2.11. Conclusions on Energy Dissipation

The ratio of L (length of the flow) to eta (Kolmogorov length scale) is of the order of Reynolds number to the power 3/4. This implies that the length scale at which dissipation occurs is much smaller than the characteristic length scale.

In turbulent flow, the characteristic length scale (L) is the large scale of the flow, while the Kolmogorov scale (eta) is much smaller and represents the scale at which viscosity dissipates energy. The equation L/eta ∝ Re^(3/4) indicates that as the Reynolds number increases, the ratio of L to eta also increases, meaning that the small scale at which energy is dissipated becomes negligible compared to the larger flow scales.

Think of a waterfall (L) cascading into a pond. As the large falls hit the water, small ripples (eta) disperse the energy. The big waterfall represents the larger flow, while the tiny ripples show the small-scale dissipation of energy in the water. The larger the waterfall, the smaller and more numerous the ripples need to be to represent the energy loss.

Energy is passed from large vortices to smaller vortices, with dissipation occurring at the Kolmogorov length scale. Eddies transfer energy until they reach this scale where it is dissipated as heat.

In turbulent flows, energy is transferred from larger eddies (coarse structures in the flow) to smaller eddies (finer structures) in a cascading manner. As this transfer happens, the energy of the flow decreases until it reaches the Kolmogorov scale, where viscosity takes over and the energy is dissipated as heat, effectively 'smoothing out' the turbulence.

Imagine a snowball fight where larger snowballs (big eddies) break down into smaller bits of snow (small eddies). As they break apart, the energy they had as large snowballs is transferred to the smaller bits until all the energy is lost to the ground (dissipation), making the snowballs less effective as they get smaller.

The computational domain must be significantly larger than the characteristic length scale L, and the grid size must be smaller than the Kolmogorov length scale for effective simulation.

When simulating turbulent flows with computational models, it's crucial to ensure that the area being simulated is significantly larger than the sizes of the large structures (L). At the same time, the size of each grid cell used in the simulation must be smaller than the Kolmogorov scale (eta) to accurately capture the smaller energy-dissipating structures. This requirement leads to a computationally demanding process, as many computations are needed.

Consider trying to model a city (L) with a map. If the city is represented by a large enough map, but each square on the map (grid size) is too large, you won’t capture the details of the roads and buildings (smaller structures). Just like in simulations, a map that is too coarse will miss essential details; a well-planned map that is detailed enough (smaller than the city’s features) is necessary for accurate understanding.

For three-dimensional simulation of turbulent flow, at least L/eta^3 or Re^(9/4) grid points are needed, leading to a substantial computational cost, especially with high Reynolds numbers.

In direct numerical simulation (DNS) of turbulent flows, the number of grid points required increases drastically with higher Reynolds numbers. This is because the relationship L/eta^3 indicates that more points are needed to accurately resolve the smallest scales of turbulence. For instance, a Reynolds number as high as 10^4 may require billions of grid points, which can exceed current computational limits.

Imagine trying to analyze the finest details of a complex tapestry. The finer the detail becomes, the more threads (grid points) you need to understand the full picture. If each stitch represents a data point in simulation, a tapestry with a large pattern needing high detail requires an enormous number of stitches to represent it accurately, highlighting the challenges of accurate simulation in turbulence modeling.

--