Simulating a 3D Rayleigh-Bénard convection?

[sidmishfpl1801]

Can anyone please guide me how can one run simulations for Rayleigh Bernard convection.
like how to

1. Define the simulation domain
2. Define the initial conditions:
   3.Define the boundary conditions:
   4.Choose simulation parameters: such as the fluid viscosity and thermal conductivity, the timestep size, and the number of iterations to run.
   5.Set up the simulation in FluidX3D:
   6.Analyze the simulation results:

[randomwangran]

Try this: https://youtu.be/CgeEg0mnIY8

[ProjectPhysX]

Hi @qwerty740,

there is a Rayleigh-Benard setup already present in the samples. I can try to better explain it with comments.

in src/defines.hpp, comment:

//#define BENCHMARK

and uncomment

#define VOLUME_FORCE
#define TEMPERATURE
#define INTERACTIVE_GRAPHICS

Here is the more detailed setup:

void main_setup() { // Rayleigh-Benard convection
	// ######################################################### define simulation box size, viscosity and volume force ############################################################################

	/*
	// define simulation parameters in SI units
	const float si_L = 1.0f; // length scale [m]
	const float si_u = 1.0f; // some velocity [m/s]
	const float si_nu = 1.48E-5f; // kinematic shear viscosity [m²/s]
	const float si_rho = 0.997E3f; // density [kg/m³]
	const float si_T_avg = 273.0f; // average temperature in Kelvin [K]
	const float si_T_hot = 293.0f; // hot side temperature in Kelvin [K]
	const float si_T_cold = 253.0f; // cold side temperature in Kelvin [K]
	const float si_alpha = ...; // thermal diffusion coefficient [m²/s]
	const float si_beta = ...; // thermal expansion coefficient [1/K]

	// define 4 independent simulation parameters in LBM units
	const float L = 256.0f; // length scale
	const float u = 0.1f; // velocity
	const float rho = 1.0f; // density (always has to be 1 in LBM units)
	const float T_avg = 1.0f; // average temperature (always has to be 1 in LBM units)

	const float w = (float)L/3.0f;
	units.set_m_kg_s(L, u, rho, si_L, si_u, si_rho); // can be used to convert evertything except temperature
	const float K = si_T/T; // Kelvin unit conversion factor

	// convert remaining quantities to LBM units
	const float nu = units.nu(si_nu); // kinematic shear viscosity
	const float T_hot = si_T_hot/K;
	const float T_cold = si_T_cold/K;
	const float alpha = si_alpha*units.si_t(1.0f)/sq(units.si_x(1.0f));
	const float beta = si_beta*K;
	/**/

	// next is the LBM constructor; here you set all simulation pparameters in LBM units
	LBM lbm(/*resolution in x*/ 256u, /*resolution in y*/ 256u, /*resolution in z*/ 64u, /*kinematic shear viscosity (nu) in LBM units*/ 0.02f, /*volume force (gravity) in x*/ 0.0f,  /*volume force in y*/ 0.0f, /*volume force in z*/ -0.001f, /*surface tension (unused here, leave at 0*/ 0.0f, /*thermal diffusion coefficient in LBM units*/ 1.0f, /*thermal expansion coefficient in LBM units*/ 1.0f);

	// initialize geometry and boundaries in simulation box; the following loop iterates over the entire grid, and at each grid point (x, y, z)=n you can set density lbm.rho[n] (=1 if not set otherwise), velocity lbm.u.x[n] / lbm.u.y[n] / lbm.u.z[n] (=1 if not set otherwise) and temperature lbm.T[n] (=1 if not set otherwise)
	// #############################################################################################################################################################################################
	const ulong N=lbm.get_N(); const uint Nx=lbm.get_Nx(), Ny=lbm.get_Ny(), Nz=lbm.get_Nz(); for(ulong n=0ull; n<N; n++) { uint x=0u, y=0u, z=0u; lbm.coordinates(n, x, y, z);
		// ########################################################################### define geometry #############################################################################################
		lbm.u.x[n] = random_symmetric(0.015f); // initialize velocity field with some noise
		lbm.u.y[n] = random_symmetric(0.015f);
		lbm.u.z[n] = random_symmetric(0.015f);
		if(z==1u) { // set temperature boundaries at bottom of simulation box
			lbm.T[n] = 1.75f;
			lbm.flags[n] = TYPE_T;
		} else if(z==Nz-2u) { // set temperature boundaries at top of simulation box
			lbm.T[n] = 0.25f;
			lbm.flags[n] = TYPE_T;
		}
		if(z==0u||z==Nz-1u) lbm.flags[n] = TYPE_S; // very top and bottom need to be solid; lateral boundaries remain periodic
	}	// #########################################################################################################################################################################################

	lbm.run(); // run simulation with interactive graphics mode, don't write any data to the hard drive

	// alternatively: analyze simulation data
	//lbm.run(0); // initialize simulation
	//while(lbm.get_t()<10000u) {
	//	lbm.T.write_device_to_vtk(); // write temperature field to binary .vtk file in bin/export/T-000000000.vtk with the number automatically being set to the LBM time step
	//	//lbm.T.read_from_device(); // alternatively: look at only certain temperature values on the grid
	//	//println(to_string(lbm.T[lbm.index(x, y, z)])); // print temperature at coordinates (x, y, z) in console
	//	lbm.run(10u); // run some number of time steps
	//}
} /**/

Then compile and run, and press P to start the simulation and 3 to visualize temperature-colored streamlines.

[randomwangran]

checkout the paper for comparision:

DOI 10.1140/epje/i2012-12058-1