UFR 3-14 Best Practice Advice
Flow over surface-mounted cube/rectangular obstacles
Underlying Flow Regime 3-14 © copyright ERCOFTAC 2004
Best Practice Advice
Best Practice Advice for the UFR
There is now substantial evidence, of the kind summarised above, that steady RANS models cannot be relied on for accurate computations of flow around single, surface-mounted bluff bodies. (They fare even less well, incidentally, for isolated bodies which lead to strongly periodic wakes - vortex shedding, for example). Even with the necessary modifications to counteract the standard k-ε model's failing in regard to production of excess turbulence in stagnation regions, solution of the flow field shows significant quantitative differences from experiment. The length of the downwind separation region, for example, is likely to be over-predicted by at least 35%, with consequently large errors in local flow velocities in the near wake. Likewise, body surface pressures will also be seriously in error - particularly near the sharp edges if the body and if the standard k-ε model is used, since this can lead to spurious shear layer attachment on the top surface.
Model modifications to prevent production of excess turbulence energy in regions of strong strain can lead to rather better surface pressure predictions (despite the rather worse flow field predictions). There seems little to gain by employing higher-order (steady) RANS methods although there is limited evidence that unsteady RANS can be more successful.
In contrast, LES techniques have been shown to perform generally very much better, perhaps largely because resolution of the large-scale unsteady motions produces the correct overall, time-averaged flow. However, even LES cannot be relied upon without careful attention to details of the gridding and the sub-grid model and results from the LES computations performed thus far vary amongst themselves almost as much as those obtained with simpler methods. One of the very best LES simulations of this particular UFR has been reported by Shah & Ferziger (1997) and even these authors recognise that 'it is possible for LES to yield results that are incorrect qualitatively as well as quantitatively'. At least some of the difficulties, whatever model is used, arise in the near-wall regions and although there are methods which avoid having to impose wall-functions, these have not yet been shown to be any more reliable. Note also that aerodynamically rough walls produce their own set of difficulties and, although this naturally has not been discussed in the context of the cube-in-a-channel UFR, it will obviously be an issue in nearly all computations of genuine Wind Engineering problems.
If the requirement is for estimates of the fluctuating features of the flow (like the minimum, short-term-average roof pressure, for example) one really has little choice but to go to an LES-based method, but there is no evidence yet that the fluctuating features can be accurately predicted. This is partly because there are hardly any genuinely useful experimental data on the unsteady nature of the flow. (We mean 'useful' in the sense that the data form part of a database sufficiently complete to allow setting up a numerical experiment for comparative purposes).
It is concluded that this UFR is very challenging. Even though it is in many respects simpler than most industrial problems of its type, there remain serious difficulties in generating accurate solutions for both surface pressures and flow field data simultaneously. Steady RANS-based methods will always produce unreliable results which may even be qualitatively wrong in some respects (e.g. yielding flow reattachment where none exists). LES has begun to demonstrate its potential, but is not without its difficulties. With the continuing rapid improvements in computer power, solution time and storage requirements are increasingly less serious relative to the more fundamental ones concerning model details and these should be addressed as a matter of some urgency
© copyright ERCOFTAC 2004
Contributors: Ian Castro - University of Southampton