We’ve talked about a decent number of topics related to wind tunnel testing over the past few weeks:
- What wind tunnels are and the forms they take
- The kinds of data tunnels provide and how this data is useful to us
- The main method of force and moment data collection, the balance
- Why testing a scale model still produces valid aerodynamic data
I know some of you might have hopped in partway through, so if you missed anything you can peruse the archive here.
It’s about time we talked about the elephant in the room, though.
With all the advances in CFD and computing technology, such as using GPUs for faster processing and AI for efficient geometry selection…do we even still need wind tunnels?
It’s precisely because CFD is a computer simulation that we need tunnels.
CFD programs work by taking a region of fluid, separating it into millions of tiny cells, and then propagating its initial conditions through each cell using the equations of conservation of mass, momentum, and energy.
Modern CFD is a wonder, but it’s still just like any computer program. It is entirely dependent on the selected settings, the starting conditions, and the physics underpinning the simulation.
These simulations are all based on our understanding of fluid dynamics. The thing is, we don’t understand all of fluid dynamics.
We can’t properly mathematically describe turbulence because it’s inherently nonlinear, random, and unsteady. We can generally account for the effects and estimate the results, but we can’t precisely solve the physics of any turbulent condition. (This is why I go easy on meteorologists: they’re making their best guesses of an inherently unpredictable phenomenon.)
This is why wind tunnels will remain useful for a while to come. You don’t have to pick a solver or set up different mesh cell shapes on surfaces to make sure the flow resolves properly.
You get a model of your aircraft, stick it on the balance, blast some air at it, and see what happens.
Physics doesn’t lie. It doesn’t predict nonsensical behavior just because you didn’t set your Mach number correctly. It doesn’t decide that a flow is too turbulent to solve. The air just does what it’s gonna do, based on what you put in front of it, and the balance collects the evidence.
Like I’ve said before, it’s best to think of CFD and wind tunnel data not as an either-or option, but as complementary data sources.
Your wind tunnel data provides the truth, which you can use to fine-tune your CFD until it too solves the same conditions and gets the same results. Any future changes can then rely just on the simulations, instead of needing another physical test.
That initial investment in a wind tunnel entry makes for much cheaper, faster analysis later down the line, because you know your CFD data can be believed.
So to answer my own question: yes, we still need wind tunnels. No, I don’t think they’ll be replaced anytime soon. Eventually someone will solve turbulence, and maybe then all the world’s wind tunnels will fall silent. But it is not this day.