OpenFOAMとCFXの乱流モデルk-ω SSTの違いについて

1,596 views
Skip to first unread message

bobna...@gmail.com

unread,
Mar 3, 2016, 9:09:03 AM3/3/16
to OpenFOAM
はじめまして

OpenFOAM初心者の佐藤と申します。


現在、OpenFOAM(v2.4.0)の検証のため、ポペット弁を対象にCFX(v15.0)との比較を行っていますが、
CFXに比べて圧力損失がかなり高い結果となりました。

オリフィス絞りで近似すると流量係数0.1以上の差で、最大流速はOpenFOAMが10%以上大きくなって
います。

ちなみに、乱流モデルは両者ともk-ω SSTを使用しています。
(ただし、OpenFOAMは高Re型モデル、CFXは低Reモデル型といった違いがあるようです。)

メッシュモデルは、CFXの低Re型モデルに合わせて第一格子はy+=1前後になるように設定しています。

同じメッシュモデルを使用したため、高Re型モデルのOpenFOAMでは壁関数は使用していません。


なお、乱流モデルk-ω SSTを調べてみると、「CFX(v14.0)はMenterの2003年の論文を基にしたモデルが、
OpenFOAM(v2.1.0)はMenterの2001年の論文を基にしたモデルが実装されているため、OpenFOAM
は境界層の粘性底層付近で渦動粘性係数を大きく予測する傾向がある」といった論文情報がありました。

今回圧力損失がCFXに比べて高くなった原因も、OpenFOAM(v2.4.0)がMenterの2001年の論文を基に
したモデルが実装され、CFXとは渦動粘性係数の算出方法が異なっていることによるものではないかと
考えましたが如何でしょうか?

色々調べてみましたが、OpenFOAM(v2.4.0)の渦動粘性係数の算出方法の基となる論文が分かりませ
んでしたので、ご存知の方がいらっしゃいましたら、ご教授をよろしくお願いいたします。

Youhei Takagi

unread,
Mar 3, 2016, 7:56:28 PM3/3/16
to open...@googlegroups.com
佐藤様

高木と申します。

$WM_PROJECT_DIR/src/turbulenceModels/incompressible/RAS/kOmegaSST/kOmegaSST.H
のコメント部分で、

Description
    Implementation of the k-omega-SST turbulence model for incompressible
    flows.

    Turbulence model described in
    \verbatim
        Menter, F., Esch, T.,
        "Elements of Industrial Heat Transfer Prediction",
        16th Brazilian Congress of Mechanical Engineering (COBEM),
        Nov. 2001.
    \endverbatim

    with updated coefficients from
    \verbatim
        Menter, F. R., Kuntz, M., and Langtry, R.,
        "Ten Years of Industrial Experience with the SST Turbulence Model",
        Turbulence, Heat and Mass Transfer 4, 2003,
        pp. 625 - 632.
    \endverbatim

    but with the consistent production terms from the 2001 paper as form in the
    2003 paper is a typo, see
    \verbatim
        http://turbmodels.larc.nasa.gov/sst.html
    \endverbatim

    and the addition of the optional F3 term for rough walls from
    \verbatim
        Hellsten, A.
        "Some Improvements in Menter’s k-omega-SST turbulence model"
        29th AIAA Fluid Dynamics Conference,
        AIAA-98-2554,
        June 1998.
    \endverbatim

    Note that this implementation is written in terms of alpha diffusion
    coefficients rather than the more traditional sigma (alpha = 1/sigma) so
    that the blending can be applied to all coefficuients in a consistent
    manner.  The paper suggests that sigma is blended but this would not be
    consistent with the blending of the k-epsilon and k-omega models.

    Also note that the error in the last term of equation (2) relating to
    sigma has been corrected.
    The default model coefficients correspond to the following:
    \verbatim
        kOmegaSSTCoeffs
        {
            alphaK1     0.85;
            alphaK2     1.0;
            alphaOmega1 0.5;
            alphaOmega2 0.856;
            beta1       0.075;
            beta2       0.0828;
            betaStar    0.09;
            gamma1      5/9;
            gamma2      0.44;
            a1          0.31;
            b1          1.0;
            c1          10.0;
            F3          no;
        }
    \endverbatim

と記載されています。

これらの論文とソースコードを比較しながら検証されるのが良いかと
思います。

以上ご参考までに。

2016年3月3日 23:09 <bobna...@gmail.com>:

--
このメールは Google グループのグループ「OpenFOAM」に登録しているユーザーに送られています。
このグループから退会し、グループからのメールの配信を停止するには openfoam+u...@googlegroups.com にメールを送信してください。
このグループに投稿するには open...@googlegroups.com にメールを送信してください。
https://groups.google.com/group/openfoam からこのグループにアクセスしてください。
その他のオプションについては https://groups.google.com/d/optout にアクセスしてください。

nakagawa

unread,
Mar 3, 2016, 7:58:02 PM3/3/16
to OpenFOAM
佐藤様

kOmegaSSTLowRe turbulence model を開発・公開している人がいます。
OF2.4系用ブランチ

低レイノルズ数型に適したメッシュであれば,上記を使う方が良いのではないでしょうか。

お調べになった論文の出典を記載していただけると,多くの人の参考になると思います。差し障りなければ,ぜひ,情報を共有してください。
私の知り合いの初心者も,同じようなことを調べています。情報を公開・共有していただけると助かります。

最近のOpenFOAMソースコードでは,クラスのヘッダ部分に,文献情報等を書いてくれているものが増えているように思います。下記を見ると,文献の詳しい情報も記載されています。

Description
    Implementation of the k-omega-SST turbulence model for
    incompressible and compressible flows.
    Turbulence model described in:
    \verbatim
        Menter, F. R. & Esch, T. (2001).
        Elements of Industrial Heat Transfer Prediction.
        16th Brazilian Congress of Mechanical Engineering (COBEM).
    \endverbatim
    with updated coefficients from
    \verbatim
        Menter, F. R., Kuntz, M., and Langtry, R. (2003).
        Ten Years of Industrial Experience with the SST Turbulence Model.
        Turbulence, Heat and Mass Transfer 4, ed: K. Hanjalic, Y. Nagano,
        & M. Tummers, Begell House, Inc., 625 - 632.
    \endverbatim
    but with the consistent production terms from the 2001 paper as form in the
    2003 paper is a typo, see
    \verbatim
    \endverbatim
    and the addition of the optional F3 term for rough walls from
    \verbatim
        Hellsten, A. (1998).
        "Some Improvements in Menter’s k-omega-SST turbulence model"
        29th AIAA Fluid Dynamics Conference, AIAA-98-2554.
    \endverbatim
    Note that this implementation is written in terms of alpha diffusion
    coefficients rather than the more traditional sigma (alpha = 1/sigma) so
    that the blending can be applied to all coefficuients in a consistent
    manner.  The paper suggests that sigma is blended but this would not be
    consistent with the blending of the k-epsilon and k-omega models.
    Also note that the error in the last term of equation (2) relating to
    sigma has been corrected.
    Wall-functions are applied in this implementation by using equations (14)
    to specify the near-wall omega as appropriate.
    The blending functions (15) and (16) are not currently used because of the
    uncertainty in their origin, range of applicability and that if y+ becomes
    sufficiently small blending u_tau in this manner clearly becomes nonsense.


なかがわ


2016年3月3日木曜日 23時09分03秒 UTC+9 Sato:
Message has been deleted

Sato

unread,
Mar 4, 2016, 7:46:47 AM3/4/16
to OpenFOAM

高木様、なかがわ様

丁寧なご回答、有難うございます。
大変参考になりました。

教えて頂いた、OpenFOAMソースコードのリンク先の内容を確認しました。

ソースコードと論文やCFXの資料等と比較してみます。


また、低Re型のk-ω SSTも開発・公開されているとのこと、試してみます。


私の方で調べたOpenFOAMとCFXの乱流モデルの違いに関する論文ですが、
「自動車室内温熱環境予測におけるOpenFOAMの実用性」という空気調和・衛生工学会の論文です。
http://ci.nii.ac.jp/naid/110009969874


nakagawa

unread,
Mar 4, 2016, 10:19:29 AM3/4/16
to OpenFOAM
佐藤様

なかがわです。
情報の共有ありがとうございました。

なかがわ

2016年3月4日金曜日 21時46分47秒 UTC+9 Sato:
Reply all
Reply to author
Forward
0 new messages