I'm working on a CFD multiphase problem where I'm obtaining a converged solution (monitored variable). I have two questions.
1. Can the variable chosen to monitor convergence be volume averaged entity ( volume average energy dissipation rate) or a variable at a point (velocity at a point).
2. Though I'm getting a converged solution, the solution is not grid independent. i have three grids to compare the solution. (Fine, Medium and coarse). The error in solution between the fine and medium mesh is more than that of the medium and coarse grid. can anyone help out and let me know what can be the possible reason?
1. Using a domain averaged parameter is generally a good bet, as it intrinsically accounts for regions of high sensitivities/gradients, which might be helpful in deciding whether the results can be trusted.
2. There are numerous reasons why simulation results might not be trustworthy. Grid independence test is used to eliminate spatial discretization as reason for the discrepancy in the result. My advice would be to go for even finer meshes check whether the precision is achieved.
It is difficult to comment further, while the simulation method remains unnamed.
Generally speaking, CFD is based on turbulence closures that are developed in order to express unknown terms as a function of known parameters. Thus models are predictive only to the extent of the accuracy of their physical content and their validations with respect to theoretical solutions or experimental data. See references within the following research project: https://www.researchgate.net/project/Single-Phase-and-Multiphase-Turbulent-Flows-SMTF-in-Nature-and-Engineering-Applications
However and whatever the problem, we should first teach our students that there are five basic principal requirements that must be respected in model development.
1. Tensorial nature of the unknown terms: the closure of modeled terms must be carried out using tensors having the same properties as the original thems: tensor order, tensor properties (symmetry, anti-symmetry, null trace, etc).
2. Material indifference: The laws of closure should not depend on the coordinate system. This means that the laws are preserved during a frame change.
3. Dimensional analysis: The terms generated must be dimensionally correct. We then obtain the form necessary for each term up to a numerical coefficient.
4. Asymptotic behavior: the must-have solutions that relax towards expected asymptotic physical and theoretical solutions
5. Conditions of feasibility: The methodological principles set out above do not guarantee the feasibility conditions. Checking them helps prevent the development of physically unacceptable behaviors such as the appearance of negative energy, negative absolute pressure or whatever other incoherencies
- Badges
- Science topic
- Similar topics
- Mathematics