Computational Fluid Dynamics, or CFD, is a computer modelling technique that analyzes the flow of fluid. And not just liquid fluids like water or oil, but any substance that can flow and move to fill the space it’s in. This means fluids are also gases like air or helium, but also more counterintuitive things like sand draining in an hourglass or concrete in a mixing truck. The list of applications CFD is used for is too long to list here, but some more examples are simulating airplanes or cars in a wind tunnel to design bodies without having to build them first. Or using CFD to aid in the design of more efficient combustion engines and wind turbines. An even more exotic use is simulating blood flow in the human body to aid in the diagnosis and treatment of cardiovascular disease.  These scenarios all have to do with fluids, and so all of them can be simulated using CFD.

At Smith Board Co, were not as complex as simulating a whole humans blood flow, but we can and do use CFD to simulate how our boards behave in the water. We can simulate different board sizes and then analyze that data to understand how the width or length affects speed and handling. We can do the same for different shapes, rockers or anything else you can think of changing. In fact, the Curie was made because we were simulating different and weird ideas and found that a peanut shape “pushed” water away at the rear end and resulted in significantly increased speed and control compared to a more conventional shape. Figure 1 shows how the water flows slightly outward as it flows past the end of the board, whereas conventional boards with single curve “hug” the water so it meets at the back tip of the board. This leads to vortexes, decreased speed, and decreased control with a more conventional shape. CFD allowed us to discover this without having to build a whole board just to test our random idea! And It’s not just us, CFD has allowed aerospace companies to simulate new ideas for wings and rocket designs that never would have been built just for being a little weird.

How's it Work?

So, we’ve covered a little of how CFD is used, but how’s it work? In high level gobbledygook, CFD uses numerical analysis and algorithms to solve and analyze problems involving fluid flows by discretizing the equations that govern fluid flow. In less technical terms, that essentially means that we can use a computer to solve the same equation over and over at every different point in our fluid. Think of breaking our 3D volume of liquid into a bunch of tiny cubes. If our fluid is flowing, then some portion of that fluid will be flowing through each of these tiny cubes. This can be seen in Figure 2, where the cube represents one of these small volumes of fluid. The arrows pointing into the cube represent the speed and volume of fluid flowing in, and the arrows pointing away from the cube represent the speed and volume of fluid flowing out. Inside that cube, the computer solves an equation (Navier-Stokes equation) that tells us how much of those incoming arrows should go to each of the outgoing arrows. This picture only shows a single cube, but in an actual simulation, these cubes are stacked one on top of another, to fill up the whole volume. That means that the arrows coming into this cube come from another cube and the arrows going out of this cube go into another cube and on and on and on. To solve the equation in the current cube, all the adjacent cubes that contribute an incoming arrow have to be solved before the equation can be solved for the current cube.

 

Since fluid flow happens over time, in order to incorporate time into our simulation we have to repeat the step of solving every single cube for every time step. To put all of this into perspective, if we split our volume into 1000 tiny cubes and run our simulation for an hour with a time step of one second, we’d be solving the fluid dynamics equation 3.6 million times. To get the detail that is needed for most simulations though, the volume is split into millions of little cubes and the time step is one millisecond (.001 second). That means an average simulation is solving billions upon billions of equations. Because of the sheer number of calculations that have to be performed, CFD simulations, even run on powerful computers, usually take hours or days to complete. If we had to solve this by hand, even with thousands of people this would have taken longer than building the Pyramids.

Understanding fluid flow is hard. The fluid can flow up, down, diagonal, any direction really. And all the while flowing past or interacting with other bits of fluid moving in just the same any which way. If you’ve heard that a butterfly can flap its wings and cause a hurricane half a world away, the same thing is happening here. A single bit of fluid can set off a chain reaction that changes how every other bit moves. Because of this predicting it at a high level has been impossible for as long as scientists have been trying. Same for the butterfly and weather come to think of it. By breaking that big volume that the fluids flowing through into small, discretized (think the small cubes from before) parts, CFD allows us to use the power of computers to model what will happen rather than trying to predict it. Because of this, designs can be tested before they are made. Ideas can be implemented with the click of a mouse rather than the production of a prototype. And understanding and insight can be discovered where once it wouldn’t have had a chance.