This commit introduces a special case for the restart file writing
code. In particular we no longer write dynamic data vectors (e.g.,
*WEL, *GRP, *CON) to the restart file if we're being asked to output
the initial condition (i.e., report step/sequence number 0).
Add the report step as a new explicit parameter to createInteHEAD()
instead of inferring this value from the "sim_step" and output
dynamic sizes as zero for report_step=0. While here, also correct
an omission from earlier commit 7986e99e. We must ensure that the
maximum number of wells in the field (intehead[163]) is at least as
large as the current number of active wells (intehead[16]).
Update "intehead" test to account for report_step being passed
directly as an argument to InteHEAD::stepParam().
This commit adds support for persisting the *VIT cumulative summary
vector at the well, group, and field levels in the simulation
restart files. Connection level currently not supported.
This outputs a numeric code, an integer value, for all configured
wells at all times and enables more visual inspection of how the
wells' active controls change over time--either in response to
explicit target modes in the input or as a result of limits becoming
active.
Note that due to the fall-back option of selecting the input value
if no dynamic value is available, we must ensure that the test code
does not specify well types (producer/injector) that conflict with
the keywords used in the backing data set. Update the relevant
tests accordingly.
this will always return the global field props, even in parallel.
necessary since certain operations related to IO need to full global
view on the root process, even when running in parallel
also virtualize the EclipseGrid accessor to add assert in parallel
to ensure only root-process access
motivation: an upcoming parallel frontend to the field props manager.
templated functions cannot be virtualized, and thus having these exposed
would give a great chance of confusing the users in the downstream
code, where properties would be caught from the (potentially) empty
backend instead of from the frontend.
- In the case of several tokens separated with +/- the final tree is assembled
from left, was previously from the right. This fixes evaluation bug with
expressions like A - B - C which were eventually evaluated as A - (B - C).
- Improved handling of different types; both transiton from scalar to sets of
wells and groups, and also groups in general. In particular you can now
assign to a group variable:
DEFINE GUXXX WOPR * 0.25
- Actually hook up the evaluation of group variables from the Summary code.
