There are multiple explanations for this kind of error message.
You should first carefully check your simulation setup in terms of models and boundary conditions. If this is the case, I guess you would probably see the error occurrence in the simulation first steps.
After you guarantee everything is physically consistent, you can try to use classical numerical stabilization techniques, such as reducing the time step, use smooth ramp functions for boundary conditions, reduce under-relaxation factors (maybe increasing the maximum iteration number). Check out the manual for specific recommendations based on the models you are using.
Another good thing to do is to plot the simulation residuals and verify if you can localize the source of error (it always can be related with a poor quality mesh).
There are multiple explanations for this kind of error message.
You should first carefully check your simulation setup in terms of models and boundary conditions. If this is the case, I guess you would probably see the error occurrence in the simulation first steps.
After you guarantee everything is physically consistent, you can try to use classical numerical stabilization techniques, such as reducing the time step, use smooth ramp functions for boundary conditions, reduce under-relaxation factors (maybe increasing the maximum iteration number). Check out the manual for specific recommendations based on the models you are using.
Another good thing to do is to plot the simulation residuals and verify if you can localize the source of error (it always can be related with a poor quality mesh).
All above answers are related and give a good suggestion. My suggestion is to read more about it. Because every case has different reasons. After reading you may find the reason that cause this error in ( Algebraical Multi-Grid ).
Link : http://www.cfd-online.com/Forums/fluent/38074-divergence-detected-amg-solver-temperature.html
Link : Please read previous threads about this on Research Gate.