During runs with multiple wells optimized, there is excessive utilization of memory. The memory usage scales with the number of wells, which suggest there is something in the Well Index Calculator that does not get deleted.

Had the same issue previously, in a previous version of the WIC. That was solved by adding a destructor

Hi,

When I try to run FieldOpt using Flow simulator, I always encounter the following error,

"
Output deck directory not found. Copying input deck:/../FieldOpt/examples/Flow/5spot -> ../FieldOpt/examples/Flow/5spot/output_result/fo_driver_2vert_controls/5spot

"

This is the bash file I used:

export FIELDOPT_BUILD=../FieldOpt/FieldOpt/build/bin
export FIELDOPT_BIN=../FieldOpt/FieldOpt/build/bin/FieldOpt
export FIELDOPT_OUT=../FieldOpt/examples/Flow/5spot

declare -a CASES=(
"fo_driver_2vert_controls"
)

Set case and path variables

CASE_IDX=$(expr $1 - 1)
CASE=${CASES[$CASE_IDX]}
CURRENT_DIR=$(pwd)
DRIVER_PATH=${CURRENT_DIR}/${CASE}.json
OUTPUT_DIR=${FIELDOPT_OUT}/output_result/${CASE}
DECK_PATH=${CURRENT_DIR}/5SPOT.DATA
GRID_PATH=${CURRENT_DIR}/5SPOT.EGRID
SCR_PATH=${FIELDOPT_BUILD}/execution_scripts/bash_flow.sh

Delete and re-create ouput subdirectory

mkdir -p ${OUTPUT_DIR}

echo "Output: " ${OUTPUT_DIR}
echo "Driver: " ${DRIVER_PATH}
echo "Deck: " ${DECK_PATH}
echo "Grid: " ${GRID_PATH}

${FIELDOPT_BIN} ${DRIVER_PATH} ${OUTPUT_DIR} -r serial -v3 -g ${GRID_PATH} -s ${DECK_PATH} -e ${SCR_PATH}

Thanks advance for your help. By the way, do I need to set the output path for the Flow simulator?

best regard

Cong Xiao

While using Flow, if the simulation fails previous available results will be read instead of sending an interrupt signal and putting objective function value to sentinel value. Will cause PSO/GA/EGO and similar algorithms that rely on previous results to work erroneously

Hi, FieldOpt developers,

  When I testing FieldOPt,  I find some code error in FieldOpt.  In the code : abstract_runner.cpp, line 116             
     simulator_ = new Simulation::ECLSimulator(settings_, model_);

In this part, Flow simulator will be used, however, the class "ECLSimulator" is called here. I do not know this is the case, or some error. Please check it.

best regard

When generating segmented (compartmentalized wells), some well blocks will be added to more than one segment. This results in non-realistic simulator input (in that, e.g., two well blocks may be producting to two different compartments. Just one would be possible, but not two, and also the measured depths in the tables are strange).

Write unit-tests for spline-defined wells in the Model-library.

• The class to be tested is Model::Wells::Wellbore::WellSpline
• The WellIndexCalculator should be considered as a "black box" in these tests.
• The spline is defined by two (x, y, z) points, and the class returns a set of well block objects (Model::Wells::Wellbore::WellBlock). The location of these ((i, j, k) coordinates) and the well index should be checked against data from RMS.
• Reservoir models (HORZWEL and 5Spot) are easily accessible through a TestResource.

The check to see whether a cell is active or not does not work properly. It returns true for a lot of inactive cells.

Cases that have their objective function value set by the bookkeeper instead of simulating them are never applied to the model in order to save time (when applying the case to the model the well index is calculated, which can sometimes be costly). This means that the case is never written to the extended json log.

I'm creating a Vagrant setup to spawn and automatically and provision a Virtual Machine (VM) for FieldOpt.
This will enable automatic provisioning (installing dependencies etc.), making the necessary libraries and copying the necessary files to where they belong.

However, at step 6. Build FieldOpt i ran in to the following error after running 'make' inside FieldOpt-build/

[ 28%] Built target simulation_automoc [ 34%] Built target runner Scanning dependencies of target simulation [ 34%] Building CXX object Simulation/CMakeFiles/simulation.dir/simulator_interfaces/adgprssimulator.cpp.o In file included from /vagrant/FieldOpt/Simulation/simulator_interfaces/adgprssimulator.cpp:23:0: /vagrant/FieldOpt/Simulation/results/adgprsresults.h:2:25: warning: missing terminating " character [enabled by default] #define ADGPRSRESULTS_HL" ^ /vagrant/FieldOpt/Simulation/results/adgprsresults.h:2:25: warning: ISO C99 requires whitespace after the macro name [enabled by default] [ 35%] Building CXX object Simulation/CMakeFiles/simulation.dir/results/adgprsresults.cpp.o In file included from /vagrant/FieldOpt/Simulation/results/adgprsresults.cpp:1:0: /vagrant/FieldOpt/Simulation/results/adgprsresults.h:2:25: warning: missing terminating " character [enabled by default] #define ADGPRSRESULTS_HL" ^ /vagrant/FieldOpt/Simulation/results/adgprsresults.h:2:25: warning: ISO C99 requires whitespace after the macro name [enabled by default] /vagrant/FieldOpt/Simulation/results/adgprsresults.cpp: In member function ‘virtual void Simulation::Results::AdgprsResults::ReadResults(QString)’: /vagrant/FieldOpt/Simulation/results/adgprsresults.cpp:21:35: error: invalid use of incomplete type ‘class QStringList’ if (file_path.split(".vars.h5").length() == 1) ^ In file included from /usr/include/qt5/QtCore/QString:1:0, from /vagrant/FieldOpt/Simulation/results/results.h:4, from /vagrant/FieldOpt/Simulation/results/adgprsresults.h:3, from /vagrant/FieldOpt/Simulation/results/adgprsresults.cpp:1: /usr/include/qt5/QtCore/qstring.h:81:7: error: forward declaration of ‘class QStringList’ class QStringList; ^ make[2]: *** [Simulation/CMakeFiles/simulation.dir/results/adgprsresults.cpp.o] Error 1 make[1]: *** [Simulation/CMakeFiles/simulation.dir/all] Error 2 make: *** [all] Error 2
Any ideas on how to solve this? Looks like an issue when refering to a certain part of the Qt5 library.