Bug. When engaging permeability change due to precipitation or dissolution of salt by specifying IE(11) > 0, the code calculates and reports permeability change (parameter "k-red" in the output file), but the change is not actually implemented in the flow calculations.

Fix. The problem is caused by a mismatch of names in subroutine MULTI of TOUGH2 module t2f.f. To correct the problem, change the statement

      if(eosn(1).eq.'EWASG     ') then

in line 2911 to

      if(eosn(1).eq.'EWASG     '.or.eosn(1).eq.'ECO2N     ') then

Bug. In eos9.f, two parameters used for internal generation of gravity-capillary equilibrium initial conditions were not initialized when no 'REFCO' specifications were used.

Fix. In the DO 33 N=1,NM loop near the top of subroutine EOS, add two lines of coding to initialize parameters ZREF and MRP, as follows.

        DO 33 N=1,NM
        if(mat(n).eq.'REFCO') then
        ...
        ...
        else
           ZREF=0.
           MRP=0
        endif

CO2TAB fluid property table for module eco2n.f; routine co2tab3.f for generating CO2 property tables

Bug. For low temperatures (< 25 deg-C), the density, viscosity, and specific enthalpy data in the CO2TAB fluid property table supplied with the ECO2N module had lower accuracy at the very highest pressure point (600 bar). Errors reached 3.5 % for density, almost 10 % for viscosity, and 1.5 % for specific enthalpy.

The co2tab3.f routine that was supplied with the ECO2N package for generating fluid property tables has deteriorating accuracy at low temperatures, when pressures approach or exceed 600 bar.

Fix. The deterioration in accuracy arose from a flaw in the search interval for the iteration used in co2tab3.f to find the real gas compressibility factor Z. Click here for downloading an improved routine co2tab3.f for generating CO2TAB, in which the search interval was revised to avoid this problem. The revised co2tab3.f follows the iterative bisection process for determining Z with a Newtonian iteration.

Click here to download the updated CO2TAB file generated with the revised co2tab3.f. Please let us know of any problems by e-mail to K_Pruess@lbl.gov. 

Note that the updated CO2TAB has small differences from the previous version throughout, even for the table points at lower pressures that had satisfactory accuracy. The differences arise because the iteration process for Z is now slightly different. This will have small effects on simulation results, but it may lead to some changes in the behavior of a simulation run, where occasionally convergence now is achieved in a different number of iterations, so that subsequently the automatic time step control may lead to a different choice of delta-t. Once that happens, time truncation errors will of course be different. For testing proper installation of ECO2N by comparing results for sample problems with those given in the user's guide, we recommend using the original CO2TAB.

Bug. Occasionally severe convergence problems are encountered when using the MINC method to simulate multiphase flow problems in fractured reservoirs.

Fix. The problem occurs because of a conditional switching of relative permeabilities at interfaces in which a nodal distance is zero, as is the case for fracture nodal distances generated by the MINC method. It is recommended to simply comment out (or delete outright) the following two lines in subroutine MULTI, module t2f.f:

        IF(D1.EQ.0..AND.REL20.NE.0.) REL1=REL2
        IF(D2.EQ.0..AND.REL10.NE.0.) REL2=REL1

(These two lines occur a few lines after the following line:)

C-----PERFORM APPROPRIATE UPSTREAM WEIGHTING FOR MOBILITIES.

Note that D=0 triggers other switches as well that should not be changed. These other switches are "static," depending only on input parameters, not on parameters such as relative permeabilities that may change during a simulation.

Bug. Subroutine DIFFU did not have a "return" statement, causing bugs on some compilers.

Fix. Insert a "return" statement in front of "END" in subroutine DIFFU.

Bug. The calculation of internal energies in BALLA uses the variable NLOC which is undefined. (This will only affect the reporting of global energy balances, not simulation results for grid blocks.)

Fix. In subroutine BALLA, module ewasg.f, near the top of the DO 10-loop, insert a statement

        nloc=(n-1)*nk1

in front of

        NLOC2=(N-1)*NSEC*NEQ1

Bugs. In single-phase gas conditions,

(1) vapor pressure is undefined when no non-condensible gas is present, and

(2) gas phase enthalpy calculation has a divide by zero when no vapor is present.

Fixes.

(1) Before line 1061 (CALL SUPST(TX,PX,DS,US)) insert the following line

             ps=px

(2) Replace line 1072 (PAR(NLK2+5)=X3*HAIR+(1.E0-X3)*(US+PS/DS)) with the following two lines

          PAR(NLK2+5)=X3*HAIR
          if(ds.gt.0.) PAR(NLK2+5)=PAR(NLK2+5)+(1.E0-X3)*(US+PS/DS)

Bug. Affects EWASG-use only. On November 15, 2000 a revision was introduced whereby permeability change due to solid precipitation is now transmitted through location PAR(...+3) in the solid phase section of the PAR-array, instead of the previously used location PAR(...+6). In order to work properly this revision must also be implemented in subroutines PHASD, PHASF, and GCOR of module t2f.f, which had not been mentioned in the original suggestion of November 15.

Fix. In module t2f.f, in lines 3718 (PHASD), 3836 (PHASF), and 4202 (GCOR), change coding from PAR(...+6) to PAR(...+3).

Bug. When using fully-coupled multiphase diffusion (default; MOP(24)=0) with the EWASG fluid property module, diffusive fluxes are erroneously printed as 0. This error affects only the printing of fluxes; the simulation uses proper fluxes, and the results calculated for mass fractions etc. are all o.k.

Fix. Subroutine OUTDF in file t2f.f has two "PRINT 5071 ..." statements, the first of which is engaged for MOP(24)=1, while the second is active for MOP(24)=0. In the second statement, change the argument of array FDIF from II(KK,NPH,N) to II(KK,2,N).

Bug. There have been a number of instances where the preconditioned conjugate gradient package T2CG2 gave erratic or unreliable results, occasionally failing altogether. Such failures have occurred when the Jacobian matrix had a large and varying number of zero diagonal elements, and when a grid block was involved in many connections.

Fix. A revised version of T2CG2 is available for downloading. Tests of this new version have shown satisfactory performance for problems where the previous version failed, but tests so far have been far from exhaustive. Click here to download a beta-testing version of the revised module (file t2cg22.f, which replaces t2cg2.f of the TOUGH2 V 2.0 package), and a revised INCLUDE file to go with this module (file T2). Please let us know of any problems by e-mail to K_Pruess@lbl.gov.