computing water-hammer flows with a two-fluid model · the two-fluid model & water hammer main...

34
LaMSID - EDF R&D, AMA LATP, University of Provence Computing water-hammer flows with a two-fluid model AMIS 2012 Yujie LIU 21 Juin 2012 PhD supervisors : Fabien Crouzet, EDF R&D, AMA Fr´ ed´ eric Daude, EDF R&D, AMA Pascal Galon, CEA Saclay Philippe Helluy , IRMA Strasbourg Jean-Marc H´ erard, EDF R&D, MFEE Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 1 / 32

Upload: others

Post on 22-Oct-2020

6 views

Category:

Documents


0 download

TRANSCRIPT

  • LaMSID - EDF R&D, AMA LATP, University of Provence

    Computing water-hammer flows with a two-fluidmodel

    AMIS 2012

    Yujie LIU

    21 Juin 2012

    PhD supervisors : Fabien Crouzet, EDF R&D, AMAFrédéric Daude, EDF R&D, AMAPascal Galon, CEA SaclayPhilippe Helluy , IRMA StrasbourgJean-Marc Hérard, EDF R&D, MFEE

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 1 / 32

  • Outline

    1 Introduction

    2 The two-fluid model & water hammer

    3 Computation of the two-fluid modelEvolution stepRelaxation step

    4 Verification of the two-fluid model

    5 Conclusions and perspectives

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 2 / 32

  • Introduction

    Water-hammer transient flow in piping systems of a nuclear power plant

    Water-hammer (hydraulic shock) :sudden change of direction or velocity ! large changes in pressureRisk : damage to pipes, supports, and valve ; power plant availability

    Figure: Water-hammer ‘classical’ Figure: Are line Civaux (2011)

    Objective : Water-hammer simulation(SITAR Project )

    Modeling water-hammer flows (Assessing/comparing the di↵erent models)

    Computing the model with fluid-structure interaction (FSI) (Europlexus)

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 3 / 32

  • Introduction

    Modeling water-hammer flows

    Complex phenomenon : column separation/cavitation...

    Phenomenon characteristics :

    compressible two-phase flows(water/steam)unsteady flowsshock wavesfast transient

    �!�!�!�!

    two-phase flow modelinghyperbolic systemunique jump conditionsexplicit scheme

    Di↵erent two-phase flow models :

    two-fluid approach :the two-fluid model (7 equations)the two-fluid single-pressure model (6 equations)

    homogeneous approach :the five-equation models (5 equations)homogeneous relaxation model (4 equations)homogeneous equilibrium model (3 equations)

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 4 / 32

  • The two-fluid model & water hammer

    Outline

    1 Introduction

    2 The two-fluid model & water hammer

    3 Computation of the two-fluid modelEvolution stepRelaxation step

    4 Verification of the two-fluid model

    5 Conclusions and perspectives

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 5 / 32

  • The two-fluid model & water hammer

    Governing equations of the two-fluid model

    7 PDE+ closure (P

    I

    , VI

    )+ EOS(equation of state) : "'

    (⇢'

    , p'

    ) + source terms

    8

    >

    >

    >

    >

    >

    >

    >

    <

    >

    >

    >

    >

    >

    >

    >

    :

    @t

    ↵v

    +VI

    @x

    ↵v

    = �v

    @t

    (↵'

    ⇢'

    ) +@x

    (↵'

    ⇢'

    u'

    ) = �'

    , ' = l, v

    @t

    (↵'

    ⇢'

    u'

    ) +@x

    ↵'

    ⇢'

    u2'

    + ↵'

    p'

    � PI

    @x

    ↵'

    = �'

    Uint

    +D'

    @t

    (↵'

    ⇢'

    e'

    ) +@x

    [(↵'

    ⇢'

    e'

    + ↵'

    p'

    )u'

    ] + PI

    @t

    ↵'

    = �'

    Hint

    +D'

    Uint

    +Q'

    (1)with e

    '

    = "'

    (⇢'

    , p'

    ) + u2'

    /2, Uint

    = (ul

    + uv

    )/2 and Hint

    = ul

    ⇤ uv

    /2,

    �v

    �'

    D'

    Q'

    mass transfermomentum transferenergy transfer

    pressure relaxationchemical potential relaxationvelocity relaxationtemperature relaxation

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 6 / 32

  • The two-fluid model & water hammer

    Choice of closure laws for water-hammer flows

    Interfacial pressure/velocity :

    PI

    = pl

    , VI

    = uv

    (Baer-Nunziato)

    EOS :

    Liquid : Sti↵ened Gas (SG) "l

    (⇢l

    , pl

    ) =pl

    + �l

    (pl

    )1⇢l

    (�l

    � 1)Vapor : Perfect Gas (GP) "

    v

    (⇢v

    , pv

    ) =pv

    ⇢v

    (�v

    � 1)

    Source terms (Entropy-consistent interfacial closure) :

    pressure relaxation : �v

    = (⌧p

    )�1↵l

    ↵v

    |Pl

    |+|Pv

    |(P

    v

    � Pl

    ),X

    '=l,v

    �'

    = 0

    velocity relaxation : Dv

    = (⌧u

    )�1m

    l

    mv

    ml

    +mv

    (ul

    � uv

    ),X

    '=l,v

    D'

    = 0,

    with ⌧k

    , k =p,u, time scales for pressure/velocity relaxations,m

    '

    = ↵'

    ⇢'

    , ' = l, v, partial masses.

    Remark :

    liquid/vapor : PI

    = pl

    , (pl

    )1 >> (pv)1gas/particles : P

    I

    = pg

    , (pg

    )1

  • The two-fluid model & water hammer

    Main properties of the two-fluid model

    Property 1 Hyperbolicity and structure of wavesThe system (1) is hyperbolic unless |u

    l

    � uv

    |= cl

    . It admits 7 realeigenvalues :

    �1,2 = uv ,�3 = uv � cv ,�4 = uv + cv ,�5 = u

    l

    ,�6 = ul

    � cl

    ,�7 = ul

    + cl

    Fields associated with �1,2,5 are linearly degenerate (LD).Other fields are genuinely nonlinear (GNL).

    Property 2 Jump conditionsUnique jump conditions hold within each field associated with �

    k

    .↵l

    is uniform apart from the field associated with �1,2 = uv .Jump conditions in other fields correspond to single phase jump relations.

    Property 3 Entropy inequalityDefine the entropy ⌘(W ) = m

    l

    sl

    +mv

    sv

    and the entropy fluxf⌘

    (W ) = ml

    sl

    ul

    +mv

    sv

    uv

    ; then smooth solutions W of (1) are such that :

    0 @t

    (⌘(W )) + @x

    (f⌘

    (W ))

    with the constraint c2'

    @p' (s') + @⇢' (s') = 0 for entropies s' .

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 8 / 32

  • Computation of the two-fluid model

    Outline

    1 Introduction

    2 The two-fluid model & water hammer

    3 Computation of the two-fluid modelEvolution stepRelaxation step

    4 Verification of the two-fluid model

    5 Conclusions and perspectives

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 9 / 32

  • Computation of the two-fluid model

    A fractional step method complying with the entropy inequality :

    At time interval [tn, tn +�t], given initial values Wn

    Evolution step : Wnconvective e↵ects����������������������!

    hyperbolic homogeneous systemW̃

    8

    >

    >

    <

    >

    >

    :

    @t

    ↵v

    +VI

    @x

    ↵v

    = 0@t

    (↵'

    ⇢'

    ) +@x

    (↵'

    ⇢'

    u'

    ) = 0@t

    (↵'

    ⇢'

    u'

    ) +@x

    ↵'

    ⇢'

    u2'

    + ↵'

    p'

    � PI

    @x

    ↵'

    = 0@t

    (↵'

    ⇢'

    e'

    ) +@x

    [(↵'

    ⇢'

    e'

    + ↵'

    p'

    )u'

    ] + PI

    @t

    ↵'

    = 0

    (2)

    Relaxation step : W̃source terms���������! Wn+1

    8

    >

    >

    >

    >

    >

    >

    >

    <

    >

    >

    >

    >

    >

    >

    >

    :

    @t

    ↵l

    = �l

    @t

    (↵l

    ⇢l

    ) = 0@t

    (↵l

    ⇢l

    ul

    ) = Dl

    @t

    (↵l

    ⇢l

    el

    ) + PI

    @t

    (↵l

    ) = Dl

    Uint

    @t

    (↵v

    ⇢v

    + ↵l

    ⇢l

    ) = 0@t

    (↵v

    ⇢v

    uv

    + ↵l

    ⇢l

    ul

    ) = 0@t

    (↵v

    ⇢v

    ev

    + ↵l

    ⇢l

    el

    ) = 0

    (3)

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 10 / 32

  • Computation of the two-fluid model Evolution step

    Evolution step : an extension of Rusanov scheme

    Homogeneous system :

    @t

    W + @x

    (f(W )) + h(W )@x

    ↵v

    = 0 (4)

    Extension of Rusanov scheme (SR1) :

    hi

    (Wn+1i

    �Wni

    ) +�tn(Fni+1/2 � F

    n

    i�1/2) +�tn

    (↵1)n

    i+ 12� (↵1)

    n

    i� 12

    �ni

    = 0

    (5)Flux :

    Fni+1/2 =

    12

    f(Wni

    ) + f(Wni+1)� ri+1/2(Wni+1 �Wni )

    ri+1/2 = max(ri, ri+1), r spectral radius

    (6)

    Non-conservative terms :

    (↵l

    )n

    i+ 12= 12

    (↵l

    )ni

    + (↵l

    )ni+1

    , �ni

    = h(Wni

    ) (7)

    .Property 4The scheme (5) preserves positive values of partial masses m

    '

    and void fractions↵'

    , if

    �tn (ri+1/2 + ri�1/2) 2CFLhi, CFL 2]0, 1[

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 11 / 32

  • Computation of the two-fluid model Evolution step

    The second-order Rusanov scheme (SR1-ORDER2 )

    Time scheme : second-order Runge-Kutta methodSpace discretisation : minmod reconstruction (Y = (↵

    v

    , uv

    , pv

    , sv

    , ul

    , pl

    , sl

    )t)

    If (�ni+1 � �ni )(�ni � �ni�1) > 0,

    (�ni

    )+ = �ni

    + �ni

    hi2 , (�

    n

    i

    )� = �ni

    � �ni

    hi2 ,

    �ni

    = sign(�ni+1 � �ni )min

    2

    ?

    ?

    ?

    ?

    �ni+1��ni

    hi+1+hi

    ?

    ?

    ?

    ?

    , 2

    ?

    ?

    ?

    ?

    �ni ��ni�1

    hi+hi�1

    ?

    ?

    ?

    ?

    Else, (�ni

    )+ = (�ni

    )� = �ni

    (8)

    Flux :

    Fni+1/2 =

    1

    2

    n

    f(W ((Y ni

    )+)) + f(W ((Y ni+1)

    �))� ri+1/2

    n

    W ((Y ni+1)

    �)�W ((Y ni

    )+)oo

    (9)Non-conservative terms :

    h(W )@x

    ↵v

    �!

    (↵1)n

    i+ 12� (↵1)

    n

    i� 12

    �ni

    hi

    (10)

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 12 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the schemes : test cases

    Test case 1 (TT, JCP, 2010 ; SWK, JCP, 2006) [6, 4]2 contact discontinuities, 1 rarefaction, 3 shocksEOS : GP(vapor), SG (liquid)

    Test case 2 (constructed),A moving void fraction waveEOS : GP(vapor), SG (liquid)

    VI

    Figure: Structure of fields of case 2

    Test case 3 (constructed),A moving void fraction waveEOS : GP(vapor), SG (liquid)Large variation of ↵

    v

    : 0.05 ! 0.95, equilibrium of other variables

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 13 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the first/second-order Rusanov schemes : test case 1

    −0.5 0 0.5

    −0.4

    −0.2

    0

    0.2

    0.4

    0.6

    0.8

    x

    α1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    −0.5 0 0.5

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    1.4

    1.6

    1.8

    2

    x

    ρ1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    −0.5 0 0.5

    −0.6

    −0.4

    −0.2

    0

    0.2

    0.4

    0.6

    0.8

    x

    u1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    −0.5 0 0.5

    −0.5

    0

    0.5

    1

    1.5

    2

    2.5

    3

    x

    p 1

    Exact solutionSR1−200−cellsSR1−500−cellsSR1−ORDER2−200−cellsSR1−ORDER2−500−cells

    −0.5 0 0.5

    1650

    1700

    1750

    1800

    1850

    1900

    1950

    2000

    2050

    x

    ρ2

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    −0.5 0 0.5

    −0.3

    −0.25

    −0.2

    −0.15

    −0.1

    −0.05

    0

    x

    u2

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    −0.5 0 0.5

    −600

    −400

    −200

    0

    200

    400

    600

    800

    1000

    x

    p 2

    Exact solutionSR1−200−cellsSR1−500−cellsSR1−ORDER2−200−cellsSR1−ORDER2−500−cells

    Figure: SR1, SR1-ORDER2 ; with 200 & 500 cells, CFL = 0.49, t = 0.15.

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 14 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the first/second order Rusanov scheme : test case 1

    Rate of convergence observed :

    SR1 :1

    2, SR1-ORDER2 :

    2

    3

    −5 −4.5 −4 −3.5 −3 −2.5 −2−4.5

    −4

    −3.5

    −3

    −2.5

    −2

    −1.5

    −1

    −0.5

    ln(h)

    ln(e

    rror

    )

    α

    1−SR1

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    α1−SR1−ORDER2

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    C0 h 1/2

    C h 2/3

    Figure: L-1 norm of the error for varibles ↵v

    , ⇢'

    , u'

    , p'

    ,' = l, v traced in logarithmicscale, SR1, SR1-ORDER2, CFL = 0.49, t = 0.15

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 15 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the first/second order Rusanov scheme : test case 2

    0 200 400 600 800 1000

    −0.3

    −0.2

    −0.1

    0

    0.1

    0.2

    0.3

    0.4

    0.5

    x

    α1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    0 200 400 600 800 1000

    8.5

    9

    9.5

    10

    10.5

    11

    11.5

    12

    x

    ρ1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    0 200 400 600 800 1000

    9.75

    9.8

    9.85

    9.9

    9.95

    10

    10.05

    x 105

    x

    p 1

    Exact solutionSR1−200−cellsSR1−500−cellsSR1−ORDER2−200−cellsSR1−ORDER2−500−cells

    0 200 400 600 800 1000

    12

    12.5

    13

    13.5

    14

    14.5

    15

    15.5

    16

    16.5

    17

    x

    u1

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    0 200 400 600 800 1000

    999.86

    999.88

    999.9

    999.92

    999.94

    999.96

    999.98

    1000

    1000.02

    x

    ρ2

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    0 200 400 600 800 1000

    2

    3

    4

    5

    6

    7

    8

    9

    10

    x

    u2

    Exact solution

    SR1−200−cells

    SR1−500−cells

    SR1−ORDER2−200−cells

    SR1−ORDER2−500−cells

    0 200 400 600 800 1000

    6.5

    7

    7.5

    8

    8.5

    9

    9.5

    10

    10.5

    11

    x 105

    x

    p 2

    Exact solutionSR1−200−cellsSR1−500−cellsSR1−ORDER2−200−cellsSR1−ORDER2−500−cells

    Figure: SR1, SR1-ORDER2 ; 200 & 500 cells, CFL = 0.49, t = 0.25

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 16 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the first/second order Rusanov scheme : test case 2

    Rate of convergence observed :

    SR1 :1

    2, SR1-ORDER2 :

    2

    3

    −5 −4.5 −4 −3.5 −3 −2.5 −2−8

    −7

    −6

    −5

    −4

    −3

    −2

    −1

    ln(h)

    ln(e

    rror)

    α

    1−SR1

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    α1−SR1−ORDER2

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    C0 h 1/2

    C h 2/3

    Figure: SR1, SR1-ORDER2, CFL = 0.49, t = 0.25

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 17 / 32

  • Computation of the two-fluid model Evolution step

    Evolution step : a fractional step method

    8

    >

    >

    <

    >

    >

    :

    @t

    ↵v

    + uv

    @x

    ↵v

    = 0@t

    (↵'

    ⇢'

    ) = 0, ' = l, v@t

    (↵'

    ⇢'

    u'

    ) = 0, ' = l, v@t

    (↵'

    E'

    )� pl

    uv

    @x

    ↵'

    = 0, ' = l, v

    (11)

    8

    >

    >

    <

    >

    >

    :

    @t

    ↵v

    = 0@t

    (↵'

    ⇢'

    ) + @x

    (↵'

    ⇢'

    u'

    ) = 0, ' = l, v@t

    (↵'

    ⇢'

    u'

    ) + @x

    ↵'

    ⇢'

    u2'

    + @x

    (↵'

    p'

    )� pl

    @x

    ↵'

    = 0, ' = l, v@t

    (↵'

    E'

    ) + @x

    [(↵'

    E'

    + ↵'

    p'

    )u'

    ] = 0, ' = l, v

    (12)

    PropertySystem (11) is hyperbolic. It admits seven real eigenvalues. All fields are LD.

    �1 = uv ,�2�7 = 0

    System (12) is hyperbolic unless : |ul

    | = cl

    , or |uv

    | = cv

    . It admits seven realeigenvalues. The 1, 2, 5-fields are LD and other fields are GNL.

    �1 = 0,�2 = uv , �3 = uv � cv , �4 = uv + cv ,�5 = u

    l

    , �6 = ul

    � cl

    , �7 = ul

    + cl

    .

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 18 / 32

  • Computation of the two-fluid model Evolution step

    Evolution step : a fractional step method

    First step : Wn ! W ⇤8

    >

    >

    <

    >

    >

    :

    @t

    ↵v

    + uv

    @x

    ↵v

    = 0@t

    (↵'

    ⇢'

    ) = 0, ' = l, v@t

    (↵'

    ⇢'

    u'

    ) = 0, ' = l, v@t

    (↵'

    E'

    )� pl

    uv

    @x

    ↵'

    = 0, ' = l, v

    (13)

    ↵v

    : Rusanov flux formulation(↵

    v

    Ev

    + ↵l

    El

    )⇤i

    = (↵v

    Ev

    + ↵l

    El

    )ni

    (sl

    )⇤i

    = (sl

    )ni

    (14)

    Second step :W ⇤ ! Wn+1 : generic Rusanov formulation

    PropertyThe scheme preserves positive values of partial masses m

    '

    and void fractions ↵'

    ifthe CFL like condition holds on. The scheme preserves the mean total energy.

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 19 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the fractional step method : test case 1

    Rate of convergence observed :

    PFRAC3 :1

    2,

    −5 −4.5 −4 −3.5 −3 −2.5 −2

    −3.5

    −3

    −2.5

    −2

    −1.5

    −1

    −0.5

    ln(h)

    ln(e

    rror

    )

    α

    1−SR1

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    α1−PFRAC3

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    C0 h1/2

    Figure: SR1, PFRAC3, CFL = 0.49, t = 0.15.

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 20 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the fractional step method : test case 2

    Rate of convergence observed :

    PFRAC3 :1

    2,

    −5 −4.5 −4 −3.5 −3 −2.5−8

    −7

    −6

    −5

    −4

    −3

    −2

    −1

    0

    ln(h)

    ln(e

    rror)

    α

    1−SR1

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    α1−PFRAC3

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    C0 h1/2

    Figure: SR1, PFRAC3 CFL = 0.49, t = 0.25 .

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 21 / 32

  • Computation of the two-fluid model Evolution step

    Verification of the fractional step method : test case 3

    Rate of convergence observed :

    PFRAC3 :1

    2,

    −5.5 −5 −4.5 −4 −3.5 −3 −2.5−5

    −4

    −3

    −2

    −1

    0

    1

    ln(h)

    ln(e

    rror)

    α

    1

    ρ1

    u1

    p1

    ρ2

    u2

    p2

    C0 h1/2

    Figure: PFRAC3, CFL = 0.49, t = 0.25 .

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 22 / 32

  • Computation of the two-fluid model Relaxation step

    Relaxation step

    W̃velocity relaxation�������������! W ⇤ pressure relaxation�������������! Wn+1

    i

    (15)

    Computing the ordinary di↵erential equations :

    Velocity relaxation : drag terms

    8

    >

    >

    >

    >

    >

    >

    >

    <

    >

    >

    >

    >

    >

    >

    >

    :

    @t

    ↵l

    = 0@t

    (↵l

    ⇢l

    ) = 0@t

    (↵l

    ⇢l

    ul

    ) = Dl

    @t

    (↵l

    ⇢l

    el

    ) = Dl

    Uint

    @t

    (↵v

    ⇢v

    + ↵l

    ⇢l

    ) = 0@t

    (↵v

    ⇢v

    uv

    + ↵l

    ⇢l

    ul

    ) = 0@t

    (↵v

    ⇢v

    ev

    + ↵l

    ⇢l

    el

    ) = 0(16)

    Proposed in [2](HH, CaF, 2012).

    Pressure relaxation :

    8

    >

    >

    >

    >

    >

    >

    >

    <

    >

    >

    >

    >

    >

    >

    >

    :

    @t

    ↵v

    = �v

    @t

    (↵v

    ⇢v

    ) = 0@t

    (↵v

    ⇢v

    uv

    ) = 0@t

    (↵v

    ⇢v

    ev

    ) + pl

    @t

    ↵v

    = 0,@t

    (↵l

    ⇢l

    ) = 0@t

    (↵l

    ⇢l

    ul

    ) = 0@t

    (↵l

    ⇢l

    el

    ) + pl

    @t

    ↵l

    = 0,(17)

    Proposed in [1](GHHN, M2AN, 2010)+ Extension to new EOS

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 23 / 32

  • Computation of the two-fluid model Relaxation step

    Pressure relaxation : extension to EOS GP(vapor)+SG(liquid)

    The ordinary di↵erential equation (17) is :

    8

    >

    >

    >

    <

    >

    >

    >

    :

    @t

    ↵v

    = �v

    @t

    (mv

    ev

    ) + pl

    @t

    ↵v

    = 0@t

    (ml

    el

    )� pl

    @t

    ↵v

    = 0@t

    (↵'

    ⇢'

    ) = 0@t

    (↵'

    ⇢'

    u'

    ) = 0

    (18)

    An implicit scheme calculates (pn+1l

    , pn+1v

    ,↵n+1v

    ), solution of

    8

    <

    :

    ↵n+1v

    � ↵⇤v

    +�t(✓)�1↵n+1l

    ↵n+1v

    (pn+1l

    � pn+1v

    ) = 0(m

    l

    el

    )n+1 � (ml

    el

    )⇤ � Pn+1I

    (↵n+1v

    � ↵⇤v

    ) = 0(m

    v

    ev

    )n+1 � (mv

    ev

    )⇤ + Pn+1I

    (↵n+1v

    � ↵⇤v

    ) = 0(19)

    Property 6Assume that the EOS of the gas phase is GP, and that of the liquid phase is SG,then the scheme (19) admits a unique relevant solution (pn+1

    l

    , pn+1v

    ,↵n+1v

    ) such

    that pv

    > 0, pl

    + (pl

    )1 > 0 and ↵n+1v

    lies in [↵M

    ,↵m

    ] 2 [0, 1], with

    ↵M =�v

    � 1�v

    ↵0v

    , ↵m =1 + (�

    l

    � 1)↵0v

    �l

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 24 / 32

  • Computation of the two-fluid model Relaxation step

    Verification of the pressure relaxation scheme

    Slope of curves : 0.56Rate of convergence expected : 1‘Exact solution’ : Matlab ODE45 approach (AbsTol : 1e-14)

    −5.5 −5 −4.5 −4 −3.5 −3 −2.5 −2−3.5

    −3

    −2.5

    −2

    −1.5

    −1

    −0.5

    0

    ln(h)

    ln(e

    rro

    r)

    α

    l

    pl

    pv

    Figure: Curves of convergence of the pressure relaxation scheme.

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 25 / 32

  • Verification of the two-fluid model

    Outline

    1 Introduction

    2 The two-fluid model & water hammer

    3 Computation of the two-fluid modelEvolution stepRelaxation step

    4 Verification of the two-fluid model

    5 Conclusions and perspectives

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 26 / 32

  • Verification of the two-fluid model

    Simpson Experiment (in process)S A.R., PhDThesis, 1986 [5]

    Reservoir

    Tuyau 36 m

    u

    Figure: Simpson ExperimentFigure: Pressure at the valve ofpipe

    Combination of many di↵erent phenomena :

    Reflection of waves, vaporous cavitation

    Regime transition : almost single-phase flow ! mixture of water/steamFluid structure interaction

    Preliminary questions : Is the two-fluid model able to

    retreive the single-phase phenomena at the beginning of experiment ?

    capture the speed of sound in the liquid/vapor mixture ?

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 27 / 32

  • Verification of the two-fluid model

    Single-phase phenomena of Simpson experiment

    0.9999999990.999999990.99999990.9999990.999990.99990.9990.990.94

    5

    6

    7

    8

    9

    10x 10

    5

    αl

    pres

    sion

    (p)

    Pression−formule de JoukowskyResu numerique 100 maillesResu numerique 1000 maillesResu numerique 10000 mailles

    Figure: Pressure magnitude of the first shock wave : p = ↵l

    pl

    + ↵v

    pv

    according todi↵erent values of ↵

    l

    Joukowsky formula : incompressible single-phase estimation

    �p = ⇢l

    cl

    �u (20)

    Numeric : ⌧p

    = 10�10s, ⌧u

    = 10�10s,

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 28 / 32

  • Verification of the two-fluid model

    Speed of sound in the liquid/vapor mixture

    0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

    500

    1000

    1500

    αl

    c

    Formule de WallisExperience 250 HZExperience 500 HZResu numérique avec 1000−maillesResu numérique avec 10000−mailles

    0.3 0.4 0.5 0.6 0.7 0.8 0.9 120

    30

    40

    50

    60

    70

    80

    90

    100

    αl

    c

    Formule de WallisExperience 250 HZExperience 500 HZResu numerique avec 1000−maillesResu numérique avec 10000−mailles

    Figure: Numerics, Wallis, experiment for di↵erent values of ↵l

    Wallis formula :1

    mcWallis

    2=

    ↵l

    ⇢l

    cl

    2+

    ↵v

    ⇢v

    cv

    2(21)

    Experiment : [3] (Karplus, H. B.-1958)

    Numerics : ⌧p

    = 10�10s, ⌧u

    = 10�10s,

    Remark : There is no definition of speed of sound for the mixture for the

    two-fluid model , here cnum

    = dispulse

    /time

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 29 / 32

  • Conclusions and perspectives

    Outline

    1 Introduction

    2 The two-fluid model & water hammer

    3 Computation of the two-fluid modelEvolution stepRelaxation step

    4 Verification of the two-fluid model

    5 Conclusions and perspectives

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 30 / 32

  • Conclusions and perspectives

    Conclusions and perspectives

    . Two-fluid model :stable convection schemes + relaxation schemes (p, u)rather expensive (3D computation ! ! !)

    good approximation of speed of sound in the mixturesingle-phase phenomena retrieved

    masse transfer relaxation scheme (cavitation) to be testedmore complete realistic calculations to be leaded

    . The five-equation models

    . Comparison of models + FSI

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 31 / 32

  • Conclusions and perspectives

    Thank you for your attention !

    Any questions ?

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 32 / 32

  • Conclusions and perspectives

    T. Gallouët, P. Helluy, J.-M. Hérard, and J. Nussbaum.

    Hyperbolic relaxation models for granular flows.

    Math. Model. and Numer. Anal., 44(2) :371–400, 2010.

    J.-M. Hérard and O. Hurisse.

    A fractional step method to compute a class of compressible gas-liquide flows.

    Computers & Fluids, 55 :57–69, 2012.

    H. B. Karplus.

    The velocity of sound in a liquide containing gas bubbles.

    Armour Research Foundation of Illinois Institute of Technology, C00-248, 1958.

    D. Schwendeman, C. Wahle, and A. Kapila.

    The Riemann problem and a high-resolution Godunov method for a model ofcompressible two-phase flow.

    Journal of Computational Physics, 212(2) :490–526, 2006.

    A. R. Simpson.

    Large water hammer pressures due to column separation in sloping pipes

    (transient cavitation).

    PhD thesis, University of Michigan, 1986.Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 32 / 32

  • Conclusions and perspectives

    S. Tokareva and E. Toro.

    HLLC-type Riemann solver for the BaerNunziato equations of compressibletwo-phase flow.

    Journal of Computational Physics, 229 :3573–3604, 2010.

    Yujie LIU (EDF R&D, AMA) Water-hammer/Two-fluid model AMIS 2012 32 / 32

    IntroductionThe two-fluid model & water hammer Computation of the two-fluid modelEvolution step Relaxation step

    Verification of the two-fluid model Conclusions and perspectives