# Why is Rocket Nozzle CFD done in 2D instead of 3D?

Why do I see rocket nozzle CFD like the one here in this SpaceX video done in 2D?

Real Fluid flow is three dimensional and involves an x, y and z axis. Is fluid flow done in 2D only because computers are not powerful enough to run 3d fluid flow simulation? Or is CFD done in 2D because that gives an accurate enough real world approximation? Or is it a bit of both.

• Freshman majoring in Aerospace engineering.
• Which part of the video are you asking about? People don't have time to sit through a 45 minute video about the SpaceX Mars project just to find a picture of a CFD simulation. Mar 24 '19 at 9:50

## 2 Answers

It isn't really done in 2D. It is done in pseudo 3D. The actual annular areas are factored in. It is primarily based on a uniformity assumption that the gas dynamics are the same all around the nozzle for any constant axial and radial coordinate value. But all the physics are correct for 3D. In the analytic kernel, the uniformity assumption allows you to extract the azmuthal terms because they are constant, and typically very small in this case compared to the axial and radial terms.

Note that gas injection thrust vectoring destroys this symmetry and is used to steer a fixed nozzle rocket.

I think the full answer is a little of both. When choosing to use CFD for any analysis there is always a trade-off between accuracy vs. cost.

Let's take the rocket nozzle for example. The real physics of the problem include effects such as unsteadiness, turbulence which is inherently 3D, combustion, reacting chemicals for the combustion products etc. These can indeed be simulated with today's computers but as you add physics and complexity the computational cost rises. This sets the limit on the cost side.

However, let's say we really just want/need to understand if the nozzle is over/under expanded. Well then a simple 2D axisymmetric analysis (which is far cheaper computationally) will suffice. If we do care about things like how the reacting combustion products affect nozzle efficiency then those physics need to be included and consequently the computation cost rises as our level of accuracy increases. This sets the limit on the accuracy side. That is, what level of answer is "good enough"?

Rounding back to your question, it is a bit of both where the level of analysis is the result of the balance of cost and accuracy desired.