Are there alternatives?Despite the increases in computational powerand progress in numerical techniques, it is currentlynot possible to resolve the small scales athigh Reynolds number. Even with computationalpower doubling every 18 months, it wouldtake several decades before a model-free, directnumerical simulation (DNS) of the simplest flowcase of non-decaying isotropic turbulence couldbe performed with the separation of scales 10 5equivalent to the wind tunnel proposed here(tremendous resolution requirements for a comparablesimulation – number of grid pointsgreater than 10 15 ).“Despite the impressive developments in ComputationalFluid Dynamics over recent decades, [...] many of the fundamentalquestions still facing the turbulence communitycannot possibly be answered in the foreseeable futurewithout recourse to laboratory work.”“<strong>The</strong>re are now a number of relatively small-scale facilitiesdesigned to allow study of very high Reynolds numberflows (via use of very high pressures, exotic gases, or whatever)butt these are all inherently limited by the technicaldifficulties in developing and applying appropriate instrumentation.In my view, there is much to be said for theonly possible alternative approach: going to much largerscales, so that existing, well-developed techniques can beused.Your proposal is therefore very exciting and, if I maysay so, very well thought out.”Ian P. Castro, Professor of Fluid Dynamics, University ofSouthampton, UK.“Computers cannot (and will not be able to in the foreseeablefuture) explore the high Reynolds numbers thatare required to understand the basic properties of turbulence.”Zellman Warhaft, Professor of Mechanical and AerospaceEngineering, Cornell University, USA.“<strong>The</strong>re are many important questions about the physics oflarge Reynolds number turbulence that could beanswered with highly accurate measurements in your proposedtunnel and nowhere else. I can envisage quite a fewexperiments in your tunnel that will help validate my DNS[direct numerical simulation] results.”Said Elgobashi, Professor and Chair, Mechanical andAerospace Eng., University of California, Irvine, USA.“I am convinced that the turbulence community reallyneeds a large-scale wind tunnel, with sufficient separationbetween the large and small scales, with sufficiently lowturbulence intensity, and in which the smallest scales canbe probed with no special difficulty even for the largestReynolds numbers. Such a tunnel (none of the presentlyavailable ones possess such specifications because theyare either too small, or, for the largest ones, they are recycledindustrial tunnels with insufficiently clean and controlledaerodynamic conditions) will be of considerablevalue for testing new ideas or new theories, and even forproviding basic clean data on which the present theorieswill be tested”Professor Fabien Anselmet, Director of Research at CentreNational de la Recherche Scientifique (CNRS), <strong>Turbulence</strong>Research Group, University Aix-Marseille, France.6<strong>The</strong> <strong>Nordic</strong> <strong>Wind</strong> <strong>Tunnel</strong>
What makes the <strong>Nordic</strong> <strong>Wind</strong> <strong>Tunnel</strong> unique?<strong>The</strong> <strong>Nordic</strong> <strong>Wind</strong> <strong>Tunnel</strong> overcomes the shortcomingsof present research facilities and is proposedfor construction at Chalmers Universityof Technology. It would be• wide enough to remove the effect of sidewalls on the energetic turbulence scales• fast enough and large enough to get the necessaryhigh Reynolds numbers, yet still resolvethe dissipative scales• long enough and with low enough backgrounddisturbances to obtain the necessarydownstream development times and will thusprovide an experimental facility capable ofresolving some of the oldest questions in turbulencewhile also testing conclusively new ideas.In the unique inverse design process it was askedwhat length and time scales needed to beresolved to conduct “meaningful” measure-ments (e.g., resolution of wall layer to obtainshear stress) and what Reynolds numbers neededto be achieved in the experiments to be performed(e.g., zero pressure gradient boundarylayer, far wake, decay of isotropic turbulence) tohelp resolve fundamental questions and sortcompeting theories. <strong>The</strong>se length and timescales and Reynolds number criteria depend onthe flow being measured. <strong>The</strong> size (length andcross-sectional area of the test section) and performance(maximum free stream velocity) of theproposed wind tunnel facility were then determinedby what can be resolved with existingprobes.<strong>The</strong> small scales of the flow become smaller withincreasing flow speed or pressure. To quoteProfessor Tony Perry, University of Melbourne,who realized the dilemma and – before hisuntimely death – stated that: ``Big and slow - isthe way to go''.Example of one design criterion:For wall-bounded turbulent flows therequirement was that one viscous lengthscale can still be resolved (using a micro-LDA or a micro-PIV system, with a measuringvolume height of 10 µm) whileachieving a Reynolds number based onmomentum thickness of at least 100,000.Other “base case flows”, e.g. decaying turbulenceor wake flows, also provided furthersize and flow quality criteria.<strong>The</strong> <strong>Nordic</strong> <strong>Wind</strong> <strong>Tunnel</strong>7