Switch to storing a complete fluid-matrix derivative in the
compr_quantities rather than the total compressibility. Maintain the
"volume discrepancy" field. Also, add traditional memory management
functions.
Functions csrmatrix_write_stream() and vector_write_stream() subsume
the output responsibilities of csrmatrix_write() and vector_write(),
respectively, but do not deal with opening or closing streams. This
allows, e.g., a call such as
csrmatrix_write_stream(A, stderr);
for debugging purposes.
Re-implement csrmatrix_write() and vector_write() in terms of
csrmatrix_write_stream() and vector_write_stream(), respectively.
Introduce two intermediate variables, c1, and c2, to hold cell numbers
during the building of the cell<->cell neighbourhood. This reduces
the statement complexity--both for the human reader and the compiler...
Specifically, GCC in strict ISO C90 mode warns of:
cfs_tpfa.c: In function ‘cfs_tpfa_impes_maxtime_cell’:
cfs_tpfa.c:1086: warning: ISO C90 forbids mixed declarations and code
Specifically, allocate storage for 'masstrans' and 'gravtrans' terms
per well completion (perforation), as well as compressible
transmissibilities (scalar per completion). Calculate compressible
quantities by treating each completion as an interface.
Introduce a new structure, 'completion_data', to collect static and
dynamic discretisation data pertaining to each completion (e.g.,
productivity indices, gravity potentials and density ratio
operators). Pass this structure, rather than individual fields, into
affected CFS_TPFA entry points.
Compile tested only.
Specifically, rename the obtuse structure 'disc_data' to the more targeted
'densrat_util' and hoist the compressible terms 'ctrans' and 'P' into
the 'cfs_tpfa_impl' structure. Moreover, rename the remaining fields
into something that makes sense in (almost) isolation. Update
compute_densrat_update() and cfs_tpfa_construct() accordingly.
This is in preparation of adding compressible well terms.
Specifically, only assemble gravity contributions on internal faces or
external Dirichlet faces. Moreover, pay attention to direction of
gravity flux (in/out of cell) during assembly.
Specifically, split detection of prescribed reservoir pressure values
from prescribed well (i.e., BHP) pressure values. Previously, we
would not even assemble any well contributions if there were any
prescribed reservoir pressure values. Moreover, the return value from
assemble_cell_contrib() was exactly opposite of its intended value...
Callers may pass NULLs in absence of wells in any given model. This
implementation assembles an equation for each well, irrespective of
well control type (BHP or RATE), and assumes that productivity indices
and perforation pressure drops account for multiphase effects.
We will generate square matrices only in the foreseeable future and
there is no need to maintain the number of columns (the 'n' member)
separately. Update all users accordingly.
Following cset ba5d27f90d7a there is no need to compute the interface
pressure values after every linear solve. Re-factor
cfs_tpfa_press_flux() to only compute cell pressures and interface
fluxes, and move interface pressure value computation to cfs_tpfa_fpress().
Having a 'MAT_SIZE_T' that is sometimes an int and sometimes an
mwSignedIndex is asking for trouble. The practical impact for OPM is
low, though, as this issue affects only the MEX interface.
Specifically, move the evaluation of cell transmissibilities into
cfs_tpfa module (from original MEX implementation), and create a new
structure, 'struct compr_quantities', to hold the 'RB^{-1}' data and
(upwind) phase mobilities &c.
Specifically, introduce utilities compute_fpress() and
compute_flux(). The former is needed to implement the latter across
external boundary faces. Moreover, interface pressure values are
needed to evaluate fluid properties on faces (specifically R/B).
Add small gateway routine, cfs_tpfa_fpress(), to allow callers to
recover interface pressure values. Re-implement cfs_tpfa_press_flux()
in terms of compute_fpress() and compute_flux().
Also, add fields 'fpress' and 'fpaccum' to struct cfs_tpfa_impl.
We may wish to generalise this part into real utility functions but
for now, leave the functionality where it is most directly needed.
Rename the functions to highlight relationship to compr. tpfa.
We were already computing the inverse IP, but now centralise the
gpress as well. Moreover, the mim_ip_*_update() functions will assist
in the pressure solvers accepting effective inner products and gravity
pressures only.
and non-zeros. Note that this routine is for allocation only.
Caller must create and manage sparsity structure if this is being
used in a global assembly process.
This routine finds utility in MsMFEM BF construction.
Bård Skaflestad