udell-fitch test
Hi,
I wanted to re-create one of the original Udell-Fitch input file. Copying the input from the original tough manual, the attached is what I have. When I run it, I get an initial setting error. Do you have this input available for TOUGH2? or what should I fix?
$$$$$$$$$$$$$ LIQUID PHASE EVOLVES AT ELEMENT *LBD 0* $$$$$
!!!!!!!!!!! ERRONEOUS DATA INITIALIZATION !!!!!!!!!!
8 replies
-
The input file for the Udell and Fitch problem is for TOUGH (Pruess, 1987), not for TOUGH2 (Pruess, 1991). The difference is in the primary variables (as indicated by the error message). You can either change the primary variables to TOUGH2 style (i.e., change the second primary variable in PARAM.4 from X2=Sg=0.5 to X2=Sg+10=10.5), or simply set MOP(19)=1, as described in the footnote to Table 9 in the TOUGH2 manual.
Stefan
-
Two short addenda to my previous reply:
(1) The difference in primary variables between TOUGH and TOUGH2 is the result of a clever and elegant design decision by Karsten Pruess; it requires that temperature be the last primary variable; this is described in the second paragraph of Section 6.3 of the manual.
(2) The error message !!!!!!!!!!! ERRONEOUS DATA INITIALIZATION !!!!!!!!!! always refers to an error in specifying primary variables in either PARAM.4, INDOM, or INCON. Specifying thermodynamically correct initial conditions for the desired phase state is a difficult task and thus one of the main input errors. I strongly recommend to always consult the table with the primary variables for the given EOS (i.e., the appropriate subsection of Section 6 in the manual) to make sure the correct primary variables are specified. This also leads to the footnote mentioned in my reply, which resolves the issue.
Stefan
-
Thank you. it seems like my reference for the original input for this problem was missing INCON and INDOM. I guess BC doesn't have to be assigned (zero flux), but IC should be.
-
Heeho,
Please read in the manual about the relationship between PARAM.4 (default initial conditions), INDOM (material-specific initial conditions), and INCON (element-specific initial conditions), each overwriting the other. In short, your reference was correct, i.e., nothing was missing.
Also, read about specifying boundary conditions. Neumann BCs are specified in block GENER, while Dirichlet BCs are specified as ICs for inactive elements or elements with a very large volume.
Stefan
-
I'm getting more specific details to complete this model.
-
You are absolutely correct. I just have convergence issues now. The simulation only reaches about ~9 days (100 time steps) and if I increase number of time steps it breaks down around 20 days.
-
I'm using EOS3 by the way.
-
Heeho,
The input file you created is for the original TOUGH, not TOUGH2. The two codes differ in the way vapor diffusion is implemented - a process that is important at the hot end of the heat pipe. I noticed that you made some changes to the input (e.g., PARAM.1) that are inconsistent with TOUGH2. You would also need a DIFFU block. At any rate, I propose one of the following:
(1) If you are interested in heat pipes, consider the radial heat pipe problem rhp, which is Problem No. 2 in the TOUGH2 User's Guide.
(2) If you are interested in converting TOUGH to TOUGH2 files, check Appendix D and adapt the parameters to reproduce the diffusion as implemented in TOUGH.
(3) If you are interested in the Udell-Fitch test problem, read the original paper and develop a corresponding TOUGH2 problem from scratch as a verification case.
I consider (2) and (3) good exercises.
Stefan