Pulsed Mass Flow Rate

PWeller · February 5, 2015, 05:52

I am trying to achieve a pulsed mass flow rate of 3.0625E-4 kg/s of air from a thin [500 micron] outlet into atmospheric conditions, over a time period of 20 seconds, with a pulse of 2 seconds, 0.5 secs from the start, with a gap of 6 seconds between pulses.

I believe my physics models are correct; constant density, coupled flow, gas-air, gradients, implicit unsteady, laminar, three dimensional.

I have attempted to create a table(time) in excel and saved as .csv, that provides the massflowrate at 0.1 second intervals.

I having problems getting this to simulate as it tells me;

Bi-Conjugate Gradient Stabilized solver did not converge !
* CFL 50 -> 5...

I think the issue may be:
the courant number is too high, set at 50
under my mass flow rate boundary condition>physics values>mass flow rate>table(time); Table:Time =Time, Interpolation = Spline, Table:data =Time, Table = Mass Flow Rate[.csv file name]

any help would be great thanks

me3840 · February 9, 2015, 00:46

Check the convective courant number in the channel and make sure your mesh is very fine there. I would bet your timestep is too high.

PWeller · February 11, 2015, 11:37

Thanks for the advice. i have decreased my courant number from 50 to 5, and also decreased my time step from 0.1 sec to 0.025 secs.

It seems to have corrected the initial error that was happening, but my residuals are still increasing, leading to a floating point exception.

is it worth me changing the temporal discretization from 1st to 2nd order, and should i also include a linear courant number ramp?

Thanks

me3840 · February 11, 2015, 21:07

What is the convective courant number in the channel?

PWeller · February 12, 2015, 04:21

the convective courant number in the channel is very high, around 35000, and increasing with every iteration.

me3840 · February 12, 2015, 17:58

Yeah, that's far too high and probably your problem. You need to resolve that with more cells.

PWeller · February 17, 2015, 10:46

I have increased the accuracy of the cells in the channel, which has reduced the convective courant number in the middle of the channel, but is still high on the boundary edges. I have included 6 prism layers in the channel that are very small, considering the channel is 500 microns wide.

I have also reduced my time step down to 0.005 seconds.

What are your thoughts?

Thanks

February 5, 2015, 05:52	Pulsed Mass Flow Rate	#1
PWeller New Member Paul Weller Join Date: Feb 2014 Posts: 5 Rep Power: 12	I am trying to achieve a pulsed mass flow rate of 3.0625E-4 kg/s of air from a thin [500 micron] outlet into atmospheric conditions, over a time period of 20 seconds, with a pulse of 2 seconds, 0.5 secs from the start, with a gap of 6 seconds between pulses. I believe my physics models are correct; constant density, coupled flow, gas-air, gradients, implicit unsteady, laminar, three dimensional. I have attempted to create a table(time) in excel and saved as .csv, that provides the massflowrate at 0.1 second intervals. I having problems getting this to simulate as it tells me; Bi-Conjugate Gradient Stabilized solver did not converge ! * CFL 50 -> 5... I think the issue may be: the courant number is too high, set at 50 under my mass flow rate boundary condition>physics values>mass flow rate>table(time); Table:Time =Time, Interpolation = Spline, Table:data =Time, Table = Mass Flow Rate[.csv file name] any help would be great thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CFX turbo - Mass flow rate	Vashishth Patel	CFX	7	April 3, 2014 20:33
Mass Flow rate in spray water modeling	Behnam Ghadimi	FLUENT	0	June 8, 2013 16:05
Mass Flow rate in spray water modeling	Behnam Ghadimi	Main CFD Forum	0	June 8, 2013 15:48
negative global mass flow rate	Gimli	FLUENT	0	April 21, 2006 07:17
mass flow rate error	Masood	FLUENT	0	May 22, 2005 00:32

February 9, 2015, 00:46		#2
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Check the convective courant number in the channel and make sure your mesh is very fine there. I would bet your timestep is too high.

February 11, 2015, 11:37		#3
PWeller New Member Paul Weller Join Date: Feb 2014 Posts: 5 Rep Power: 12	Thanks for the advice. i have decreased my courant number from 50 to 5, and also decreased my time step from 0.1 sec to 0.025 secs. It seems to have corrected the initial error that was happening, but my residuals are still increasing, leading to a floating point exception. is it worth me changing the temporal discretization from 1st to 2nd order, and should i also include a linear courant number ramp? Thanks

February 11, 2015, 21:07		#4
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	What is the convective courant number in the channel?

February 12, 2015, 04:21		#5
PWeller New Member Paul Weller Join Date: Feb 2014 Posts: 5 Rep Power: 12	the convective courant number in the channel is very high, around 35000, and increasing with every iteration.

February 12, 2015, 17:58		#6
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Yeah, that's far too high and probably your problem. You need to resolve that with more cells.

February 17, 2015, 10:46		#7
PWeller New Member Paul Weller Join Date: Feb 2014 Posts: 5 Rep Power: 12	I have increased the accuracy of the cells in the channel, which has reduced the convective courant number in the middle of the channel, but is still high on the boundary edges. I have included 6 prism layers in the channel that are very small, considering the channel is 500 microns wide. I have also reduced my time step down to 0.005 seconds. What are your thoughts? Thanks