files-2-Lectures_Lec18.ppt

Embed Size (px)

Citation preview

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    1/19

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    2/19

     

       C   h  a  p   t  e  r   6

     Example: Turbulent flow in a pipe

    +et, fluid property (e.g., temperature or composition) at

     point 1

      fluid property at point 2

    u

     y

    %ssume t!at t!e velocity profile is flat-, t!at is, t!e velocity

    is uniform over t!e cross"sectional area. T!is situation is

    analyed in /&ample .0 and Fig. ..

    Fluid n Fluid ut

    Figure .0

    oint 1 oint 2

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    3/19

     

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    4/19

     

       C   h  a  p   t  e  r   6

     Example 6.5

    (a) relating t!e mass flow rate of li$uid at 2, w2, to t!e mass flow

    rate of li$uid at 1, wt,

    (b) relating t!e concentration of a c!emical species at 2 to t!e

    concentration at 1. %ssume t!at t!e li$uid is incompressible.

    Solution

    (a) First we ma5e an overall material balance on t!e pipe

    segment in $uestion. #ince t!ere can be no accumulation(incompressible fluid),

    material in 6 material out   ( ) ( )1 2w t w t  ⇒ =

    For t!e pipe section illustrated in Fig. .0, find t!e transfer

    functions:

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    5/19

     

       C   h  a  p   t  e  r   6 (b) bserving a very small cell of material passing point 1 at time

    t , we note t!at in contains Vc1(t ) units of t!e c!emical species of

    interest w!ere V  is t!e total volume of material in t!e cell. f, attime t 7 , t!e cell passes point 2, it contains units of

    t!e species. f t!e material moves in plug flow, not mi&ing at all

    wit! ad8acent material, t!en t!e amount of species in t!e cell is

    constant:

    utting ("3) in deviation form and ta5ing +aplace transforms

    yields t!e transfer function,

    ( )

    ( )2

    1 1

    W s

    W s

    =′

    '   ( )2 'Vc t  +

    ( ) ( )2 1' ("3)Vc t Vc t  + =

    or 

    ( ) ( )2 1' ("31)c t c t  + =

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    6/19

     

       C   h  a  p   t  e  r   6

    %n e$uivalent way of writing ("31) is

    ( ) ( )2 1 ' ("32)c t c t  = −

    if t!e flow rate is constant. utting ("32) in deviation form and

    ta5ing +aplace transforms yields

    ( )

    ( )2 '

    1("33)

     sC s

    eC s

    −′

    =′

    Time Delays (continued)

    Transfer Function 9epresentation:

    ( )

    ( )' ("2) s

    Y se

    U s

    −=

     ;ote t!at !as units of time (e.g., minutes, !ours)'

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    7/19

     

       C   h  a  p   t  e  r   6

    Polynomial Approximations to

    For purposes of analysis using analytical solutions to transfer

    functions, polynomial appro&imations for are commonly

    used. /&ample: simulation software suc! as

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    8/19

     

       C   h  a  p   t  e  r   6

    2. Padé pproxi!ations:

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    9/19

     

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    10/19

     

    Approximation of Higher-rder Transfer

    !unctions

    '

    1' ("0*) s

    e s− ≈ −

    n t!is section, we present a general approac! forappro&imating !ig!"order transfer function models wit!

    lower"order models t!at !ave similar dynamic and steady"state

    c!aracteristics.

    n /$. "4 we s!owed t!at t!e transfer function for a time

    delay can be e&pressed as a Taylor series e&pansion. For small

    values of s,

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    11/19

     

    B %n alternative first"order appro&imation consists of t!e transfer

    function,

    '

    '

    1 1("0)

    1'

     s

     se

     se

    − = ≈+

    w!ere t!e time constant !as a value of 

    B /$uations "0* and "0 were derived to appro&imate time"

    delay terms.

    B Cowever, t!ese e&pressions can also be used to appro&imate

    t!e pole or ero term on t!e rig!t"!and side of t!e e$uation by

    t!e time"delay term on t!e left side.

    ' .

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    12/19

     

    S"ogestad#s $half rule%

    B #5ogestad (22) !as proposed a related appro&imation met!od

    for !ig!er"order models t!at contain multiple time constants.

    B Ce appro&imates t!e largest neglected time constant in t!e

    following manner.

    B ne !alf of its value is added to t!e e&isting time delay (if any)and t!e ot!er !alf is added to t!e smallest retained time

    constant.

    B Time constants t!at are smaller t!an t!e largest neglected time

    constant- are appro&imated as time delays using ("0).

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    13/19

     

     Example 6.4

    onsider a transfer function:

    ( )   ( )( ) ( ) ( )

    .1 1("0A)

    0 1 3 1 .0 1

     $ s% s

     s s s− +=

    + + +

    Derive an appro&imate first"order"plus"time"delay model,

    ( )'

    (")E 1

     s $e% s

     s

    =+

    %

    using two met!ods:

    (a) T!e Taylor series e&pansions of /$s. "0* and "0.

    (b) #5ogestads !alf rule

       C   h  a  p   t  e  r   6

    ompare t!e normalied responses of %( s) and t!e appro&imate

    models for a unit step input.

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    14/19

     

    Solution

    (a) T!e dominant time constant (0) is retained. %pplying

      t!e appro&imations in ("0*) and ("0) gives:

    .1.1 1 ("1) s s e−− + ≈

    and

    3 .01 1 ("2)3 1 .0 1

     s se e s s

    − −≈ ≈+ +

    #ubstitution into ("0A) gives t!e Taylor series

    appro&imation, ( ) :TS 

    % s%

    ( ).1 3 .0 3.

    ("3)0 1 0 1

     s s s s

    TS 

     $e e e $e% s

     s s

    − − − −

    = =+ +

    %

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    15/19

     

    (b) To use #5ogestads met!od, we note t!at t!e largest neglected

    time constant in ("0A) !as a value of t!ree.

    ' 1.0 .1 .0 2.1= + + =

       C   h  a  p   t  e  r   6

    B %ccording to !is !alf rule-, !alf of t!is value is added to t!e

    ne&t largest time constant to generate a new time constant

     B T!e ot!er !alf provides a new time delay of .0(3) 6 1.0.B T!e appro&imation of t!e 9C ero in ("1) provides an

    additional time delay of .1.B %ppro&imating t!e smallest time constant of .0 in ("0A) by

    ("0) produces an additional time delay of .0.B T!us t!e total time delay in (") is,

    E 0 .0(3) .0.= + =

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    16/19

     

    and %&s' can be appro&imated as:

    ( )2.1

    ("4).0 1

     s

    S( 

     $e% s

     s

    =+

    %

    T!e normalied step responses for %&s' and t!e two appro&imate

    models are s!own in Fig. .1. #5ogestads met!od provides

     better agreement wit! t!e actual response.

       C   h  a  p   t  e  r   6

    Figure .1

    omparison of t!e

    actual and

    appro&imate modelsfor /&ample .4.

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    17/19

     

     Example 6.5

    onsider t!e following transfer function:

    ( )   ( )( ) ( ) ( ) ( )

    1 ("0)12 1 3 1 .2 1 .0 1

     s

     $ s e% s s s s s

    −−= + + + +

    Gse #5ogestads met!od to derive two appro&imate models:

    (a) % first"order"plus"time"delay model in t!e form of (")

    (b) % second"order"plus"time"delay model in t!e form:

    ( ) ( ) ( )

    '

    1 2 (")E 1 E 1

     s $e

    % s  s s

    =+ +

    %

    ompare t!e normalied output responses for %&s' and t!e

    appro&imate models to a unit step input.

       C   h  a  p   t  e  r   6

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    18/19

     

    Solution

    (a) For t!e first"order"plus"time"delay model, t!e dominant time

    constant (12) is retained.

    3.' 1 .2 .0 1 3.*02

    3.E 12 13.0

    2

    = + + + + =

    = + =

       C   h  a  p   t  e  r   6

    B ne"!alf of t!e largest neglected time constant (3) is allocated

    to t!e retained time constant and one"!alf to t!e appro&imate

    time delay.

    B %lso, t!e small time constants (.2 and .0) and t!e ero (1)are added to t!e original time delay.

    B T!us t!e model parameters in (") are:

  • 8/17/2019 files-2-Lectures_Lec18.ppt

    19/19

     

    (b) %n analogous derivation for t!e second"order"plus"time"delay

    model gives:

    1 2

    .2' 1 .0 1 2.10

    2

    E 12, E 3 .1 3.1

    = + + + =

    = = + =

    n t!is case, t!e !alf rule is applied to t!e t!ird largest time

    constant (.2). T!e normalied step responses of t!e original and

    appro&imate transfer functions are s!own in Fig. .11.   C   h  a  p   t  e  r   6