H.O.S. and time?
 

A couple of questions... did the HOS momentum handling fix get into 3.6.0 or 3.6.1, and are there plans to include higher order time schemes? Has anyone here done this in, say, GROUND?

Thank You Very Much, PattiMichelle

Re: H.O.S. and time?
 

A fix has gone into HOS recently, see below. I am told that there are no plans to include HOS for time.

DESCRIPTION OF PROBLEM: The user reports that the use of a higher-order scheme in the file GXHOCS.FOR results in rapid divergence for three-dimensional cases.

Library case N109 diverges immediately producing absurd values.

CAUSE OF THE PROBLEM TO BE FILLED IN WHEN KNOWN: The problem was traced to the following line in Subroutine GXHOEN is GXHOCS.FOR:

C N109 fails because f(l0rua+ixy+nxny) has 1e19 on 1st sweep

RUA = 0.5*(RUA + F(L0RUA+IXY+NXNY))

when solving for W1 and dealing with the staggered north-cell face contribution to the higher-order deferred-correction source term. The error is that because the mass flux store at IZ+1, namely: F(L0RUA+IXY+NXNY), contains 1E19 on the 1st sweep. The problem occurs because the stores CON1E and CON1N are initialised to 1E19 by EARTH, whereas at one time they were initialised to 0.0. The easiest solution to the problem is to initialise CON1E and CON1N in GXHOCS.FOR.

It can be seen from the comment line that someone has already discovered that Library case N109 fails because of this problem. However, if you correct the problem, the library case still diverges rapidly and the reason is that immediately before entering GXHOHL, the indices of three variables are shifted by one slab, but then never put back after leaving GXHOHL. This is necessary because the correct indices are needed when entering GXHOEN for the east-face contribution to the deferred-correction source term. The solution to the problem is to re-set the indices after leaving GXHOHL.

ACTION: Correct the two problems, as per the modified gxhocs.htm in g:\d_36bugs\grex\001.

PASSED TO DEVELOPMENT ON (date): 10.05.04 Passed to DBS for action Copied to HQQ, JZW, WZH, JCL for information

STATUS (ie corrected or outstanding): Earexe dated 12.05.50 gives correct solution to N109

Re: H.O.S. and time?
 

Thanks :)