2.12. Dimensions of Computational Domain

To sustain the fluctuations, the supply of kinetic energy must be balanced by the dissipation of turbulent energy. This is an important consideration in understanding the dynamics of turbulent flows.

In turbulent flows, the kinetic energy generated needs to be managed. This balance means that the energy created through turbulence must equal the energy that is lost due to dissipation processes like heat. If the energy created exceeds the energy dissipated, the turbulence may escalate uncontrollably.

Imagine a crowded dance floor where people are pushing against each other (creating turbulence). If more people keep entering the floor (supply of kinetic energy) without anyone leaving, it gets chaotic. However, if a few people step off the dance floor (energy dissipation), the chaos can be balanced, preventing a congestion disaster.

From the above consideration, what we get is L/eta is of the order of Reynolds number to the power 3/4, where eta is the Kolmogorov length scale, which is the length scale at which the energy is dissipated.

The ratio of the characteristic length scale (L) to the Kolmogorov length scale (η) indicates how the energy dissipation scales with the Reynolds number (Re). As the Reynolds number increases (indicating more turbulence), the size at which energy dissipates (η) decreases significantly compared to L. This suggests that turbulent structures are much smaller than the general flow structures.

Think of a river (L) with very tiny pebbles (η) on the riverbed. As the water flows faster (high Reynolds number), the pebbles create turbulence but are much smaller compared to the river itself. The energy that disrupts the flow is dissipated at these tiny scales, showing how turbulent dynamics can differ greatly in size from the flow itself.

To simulate all the scales in turbulent flow, the computational domain must be sufficiently larger than the characteristic length scale, L, and the grid size must be smaller than the Kolmogorov length scale, η.

When performing simulations of turbulent flow, the area being studied must encompass enough space to capture larger flow structures. At the same time, the resolution of the simulation (the grid size) needs to be fine enough to account for the tiny eddies where dissipation occurs. This double requirement ensures that the complexities of turbulent flows are accurately represented without missing critical interactions.

Imagine trying to model a large forest fire. If your model (computational domain) is only the size of a small room, you won’t see how the fire spreads (the characteristic length scale). However, if your model is the right size but your resolution (grid size) is too coarse, you wouldn’t capture the tiny embers that drift away and start new fires (Kolmogorov scale). Both size and detail matter for a good simulation.

For a three-dimensional simulation of turbulent flow, at least L/eta to the power of 3 grid points will be required, indicating a high computational cost.

The number of grid points necessary for the simulation increases dramatically with the Reynolds number. This relationship suggests that modeling more turbulent flows requires exponentially more computational resources since the grid must refine to capture smaller scales of turbulence effectively.

Consider building a 3D model of a city. To capture the tiniest details like street signs and trees (small turbulence scales), you need thousands of tiny blocks. If you want to represent a bustling metropolis accurately (high Reynolds number), you’ll need an enormous number of blocks (grid points) to account for all the buildings and details, which can overwhelm typical computer capacity.

--