Welcome back to Sports & Building Aerodynamics in the week on Computational Fluid Dynamics. In the second module on Computational Wind Engineering, we start again with a module question. Why can steady RANS be sufficient for the assessment of pedestrian-level wind comfort and wind danger, possibly obviating the need to turn to LES? Is that because of A) Mean wind speed can be predicted with sufficient accuracy by RANS. B) Mean wind speed and turbulence intensity can be predicted with sufficient accuracy by RANS. Or C) Mean wind speed in high wind speed areas can be predicted with sufficient accuracy by RANS. Or finally D) Mean wind speed and turbulence intensity in high wind speed areas, can be predicted with sufficient accuracy by RANS. Please hang on to your answer and we'll come back to this question later in this module. At the end of this module, you will understand the advantages of CWE. You'll understand the sense or nonsense of the numerical wind tunnel. You'll understand the history and progress in CWE as shown by symposia and overview papers. And you'll understand some concerns about RANS versus LES. And finally, some interesting findings from a literature study on CFD simulation of pedestrian-level wind conditions. And this presentation also is part of a modified version of my keynote presentation, at the European-African Conference on Wind Engineering. And, you can read more about this overview in this article. So, CWE is definitely complementary to other and more traditional areas of wind engineering, such as full-scale on-site experimentation and reduced-scale wind-tunnel testing. And there are some particular advantages of CWE, when used in combination with those methods. It's fully controllable, as you also have in the wind tunnel but not in on-site experimentation. You can get whole-flow field data that is even extremely difficult or impossible to obtain, with methods such as Particle-Image Velocimetry, and Laser-Induced Fluorescence. The simulations can be performed at full scale, so, no similarity issues so you can avoid potentially incompatible similarity requirements, and there's seemingly a wider range of problems that can be studied, although this should be stated very carefully because the problems that will be listed here are almost equally challenging to predict with CFD as they are in the wind tunnel. Examples are wind flow in atmospheric boundary layers with stable and unstable stratification. Wind flow over very extensive areas that would require very large scaling factors to put them into a wind tunnel. Wind flow for study areas that are quite large and where relatively small flow features and length scales are important. Just as for example natural ventilation through relatively small openings. If you would put a building with small ventilation openings in a wind tunnel, they might scale down too far, so that the flow becomes laminar or transitional and not turbulent like in reality, and clearly similarity will be violated. So this kind of studies you cannot do in wind tunnels. Then there are multiphase flow problems such as transport of sand, dust, hail and rain. Very challenging in wind tunnels, but also not that straightforward in CFD. There's buoyant flows, such as pollutant dispersion, but also with buoyancy-driven natural ventilation. And finally, meteorological phenomena such as tornadoes and downbursts. So, CWE definitely provides many opportunities. But the question is, does it provide a numerical wind tunnel? Well actually several of such claims have been made in the past, mainly by non-wind engineers. But the wind engineering community, has actually systematically and throughout the past decades, always denounced this label, and warned that many CWE problems are far too complex to be tackled by CFD alone, and instead the synergy between experiments and CFD should be exploited. So, the conclusion is, there's no numerical wind tunnel, because of the disadvantages of CFD, which are that it's very sensitive to the wide range of computational parameters that needs to be selected. The quality of the results also depends to a very large extent on the expertise of the user. And verification and validation remain very important, and therefore also experiments remain indispensable. So let's look at some quotes on the numerical wind tunnel. This is a quote by Leschziner stating that: "This", and he refers to computer technology, "evolution has given rise to the rather radical view, expressed predominantly among the US aerodynamics fraternity, that the wind tunnel is destined to become a convenient storage cabinet for computer output. A moment's contemplation leads to the conclusion that this view reflects a rather narrow interpretation of CFD, focusing on the particular type of flows, most relevant to high-speed external aerodynamics and some turbomachinery applications." There's a quote by Leitl and Meroney stating that: "Using numerical codes like Fluent can help to design and setup wind tunnel experiments, hence reducing the time required to optimize a physical model, and expensive pre-runs in a wind tunnel. With a numerical simulation critical points like source design for dispersion stimulation can be examined, and boundary conditions can be modified." A quote by Stathopoulos saying: "Nevertheless, there seems to an ever-increasing confidence in the results obtained by CFD codes, and more and more papers propagate the idea that the numerical wind tunnel does exist today, and produces results ready to be used by practitioners. In the author's opinion, this is at best premature and at worst dangerous, with the exception of very limited cases." The quote by Baker stating that: "Applications will become widespread in areas where wind velocities rather than surface pressures are required, such as the assessment of pedestrian comfort. These trends may well lead to the concentration of boundary layer wind tunnel testing for complex structures, into a smaller number of institutions over the next few decades." There's quote by Cochran and Derickson, stating that: "The biggest remaining challenge for CWE is the treatment of peak structural wind loads and peak cladding pressures on buildings. Continued hybrid use of wind tunnels and CFD with cross-comparison validation between wind-tunnel (or full-scale) results, will be essential to gain confidence in the methodology." Let's look at the progress in Computational Wind Engineering over the past decades as demonstrated by CWE symposia. And concerning this overview, I would like to refer to this excellent and very comprehensive review paper explaining the history, progress, and prospects, of the IAWE, so that's the International Association of Wind Engineering, that was compiled by previous IAWE president Giovanni Solari. And this paper in a very detailed way outlines the start and the emergence of the association. And this table is actually provided in the paper and I have added the most recent conferences in green at the bottom and this is an overview of international conferences in wind engineering, where actually the eigth one, as mentioned before, was a very particular one, because it was for the first time that a session was devoted to CFD. And this was also fully acknowledged by Solari in his paper stating that: "In this renewed framework, the first session devoted to CFD by an ICWE, deserves a special mention for the impressive growth that this matter would exhibit in the forthcoming years. And then there was the very successful beginning: the first symposium on Computational Wind Engineering, in 1992, Tokyo, Japan, organized by Murakami assisted by Matsumoto and Mochida. And this indeed was the conference where CWE really took off. Murakami stated that: "CFD developed mainly in the fields of mechanical and aeronautical engineering. Great success has been achieved in predicting such relatively simple flows as channel flow, air flow around a wing, etc. However, problems concerning air flow in wind engineering are far more complicated. In many cases, the current CFD technology borrowed from other fields, is thus inadequate and new research and development are urgently required in this field." This is a quote by Solari from his extensive review paper, stating that: "The CWE highlighted a new fundamental reality: the advent of CFD and, more generally, the growth of computational tools as an alternative or an integration to analytical methods, wind tunnel tests, and full-scale measurements. It was so successful, that following the example of the First CWE, these conferences were organized, close to the BBAAs," that is, Bluff Body Aerodynamics and Applications, also in 1996, and in 2000, respectively by Meroney and by Baker." This is a quote by Stathopoulos also emphasizing the importance of this first conference: "The importance of the symposium cannot be overemphasized. For the first time, it joined wind engineering delegates with classical aeordynamicists, who were using CFD rather routinely to address and solve aeronautical problems. As these problems are very different from wind engineering problems, the first CWE symposium led to a very fruitful interaction between the different groups of delegates." And then we have the successful sequels. In 1996 in Fort Collins, Colorado, organized by Meroney and Bienkiewicz. In 2000 in Birmingham, UK,organized by Baker, 2006, Yokohama, Japan, organized by Murakami, Matsumoto, and Tamura, in 2010, organized by Huber, Blocken, and Stathopoulos, and the most recent one in 2014 will take place in Hamburg, Germany. And many topics were dealt with at these symposia. Of course, general topics such as simulation of the atmospheric boundary layer, bluff body aerodynamics, turbulence modeling, so the more fundamental topics. But also applications in environmental wind engineering, in structural wind engineering, wind energy, and in other topics, such as sports and vehicle aerodynamics. The progress in CWE in the past decades can also be indicated by review papers. And rather than providing you with a full list of review papers that you can read in the article, I would like to highlight a few. And this is, to the best of my knowledge, the first overview paper, review paper in Computational Wind Engineering by Murakami published already in 1990. And then there is another very extensive paper, also a review paper by Murakami, published in 1997. And this was especially a very complete and almost timeless paper, because it addressed in great detail why CWE is so more difficult than some other areas in CFD. It addressed inflow boundary conditions for LES in great detail, already at that stage of CWE development. And then it provides an impressive list of application examples, all of which are still intensively studied today and are still very relevent problems, such as velocity and temperature fields around the human body. And not just any simulation, but a simulation with grid resolutions actually fine enough to resolve the laminar sublayer, which is the way it should be done. Then velocity and pressure fields around bluff bodies, be it buildings or bridge sections. Coupled fluid-structure analysis, pollutant dispersion around buildings, pedestrian wind conditions around a high-rise building, analysis of outdoor climate within city blocks including also convection, radiation, and conduction. Mesoscale analysis of city and regional climates, and so on. So a very impressive list of application examples. There are however, looking back at a few papers over the past decades in CWE, there are some missing review papers, one could argue. And that's about CFD simulation of the neutral, stable and unstable ABL, wind and acoustics, wind loads on solar panels, wind energy in the built environment, sports aerodynamics, and wind-borne debris. I would like to offer you some quotes on Large Eddy Simulation versus RANS approaches in CWE. This is a quote by Ferziger stating that: "If it turns out that LES can be done on very coarse grids, it will be one of the few times that nature has been kind to us with regard to turbulent flows." A quote by Leschziner saying: "In the event of peak wind and pressure loading having to be determined, a statistical framework, as in RANS, is obviously inappropriate. In this case, the only alternative route is Large Eddy Simulation." A quote by Stathopoulos stating that: "It should be stressed however, that LES as a procedure of turbulence modeling is going to be truly useful only if it reaches the stage of producing peak instantaneous pressure coefficients, with some reasonable accuracy." This is a quote by Tamura, stating that: "It should be noted that, in order to use CFD for wind load estimation, an accurate time-dependent analysis, such as LES, is definitely required, because it enables prediction of peak-type of quantities such as a peak pressure or maximum response of a building and a structure." Finally, in this module, I would like to share with you some important findings from a literature review that I made on CFD simulation of pedestrian-level wind conditions. Which is also a topic that will be dealt with, in more detail, in week 4. And one of the remarkable studies in this field was the validation study by Yoshie and Tominaga and their co-authors, where CFD simulations and comparisons with wind-tunnel experiments were made for a very impressive range of example building configurations. Ranging from an isolated building, to an actual complex urban area. And they found that when comparing wind speed ratios in the experiment versus the CFD simulation, that a remarkably close agreement was found between experiments and simulations for the high wind speed ratios, which are exactly the ones that we're interested in if you want to focus on pedestrian-level wind comfort and wind danger. However, large deviations were obtained with these RANS turbulence models when comparing the simulated and the experimental values for the low wind speed ratios; deviations up to a factor five. Then later there was a comparison and a validation study, by Blocken and Carmeliet, who actually found a very similar thing. As you can see here, in the graph wind tunnel versus CFD, a very good agreement both quantitatively and qualitatively in high wind speed regions, but a rather poor agreement with deviations of up to factor four to five in the low wind speed regions. And that was found actually for different building configurations and different wind directions, systematically, the same finding. And the same finding actually has also been confirmed by the study of my PhD student Wendy Janssen who very patiently measured wind speed at different positions at the Campus of Eindhoven University of Technology. Then she made an extensive computational grid, and then also compared CFD simulations and measurements, finding the same result, and this time for on-site experimentation. And these are some images of the mesh that was used at that time. So finally reflecting on this finding and also taking the quotes that have been mentioned before. Well this quote by Baker is actually quite applicable, because it says that indeed CWE might be successful when looking at mean wind velocities, and this is indeed what appeared from this previous literature review of quite recent papers. And then there is this quote by Ferziger stating that: "If LES can be done on very coarse grids it will be one fo the few times that nature has been kind to us with regard to turbulent flows." Well in view of the literature findings reported just before, I would like to modify this quote, stating that: "Pedestrian-level wind comfort might be one of the few topics in CWE where nature is kind to us, with regard to turbulent flows." Because it is one that we can address with a sufficient level of accuracy with steady RANS simulations. Of course taking into account important best practice guidelines. So let's go back to the module question. Can steady RANS be sufficient for the assessment of pedestrian-level wind comfort and wind danger, possibly obviating the need to turn to LES? And why is that. Well that is, because indeed mean wind speed, not turbulence intensity, but only mean wind speed, in high wind speed areas, can be predicted with sufficient accuracy by RANS. In this module, we've learned about the advantages of CWE, the sense or nonsense of the numerical wind tunnel, and especially the nonsense, the history and progress in CWE as shown by symposia and overview papers, some concerns about RANS versus LES. And we've looked at some interesting findings from a literature study on CFD simulation of pedestrian-level wind conditions. This actually concludes our third week on Computational Fluid Dynamics, in which we've learned about possibilities and limitations of CFD. The main approaches for solving and modeling turbulent flow. We've addressed some basic aspects of discretization, and we've looked at the complexity of near-wall modeling. We've stressed the errors in CFD and the importance of verification and validation, as part of the important best practice. And we've looked at some past achievements and future challenges in Computational Wind Engineering. This week might have been a rather tough week, but it provides the basic background for the rest of the course, and I'm sure you will benefit from it in the coming weeks that will deal with applications. And we will start the next week, week four, with building aerodynamics. Thank you again for watching, and we hope to see you again, in the next week.
