![]() | |
Fig.A-3
Experimental aerodynamic research
- as an example, a study of flow past an automobile is shown. Stream- lines are visualised by smoke particles carried with the air in a wind tunnel (Experiment performed by Takagi, Nissan Motor Comp., Japan). | |
![]() | |
Fig .A-4
The basis of a numerical solution is the division of the investigated flowfield into parts having finite dimensions: in this case, a two- -dimensional computation of flow past an aero- plane wing (airfoil ONERA M6) is performed with an unstructured adaptive grid, which becomes automatically more dense in areas where flow is more complicated (here in the region with higher velocity gradients). | |
![]() | |
Fig .A-5
Computer solutions:An exact solution of such tasks as flow past a complete aeroplane takes about 5,000 hours on a supercomputer with speed of the order of TFlops (and it is necessary to note that there are only two or three such computers around the world). Using a common large com- puter with speed around 150 MFlops is out of question - the solution time would be around 3,000 years. The only way is to use some simplification (e.g. turbulence modelling). This, however, means that the results are not perfectly reliable and it is necessary to verify them experimentally. | |
x, y, z
. The single spatial coordinate,
along which the variables will usually vary, is
here not just ![]() | |
Fig .A-6 For solution of a hydraulic system (such as the simple pipeline shown here) it is useful to start from properties of the elements comprising the system. | |
click
Vaclav TESAR : "BASIC FLUID MECHANICS"