which properties are more correct(and well-posed) to be specified at the boundary?
 

hi all my friends
Its a overall question.Since im using groovyBC i have a freedom on electing what properties to select for specificatin at inlet and outlet boundary.
From characteristics we know at inlet we should specify some properties ourselves that transfer information into domain and some is assigned from inside the domain themselves! :)
Suppose We have 3 property(U,T,p) its advised from Riemann invarients that U should specify from outside(our duty!)and also one of others two(p or T) at outflow patch only one of T or (better)p should be specified by ourselves.am i right untill now?!
But when i assign U and p divergence occured also it was theoretically correct.but when i assigned p and T only and let U be assigned by the flow from inner the domain it converged(but the results were not correct exactly)
Altough it was not theoretically true.
What idea do you dear reader have about this situation?that has made me very thinkful about it.whats your opinion?
Submit it to me.thanks.

If Your case is subsonic:
- You must set U,T at inlet, p - at outlet
- Or - p at outlet and inlet and T at inlet

thank you dear Matvej for you quick answer.
Its more stable from your experience?is there any other proper sets? in my work p at one of two outlet patches should be calculated because all the run is performed to obtain pressure at this outlet correctly.so i can only set T on this patch.i need your advices.
Thanks.

BC, which i specified is stable.
Can You describe Your task with draft scheme without description of real device (i need it only to understand Your case)?
Can You post simple figure with description how fluid enters Your domain? How it leaves it?

yes.imagine a tube like shock tube.but left and right walls of it gradually change to environments according to time.two environment are inflow at left and two outflows are at right.at the first time fluid is homogeniously in tube with zero velocity and specified T and p then the right side is opened to the outside where the flow is going out from the tube by expansion fans.after reaching the expansion fan to the left,left wall is replacing by environment cell by cell.the fluid is flowing in by the vaccum emancipated by fans.after reaching the fan to the right the right patch gradually becomes wall.after a while the left patch is also closed.after that again the left side is opened to environment 3 with high pressure and temperature and a shock wave is generated because of high pressure differences.the shock is moving to right and reflects after reaching right the right wall became open again and left environment turns closed and one cycle is finished.
For more accurate results this cycle should repeat several for example 10 times.but it crashes at the first micro seconds.
Ill send you a figure of it soon.

so the pressure at one of the two outputs has to be obtained.then if i assign only temperature at it is it stable?

hello.i put the case im working on in another thread.could you have a look at it please?"an uncovered error.please help" thanks(i meant "odd" from uncovered.sorry about poor english)