[naveen2790@yahoo.co.in]

Hi All,

I am doing a simulation of pneumatic conveying of particles in a cylindrical pipe. I am using Euler model with explicit formulation. My solution is not being converged in first 4 time steps. After that it is being converged in each time step.
Order of residual is 10^-5.
I also want to know how many time step should I run.

[LuckyTran]

Originally Posted by naveen2790@yahoo.co.in

Hi All,

I am doing a simulation of pneumatic conveying of particles in a cylindrical pipe. I am using Euler model with explicit formulation. My solution is not being converged in first 4 time steps. After that it is being converged in each time step.
Order of residual is 10^-5.
I also want to know how many time step should I run.

The first few time steps are usually the hardest to converge because the solution is responding to a sudden change in conditions (boundary conditions, forces, etc). Unless you need a time-accurate simulation these initial time-steps my be ignored. If you need a time-accurate simulation where all time-steps are converged, you should try increasing the number of iterations per time-step (but only for those initial time-steps) or decrease the dt for the first few time-steps.

It's entirely up to you how many time-steps to run since it is your problem. What do you need from your simulation? What are you hoping to gain, understand, or quantify?

[naveen2790@yahoo.co.in]

Hi Lucky,
first of all I would like to thank for your valuable reply.

I am trying to get pneumatic conveying pressure drop in bend.

[raghulvr]

Ideally,the number of iterations per time step can be set to 20.For your simulation,iterations must converge within the each and every time step in order to have high accuracy in yoir results

[SabarishCEG]

Hello all, I am trying to simulate a transient flow with a time dependent boundary condition (inlet velocity). For this case what should be the initial value of my solution? I

[raghulvr]

First of all explain your problem clearly,then ask a question regarding what sort of difficulty you have.whenever you post any query in this forum,be as specific as possible

[SabarishCEG]

Hello Raghulvr
The simulation that i am doing is flow over a bluff body. For this simulation the inlet boundary condition is velocity inlet which is time varying. I wrote a UDF for that. I would just like to know what the intial conditions should be. I refered a few literature and it said that for finding the initial conditions first run a steady case and determine them.The flow in my case is linearly accelerated with time. The velocity is zero at time zero and it increases with time upto the value of 25m/s. I just do not know at what velocity should the steady case be run whether at 25m/s or st some intermediate value.

[raghulvr]

Initialization will be taken care by the solver.Moreover, you have already written a UDF that governs the variation of velocity with time.so just go ahead with your initialization.Yes, first initialise your solution with hybrid initialization.that will take care of the starting values that will be needed for your solution.after iterations have started,just monitor the residuals whether they are steadily decreasing.

[ghost82]

It depends on what you want to study....
Do you want to study the transient from a steady state solution? Then run the steady state solver and use that converged solution as starting solution for the transient analysis.
Do you want to study the transient starting from your fluid at rest?Then inizialize your flow with zero velocity and all the other variables at time zero and start from the beginning the transient analysis.

There is no a reply to your question, only you know what you have to study.

Daniele

[Shamoon Jamshed]

Hello experts,

I am doing a simulation with coupled heat transfer. I did it with success for some cases with coupled solver and pseudo time stepping. However, for one case , it diverged with coupled. I tried different arrangements, like FMG , or hybrid or direct iterations but it diverges. When I take SIMPLEC technique for this problem it takes ages for PRessure monitor to converge. (ITs like a pipe flow). While I want to use coupled because it converges very fast. Can someone help in my problem please?

Regards,

[Mubashir]

Hello everyone , i am doing a simulation of pcm melting a 2d complex shape with a fin inside it.. one side is at constant temperature pf 343 K and other sides are insulated ..
The issue i am facing is that some timesteps are converging and some are not .

[Shamoon Jamshed]

Originally Posted by Mubashir

Hello everyone , i am doing a simulation of pcm melting a 2d complex shape with a fin inside it.. one side is at constant temperature pf 343 K and other sides are insulated ..
The issue i am facing is that some timesteps are converging and some are not .

See the number of iterations it took for converged residuals, it must be less than number of iterations you gave. So specify that much number of iterations in the panel, in this manner it will give residual less time to go more below the convergence criteria, hence in the next time iteration the residuals will go to new iterations and hopefully will converge. (Just specify less iteration numbers in thepanel). also , reduce Under relaxation factors.