The last category of tools we’ll talk about is probably the heftiest one, for a number of reasons. These are the 3D viscous tools.
When you hear “aerodynamic analysis,” these are likely what springs to mind: computer-generated images of aircraft or cars, with swirling colored lines streaming over them.
With a few exceptions, this category of tools is dominated by computational fluid dynamics solvers, or CFD. CFD actually can do everything that the previous tools could—you can run inviscid or viscous 2D simulations, and definitely inviscid 3D ones.
But as we’ve established, there are cheaper and faster tools for those sorts of analyses, so CFD is most often used for modeling and analyzing situations where air viscosity really matters.
So how is a CFD simulation able to model this behavior, when previous tools couldn’t?
I mentioned it briefly when talking about MSES, but CFD works by modeling the fluid moving over and around an object. The computer-generated model of your aircraft is actually treated as a solid surface that the fluid interacts with.
This fluid body is split into a mesh, made of hundreds of thousands to millions of individual cells. Each cell is basically a tiny little piece of the entire fluid being analyzed. And each cell has its own physical properties: density, pressure, speed and direction of motion, etc.
A user sets the starting properties of this flow far upstream of the geometry they want to analyze; this would be like dictating the static pressure, flow speed, temperature, and other quantities at a wind tunnel. When they hit “run” on their simulation, the software solves a series of equations to propagate those starting conditions into the next layer of cells. And then the next layer. And the next.
When the simulation reaches the cells that are close enough to start interacting with the aircraft geometry, it solves for how that interaction changes the cells’ properties. Maybe the fluid starts to move in a different direction. Or the pressure increases. And that change is further evolved in each cell of the fluid body, until the equations for each cell are solved well enough that the whole simulation can be considered to have “converged” on an aerodynamic flow solution.
This is why CFD software is expensive, and why using it is extremely computationally intensive and time-consuming. It takes time and processing power to calculate the physical properties and interactions of the, again, millions of cells in a single simulation.
A reasonably sized mesh for a UAV is usually around four million cells or so. That’s a whole lot of equations to solve!