[BLANK_AUDIO] Welcome back to Sports & Building Aerodynamics. This is week 3 on Computational Fluid Dynamics. At the end of this week, you will understand the possibilities and limitations of CFD. You will understand the main approaches for solving and modeling turbulent flow. You will understand some basic aspects of discretization, the complexity of near-wall modeling. You'll understand errors in CFD and the importance of verification and validation as part of best practice. And finally, you will understand past achievements and future challenges in Computational Wind Engineering. This week provides the basic background for the rest of the course. We start again with a quote. This time from Sophocles saying: "wisdom outweighs any wealth". These are the contents of week 3. We start with answering three basic questions about Computational Fluid Dynamics namely, what, why, and how? Then we'll have a look at the approximate forms of the Navier-Stokes equations, move on to turbulence modeling, some aspects of discretization, then near-wall modeling. After that, errors in CFD and best practice guidelines. And we conclude with two modules on Computational Wind Engineering. Let's start again with the module question. What constitutes the main power of CFD? Is that A, CFD provides whole-flow field data, B, CFD is inexpensive and fast, C, CFD is highly accurate, or D, CFD allows us to solve flow problems that cannot be solved in any other way. Please hang on to your answer and we'll come back to this later in this module. At the end of this module, you will understand what is meant by Computational Fluid Dynamics, you'll understand the possibilities and limitations of CFD, and you'll understand the general procedure or outline of a CFD simulation. Let's start first with the first question. What is Computational Fluid Dynamics? Well maybe the shortest definition is, it is solving fluid flow problems numerically. This however is a more descriptive and a very nice definition: CFD is the art of replacing the integrals or the partial derivatives (as the case may be) in the Navier-Stokes equations by discretized algebraic forms, which in turn are solved to obtain numbers for the flow field values at discrete points in time and/or space. So Computational Fluid Dynamics actually is a tool that indeed allows us to solve flow problems that do not have known analytical solutions, and that cannot be solved in any other way. So why would we use Computational Fluid Dynamics? Well, there are different reasons for doing that. First of all, for understanding and interpreting, and this is sometimes called numerical experiments. Experiments that are done purely with CFD to provide information about experiments that for some reason are not possible, or maybe in addition to experiments, a limited set of experiments. But you can also apply CFD for designing, for designing future objects on which experiments are not yet possible, simply because they do not yet exist. There are some particular advantages to CFD. It is stated often that it's relatively inexpensive and fast. Although one could argue about that, but indeed computational costs do decrease as a function of time. CFD provides so-called complete information, sometimes also called whole-flow field data. You get all information or you get information about all relevant variables in the whole computational domain. It also easily allows parametric studies, which is certainly important in design cycles. It does not have similarity constraints, and that is important because in the week on wind-tunnel testing, we noted that similarity constraints can indeed be quite restrictive. Certainly in building aerodynamics. So here CFD simulations can be performed at full scale, so no similarity problems. And then it allows numerical experiments. For example, you can study explosions, failures and so on, attacks, and these are certainly things you do not want to reproduce in reality. There are also quite some disadvantages with CFD. Accuracy and reliability are really major concerns because it's not too difficult to ruin the accuracy of a CFD simulation. The results are also very sensitive to the large number of parameters that have to be set by the user. And therefore, verification and validation are really very, very important. And validation then in turn requires experiments, and that is the reason why CFD in sports and building aerodynamics cannot be a stand-alone tool. Some examples here. Understanding and interpreting, for example can be applied in sports aerodynamics, and that we will do in week six, where, based on Computational Fluid Dynamics, validated with a limited set of experiments, we will focus in detail, on aerodynamic effects in team time-trial cycling. Another example of a numerical experiment is the one that we addressed in the first module with analyzing laminar and turbulent flow around an airfoil which gave quite surprising results. Concerning designing for example, in week four we'll see that CFD can be used for example to redesign part of the university campus, which we did here at Eindhoven University. But also, to design and to explore potential future ventilation configurations for complex buildings such as multifunctional football stadia. This is another nice example of how CFD has been very effective and very efficient in designing. And this is the HiMAT project and on that I would like to briefly read this quote to you. So in the late 1970s, this approach, and that refers to the use of supercomputers to solve aerodynamic problems, began to pay off. One early success was the experimental NASA aircraft called HiMAT, Highly Maneuverable Aircraft Technology, designed to test concepts of high maneuverability for the next generation of fighter planes. Wind-tunnel tests of a preliminary design for HiMAT showed that it would have unacceptable drag at speeds near the speed of sound. If built that way, the plane would be unable to provide any useful data. The cost of redesigning it in future wind-tunnel tests would have been around $150,000, and would have unacceptably delayed the project. Instead, the wing was redesigned by a computer at a cost of only $6,000. And this indeed is a very nice indication of how CFD can make a difference when supported in the right way and when used in the right way to supplement experimental testing. Final question on this module, how? Well generally, a CFD simulation consists of three steps. There's a preprocessing step, a solution step, and a postprocessing step. And let's just briefly run through these steps together for an example, aerodynamics of a single cyclist, which we will focus on more, in much more detail in week six. But here just as an illustration of the procedure. First, we have to select target variables. What are we interested in? Well for a cyclist that will be aerodynamic drag, the drag force expressed in Newton for example. Then we have to select approximate equations to describe the physics of the flow. That is because the full Navier-Stokes equations are too complex, and it would be too computationally intensive to solve them completely, exactly. So we're going to apply a simplified approach, the Reynolds-averaged Navier-Stokes equations, with some turbulence models. Then we select the model geometry, in this case by 3D laser scanning, smoothing out some details. We select the computational domain, put the model in the computational domain, apply appropriate boundary conditions, we generate the computational grid. And this is a step that really is very important and generally also takes the largest part of the user time in a CFD simulation. So a careful computational grid has to be made, based on best practice guidelines. And then we can start with the solution. We have to first initialize the flow field. Then we select numerical approximations and solution algorithms. And this is important because what we do, in the control volume method, and that's the method that we will use here, is that we're going to integrate the governing equations, and also the turbulence model equations for every control volume, that way we are going to arrive at discretized algebraic forms, and those we are going to solve iteratively for every control volume. And then of course you also have to select convergence criteria, also there we have to be very careful. And then finally we can start postprocessing. Visualizing data, which can be for the cyclist for example, air speed around the cyclist, but also pressure on the body or pressure fields around the body, and this way we gain a lot of information about aerodynamic processes in cycling. Finally, last but certainly not least, error analysis is important. We have to demonstrate how accurate our results are. And if not, if they are not accurate enough, we have to step back into the cycle, and repeat some important steps. So solution verification and validation here are truly essential. Let's go back to the module question. What constitutes the main power of CFD? Well it's definitely true that CFD provides whole-flow field data. But this is also something that could be obtained from for example PIV measurements, or sand erosion techniques in the wind tunnel. You could argue about the fact that CFD is inexpensive and fast, and its definitely not always highly accurate. This really depends on the type of problem and also on the expertise of the user and on the type of models that he or she will apply. But it's definitely true that CFD allows us to solve problems that cannot be solved in any other way. In this module, we've learned about what is meant by Computational Fluid Dynamics. We've also learned about the possibilities and limitations of CFD, and we've had a short look at a general procedure or outline of a CFD simulation. In the next module, we're going to focus on the difference between RANS, URANS, and LES. And we're going to see why steady RANS and LES are the most often used approaches in sports and building aerodynamics, as opposed to unsteady RANS and Detached-Eddy Simulation. Thank you for watching, and we hope to see you again in the next module. [BLANK_AUDIO]