Learning Forum

[omkarmadhekar]

I m trying to simulate a nebulizer head nDensity-based solver steady-state condition. nModel - Turbulent K-omega- SSTnMaterial air Ideal gas nBoundary conditions inlet and outlet are given as mass flow rates. n Solution method Implicit ( ALl the parameters are as default)nControl method defaultnResiduals continuity 1e-6 rest other are defaultnHybrid initialization nn

[Atgeirr]

I am by no means an experienced Fluent user but I have had residuals look similar to yours, so I'll try to explain what I learned so far.nFirst, looking for convergence by only residuals is not always the best practice. Try and also plot mass flow rate flux and/or other parameters you know are important for your case. If these plots remain flat it's also an indication that the simulation is converged. Residuals aren't the be all end all of finding convergence, it's just an easy way for the program to measure convergence.nOscillating residuals can hint at your case being of transient nature. You haven't specified if you are simulating steady-state or transient, however, it could be an indication. nK-omega SST is one of the harder viscous models to converge, especially continuity. It needs accurate ''first guess'' inputs to successfully converge in a timely manner. Try using the K-epsilon model to find those first guesses more accurately.nDo you have any test data available to check your simulation against? If so, check your simulation against that data to see if you're getting comparable results. If results are comparable, you've done a good job whether or not your residuals make you think otherwise.nK-omega SST also requires pretty dense meshing and a correct Y+ (first layer height) value.nn

[omkarmadhekar]

No, I don't have any test data to check my simulation. n

[omkarmadhekar]

My meshing is dense with almost 0.5 million cells and a skewness factor of 0.81 My simulation is to be solved in density-based solver n

[Rob]

Not sure if you need the density based solver, pressure based is good to about Mach 5 if you use the pressure based coupled solver. Historically we switched to density based at about Mach 2-3 but that's changed with the more recent builds. n0.5M cells doesn't tell us if the mesh is dense, or whether it's focused in the right areas. Look at variable plots, as suggests, are these level? Also try reducing the Courant Number and time step. You're resolving the flow in time and space with a transient run. n

[omkarmadhekar]

I m solving the simulation in a steady state. n

[Rob]

In which case switch to transient if the error is high: you've picked up some transient effects in the solution. n