...
Initial conditions are the starting values in the domain. The closer the initial conditions are to the result values, the less time the simulation will take to solve. You should have an idea of what values you’re expecting but if you don’t know or don’t care to estimate them, you can leave the default values as isthey are. Expand the Initial conditions section by clicking the plus sign in a box on the left. I haven’t tried changing Gauge pressure , since the average pressure throughout the domain is probably close to can probably be left as zero. For velocity , most of our simulations were run at 20 m/s (but if you are interested in a different speed simply use your speed instead). This means that a lot most of the domain will have air moving close to a speed of 20 m/s, with some regions that will have with faster moving air and some other regions with slower moving air. So I use an initial velocity of -20 m/s in the z direction. For κ and ω, there are ways to estimate these values that can be found online. The values first estimate that I found were gave values of 0.303 m2/s2 and 2.871 s-1, however I have also seen an estimate that gives values closer to 4e-4 m2/s2 and 20 s-1. Both sets of values work, although the 2nd set of values are supposedly more appropriate for external aerodynamics, according to the source that published them.
| Expand | ||||
|---|---|---|---|---|
| info||||
It might be worth reviewing what’s happening in the domain. When driving a car with no wind, the air is mostly stationary and the car is moving at a velocity u. (Not all the air is stationary though, as some air is displaced by the car’s motion.) However, you may have noticed that I talk about the air moving, rather than the car moving. When simulating something like this, it’s common to use the car as your frame of reference. This means that the car will have a velocity of 0, while the air (and the road) will be travelling with a velocity of -u. |
Boundary Conditions
Boundary conditions are conditions that stay constant during the simulation. Every surface needs a boundary condition to solve the system, so we’ll create 6. Create a Velocity inlet, a Pressure outlet, 3 Walls, and a Symmetry condition.
...
Advanced Concepts and Numerics
Advanced concepts probably aren’t needed for nowsimple simulations. These will only be needed for rotating wheels, ventilation fans, or other complicated features.
Numerics is another section that I would consider as advanced settings. The defaults are selected to work for a broad range of simulations, and you shouldn’t need to make changes often.
| Expand | ||
|---|---|---|
| ||
With that being said, I will record some Numerics settings that have worked well for me (as in better than the defaults). It’s up to you if you want to use these settings. I’ve had the best convergence with:
|
Simulation Control
Simulation control is where we tell the program how long to run the simulation for, along with a few other things. An End time of about 1000 s is usually okay, for a lot of simulations I used 750 s or 800 s to get results faster. The Write interval can usually be matched to the End time, unless you want to see the results at multiple time points. 2e+4 s or 2.5e+4 s should be fine for the Maximum runtime. If it takes a little longer, that’s also fine. I have always used Potential flow initialization, as I have had simulations diverge frequently without it. With that being said, I tested this feature a while ago and I’ve changed several things since then so it may not be needed anymore. The rest of the settings can stay as defaults.
...
This is where we tell the program what quantities to save and report. Add a Forces and moments control and assign it to the car faces. We’re more interested in the total forces acting on the car than the moments acting on the car so don’t worry about setting an accurate center of mass. Of course if you have a good estimate for the center of mass, feel free to add it as that will give you a better estimate of the moments acting on the vehicle. The default write settings should be fine. Now add a Wall shear stress control under Field calculations. This is to help measure viscous forces on the car. You could also add any other controls if you’re interested in those results.
...
Now how do you know if a simulation is complete? It doesn’t really matter how long a simulation has been running or how many iterations it has completed. What matters is the convergence of the simulation. A common way to track convergence is to plot the simulation residuals at each iteration. SimScale automatically generates this plot and it can be found in the navigation panel, under the run that you started, then under Convergence plots. The goal is to have residuals that are as close to zero as possible. They will not actually reach zero though, so you have to end the simulation when you've achieved sufficient convergence. At most, your residuals should be 1e-3 to have any confidence that your results are realistic, ideally they should be in the 1e-4 to 1e-6 range. At residuals of 1e-5 to 1e-6 you can have moderate confidence that your predicted drag and lift values will match the real values (Assuming the rest of the simulation was set up properly)appropriately).
SimScale also has several other convergence plots for different regions of the simulation (domain, inlet, outlet, and walls), good convergence is indicated in these plots by achieving stable values that don’t change very much.
| Expand | ||
|---|---|---|
| ||
 Figure 3.1: A residual plot, the main tool for determining convergence.  Figure 3.2: A wall convergence plot, SimScale also creates a plot for the domain, inlet, and outlet. |
After the simulation finishes, you’ll want to process the results to make them easier to understand and interpret.