This was an analysis I was very excited to perform, it takes a real life example and actually lets you play around with it. While more classical problems may have critical and broad implications, I like looking into specific problems that give me a real life example which makes a lot of it more intuitive. As I’ve learnt, we start with the prettiest of pictures, here’s the big picture of the results
Mesh is a very important part of performing a CFD analysis. It dictates the quality and resolution of the solution, I think that we have more wiggle room in the solution setup than the mesh setup.
First thing we need to note is the size of our computational domain.
You see the long trail of mesh that exists after the car projection, it’s about 5 times of the length of the car. The reason it’s there is that the effects of the car zooming through the air last longer than just where the car ends. As can be seen in fig 1, the disturbances in the flow continue more than 2 car lengths, and are important for our analysis
Something to notice is that the whole mesh is triangles, the reason for this is that I was able to create a fine enough (as seen in fig 3) mesh around the geometry and decided to work smart rather than hard on the mesh formation.
a medium zoom on the car geometry, you can see the tapering of the size of the triangle as you get further away from the geometry.
a little bit more zoom, each of the front facing faces has 500 divisions, which causes the mesh to be very fine at the front as can be seen in fig 5.
B.C Boundary conditions:
- The left vertical wall is the Inlet velocity and it is situated normal to the line towards the inside of our computational domain, at the speed of 15.7 m/s, about 56 km/h.
- the right vertical wall is our environment pressure and is set to 0 gauge (1 [atm] absolute)
- Pressure 0.3
- Density 1
- Body force 1
- Momentum 0.7
We did a standard initialization with 0 for all values and ran the calculation for 2000 iterations, and received convergence around 1700 iterations with a stop condition of 0.001 (10^(-3)) for continuity, x velocity and y velocity.
These are the results I like the most, I like the contour more than the streamlines for this kind of large geometries because I like seeing the trend more than the specific local detail. I keep the local differences for another kind of answers to other kind of questions.
Here we have the overlook on the pressure profile, high pressure being created at the front of the car, and low pressure left behind it, which converges with what we would think beforehand according to Bernoulli’s equation and all those buzzwords.
The solution of this problem provides a good overlook on aerodynamics, velocity turbulence, air trail after car. Now, of course, these are not very scientifically accurate and I haven’t extracted any aerodynamic coefficient from this study nor do I have the Reynolds number at any point (although it can be easily extracted for a given point using the known viscosity, temperature, density and so on.
This is a case of giving me intuition for how the fluid behaves in certain scenarios. I believe the intuition is important when approaching problems in the future, when you know a lot of simple cases, you can take a big complex problem and break it down to smaller cases and thus simplify the problem.
For example, if you ever stopped on a highway (on the side of the road naturally) and other cars flashed by, you always feel something moving your car from side to side, so that is probably the air that the other car is pushing to its side. then when the car has passed already, it creates a low pressure trail that pulls the car, that’s kind of interesting for me.
another thing we can see here is that a vertical front of the car is not very efficient with aerodynamics, it creates a very high pressure area, meaning higher resistance, as well as the two “corners” the one at the top of the hood and another at the top of the windshield