It is probably the most asked question. I want the opinions of some experts.

Q CFD vs WT. Which result is reliable for the Design of the vehicle. If WT, then what is the role of CFD in Design process?

Waiting for reply

Sorry, my correct email address is ; wblake63@hotmail.com

(1). First of all, I don't think CFD can replace the wind tunnel testing.(period) (2). In testing, there are two environments to consider, one is the almost real test environment, and the other is the simulated test environment. The hardware can be either the real hardware or the scaled model. (3). So, in wind tunnel testing, the issue is in the simulation of the real environment. Sometimes it is easy, sometimes it is difficult, and sometimes it is impossible. (4). In the CFD approach, the specification of the flow condition is in general not a problem. The computer simply has no feeling about the inlet or free stream conditions, whether it is hot or cold. In a way, the simulation of the real environment in the CFD approach is not an issue. (5). At this point, the CFD approach is slightly better than the wind tunnel testing approach, in simulating the real environment. (6). Beyond this point, it is really hard to say which approach will produce more reliable results. (7). For simple problems, CFD approach can generate highly accurate results easily on computer. But for more complex geometry, the demand for a lot of mesh points, and the lack of reliable turbulence models always create difficulties in getting a converged solution. (8). These problems in principle can be resolved in the future, for now, these are big problems. (9). The only advantage of using CFD approach now is, the cost of computer is coming down at very fast rate due to the huge PC market. So, the use of CFD is no longer limited to the super-computer class only. Simpler problems can now be simulated on workstations or PC workstations. (10). The cost of wind tunnel testing can be imporved in the rate of data handling, that is high volume of high quality data can be obtained in shorter time. Thus the cost of testing can be reduced. (11). I think, we will be talking about the optimum use of both the wind tunnel testing and CFD simulation, instead of which one can better simulate the free stream conditions. (12). What I am saying is " We have long past the era of wind tunnel VS CFD ". Since there is no short term solutions to the shortage of wind tunnel testing and CFD simulation, we are forced to combined these two approaches in the design loops. (13). In case you forgot about the shortage of these approaches, I will list a few here: wind tunnel testing, the cost of energy, the difficulty in simulating the real environment; CFD simulation, need very large mesh size to handle real configurations, turbulence models are still not reliable for realistic problems. (14). By the way, if you know how to design a better vehicle, then you probably don't care which approach you select. But if you don't, then the situation is just like having 6000 nuclear war heads in your hands, to have or not to have, itself is the big question.

it's all a matter of economics mostly. in some industries cfd is cost effective because of the high cost of experimental models. this is the case in turbomachinery. think about how you could produce a test turbine to run at the same conditions you're designing for without doing computer simulations on it. think also of the costs of rotating rigs. i suppose your question is related to the aircraft industry in that case the windtunnel is relatively (to cfd modelling) cheap, huge amounts of data can be extracted quickly and the model are cheap and quick to build. you can put a wing into the tunnel at 8 am set a 2 hr preprogrammed set of test points which would include static and dynamic tests and have your results reduced by lunchtime and do it all over again after lunch. these results would take a big ibm SP2 probably a month or so(conservative guess, probably a few months) to get. also in the wt you can just make your wing out of lexan or steel and not worry if it going to break and with rapid prototyping you can make models in minutes.

it doesn't matter though whether you use wt or CFD if your designer is skillful the results will be good if not then they won't be. the truths of plane design haven't haven't changed for a long time it's just that we can do a better job on the details. the F105 set altitude and speed records that were not bettered by another fighter plane for a long time. even today an F16 or F15 would be hard pressed to out speed a Starfighter even though the modern planes have better engines. Kelly Johnson and his Skunkwork boys just knew what they were doing

I believe it depends on your particular design problem. CFD and experiments both have their individual strength, as well as drawback...

In general each has its advantages and disadvantages; one can complement the other. Advantages of WT: higher fidelity; quicker generation and reduction of data. Disadvantages of WT:longer lead time and costlier model building; costlier of maintenance and running; wall effects; Reynold's number matching; matching free stream turbulence. Advantages of CFD: cheaper and quicker building of model; more data (velocity, density and pressure at every grid point); match Reynold's number and no wall effects. Disadvantages of CFD: discretization error; turbulence models; cannot grid every nook and cranny of model; convergence to solution time. One can use wind tunnel data to calibrate the CFD model and one can use the CFD solution as a guide to testing and instrumention of the WT model. Many times the CFD is used as the initial winnowing of possible designs which will cut down the number of WT tests. This same synergy can exist between CFD/WT/Flight test.

The answer of course is that you want both. A typical design cycle might include some preliminary investigations using CFD to confirm and help further develop intuitive design decisions, followed by experimental evidence that your CFD is in the ballpark. If not, you might refine your model and try again. The final phase should be an experimental reality check. That's the basic idea.

Bob