CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FloEFD, FloWorks & FloTHERM (http://www.cfd-online.com/Forums/floefd-floworks-flotherm/)
-   -   Time Dependent FloWorks Analysis error: "Manual time step has been reduced" (http://www.cfd-online.com/Forums/floefd-floworks-flotherm/95811-time-dependent-floworks-analysis-error-manual-time-step-has-been-reduced.html)

Mochi January 3, 2012 16:08

Time Dependent FloWorks Analysis error: "Manual time step has been reduced"
 
I posted this as a reply to another thread, but I'm asking now as a general question.

I am defining a time step of 1s for a transient process where I want to see 10s of data, but FloWorks is telling me: "Manual time step has been reduced" from 1s to 1.3e-08s. Why does FloWorks want such a tiny time step? Should I adjust my physical time to end at something on the order of this small time step and adjust my periodic savings accordingly in the calculation control options?

The model has 100psi inlet and 100LPM outlet and includes a narrow passage .005" tall. The mesh is tight around the passage such that the passage is at least 4 cells tall and 10 cells long.

kimmj January 3, 2012 20:26

I found fallowing at CFdesign manual.
It can help you understand about timestep.

Inner Iterations: Because CFdesign uses an implicit method to discretize the transient flow equations, iterations must be run for every time step. This inner iteration is similar to the amount of work required for a single steady state iteration. However, the inner iterations in a transient analysis are almost always better-conditioned mathematically than a steady state iteration. For this reason, far fewer inner iterations per time step (typically 10) are required than iterations for a steady state solution.

Time Step Size: For transient flow solutions, it is important to select an appropriate time step size. A time step that is too large will result in lost detail because it exceeds the time scale of the flow. A time step that is too small will capture the flow detail, but will not be efficient because it requires more time steps than necessary to characterize the time scale of the flow.
A good guideline for the time step size is approximately 1/20th the time required for a particle of fluid to traverse the length of the device.

Mochi January 4, 2012 12:42

"A good guideline for the time step size is approximately 1/20th the time required for a particle of fluid to traverse the length of the device"

Thanks Kimmj, this helps me a lot.

A steady state velocity plot shows my min and max velocity to be 6in/s and 23,000in/s. My average velocity global goal tells me the average velocity is 640in/s. The particle flow path is about 2inches, so the time for a particle to traverse the length should be around 2/640 = .003sec.

I will try a time step size of 1/20th of that, which is .00015sec.

Ten time steps will give me a physical time between 0 and .002sec. I guess I will save every .00015sec. Is there anything else I need to be aware of regarding the Calculation Control Options window?

kimmj January 4, 2012 20:34

Quote:

Originally Posted by Mochi (Post 337864)
"A good guideline for the time step size is approximately 1/20th the time required for a particle of fluid to traverse the length of the device"

Thanks Kimmj, this helps me a lot.

A steady state velocity plot shows my min and max velocity to be 6in/s and 23,000in/s. My average velocity global goal tells me the average velocity is 640in/s. The particle flow path is about 2inches, so the time for a particle to traverse the length should be around 2/640 = .003sec.

I will try a time step size of 1/20th of that, which is .00015sec.

Ten time steps will give me a physical time between 0 and .002sec. I guess I will save every .00015sec. Is there anything else I need to be aware of regarding the Calculation Control Options window?


Consider saving steps.
I mean, If you don't have enough saving capacity(HDD), program will stop without any message..
And easy and don't haste -you need lots of time for transient analysis..^^..

Actually I don't like control time step function.
Cause, that can make not diversed result and some errors of result.
How about use flow freezing?

Boris_M January 16, 2012 17:45

Hi Kimmj and Mochi,
just saw you have posted it here too.
Just on the last comment from Kimmj.
The software does not stop if the HDD is full, it continues to calculate but cannot save the results you need. Also flow freezing is not helping if you are interested in the flow. It can help to simulate heat flow out of a hot body if the flow is not changing anymore and therefore save calculation time. If using periodical settings for the flow freezing, the flow can be updated every now and then.

Regards,
Boris

kimmj January 17, 2012 20:28

Quote:

Originally Posted by Boris_M (Post 339608)
Hi Kimmj and Mochi,
just saw you have posted it here too.
Just on the last comment from Kimmj.
The software does not stop if the HDD is full, it continues to calculate but cannot save the results you need. Also flow freezing is not helping if you are interested in the flow. It can help to simulate heat flow out of a hot body if the flow is not changing anymore and therefore save calculation time. If using periodical settings for the flow freezing, the flow can be updated every now and then.

Regards,
Boris


Hi Boris,

Anyway, many data makes hanging the PC. Also, I don't want to try again^^;;
And I agree your comment about flow freezing function.
Thanks, Clear comment..

Boris_M January 18, 2012 09:05

Hi Kimmj,
the reason for your system getting into hanging is if the data of the results are written on your system partition (C-drive). I did a calculation some time ago too where the hard drive partition was full after some time but didn't have any problems of the PC hanging as I worked on D or E drive. I could delete the older result files as they were not necessary for me, just backup files as the project was important but in transient calculations of course you cannot delete the results as you'll need them.

Boris

kimmj January 24, 2012 22:36

system dependent..
 
Quote:

Originally Posted by Boris_M (Post 339912)
Hi Kimmj,
the reason for your system getting into hanging is if the data of the results are written on your system partition (C-drive). I did a calculation some time ago too where the hard drive partition was full after some time but didn't have any problems of the PC hanging as I worked on D or E drive. I could delete the older result files as they were not necessary for me, just backup files as the project was important but in transient calculations of course you cannot delete the results as you'll need them.

Boris


Hi,
System hanging is made by many reasions. As you know, ODD processing(HDD wrighting or reading) can make hanging.
Maybe, my system(assembled PC) is unstable, so that hang or reboot, Possible.
I want to say, Program needs much resource for transient analysis, So must be careful for that kind analysis.
That's all.


All times are GMT -4. The time now is 11:24.