[Xun]

Hallo, everyone

I recently tried to simulate a acetylen-oxy combustion in CFX using the reaction model Acetylen-Air-WGS.

i successfully finished a transient simulation with total time about 3e-3 s for distribustion of pre-mixed oxy-fuel in whole space and then ignition for 1e-5 s. The simulation at this stage seemed ok, but when i used this result as the initial condition for a steady state simulation, everytime when it reached the residual target, the combustion was extinct, and it looked just like a normal flow mixing simulation.

i have tried with different timestep setting, but it is still unsolved

so i want to ask, is it impossible to solve a steady state simulation for this?

hope anyone can give me some advice, thanks!

[Xun]

could anyone pls answer my question?

[Opaque]

Are you aware that cold flow is a solution to the problem?

You need to "numerically ignite the flow", then if the stars and the moon align correctly, the solver may converge to the "hot solution"

[Xun]

Originally Posted by Opaque

Are you aware that cold flow is a solution to the problem?

You need to "numerically ignite the flow", then if the stars and the moon align correctly, the solver may converge to the "hot solution"

thanks for your reply, sir. i thought you are right. I used the result file from a succesful combustion simulation as initial condition(0.003s cold flow+ 1e-5s combustion for the ignite stage, and it could reach over 3000 °C). but i didnt expect the cold flow situation like u said.

so, i'd like to ask for one more advice, what can i do in set-up to prevent the hot flow status from turning to the cold flow? i used extinction teperature 298k in combustion model but it seemed not to work. should i set a auto ignition model with time delay?

[Xun]

and another question, errors can occur in my transient combustion simulation if i didnt use conservation target 0.01. is that normal for this kind of simulations?

[Opaque]

Originally Posted by Xun

and another question, errors can occur in my transient combustion simulation if i didnt use conservation target 0.01. is that normal for this kind of simulations?

You are referring to numerical errors, correct?

If the conservation target is not satisfied, and it is large then there is no conservation of the quantity for that transport equation and the results are suspect at the least.

Recall you are trying to conserve mass (continuity and mass fraction equations), energy, and linear momentum. If they are not conserved, what is the point of the simulation?

[Xun]

got it, sir.

pls correct me if i have misunderstandings here:
1.my domain has the boundaries of inlet, opening, and the walls, so there is no reason that the imbalance of mass surpass 100%.

2.i used also residual target 1e-4, but when iteration reached this value, the imbalance could usually reach +or-100% and that was clearly not the solution, so now i need to let every iteration also reach an imbalance target 0.01 to get the right solution.

To be honest, i rarely saw this situation before, thus i am not so sure about how meaningful were my settings in solver.
sincerely thanks to all your replies!

[Opaque]

Originally Posted by Xun

got it, sir.

pls correct me if i have misunderstandings here:
1.my domain has the boundaries of inlet, opening, and the walls, so there is no reason that the imbalance of mass surpass 100%.

Imbalance also accounts for the accumulation (time derivative) of the transport equation. Therefore, for steady-state should be tolerable small (theoretically 0) when converged, and for transient is should be tolerable per timestep.

Which case were you referring to when mentioning +/- 100% imbalances? steady or transient?

[Xun]

all the imbalance issues i mentioned, happend in my transient simulations.

for example, if i restrict no conservation target, the imbalance monitor can reach 100% after reaching residual target and may jump to -100% after several timesteps, and soon the simulation may end with errors like zero-divide.
and when i set conservation target 0.01, simulation can be finished.

i am not confident of my knowledge in combustion simulation, so i would like ask people with more experience, to make sure my conservation target setting is positive to my simulation, and this issue is not caused by setup setting.

[Opaque]

Originally Posted by Xun

all the imbalance issues i mentioned, happend in my transient simulations.

for example, if i restrict no conservation target, the imbalance monitor can reach 100% after reaching residual target and may jump to -100% after several timesteps, and soon the simulation may end with errors like zero-divide.
and when i set conservation target 0.01, simulation can be finished.

i am not confident of my knowledge in combustion simulation, so i would like ask people with more experience, to make sure my conservation target setting is positive to my simulation, and this issue is not caused by setup setting.

For transient simulations, you want low imbalances; otherwise, the errors in the previous time step pollute your transient.

Say the continuity imbalance is positive, it means your numerical error injected mass randomly into the flow while if it is negative you removed mass randomly. Neither is a good step towards an accurate simulation.

Same example can be used with energy, and so on.

[Xun]

thank you so much for your answer. i guess i am sure about the setting to let my transient simulation get converge solution.

Now my tutor wants me to simulate this combustion with longer total time, like 1 second or just show him the stable accetylene-oxy flame as result. (now what i got reasult from was only transient simulation with total time 1e-5s)
So i hope that u could give me more advices cause i am lack of experience, and so far all my attempts were failed. i thought there are two possibilities. please correct me if i said sth wrong here, and i really want to know your suggestions:

1.when i try transient simulation, the expense of transient simulation is too high, because the timestep about 1e-8 or 1e-9 is too small in comparison with total time 1 second. even worse, many of my results ended due to divergence(thats why i ask you about conservation target setting) and i didnt run the long enough, so i have no idea if the timestep later are larger in a 1 second simulation. For example, i use adaptive time step, but the simulation so far never used an increasing timestep before breakup. so i want to know your estimation, whether that timestep in combustion simulation can gradually get larger with reaction going on.

2. when i try steady state simulation, the results are just cold flow from a intial condition using the result of my 1e-5 simulation(hot flow), so how can i keep this hot flow? is it good to use an auto ignition model in order to keep it burning here? or is it a matter of timescale or other parameter in my simulation, that make the solver get a cold flow solution?

Thank you very much, sir!