Can we relate residuals to absolute error in velocity/pressure measurements?

wildfire230 · May 15, 2018, 09:36

Hi all,

I have a steady simulation with a primary velocity component of around 1e-3 m/s, and secondary velocity components (which I need to resolve accurately) that are around 1e-8 m/s. I am using simpleFoam to solve the flow, and I am facing some uncertainty about what magnitude to set the residuals in order to properly resolve the flow. For example, if I need a precision of 1e-10 m/s on the y-component of velocity, does that mean I need to wait until the initial residual for Uy is less than 1e-10?

I was just wondering if there is a simple way to estimate the needed residuals based on the needed precision in physical variables.

Thanks so much

piu58 · May 16, 2018, 00:54

If you have velocities in different directions, you have to choose the residuals to the smallest one, as you guessed.

May be it is simpler to transform your problem with some kind of coordinate transformation, to get the velocities closer. This is possible only for a simple geometry and for laminar flow.

wildfire230 · May 16, 2018, 12:02

Hi Uwe,

Thanks for the message and advice. Just so I understand, are you saying that the order of magnitude of the residual is comparable to the order of magnitude of uncertainty in the measurement? i.e. are the residuals relative errors or absolute errors? To say it another way, if your residual for Uz is 1.0e-3, does that mean the value for Uz is accurate to 3 significant digits, or that the value is accurate to 1.0e-3 m/s?

Thanks again

May 15, 2018, 09:36	Can we relate residuals to absolute error in velocity/pressure measurements?	#1
wildfire230 Senior Member Join Date: Jul 2013 Posts: 124 Rep Power: 12	Hi all, I have a steady simulation with a primary velocity component of around 1e-3 m/s, and secondary velocity components (which I need to resolve accurately) that are around 1e-8 m/s. I am using simpleFoam to solve the flow, and I am facing some uncertainty about what magnitude to set the residuals in order to properly resolve the flow. For example, if I need a precision of 1e-10 m/s on the y-component of velocity, does that mean I need to wait until the initial residual for Uy is less than 1e-10? I was just wondering if there is a simple way to estimate the needed residuals based on the needed precision in physical variables. Thanks so much

May 16, 2018, 00:54		#2
piu58 Senior Member Uwe Pilz Join Date: Feb 2017 Location: Leipzig, Germany Posts: 744 Rep Power: 15	If you have velocities in different directions, you have to choose the residuals to the smallest one, as you guessed. May be it is simpler to transform your problem with some kind of coordinate transformation, to get the velocities closer. This is possible only for a simple geometry and for laminar flow. __________________ Uwe Pilz -- Die der Hauptbewegung überlagerte Schwankungsbewegung ist in ihren Einzelheiten so hoffnungslos kompliziert, daß ihre theoretische Berechnung aussichtslos erscheint. (Hermann Schlichting, 1950)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
motorBike Residuals for SST k-omega... and mine	JR22	OpenFOAM Running, Solving & CFD	6	August 1, 2013 09:08
judging convergence through residuals	MachZero	Main CFD Forum	7	December 25, 2012 12:18

May 16, 2018, 12:02		#3
wildfire230 Senior Member Join Date: Jul 2013 Posts: 124 Rep Power: 12	Hi Uwe, Thanks for the message and advice. Just so I understand, are you saying that the order of magnitude of the residual is comparable to the order of magnitude of uncertainty in the measurement? i.e. are the residuals relative errors or absolute errors? To say it another way, if your residual for Uz is 1.0e-3, does that mean the value for Uz is accurate to 3 significant digits, or that the value is accurate to 1.0e-3 m/s? Thanks again