SimScale CAE Forum

Convective Heat Transfer Problem


#1

Good evening, my friends @roy_g @vgon_alves @jousefm ,

I am attempting to perform a simulation of convection heat transfer with time-dependent dependency, but the following errors are presented:

The main problem is the correct configuration of the simulation control, but I’m not sure. I can only do simulations with stable time dependency. Follow the project link: https://www.simscale.com/projects/Pedro_Henrique/convec-ao_natural_metade_do_biorreator/

Thanks for everything.


#2

Hi @Pedro_Henrique

The courant number shows you how rapidly the result field is changing in a single timestep in comparison to your mesh size. For example, a courant number of 0.5 would mean that fluid velocity is high enough that a particle would move (I say would move because the software models a continuum, not individual particles) half of a mesh element in a single time step and a counrant number of 1 means moving an entire mesh element in 1 time step. If you are interested, tell me and I will explain why it is bad but for now just know that your courant number should be below one.

To lower your courant number you should reduce your time step so that the time interval is smaller and the ‘particle’ therefore travels a smaller distance during the timestep. You could also reduce mesh density but this is usually a bad idea unless you have chosen an unnecessarily dense mesh to begin with.


#3

Good Morning Roy_g,

Thank you for your contributions. I’m keeping a courant number of 0.5, a delta t of 1 seconds and end time of 1000 seconds. The problem is simulation control, because I do not know the correct combination of the parameters. Can you help me?.


#4

The error that appears when you ran the simulation, suggests that the courant number is too high


#5

I chose to perform a stationary time-dependent simulation to see if I really need a time-dependent simulation. Here is a photo of the waste from this simulation. What do you think ?. Do I need to perform a transient time-dependent simulation? or so this good ?.


#6

Hi @Pedro_Henrique

Whether or not you need a transient simulation depends on what you want to get out of the simulation. Your simulation starts from some initial conditions and is subject to boundary conditions which cause changes in the system condition (i.e. your fluid heats up). Eventually the system arrives at some equilibrium where it no longer changes. This equilibrium position is its steady state position and the period from the initial condition to the steady state position is called the transient period. If you want to see how the tank heats up then you need a transient simulation. If you only care about the final steady state temperature distribution then there is no point in running a transient simulation. Note that some of your residuals are quite high. You could try reducing the relative tolerances of some of your residuals and adjust the relaxation factors.


#7

Thaks Roy_g for your contributions.


#8

Something else I should add; when you define a maximum courant number of 0.5, this number is used to guide the automatically adjustable time step but it can not guarantee that the courant number of 0.5 is not exceeded. To the best of my knowledge you should reduce your delta T and perhaps also the maximum time step if the courant number is exceeded.