If you look at your goal plots you will get your answer. The values are both continuing to increase so there is no convergence even started. It is very unusual that no convergence is being seen so there might be a serious setup issue. Do you want to post up your model?
Hi Bill Dempsey,
Thanks for your response, sorry its not possible to post my model.
I am performing natural convection transient simulation of an enclosed body with two heat sources in it, one is volumetric and other is surface heat source. for the boundary conditions, the surrounded walls are considered with ambient temperature. this is the basic set up i have done to run the simulation.
i believe that at certain time step the heat source temperature becomes constant., so i have plot goals for heat source and inside temperature of fluid. Please do correct me if i am going wrong any where.
Without seeing the model it's impossible to guess what the setup issue might be. True (as stated below) you might achieve convergence but after a very long time. Maybe you have a massive cell count? Are you doing radiation as well or just natural convection?
Have you ever achieved a result with this model? For example, steady state with no natural convection?
Can you make a simple model and post it up to share?
I think your response might be fine. Consider these items:
- It is a transient analysis - time dependent so temporal convergence isn't really a "convergence" in the sense that the calc is bad. It is about achieving quasi -steady response and this doesn't look like it has happened yet.
- What are your terminating parameters? Physical time, number of travels, etc?
- If you invoke flow freezing I suspect you will step right to the answer within your iteration or travel limit. You need to add time frames to to impose "saves" at fixed time increments to get meaningful temporal resolution. The process basically holds the convective terms constant and will use that flow field and the diffusive properties to estimate the response much quicker. And, hence, march in time much faster. So it is important to ensure that the flow has sufficiently evolved to be representative of the quasi steady response. Flow has schemes to control how this update, if enabled, to the convective terms occurs. A quality test for this is to start the flow freezing earlier or later, and how it re-synchronizes and see if you end up at the same answer. Alternatively, spend the time to run a full calc and then figure out an optimal flow freezing scheme for your particular problem. Depends on the problem (consequence, size, etc) and everything is an estimate.
Thank you Bill for your response, after reading your comment i did the required changes and ran the solver. I got the progress bar displayed with %, but with the displayed results i guess solver is not going to converge or reach a pseudo steady state.
Few challenges i am facing during run are.
1. The solver did the calculation of 50 iterations in 15 mins and its going dead slow from 60 to 70 iterations.
2. The heat source progress % is struck at 07 and 08 for two hours and is not moving further.
I am trying to rectify the above issues by refining mesh cells and running it, after doing that also there is same situation happening. please help me out on this
Please post up a simplified example model.
Are you sure there is energy balance in the system? That is a requirement for achieving a steady state result
Your goal plots appear to be "converging" to a perfectly linear slope which is what I expect to see when there is more energy input into the system than can be effectively removed.