- #1
- 1,433
- 7
When one tries to have an insight into what the turbulence regime really is, one comes across the question of: is the turbulence a deterministic process or not? Even though we are able to simulate accurately turbulence with Direct Numerical Simulation (DNS) of the Navier-Stokes Equations, it is very difficult to test the results unless done assumming an statistically steady state. What I mean is that even though the DNS and its results are showing the major turbulent coherent structures and it comes from a deterministic model (N-S equations), it is very difficult to test its results with experiments unless the numerical results are averaged and compared with those flow experimentally averaged quantities.
Nowadays fluid dynamicists are prone to think that a turbulent process is governed by a set of deterministic equations such as N-S equations. That is, the process itself seems to be deterministic (how the coherent structures evolutes in time in space seems to have a physical explanation), but it is not quite the case of how the turbulence is triggered. The onset of turbulence seems to me to be influenciated by lot of random variables. I mean, even though we are able to use DNS, the whole key stays in the Initial and Boundary Conditions of integration. Those are kind of random and extraordinarily difficult to account for. The IC and BC in a laboratory facility may be impossible to simulate because they could belong to the randomness of some environmental conditions. As the Reynolds number becomes large, the sensivity of the flow to these small perturbations grows enormously, triggering the global instability.
What about the Lorenz Attractor http://mathworld.wolfram.com/LorenzAttractor.html?? Lorenz worked out this model thinking of the instabilty of a pure convective regime (large Rayleigh Number). It seems to me that this system is a generator of randomness even it is "deterministically determined" by a set of differential equations. My question is, what is triggering this apparently chaotic behavior?? i) Is it because the proper nature of the equations? ii) Or is it because a numerical instability impossible to overcome with nowadays methods??.
If i) is true, the lorenz attractor seems to me contradictory, because then a mathematical model is not as deterministic as I thought. But if ii) is true, that is the instability is caused by the numerical integration, I think it is telling us that despites we are unable of integrate the equations properly, the process is deterministic at the bottom. The Lorenz attractor seems a great comparison with the N-S equations at high Reynolds Numbers. Are they instable per se, generators of randomness, or on the contrary are we unable of integrating the properly propagating numerical errors which trigger finally the instability??.
Nowadays fluid dynamicists are prone to think that a turbulent process is governed by a set of deterministic equations such as N-S equations. That is, the process itself seems to be deterministic (how the coherent structures evolutes in time in space seems to have a physical explanation), but it is not quite the case of how the turbulence is triggered. The onset of turbulence seems to me to be influenciated by lot of random variables. I mean, even though we are able to use DNS, the whole key stays in the Initial and Boundary Conditions of integration. Those are kind of random and extraordinarily difficult to account for. The IC and BC in a laboratory facility may be impossible to simulate because they could belong to the randomness of some environmental conditions. As the Reynolds number becomes large, the sensivity of the flow to these small perturbations grows enormously, triggering the global instability.
What about the Lorenz Attractor http://mathworld.wolfram.com/LorenzAttractor.html?? Lorenz worked out this model thinking of the instabilty of a pure convective regime (large Rayleigh Number). It seems to me that this system is a generator of randomness even it is "deterministically determined" by a set of differential equations. My question is, what is triggering this apparently chaotic behavior?? i) Is it because the proper nature of the equations? ii) Or is it because a numerical instability impossible to overcome with nowadays methods??.
If i) is true, the lorenz attractor seems to me contradictory, because then a mathematical model is not as deterministic as I thought. But if ii) is true, that is the instability is caused by the numerical integration, I think it is telling us that despites we are unable of integrate the equations properly, the process is deterministic at the bottom. The Lorenz attractor seems a great comparison with the N-S equations at high Reynolds Numbers. Are they instable per se, generators of randomness, or on the contrary are we unable of integrating the properly propagating numerical errors which trigger finally the instability??.
