time

    dimensionless time based

    on

    half

    fracture length of a vertical fracture

    t:: t = buildup time, hours

    T = reservoir temperature , 0 R (K)

    w = fracture, width, ft (m)

    xf

    = half fracture length, ft (m)

    p- = viscosity, cp (Pa .s

    P-i =

    viscosity

    at

    initial reservoir pressure, cp

    (Pa·s)

    total porosity, fraction

    hydrocarbon porosity, fraction

    JOURNAL OF

    PETROLEUM TECHNOLOGY

  • 8/9/2019 SPE-8280-PA

    9/9

    References

    1. Cinco L., H., Samaniego V., F and Dominguez A., N.:

    Transient Pressure Behavior for a Well With a Finite

    Conductivity Vertical

    Fracture,

    Soc. Pet. Eng. J (Aug.

    1978) 253-264.

    2. Agarwal, R.G., Carter, R.D., and Pollock, C.B.:

    Evaluation and Performance Prediction

    of

    Low

    Permeability Gas Wells Stimulated by Massive Hydraulic

    Fracturing, J

    Pet. Tech

    (March 1979) 362-372; Type

    Curves for Evaluation and Performance Prediction of Low

    Permeapility Gas Wells Stimulated by Massive Hydraulic

    Fracturing,

    J

    Pet. Tech.

    (May 1979) 651-654;

    Trans.

    AIME,267.

    3. van Everdingen, A.F. and Hurst, W.:

    The

    Application of the

    Laplace Transformation to Flow Problems in Reservoirs,

    Trans. AIME (1949) 186, 305-324.

    4. Hutchinson, T.S. and Sikora. V.J.: A Generalized Water

    Drive Analysis, Trans. AIME (1959) 216,169-177.

    5.

    Mueller, T.D. : Transie nt Response of Nonhomogeneous

    Aquifers, Soc Pet. Eng. J (March 1962) 33-43; Trans.

    AIME (1962) 225,33-43.

    6. Coats, K.H., Rapoport, L.A., McCord,

    J.R.,

    and Drews,

    W.P.:

    Determination of Aquifer Influence Functions From

    Field Data, J Pet. Tech. (Dec. 1964) 1417-1424; Trans.

    AIME,231.

    7.

    Jargon, 1.R. and van Poollen, H.K.:

    Unit

    Response Func

    tion From Varying Rate Data, J Pet.

    Tech

    (Aug. 1965 965-

    969; Trans. AIME, 234.

    8. Ridley,

    T.P.: The

    Unified Analysis

    of

    Well Tests, paper

    SPE 5587 presented at SPE 50th Annual Technical Con

    ferenceandExhibition, Sept. 28-0ct.l, 1975.

    9.

    Raghavan, R.: Pressure Behavior

    of

    Wells Intercepting

    Fractures,

    Proc.

    Invitational Well Testing Symposium,

    Berkeley, CA (1977).

    10. Cinco L., H. and Samaniego V., F.: Transien t Pressure

    Analysis for Fractured Wells, paper SPE 7490 presented

    at

    SPE 53rd Annual Technical Conference and Exhibition,

    Houst on, Oct. 1-3, 1978.

    OCTOBER 1980

    11. Lee,

    W.J.

    and Holditch, S.A.:

    Fracture

    Evaluation with

    Pressure Transient Testing in Low Permeability Gas Reser

    voirs. Part I: Theoretical Backgro und, paper SPE 7929

    presented at SPE Symposium on Low Permeability Gas

    Reservoirs, Denver, May 20-22,1979.

    12. Agarwal, R.G. , AI-Hussainy, R., and Ramey, H.1. Jr. : An

    Investigation of Well bore Storage and Skin Effect in Unsteady

    Liquid Flow:

    1.

    Analytical

    Treatment,

    Soc Pet. Eng. J

    (Sept. 1970) 279-290; Trans. AIME, 249.

    13. Raghavan, R.:

    The

    Effect of Producing Time on Type Curve

    Analysis, J Pet. Tech. (June 1980) 1053-1064.

    14.

    Al-Hussainy, R., Ramey, H.J. Jr., and Crawford, P.B.: The

    Flow

    of

    Real Gas Through Porous

    Media,

    J

    Pet. Tech

    (May 1966) 624-626; Trans. AIME, 237.

    15. Earlougher, Robert C. Jr.: Advances in Well Test Analysis

    Monograph Series Society

    of

    Petroleum Engineers, Dallas

    (1977) 5,24-27.

    16. Agarwal, R.G.: 'Real Gas Pseudo-Time' - A New Function

    for Pressure Buildup Analysis of

    MHF

    Wells, paper SPE

    8279 presented

    at

    SPE 54th Annual Technical Conference and

    Exhibition, Las Vegas, Sept. 23-26,1979.

    17.

    Carter, R.D.: Solutions

    of

    Unsteady-State Radial Gas

    Flow,

    J

    Pet.

    Tech

    (May 1962) 549-554; Trans. AIME, 225.

    SI Metric onversion Factors

    ep

    X 1.0*

    ell

    ft

    x

    2.831 685

    psi x

    6.894 757

    ·Conversion factor

    is

    exact.

    E-03

    E-02

    E+OO

    Pa·s

    m

    3

    kPa

    JPT

    Original

    manuscript

    received in Society

    of

    Petroleum Engineers

    office

    July

    20

    1979. Paper accepted for publication April

    18

    1980. Revised manuscript

    received Aug. 11, 1980. Paper SPE 8280) first presented at the SPE 54th Annual

    Technical

    Conference and

    Exhibition

    held in Las Vegas. Sept. 23-26. 1979.

    1719


Recommended