CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CD-adapco (
-   -   switch n.1 (

roadracer June 19, 2003 04:32

switch n.1
Hi I am running transient engine simulations, and I noticed that sometimes, expecially using high order differentiation schemes, I get the "negative density found at more than 100 cells-run stopped" message close after the exhaust port opens, but when I turn on the switch n.1 the calculation goes on. Does anybody know where this "miracolous" switch acts on? Cheers Roadracer

4xF June 19, 2003 08:16

Re: switch n.1
Switch 1 is for the use of zero-order pressure extrapolation procedure. This may help if the flow is compressible.

roadracer June 19, 2003 11:34

Re: switch n.1
Thank you. It means that the accuracy of the scheme is reduced to 0 order????

John June 19, 2003 12:26

Re: switch n.1
Not really - in fact this is the standard procedure for most cfd codes.

skipio June 24, 2003 20:12

Re: switch n.1
I assume that you get negative densities when the valve lift is very small? In which case maybe your time steps are too big and it would help if they were initially reduced and then increased leter when the valve lift increases. Also maybe your cells at the valve opening region have a big aspect ratio? In which case if you could remesh them it would also help.

roadracer June 26, 2003 08:09

Re: switch n.1
The aspect ratios of the cells are close to unity. It certainly happens when the exhaust opens (2 stroke) and reducing the time step help to improve stability, but I donīt want to have different timesteps size during the calculation due to data postprocessing difficulties.

cjtune June 29, 2003 23:20

Re: switch n.1
I suppose you want to keep the dt constant for animations and the like. Most avi makers usually can only take a const number of frames per second of saved screenshot files.

Why not use different load steps with the adv transient modules and decrease just those times steps when your piston passes through the exhause port? (unless you're doing FSI..). Just make sure the the reduced dt is some multiple of the base dt AND the total run time for that load step should be evenly divisible (no fractions) to your base dt.

roadracer July 4, 2003 05:12

Re: switch n.1
This is only one of the problems. I have to extract numerical quantities from every time step, and the excel sheet I am using to monitor the engine behavious allows only a fixed time step, and because it is a 15Mb file I donīt want to modify it, so I want to keep dt constant, or write some kind of program that extracts only the wanted data....

CJ Tune July 6, 2003 04:47

Re: switch n.1
I think you should not have any problems extracting monitored data as you would for .plot files. Since the load step dt is in multiples of a reference dt and since you know when exactly a load step will start, you can instruct your excel worksheet to skip iterations when processing output lines from the load steps with smaller dt. But perhaps I am trivialising your problem...?

roadracer July 7, 2003 09:33

Re: switch n.1
Well, the problem is that everthing has grown as a cancer (Excel sheet maily) and everything relies on the constant value of dt, so I donīt want to change it. In each case the question is: does the SW1=true affect the quality of the calculation? What does it mean "0th order pressure extrapolation"? In piso, if I well remember, the pressure is not extrapolated.

CJ Tune July 7, 2003 21:09

Re: switch n.1
Well, I'm not sure it pertains to the PISO solution algorithm. Maybe the extrapolation of pressure at outlets?

roadracer July 8, 2003 06:07

Re: switch n.1
Donīt know, but I think that it acts somewhere else, because without turning on this switch the problems are in cells near the exhaust port. Probably this procedure is related to the attach boundaries.

All times are GMT -4. The time now is 02:50.