Wolf Can Fly, Too

I guess my last entry would be a bit hard to understand if I don't explain a few terms first.

Anyway, in CFD, the basic equation that describe the flow field can be resolved to three level of completeness : DNS, which resolves the full equation set and has the highest accuracy; LES, which resolves only the large scale flow structures and use turbulence models to simulate the rest; RANS, which takes the average of the fluid flow (average the Reynolds' Stress, in case you're wondering), thus able to bypass a decent portion of the fluid equation of motion.

Keep in mind that finding analytical solution for the entire flow field would be nearly impossible. To decrease the level of complexity, the flow channel is divided into various small cells called "grids", which we can apply simple flow equations to. Think LEGO blocks: a big LEGO castle can be taken apart into smaller pieces, if we know what each pieces look like and know how it is connected to other pieces, we can link them up and reproduce the original structure. In solid mechanics this is called finite element analysis, in fluid dynamics it is called finite volume problem.

DNS, or Direct Numerical Simulation, requires the largest amount of "grids" as we want to capture all the fluid motions. This makes the method the most computationally expensive, and the most time consuming of all simulation methods.

LES, or Large Eddy Simulation, requires a more moderate amount of grids as we are only trying to solve for the larger flow structure, while the smaller structures are simulated using previously developed models. It is a rather popular model that is pretty accurate and more affordable comparing to DNS.

RANS, or Reynolds' Average Navier-Stokes equation, requires the least amount of grids. The fluid flow equation, known as Navier-Stokes equation (or simply NS equation in most journal articles)is simplified by averaging the whole equation so such that the Reynolds' stress variation term disappears (basically it is simplified in a way that we can overlook a portion of the equation). This method is very popular, but the accuracy is questionable for some flow conditions.

Anyway, why am I typing all this? Well, I guess I just ran out of stuff to write about, plus it's a way to show that I am smarter than other people (maybe).

p.s. I'll try to run a few simple simulations tomorrow so I can show pretty pictures along with this post.