Discussion 2009-12-23 (IA, DL):

  • Yes we can

  • But we didn't (or just once)
    • Refined LES Publications by S. Benhamadouche, then I. Afgan for 1 geometry : Large Eddy Simulation of flow across in-line tube bundles., Benhamadouche S., Laurence D., Jarrin N., Afgan I., Moulinec C., NURETH-11 Oct. 2005.
    • In this paper Comet and C_S were compared global expt results as rms lift and drag. C_S values were too high (x2 for CD', x3 for CL' and this was attributed to the coarse block structured mesh used in C_S). Mean CP profiles of Comet and C_S compared well however.
    • IA then continued with finer grids (1.3 Mil, 2.4 Mil) and seemed to obtain mesh convergence (he had Y+=1, Dz+ = 15).
    • IA studied gap ratios 1.2, 1.5; 1.6, 1.75 , and found some interesting exp data to validate his LES (Aiba et al. JSME 1982)
    • but only the 1st simulation was published, rest is in the thesis

Proposed simulations

Much of the Sturm4 project written in 2006 is about extending the work in Imran's thesis: unstructured near-wall cell refinement (2 to 3 or other) and rod tube bundles. It would be a pity not to take advantage of I.A. detailed knowledge of the application field (see extensive bibliography in I.A. thesis) and leave this unpublished. Having established the reliability of the cold flow simulation, we should now use it to produce a database for heat transfer and body forces.

To be investigated

  • Domain size
    • Lz=2D is perhaps just marginal enough, Lz=3D would more clearly show 2poitn correlations going to zero. (x 1.5)
    • IA's simulations contained only 2 x 2 streams and cross avenues. 3 streams x 4 avenues would be safer (x2 x 1,5)
    • typical mesh size, based on IA's 2 Mil cells => 10 Mil (2Mil x1.5 x2 x 1.5
  • Mesh types
    • Best meshing strategy: extrusion is not an option if we want to keep y+=1, dz+=10=dx+. IA's thesis has several diferrent mesh types. As is now often observed, best nearly equilatera Hexa "Gambit like3 mesh. This is also indicated by the laminar tube bundle tutorial set by Stefano Rolfo here TubeBundleTutorial solution in MahmoudAssad
  • Cases to run
    1. Inline as in IA thesis for case where most expt data is available, to check if C_S gies same results as Comet.
    2. Test grid effects on laminar case, perhaps RE = 50 or 100. Use very very fine tet or regular grid as reference solution, then test unstructured as you would use for Higher Re case. demonstrate convergenge by Richardosn extrapolation as did Imama. This is a publishable verification case.
    3. vary vertical angle of attack, discussing with steam generator or rod bundles experts (see Frank Maurel review)
    4. run staggered 45% and other angles.
    5. produce Idel'l Cik handbook of tensorial head loss coefficients and heat transfer


  1. The staggered tube bundle (Simonin O., Barcouda M. 1988),

    LES, coarse LES, and transient RANS comparisons on the flow across a tube bundle
    International Journal of Heat and Fluid Flow, Volume 24, Issue 4, August 2003, Pages 470-479
    S. Benhamadouche, D. Laurence

LES, Rollet-Miet

DNS, Moulinec

Present "fine"

Present "coarse"

N cells:

280 000

5 Million

170 000

18 000

  • Review of head loss modelling for flow in rod bundles
    Etat de l'art sur la modélisation des pertes de charge régulières pour des
    écoulements obliques en faisceau de crayons, Franck MAUREL, EDF R&D MFEE report H-I84-2007-03312-FR, 2008-09-01

Current Tags:
create new tag
, view all tags
Topic revision: r3 - 2009-12-29 - 15:13:38 - DominiqueLaurence
Main Web
22 Jul 2018


Manchester CfdTm

Ongoing Projects


Previous Projects


Useful Links:

User Directory
Photo Wall
Upcoming Events
Add Event

Computational Fluid Dynamics and Turbulence Mechanics
@ the University of Manchester
Copyright © by the contributing authors. Unless noted otherwise, all material on this web site is the property of the contributing authors.