combustion model for deflagration of methane-air-mixture

- January 25, 2020 at 2:17 pm
  Sandro
  Subscriber
  Hello, everybody,
  
  I am working on the simulation of an combustion of methane-air mixture with FLUENT. It is a stationary mixture which is to be ignited.
  
  There have been some problems with the calculation, which I could not yet solve myself:
  
  element size 0.2... 2 mm (time step length 1e-07 s to 5e-06 s)
  
  There is always an abort with the error message "Floating point exception”
  - The reaction velocity depends on the time step length: the shorter the time step, the slower the reaction respectively the area with high temperature becomes smaller and smaller after the patch, temperature decreases - reaction front does not spread
  element size 20 mm (time step length 1e-05 s to 5e-05 s)
  
  The reaction velocity depends on the time step length: the shorter the time step, the slower the reaction respectively the area with high temperature becomes smaller and smaller after the patch, temperature decreases - reaction front does not spread
  - Ignition: the longer the time step, the larger the temperature patch (energy input) must be to get the reaction startet
  - No pressure increase if fluid is surrounded on all sides by wall. My expectation was that the pressure would increase strongly due to the closed fluid space. However, this was not the case. The pressure was in the same order of magnitude as in the "open" model.
  Does anyone have experience with transient combustion simulation - especially deflagrations? I am grateful for any advice - basic problems, choice of turbulence and combustion models, physical parameters, solver settings, etc.
  
  Thanks a lot
  
  Best regards
  
  Sandro
- January 27, 2020 at 5:42 am
  
  DrAmine
  Ansys Employee
  
  Deflagration processes are expensive and require small time steps. What happens if you decrease time step further? Which models are on?
- January 27, 2020 at 1:33 pm
  Sandro
  Subscriber
  Many thanks for the quick answer.
  
  I'm going to calculate the model with 20 mm elements using smaller time steps than 1e-05 during next days.
  
  I use the following settings:
  - Transient, pressure-based
  - Energy equation is on
  - Turbulence model: k-omega
  - Species transport with eddy-dissipation-concept (with default settings from Ansis)
  - Material: methane-air-mixture (with default settings from Ansys)
  - Setting the mass fractions of methane and oxygene by patching the value after initialization
  - Ignition by patching a area with high temperature (2700 K) after initialization
  Best regards
  
  Sandro
- January 27, 2020 at 2:22 pm
  
  DrAmine
  Ansys Employee
  
  Sound Good. I guess with SBES turbulence model (Hybrid LES-RANS) your probably reach better result. However the grid will require to be a LES quality.
- February 3, 2020 at 9:31 am
  Sandro
  Subscriber
  Hello Amine,
  
  I have tested the proposed variants. I have received the following results
  1. Simulation with smaller time steps:
  - Used time steps: 1e-06 s to 8e-06 s
  - No adjustment in setup - parameters and turbulence model / combustion model as before
  - Results:
    
    The mixture does not ignite - the reaction does not start, the high temperature area (patch) mixes with the low temperature areas and the temperature drops
    
    I repeated my test with a time step of 1e-05 s - this seems to be the smallest possible time step where ignition occurs under the prevailing conditions.
  1. Simulation with adaptation of the turbulence model
  - In my model, I only adjusted the turbulence model: Viscosity - k-omega - BSL - SBES (or rather k-omega - GEKO - SBES), all other parameters as default.
  - A message appears: Recommendation to set "bounded second order implicit" for transient formulation (I did).
  - The simulation runs basically, the mixture does not ignite - the high temperature area of the patch mixes very quickly with the rest of the mixture.
  - If all boundaries of the fluid space are defined as "wall", the calculation terminates with the error "floating point exception".
  Best regards
  
  Sandro
- February 3, 2020 at 11:06 am
  
  DrAmine
  Ansys Employee
  
  Focus one variant and check the chemistry. You can even start from non-premixed combustion or the simple EBU model which could provide a good initial guess for the stiff chemistry solver. Option Relax to chemical equilibrium might be helpful.